Intelligent Computing in Engineering: Select Proceedings of RICE 2019 (Advances in Intelligent Systems and Computing, 1125) 9811527792, 9789811527791

Citation preview

Advances in Intelligent Systems and Computing 1125

Vijender Kumar Solanki Manh Kha Hoang Zhonghyu (Joan) Lu Prasant Kumar Pattnaik   Editors

Intelligent Computing in Engineering Select Proceedings of RICE 2019

Advances in Intelligent Systems and Computing Volume 1125

Series Editor Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland Advisory Editors Nikhil R. Pal, Indian Statistical Institute, Kolkata, India Rafael Bello Perez, Faculty of Mathematics, Physics and Computing, Universidad Central de Las Villas, Santa Clara, Cuba Emilio S. Corchado, University of Salamanca, Salamanca, Spain Hani Hagras, School of Computer Science and Electronic Engineering, University of Essex, Colchester, UK László T. Kóczy, Department of Automation, Széchenyi István University, Gyor, Hungary Vladik Kreinovich, Department of Computer Science, University of Texas at El Paso, El Paso, TX, USA Chin-Teng Lin, Department of Electrical Engineering, National Chiao Tung University, Hsinchu, Taiwan Jie Lu, Faculty of Engineering and Information Technology, University of Technology Sydney, Sydney, NSW, Australia Patricia Melin, Graduate Program of Computer Science, Tijuana Institute of Technology, Tijuana, Mexico Nadia Nedjah, Department of Electronics Engineering, University of Rio de Janeiro, Rio de Janeiro, Brazil Ngoc Thanh Nguyen , Faculty of Computer Science and Management, Wrocław University of Technology, Wrocław, Poland Jun Wang, Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, Shatin, Hong Kong

The series “Advances in Intelligent Systems and Computing” contains publications on theory, applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually all disciplines such as engineering, natural sciences, computer and information science, ICT, economics, business, e-commerce, environment, healthcare, life science are covered. The list of topics spans all the areas of modern intelligent systems and computing such as: computational intelligence, soft computing including neural networks, fuzzy systems, evolutionary computing and the fusion of these paradigms, social intelligence, ambient intelligence, computational neuroscience, artificial life, virtual worlds and society, cognitive science and systems, Perception and Vision, DNA and immune based systems, self-organizing and adaptive systems, e-Learning and teaching, human-centered and human-centric computing, recommender systems, intelligent control, robotics and mechatronics including human-machine teaming, knowledge-based paradigms, learning paradigms, machine ethics, intelligent data analysis, knowledge management, intelligent agents, intelligent decision making and support, intelligent network security, trust management, interactive entertainment, Web intelligence and multimedia. The publications within “Advances in Intelligent Systems and Computing” are primarily proceedings of important conferences, symposia and congresses. They cover significant recent developments in the field, both of a foundational and applicable character. An important characteristic feature of the series is the short publication time and world-wide distribution. This permits a rapid and broad dissemination of research results. ** Indexing: The books of this series are submitted to ISI Proceedings, EI-Compendex, DBLP, SCOPUS, Google Scholar and Springerlink **

More information about this series at http://www.springer.com/series/11156

Vijender Kumar Solanki Manh Kha Hoang Zhonghyu (Joan) Lu Prasant Kumar Pattnaik •

•

Editors

Intelligent Computing in Engineering Select Proceedings of RICE 2019

123

•

Editors Vijender Kumar Solanki CMR Institute of Technology Hyderabad, India

Manh Kha Hoang Hanoi University of Industry Ha Noi, Vietnam

Zhonghyu (Joan) Lu University of Huddersfield Huddersfield, UK

Prasant Kumar Pattnaik KIIT University Bhubaneswar, India

ISSN 2194-5357 ISSN 2194-5365 (electronic) Advances in Intelligent Systems and Computing ISBN 978-981-15-2779-1 ISBN 978-981-15-2780-7 (eBook) https://doi.org/10.1007/978-981-15-2780-7 © Springer Nature Singapore Pte Ltd. 2020 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

The 4th International Conference on Research in Intelligent and Computing in Engineering, popularly known as RICE 2019, was held on August 08–09, 2019 in Hanoi University of Industry (HaUI), Hanoi, Vietnam. The Fourth edition of RICE 2019, organized by the Electronic Engineering Faculty of the HaUI, provides an international forum which brings together the researchers as well as the industry practitioners, who are actively involved in the research in fields of intelligent computing, data science, or any other emerging trends related to the theme covered by this conference. RICE 2019 provided an opportunity to account state-of-the-art works, to exchange ideas with other researchers, and to gather knowledge on advancements in informatics and intelligent systems, technologies, and applications. This conference has technical paper sessions, invited talks, and panels organized around the relevant theme. RICE 2019 was the event where the author had the opportunity to meet some leading researchers, to learn about some innovative research ideas and developments around the world, and to become familiar with emerging trends in Science and Technology. RICE 2019 received a huge response in terms of submission of papers across the countries. RICE 2019 received papers from various countries outside Vietnam such as India, China, Russia, Australia, New Zealand, and many more. The Organizing Committee of RICE 2019 constituted a strong international program committee for reviewing papers. A double-blind review process has been adopted. The decision system adopted by EasyChair has been employed and 118 papers have been selected after a thorough double-blind review process. The proceedings of the conference will be published as one volume in Advances in Intelligent Systems and Computing, Springer, indexed by ISI Proceedings, EI-Compendex, DBLP, SCOPUS, Google Scholar, and Springerlink. We convey our sincere gratitude to the authority of Springer for providing the opportunity to publish the proceedings of RICE 2019. To realize this conference in 2019, we really appreciate Hanoi University of Industry to host the conference and to be continuously supporting the organization team during the preparation as well as 2 days of the conference. In addition, we v

vi

Preface

would like to give a special thanks to Vintech City, a member of Vingroup, that has supported the conference as a diamond sponsor. We would also like to thank the financial support of ASIC Technologies to RICE 2019. Without their support, this conference would have not been successful as the first time being held in Vietnam. Our sincere gratitude to all keynote address presenters, invited speakers, session chairs, and high officials in India and Vietnam for their gracious presence in the campus on the occasion. We would like to thank the keynote speaker as Prof. Vijender Kumar Solanki, CMR Institute of Technology, Hyderabad, TS, India; Dr. Le Hoang Son, VNU, Hanoi Vietnam; Dr. Kumbesan, Australia; Dr. P K Pttanaik, KIIT Bhubaneswar, Odisha, India; Dr. Rashmi Agarwal, MRIIS, Haryana, India for giving their excellent knowledge in the conference. We would like to thank the reviewers for completing a big reviewing task in a short span of time. We would also like submit our sincere thanks to the program committee members such as Dr. Le Van Thai, Dr. Hoang Manh Kha, Dr. Nguyen Thi Dieu Linh, Dr. Phan Thi Thu Hang, Dr. Tong Van Luyen—Electronic Engineering Faculty of the HaUI; Prof. Tran Duc Tan—Phenikaa University, Vietnam; and Dr. Raghvendra Kumar, GIET University, Gunupur, Odisha, India for their efforts to make congress success. Moreover, we would like to thank all the authors who submitted papers to RICE 2019 and made a high-quality technical program possible. Finally, we acknowledge the support received from the faculty members, scholars of Electronic Engineering Faculty of the HaUI, officers, staffs, and the authority of Hanoi University of Industry. We hope that the articles will be useful for the researchers who are pursuing research in the field of computer science, information technology, and related areas. Practicing technologists would also find this volume to be a good source of reference. Hyderabad, India Ha Noi, Vietnam Huddersfield, UK Bhubaneswar, India

Vijender Kumar Solanki Manh Kha Hoang Zhonghyu (Joan) Lu Prasant Kumar Pattnaik

Contents

Assessment of the Heart Disease Using Soft Computing Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dharmpal Singh, Sudipta Sahana, Souvik Pal, Ira Nath and Sonali Bhattacharyya The Reasons for Rail Accident in India Using the Concept of Statistical Methods: An Analytical Approach . . . . . . . . . . . . . . . . . . Dharmpal Singh, Sudipta Sahana, Souvik Pal, Ira Nath, Sonali Bhattacharyya and Srabanti Chakraborty Automatic Music Genre Detection Using Artificial Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pratanu Mandal, Ira Nath, Nihal Gupta, Madhav Kumar Jha, Dev Gobind Ganguly and Souvik Pal Role of Ad Hoc and Sensor Network for Effective Business Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ishu Varshney and Sunny Prakash Implementation of Integrated Security System by Using Biometric Function in ATM Machine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pushpa Choudhary, Ashish Tripathi, Arun Kumar Singh and Prem Chand Vashist DTSS and Clustering for Energy Conservation in Wireless Sensor Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arpana Mishra, Shubham Shukla, Akhilesh Kumar Singh and Anika Gupta Load Distribution Challenges with Virtual Computing . . . . . . . . . . . . Neha Tyagi, Ajay Rana and Vineet Kansal Mobile Ad Hoc Network and Wireless Sensor Network: A Study of Recent Research Trends in Worldwide Aspects . . . . . . . . . Ishu Varshney

1

9

17

25

33

43 51

57

vii

viii

Contents

Comparative Analysis of Clustering Algorithm for Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Smriti Sachan, Mudita Vats, Arpana Mishra and Shilpa Choudhary Concept of Cancer Treatment by Heating Methodology of Microwave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Awanish Kumar Kaushik, Smriti Sachan, Shradha Gupta and Shilpa Choudhary Novel Approach to Detect and Extract the Contents in a Picture or Image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Awanish Kumar Kaushik, Shilpa Choudhary, Shashank Awasthi and Arun Pratap Srivastava

63

73

81

LEACH with Pheromone Energy Efficient Routing in Wireless Sensor Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arpana Mishra, Shilpa Choudhary, Mudita Vats and Smriti Sachan

91

Industrialization of IoT and Its Impact on Biomedical Life Sciences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aritra Bhuiya and Sudan Jha

99

Parts of Speech Tagging for Punjabi Language Using Supervised Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Simran Kaur Jolly and Rashmi Agrawal

107

Development of Decision Support System by Smart Monitoring of Micro Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laxmi Kant Sagar and D. Bhagwan Das

117

Gender Recognition from Real-Life Images . . . . . . . . . . . . . . . . . . . . . Apoorva Balyan, Shivani Suman, Najme Zehra Naqvi and Khyati Ahlawat

127

Applications of Raspberry Pi and Arduino to Monitor Water Quality Using Fuzzy Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Padmalaya Nayak, Chintakindi Praneeth Reddy and Devakishan Adla

135

Developing a Smart and Sustainable Transportation Plan for a Large-Sized City: An Approach to Smart City Modeling . . . . . . Sushobhan Majumdar and Bikramjit Sarkar

145

Improved Data Dissemination Protocol for VANET Using Whale Optimization Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bhoopendra Dwivedy and Anoop Kumar Bhola

153

A Novel IoT-Based Approach Towards Diabetes Prediction Using Big Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Riya Biswas, Souvik Pal, Nguyen Ha Huy Cuong and Arindam Chakrabarty

163

Contents

Technical Solutions to Build Technology Infrastructure for Applications in Smart Agricultural Models . . . . . . . . . . . . . . . . . . Nguyen Ha Huy Cuong, Souvik Pal, Sonali Bhattacharyya, Nguyen Thi Thuy Dien and Doan Van Thang

ix

171

Edge Detection Through Dynamic Programming in Ultrasound Gray Scale Digital Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anju Mishra, Ramashankar Yadav and Lalan Kumar

177

Impact of Heterogeneous IoT Devices for Indoor Localization Using RSSI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bhagwan Sahay Meena, Sujoy Deb and K. Hemachandran

187

Indoor Localization-Based Office Automation System Using IOT Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bhagwan Sahay Meena, Ramin Uddin Laskar and K. Hemachandran

199

Ensemble Based Approach for Intrusion Detection Using Extra Tree Classifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bhoopesh Singh Bhati and C. S. Rai

213

Fourth Industrial Revolution: Progression, Scope and Preparedness in India—Intervention of MSMEs . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arindam Chakrabarty, Tenzing Norbu and Manmohan Mall

221

Call Admission Control in Mobile Multimedia Network Using Grey Wolf Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sanjeev Kumar and Madhu Sharma Gaur

229

Feature Classification and Analysis of Acute and Chronic Pancreatitis Using Supervised Machine Learning Algorithm . . . . . . . . R. Balakrishna and R. Anandan

241

RUDRA—A Novel Re-concurrent Unified Classifier for the Detection of Different Attacks in Wireless Sensor Networks . . . . . . . . . . . . . . . . S. Sridevi and R. Anandan

251

Health-Care Paradigm and Classification in IoT Ecosystem Using Big Data Analytics: An Analytical Survey . . . . . . . . . . . . . . . . . Riya Biswas, Souvik Pal, Bikramjit Sarkar and Arindam Chakrabarty

261

Human Activity Recognition from Video Clip . . . . . . . . . . . . . . . . . . . Rajiv Kumar, Laxmi Kant Sagar and Shashank Awasthi A Framework for Enhancing the Security of Motorbike Riders in Real Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yash Khandelwal, Sajid Anwar, Samarth Agarwal, Vikas Tripathi and Priyank Pandey

269

275

x

Contents

Fisherman Communication at Deep Sea Using Border Alert System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. R. Rajalakshmi and K. Saravanan

283

Promoting Green Products Through E-Governance Ecosystem: An Exploratory Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Arindam Chakrabarty, Mudang Tagiya and Shyamalee Sinha

297

Intervention of Smart Ecosystem in Indian Higher Education System: Inclusiveness, Quality and Accountability . . . . . . . . . . . . . . . . Arindam Chakrabarty, Mudang Tagiya and Shyamalee Sinha

305

A Study of Epidemic Approach for Worm Propagation in Wireless Sensor Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shashank Awasthi, Naresh Kumar and Pramod Kumar Srivastava

315

Adaptive Super-Twisting Sliding Mode Controller-Based PMSM Fed Four Switch Three Phase Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. Balaji and R. Ashok Kumar

327

Design of Multiplier and Accumulator Unit for Low Power Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J. Balamurugan and M. Gnanasekaran

339

Design and Implementation of IoT-Based Wireless Sensors for Ecological Monitoring System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Santhosh, Basava Dhanne and G. Upender

349

Enhancing Security in Smart Homes-A Review . . . . . . . . . . . . . . . . . . Bhuvana Janita, R. Jagadeesh Kannan and N. Kumaratharan

361

Accident Detection Using GPS Sensing with Cloud-Offloading . . . . . . . D. Srilatha, B. Papachary and N. Sai Akhila

371

Non-linear Correction of Transient Authentication System for Cloud Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Shanthi and R. Jagadeesh Kannan

379

Automatic Nitrate Level Recognition in Agriculture Industry . . . . . . . Md. Ankushavali, G. Divya and N. sai Akhila

387

Edge Detection-Based Depth Analysis Using TD-WHOG Scheme . . . . P. Epsiba, G. Suresh and N. Kumaratharan

397

Friend List-Based Reliable Routing in Autonomous Mobile Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Sivasankar and T. Kumanan

409

Construction of Domain Ontology for Traditional Ayurvedic Medicine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Gayathri and R. Jagadeesh Kannan

417

Contents

xi

Systolic FIR Filter with Reduced Complexity SQRT CSLA Adder . . . M. Gnanasekaran and J. Balamurugan

427

Design of Digital FIR Filters for Low Power Applications . . . . . . . . . . Gunasekaran and G. P. Ramesh

433

Reduced Frequency and Area Efficient for Streaming Applications Using Clock Gating and BUFGCE Technology . . . . . . . . . . . . . . . . . . N. Lavanya, B. Harikrishna and K. Kalpana

441

Survey on Modular Multilevel Inverter Based on Various Switching Modules for Harmonic Elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . Varaparla Hari Babu and K. Balaji

451

ANFIS-Based MPPT Control in Current-Fed Inverter for AC Load Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shaik Mohammad Irshad and G. P. Ramesh

459

Dynamic Load Balancing Using Restoration Theory-Based Queuing Model for Distributed Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . P. Sheeba Ranjini and T. Hemamalini

471

Design of Tree-Based MAC for High-Speed Applications . . . . . . . . . . . Joseph Prabhakar Williams, M. Madan and Narendra Prasad A Survey of Workload Management Difficulties in the Public Cloud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. Baskar, G. K. D. Prasanna Venkatesan and S. Sangeetha A Review on Multiple Approaches to Medical Image Retrieval System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lakshmi R. Nair, Kamalraj Subramaniam and G. K. D. Prasannavenkatesan

483

491

501

Efficient FPGA-Based Design for Detecting Cardiac Dysrhythmias . . . S. Kripa and J. Jebastine

511

Light Fidelity System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. Noor Alleema, Aadil Khatri, Ankur Gupta and Devika Senapatil

521

Link Quality and Energy-Aware Metric-Based Routing Strategy in WSNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vijayabaskar and T. Kumanan

533

Genetic Algorithm-Based PCA Classification for Imbalanced Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mylam Chinnappan Babu and Sangaralingam Pushpa

541

Radix-2/4 FFT Multiplierless Architecture Using MBSLS in OFDM Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Manikandan and M. Anand

553

xii

Contents

Multipath Routing Strategy for Reducing Congestion in WSNS . . . . . M. Jothish Kumar and Baskaran Ramachandran Cascaded Multilevel Inverter-Fed Soft-Start Induction Motor Using DTFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. Murugesan and R. Karthikeyan Analysis of Digital FIR Filter Using RLS and FT-RLS . . . . . . . . . . . . N. C. Sendhilkumar Prasad and G. P. Ramesh

561

569 579

Analysis of Interline Dynamic Voltage Restoration in Transmission Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Padmarasan and R. Samuel Rajesh Babu

587

Polariton Modes in Dispersive and Absorptive One-Dimensional Structured Dielectric Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. Chandrasekar and G. P. Ramesh

597

QoS-Based Multi-hop Reverse Routing in WSNs . . . . . . . . . . . . . . . . . G. Elangovan and T. Kumanan

607

Mitigation of Power Quality in Wind DFIG-Fed Grid System . . . . . . . P. T. Rajan and G. P. Ramesh

615

Detection and Avoidance of Single and Cooperative Black Hole Attacks Using Packet Timeout Period in Mobile Ad hoc Networks . . . S. G. Rameshkumar and G. Mohan

625

Brain Image Classification Using Dual-Tree M-Band Wavelet Transform and Naïve Bayes Classifier . . . . . . . . . . . . . . . . . . . . . . . . . A. Ratna Raju, Suresh Pabboju and R. Rajeswara Rao

635

Comparative Analysis of Cascaded Multilevel Inverter with Switched Capacitor-Fed Single-Phase Multilevel Inverter for Improving Voltage Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shaik Nagulmeeravali and K. Balaji Review on Induction Motor Control Strategies with Various Converter and Inverter Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . Meera Shareef Sheik and K. Balaji EROI Analysis of 2 KW PV System . . . . . . . . . . . . . . . . . . . . . . . . . . . Harpreet Kaur Channi and Inderpreet Kaur

643

653 665

Suggesting Alternate Traffic Mode and Cost Optimization on Traffic-Related Impacts Using Machine Learning Techniques . . . . . M. S. Manivannan, R. Kavitha, R. Srikanth and Veena Narayanan

673

Rare Lazy Learning Associative Classification Using Cogency Measure for Heart Disease Prediction . . . . . . . . . . . . . . . . . . . . . . . . . S. P. Siddique Ibrahim and M. Sivabalakrishnan

681

Contents

Intensify of Metrics with the Integration of Software Testing Compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Vaithyasubramanian, P. M. S. S. Chandu and D. Saravanan Petri Nets for Pasting Tiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. I. Mary Metilda and D. Lalitha Ad Hoc Wireless Networks as Technology of Support for Ubiquitous Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Amjed Abbas Ahmed An Experimental Study on Optimization of a Photovoltaic Solar Pumping System Used for Solar Domestic Hot Water System Under Iraqi Climate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mahmoud Maustafa Mahdi and A. Gaddoa A Novel Technique for Web Pages Clustering Using LSA and K-Medoids Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nora Omran Alkaam, Noor A. Neamah and Faris Sahib Al-Rammahi Enhancement in S-Box of BRADG Algorithm . . . . . . . . . . . . . . . . . . . Ahmed J. Oabid, Salah AlBermany and Nora Omran Alkaam A SECURITY Sketch-Based Image Retrieval System Using Multi-edge Detection and Scale Inverant Feature Transform Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alaa Qasim Rahima and Hiba A. Traish

xiii

693 701

709

717

727 737

747

Smart Photo Clicker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. Noor Alleema, Ruchika Prasad, Akkudalai Priyanka and Archit Bhandari

755

Design of VLSI-Architecture for 128 Bit Inexact Speculative . . . . . . . . Peddi Ramesh, M. Sreevani and G. Upender

767

Investigation of Solar Based SL-QZSI Fed Sensorless Control of BLDC Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Sundaram and G. P. Ramesh

779

Design of Hybrid Electrical Tricycle for Physically Challenged Person . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S. Swapna and K. Siddappa Naidu

789

Intravascular Ultrasound Image Classification Using Wavelet Energy Features and Random Forest Classifier . . . . . . . . . . . . . . . . . . A. Swarnalatha and M. Manikandan

803

xiv

Contents

Adaptive Thresholding Skin Lesion Segmentation with Gabor Filters and Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dang N. H. Thanh, Nguyen Ngoc Hien, V. B. Surya Prasath, Uğur Erkan and Aditya Khamparia Simple Model for Thermal Denaturing of Proteins Absorbed to Metallic Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Luong Thi Theu, Van Dung Nguyen, Pham Thi Thu Ha and Tran Quang Huy Trajectory Tracking Sliding Mode Control for Cart and Pole System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gia-Bao Hong, Mircea Nitulescu, Ionel Cristian Vladu, Minh-Tam Nguyen, Thi-Thanh-Hoang Le, Phong-Luu Nguyen, Thanh-Liem Truong, Van-Dong-Hai Nguyen and Xuan-Dung Huynh Online Buying Behaviors on E-Retailer Websites in Vietnam: The Differences in the Initial Purchase and Repurchase . . . . . . . . . . . . Nguyen Binh Minh Le and Thi Phuong Thao Hoang Combination of Artificial Intelligence and Continuous Wave Radar Sensor in Diagnosing Breathing Disorder . . . . . . . . . . . . . . . . . . . . . . . Nguyen Thi Phuoc Van, Liqiong Tang, Syed Faraz Hasan, Subhas Mukhopadhyay and Nguyen Duc Minh An Adaptive Local Thresholding Roads Segmentation Method for Satellite Aerial Images with Normalized HSV and Lab Color Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Le Thi Thanh and Dang N. H. Thanh Flexible Development for Embedded System Software . . . . . . . . . . . . . Phan Duy Hung, Le Hoang Nam and Hoang Van Thang Trajectory Tracking Pid-Sliding Mode Control for Two-Wheeled Self-Balancing Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anh Khoa Vo, Hong Thang Nguyen, Van Dong Hai Nguyen, Minh Tam Nguyen and Thi Thanh Hoang Le Evaluating Blockchain IoT Frameworks . . . . . . . . . . . . . . . . . . . . . . . Le Trung Kien, Phan Duy Hung and Kieu Ha My An Improved Approach for Cluster Newton Method in Parameter Identification for Pharmacokinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . Thang Van Nguyen, Tran Quang Huy, Van Dung Nguyen, Nguyen Thi Thu and Tran Duc Tan

811

821

829

845

853

865 873

885

899

913

Contents

An Efficient Procedure of Multi-frequency Use for Image Reconstruction in Ultrasound Tomography . . . . . . . . . . . . . . . . . . . . . Tran Quang Huy, Van Dung Nguyen, Chu Thi Phuong Dung, Bui Trung Ninh and Tran Duc Tan Intelligent Rule-Based Support Model Using Log Files in Big Data for Optimized Service Call Center Schedule . . . . . . . . . . . . . . . . . . . . . Hai Van Pham and Long Kim Cu Hybrid Random Under-Sampling Approach in MRI Compressed Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thang Van Nguyen, Tran Quang Huy, Van Dung Nguyen, Nguyen Thi Thu, Gian Quoc Anh and Tran Duc Tan Dynamics of Self-guided Rocket Control with the Optimal Angle Coordinate System Combined with Measuring Target Parameters for Frequency Modulated Continuous Wave Radar . . . . . . . . . . . . . . . Le Hai Ha, Nguyen Quang Vinh and Nguyen Tang Cuong An Approach of Utilizing Binary Bat Algorithm for Pattern Nulling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. L. Tong, Manh Kha Hoang, T. H. Duong, T. Q. T. Pham, V. T. Nguyen and V. B. G. Truong An Application of WSN in Smart Aquaculture Farming . . . . . . . . . . . Thong Nguyen Huy, Khanh Nguyen Tuan and Thanh Tran Trung A Newly Developed Approach for Transmit Beamforming in Multicast Transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . T. D. Thong, D. D. Vien and L. T. Hai Post-quantum Commutative Deniable Encryption Algorithm . . . . . . . . Nguyen Hieu Minh, Dmitriy Nikolaevich Moldovyan, Nikolay Andreevich Moldovyan, Quang Minh Le, Sy Tan Ho, Long Giang Nguyen, Hai Vinh Nguyen and Cong Manh Tran

xv

921

931

943

951

963

975

985 993

Indoor Positioning Using BLE iBeacon, Smartphone Sensors, and Distance-Based Position Correction Algorithm . . . . . . . . . . . . . . . 1007 Anh Vu-Tuan Trinh, Thai-Mai Thi Dinh, Quoc-Tuan Nguyen and Kumbesan Sandrasegaran Assessing the Transient Structure with Respect to the Voltage Stability in Large Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1017 Luu Huu Vinh Quang Cellular Automata Approach for Optimizing Radio Coverage: A Case Study on Archipelago Surveillance . . . . . . . . . . . . . . . . . . . . . . 1027 Tuyen Phong Truong, Toan Hai Le and Binh Thai Duong

xvi

Contents

Smart Bicycle: IoT-Based Transportation Service . . . . . . . . . . . . . . . . 1037 Vikram Puri, Sandeep Singh Jagdev, Jolanda G. Tromp and Chung Van Le LNA Nonlinear Distortion Impacts in Multichannel Direct RF Digitization Receivers and Linearization Techniques . . . . . . . . . . . . . . 1045 Ngoc-Anh Vu, Thi-Hong-Tham Tran, Quang-Kien Trinh and Hai-Nam Le Modified Biological Model of Meat in the Frequency Range from 50 Hz to 1 MHz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055 Kien Nguyen Phan, Vu Anh Tran and Trung Thanh Dang A Research on Clustering and Identifying Automated Communication in the HTTP Environment . . . . . . . . . . . . . . . . . . . . . 1069 Manh Cong Tran, Nguyen Quang Thi, Nguyen The Tien, Nguyen Xuan Phuc and Nguyen Hieu Minh Sequential All-Digital Background Calibration for Channel Mismatches in Time-Interleaved ADC . . . . . . . . . . . . . . . . . . . . . . . . . 1081 Van-Thanh Ta and Van-Phuc Hoang Comparison BICM-ID to Turbo Code in Wide Band Communication Systems in the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1091 Do Cong Hung, Nguyen Van Nam and Tran Van Dinh A Design of a Vestibular Disorder Evaluation System . . . . . . . . . . . . . 1105 Hoang Quang Huy, Vu Anh Tran, Nguyen Thu Phuong, Nguyen Khai Hung, Do Dong Son, Dang Thu Huong and Bui Van Dinh About Model Separation Techniques and Control Problems of Wheeled Mobile Robots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1119 Dao Phuong Nam, Nguyen Hoang Ha, Vu Anh Tran, Do Duy Khanh, Nguyen Dinh Khue and Dang Van Trong Fuzzy Supply Chain Performance Measurement Model Based on SCOR 12.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1129 Debasish Majumder, Rupak Bhattacharjee and Mrinmoy Dam Lightweight Convolution Neural Network Based on Feature Concatenate for Facial Expression Recognition . . . . . . . . . . . . . . . . . . 1141 Xiaohong Cai, Zheng Yan, Fang Duan, Di Hu and Jiaming Zhang Design and Issues for Recognizing Network Attack Intention . . . . . . . 1149 Anchit Bijalwan and Satenaw Sando Web-Based Learning: A Strategy of Teaching Gender Violence . . . . . 1157 Jyotirmayee Ojha and Deepanjali Mishra A Deep Neural Network Approach to Predict the Wine Taste Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1165 Sachin Kumar, Yana Kraeva, Radoslava Kraleva and Mikhail Zymbler

Contents

xvii

A Global Distributed Trust Management Framework . . . . . . . . . . . . . 1175 Son Doan Trung Single-Phase Smart Energy Meter—IoT Based on Manage Household Electricity Consumption Service . . . . . . . . . . . . . . . . . . . . . 1185 Thi Dieu Linh Nguyen, Ngoc Duc Tran and Thi Hien Tran Survey on Machine Learning-Based Clustering Algorithms for IoT Data Cluster Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1195 Sivadi Balakrishna, Vijender Kumar Solanki, Raghvendra Kumar and M. Thirumaran

About the Editors

Dr. Vijender Kumar Solanki is an Associate Professor in Computer Science & Engineering, CMR Institute of Technology (Autonomous), Hyderabad, TS, India. He has more than 10 years of academic experience in network security, IoT, Big Data, Smart City and IT. Prior to his current role, he was associated with Apeejay Institute of Technology, Greater Noida, UP, KSRCE (Autonomous) Institution, Tamilnadu, India and Institute of Technology & Science, Ghaziabad, UP, India. He has authored more than 40 research articles published in various journals, books and conference proceedings. He has co-edited 12 books and conference proceedings in the area of soft computing. He received his Ph.D in Computer Science and Engineering from Anna University, Chennai, India in 2017 and ME, MCA from Maharishi Dayanand University, Rohtak, Haryana, India in 2007 and 2004, respectively, and a bachelor’s degree in science from JLN Government College, Faridabad Haryana, India in 2001. He is the book series editor of Internet of Everything (IoE) : Security and Privacy Paradigm, CRC Press; Artificial Intelligence (AI): Elementary to Advanced Practices Series, CRC Press; IT, Management & Operations Research Practices, CRC Press; Bio-Medical Engineering: Techniques and Applications with Apple Academic Press, and Computational Intelligence and Management Science Paradigm, (Focus Series) CRC Press. He is the Editor-in-Chief of International Journal of Machine Learning and Networked Collaborative Engineering; International Journal of Hyperconnectivity and the Internet of Things, IGI-Global, co-editor of Ingenieria Solidaria Journal, Associate Editor in International Journal of Information Retrieval Research, IGI-Global. Dr. Manh Kha Hoang is the vice dean of Faculty of Electronics, Hanoi University of Industry, Hanoi, Vietnam. He was born in Bac Giang, Vietnam, in 1979. He received the B.E and M.E degree in Electronics and Telecommunications Engineering both from Hanoi University of Science and Technology, Vietnam, in 2002 and 2004, respectively. He obtained his Dr.-Eng. degree in Communications Engineering from the University of Paderborn, Germany, in 2016 with specialization in parameter estimation of missing data and application to indoor positioning. He has published several articles in ISI/Scopus indexed journals and xix

xx

About the Editors

conferences. He has been serving as TPC member and reviewer for several conferences such as International Conference on Advanced Technologies for Communications (ATC 2017, 2018, 2019), National Conference on Electronics, Communications and Information Technology (REV-ECIT) 2018. He is also serving as reviewer for Journal of Science and Technology. His research interests include digital signal processing, wireless communication, positioning engineering, nature-inspired optimization, smart antenna. Dr. Zhonghyu (Joan) Lu is Professor in Informatics at the University of Huddersfield (UK). Her extensive research covers information access, retrieval and visualization, XML technology, object oriented technologies, agent technology, data management systems, security issues and Internet computing. She has been an invited speaker for industrial-oriented events and published 5 academic books and more than 160 papers. Prof. Lu has acted as the founder and a program chair for the International XML Technology workshop and XMLTech (USA) for 11 years (2003-2011). She also serves as Chair of 5 separate international conferences, is a regular reviewer for several international journals, and a committee member for 16 international conferences. She specializes in XML technology and mobile computing with image retrieval through the latest wireless devices. Professor Lu serves as a member of the British Computer Society (BCS), BCS examiner of Advanced Database Management Systems, and fellow of the Higher Education Academy (UK). She is a founder and Editor-in-Chief for the International Journal of Information Retrieval Research. Dr. Prasant Kumar Pattnaik, Fellow IETE, Senior Member IEEE, is a Professor at the School of Computer Engineering, KIIT Deemed to be University, Bhubaneswar. He has more than a decade of teaching and research experience and has guided several doctoral students. Prof. Pattnaik has published a number of research articles in peer-reviewed international journals and conferences, and filed many patents. He has also edited book volumes in Springer and IGI Global. His areas of interest include Mobile Computing, Cloud Computing, Cyber Security, Intelligent Systems and Brain Computer Interface. He is an Associate Editor of Journal of Intelligent & Fuzzy Systems, IOS Press, and a book series editor of Intelligent Systems, CRC Press.

Assessment of the Heart Disease Using Soft Computing Methodology Dharmpal Singh, Sudipta Sahana, Souvik Pal, Ira Nath and Sonali Bhattacharyya

Abstract The techniques behind Data mining play a significant task in health care field. The techniques are mainly involved in prediction-based computing. A wide variety of algorithms are available in data mining. Data mining has many modern data analysis techniques like classification and prediction which may be utilized for this purpose. Classification is a data mining approach that allocates items in a group to target categories or classes. It also may be used to label a target item into any one of the classes identified. Among many available classification techniques, clustering is one of the unsupervised machine learning approaches that could be used for creating clusters as features to enhance classification models. There are various clustering algorithms available like K-mean clustering, AGNES clustering, etc. For disease prediction system K-means algorithm is used among all techniques available in prediction system. K-Mean algorithm creates clusters and groups of data properly. Many researchers have worked in the data clustering field and various clustering techniques have been used by them. In this work at first, Knowledge-based data are created by using factor analysis. After that K-mean algorithm is used on the result of knowledge-based data and different number of cluster points are assumed, then we get the Euclidean distance function with the simple matching dissimilarity measure. Once the cluster was formed, another method was used to forecast the resultant values of the data in the cluster. The proposed approach was tested on heart disease dataset and found efficient in this domain. D. Singh (B) · S. Sahana · I. Nath · S. Bhattacharyya Department of Computer Science & Engineering, JIS College of Engineering, Kalyani, India e-mail: [email protected] S. Sahana e-mail: [email protected] I. Nath e-mail: [email protected] S. Bhattacharyya e-mail: [email protected] S. Pal Department of Computer Science & Engineering, Brainware University, Kolkata, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_1

1

2

D. Singh et al.

Keywords Data mining · Prediction · Clustering · K-means · Hierarchical clustering · Fuzzy C-means

1 Introduction Being the muscular organ in the human body heart pumps blood through the blood vessels of the circulatory system [1]. Almost every organ in our body is supported by heart and is important for our survival. It measures about 250–350 g. Heart disorder is any type of trouble in heart function that causes syndrome. Liver disorder is also mentioned as cardiovascular disease [2]. Generally, chest discomfort, nausea, indigestion, heartburn or stomach pain are classic syndromes of heart disorder. Yet, no other technique or device is available that can recompense for the lack of the heart [3]. A premature heart disorder diagnosis can guide to healthier treatment and increase the chances of survival. Numerous research groups have developed diverse approaches and algorithms for heart disease. There are many terminologies which are used in detecting heart disorder to compare and contrast them comparative to data mining system. Data mining is the information discovering practice that accustomed to extract data patterns from a large quantity of data as the number of patients and treatments growing in the medical databases day by day [4]. The investigation of these medical data is hard without the computer-based investigation system which provides the automated medical diagnosis system. It supports to make effective decision in treatment cum disease detection and increases the quality of service provided to the patients. The heart disease can be caused by a variety of factors that damage the heart, such as hypertension, high cholesterol and smoking. Over time, damage to the heart results in scarring (cirrhosis), which can lead to heart failure, a life-threatening condition. Symptoms and signs of heart disorder include: Chest pain, Chest tightness, Chest pressure, Chest discomfort (angina), Shortness of breath, Pain in the neck, jaw, throat, upper abdomen or back, dizziness, light-headedness, easily getting short of breath during exercise or activity and Fatigue [5, 6].

2 Methodology Multivariate Data Analysis (MDA): Multivariate data analysis indicates to any statistical technique used to analyze data that arises from more than one variable. This essentially prototypes reality where each situation, product or decision involves more than a single variable. Masses of data in every field are the result of the information age. The ability to obtain a clear picture of what is going on and make intelligent decisions, despite the quantum of data available is a challenge. Multivariate Analysis can be used to process the information in a meaningful fashion when available information is stored in the database tables containing rows and columns [2, 3].

Assessment of the Heart Disease Using Soft Computing Methodology

3

Factor Analysis: Factor analysis is denoted as a statistical method to be used to illustrate variability among experiential, related variables in terms of a potentially lower number of unnoticed variables called factors. For example, it is probable that variations in four observed variables mainly reflect the variations in two unobserved variables. Factor analysis is experimented on the correlation matrix of the observed variables. Weighted average of the original variables is termed as factor. The factor analyst hopes to find a few factors from which the original correlation matrix may be generated. The information expanded about the dependencies between observed variables can be considered afterwards to reduce the set of variables in a dataset. Computationally, this system is comparable to low-rank estimation of the matrix of observed variables. Factor analysis has been invented in psychometrics and is used in behavioural sciences, social sciences, marketing, product management, operations research and other applied sciences that deal with large quantities of data [4] (Fig 1). Proposed Algorithm Step 1: Step 2: Step 3: Step 4: Step 5: Step 6:

Identify business goals and Identify data mining goals Assess needed data and Collect and understand data Select required data and Cleanse/format data as necessary Select algorithms and Build predictive models Train the model with sample datasets and Test and iterate Verify final model and Prepare visualization and deploy

KDD Process: People deal data mining as a synonym for another most used term, knowledge discovery from data or KDD which is a process of discovering useful knowledge from a collection of data. Major KDD application areas include fraud detection, marketing, manufacturing and telecommunication [3–6]. The following diagram represents different steps of KDD process. Steps of KDD are listed as data selection, data cleaning, data integration, data transformation, data mining, pattern evaluation, knowledge representation.

3 System Analysis and Proposed Method Software Specification: MATLAB is used to provide easy access to matrix software which was developed. It is a high-performance language which is used for technical computing which included computation, visualization and programming environment (Table 1). Proposed Method: In this present work, we had assessed heart disorder detection using data mining and soft computing techniques, and finally created a knowledge base using k-means clustering algorithm. In this experiment, we had evaluated our methodology based on a benchmark dataset https://archive.ics.uci.edu/ml/machinelearning-databases/heart-disease/. Dataset consist of 123 instances each having 65 attributes. All the instances represent a heart disorder of patient. The set of attributes like age, sex, cholesterol, chest pain, blood sugar, blood pressure and so on indicates

4

D. Singh et al.

Fig. 1 Multivariate data analysis Flowchart

Table 1 Cluster centroid location table

Record no.

Cluster no.

Cluster centroid location

1,2

2

83.88

3–11

5

267.53

12–56

4

351.15

57–99

1

395.14

100–123

3

437.75

Assessment of the Heart Disease Using Soft Computing Methodology Table 2 Knowledge base table (k-means)

5

Cluster point (k)

Euclidean distance

2

106786.13

5

7179.92

8

2347.68

16

136.80

32

8.90

64

0.24

the significant amount of disorders result from the different kind of blood test experiment. At first, we had applied multivariate data analysis, i.e. factor analysis (FA) and found that factor analysis produced better result. Therefore for the entire experiment we had applied factor analysis. After getting the total effect/cumulative effect from factor analysis we have applied k-means algorithm.

4 Result and Discussion According to the KDD process, the Total Effect values were calculated by using the factor analysis. After that K-means clustering algorithm was used on the total effect value for different K values (number of clusters) which were assumed and following results were depicted (Table 2). It has been observed that there are five variations in dataset, so that we have taken cluster point, K = 5 for further work, so the other cluster points have been ignored. Then the same process is followed for Hierarchical clustering and Fuzzy C-means clustering. The resultant knowledge base has been furnished in table as following: Case 1: Knowledge base table (k-means) Case 2: Knowledge base table (Hierarchical clustering) Case 3: Knowledge base table (Fuzzy C-means). After the formation of the knowledge base, raw data has been taken and total effect is also calculated by the procedure of factor analysis and we have calculated the new total effect based upon that to form the clusters. Then we have calculated the accuracy and errors based on the new dataset (Tables 3, 4 and 5).

5 Conclusion Heart disorder analysis is a necessary domain of research in view of detection and prevention of disorder. In this paper, the concept of multivariate data analysis, i.e. factor analysis is (FA) applied to form the total effect. After getting the total

6 Table 3 Knowledge base table (Hierarchical clustering)

Table 4 Knowledge base table (Fuzzy C-means)

Table 5 Error and accuracy analysis

D. Singh et al. Record no.

Cluster no.

Cluster centroid location

1

5

3

2

4

3

3, 4

2

2

5–11

1

1.42

12–123

3

1.80

Record no.

Cluster no.

Cluster centroid location

1, 2

3

3

3–11

2

1.55

12–54

5

1.74

55–94

4

1.75

95–123

1

1.96

Technique

Error %

Accuracy %

K-means

10.38

89.62

Hierarchical clustering

10.11

89.89

Fuzzy C-means

10.32

89.68

effect/cumulative effect from factor analysis, the concept of K-means algorithm, Hierarchical method and C-means algorithm has been applied to form the knowledge base. Thereafter, new data termed as tested data has been created to check the optimality of the model in different ways. Average percentage error has been calculated on three cases of the tested data with the help of the formed knowledge base. The percentage error of three cases are 10.38%, 10.11% and 10.32%, respectively. Thus, we can conclude that as the error is minimum in case of Hierarchical method, we will use Hierarchical method for clustering to confirm the assertive analysis. To determine the stage of disorder we will use K-means Clustering technique.

References 1. Rajalakshmi K, Dhenakaran SS (2015) Analysis of data mining prediction techniques in healthcare management system. Int J Adv Res Comput Sci Softw Eng 5(4) 2. Pakhale H, Xaxa DK (2016) A survey on diagnosis of liver disease classification. Int J Eng Tech 2(3) 3. Singh I, Gupta N (2015) An improved K-means clustering method for liver segmentation. Int J Eng Res Technol (IJERT) 4(12). ISSN 2278-0181

Assessment of the Heart Disease Using Soft Computing Methodology

7

4. Gorunescu M, Gorunescu F, Badea R, Lupsor M (2009) Evaluation on liver fibrosis stages using the K-means clustering algorithm. Ann Univ Craiova Math Comp Sci Ser 36(2):19. ISSN 1223–6934 5. Kaur R, Kaur L, Gupta S (2011) Enhanced K-mean clustering algorithm for liver image segmentation to extract cyst region. IJCA Special issue on Novel aspects of digital imaging applications DIA 6. Khan Z, Ni J, Fan X, Shi P (2017) An improved K-means clustering algorithm based on an adaptive initial parameter estimation procedure for image segmentation. Int J Innov Comput Inf Control ICIC Int © 2017 13(5). ISSN 1349–4198

The Reasons for Rail Accident in India Using the Concept of Statistical Methods: An Analytical Approach Dharmpal Singh, Sudipta Sahana, Souvik Pal, Ira Nath, Sonali Bhattacharyya and Srabanti Chakraborty

Abstract Indian Railways (IR) is one of the sources of technical growth, economic growth, and development progress in India. Its safety is not just a part of national concern but also brings the responsibility and challenges for the researcher of this country to optimize the safety using some meaningful tools. It has been observed that researchers have done analysis to provide the security/safety measure but which factor impacts the analysis most has not been assessed. Therefore, in this paper an effort has been made to analyze the causes of rail accidents occurred during the period from 2002 to 2015 using the concept of statistical tools. The concept of factor analysis has been used to select the major factor that may cause the railway accidents in the future and the least square-based method has been used to predict the number of people who might be killed in the forthcoming year due to rail accident. Keywords Indian Railways · Least square base · Linear equation · Exponential equation · Asymptotic equation · Curvilinear and logarithmic equation

D. Singh (B) · S. Sahana · I. Nath · S. Bhattacharyya Department of Computer Science & Engineering, JIS College of Engineering, Kalyani, India e-mail: [email protected] S. Sahana e-mail: [email protected] I. Nath e-mail: [email protected] S. Bhattacharyya e-mail: [email protected] S. Pal Department of Computer Science & Engineering, Brainware University, Kolkata, India e-mail: [email protected] S. Chakraborty Elitte Institute of Engineering and Management, Kolkata, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_2

9

10

D. Singh et al.

1 Introduction Indian Railways is the vertebral column of the country’s financial system and an essential part of our public fabric. It is one of the major rail networks in the globe with a sum of approximately 63,000 route km. Indian Railways carry 14 million passengers and run nearly 14,000 trains with a million tons of goods every day. Like urbanized countries, Indian Railways function under very much exigent situation of congestion, low cost of travel, longer trains, etc. The challenges posed in terms of expertise development and completion, are also thus greater, varied, and inimitable in nature. These challenges also provide vast opportunities for utilization of available human and procedural resources for expansion of safer and cost-effective railway network. India competes with nations of the world in nuclear, space, and software technologies [1–4]. It is vital that Railways’ massive technical network be wound up through doing research and advance in order to stimulate extensive transportation and profitable development. Indian Railways focus on modern technologies of monitoring, control, transportation, electronics, design and resources for railway safety [5, 6]. Technology issues on railway security and economy relate to large number of research by the researcher to provide the idea to make the railways work fruitful [7, 8]. The objective of this paper is to inference/estimate most important causes of accidents in particular month in Indian Railways to minimize the death of people. This will also be helpful to save the life of people lost in different causes of rail accidents and railways will be able to minimize it and provide safer train movement.

2 Objective of the Work In this paper, we have tried to find out the reasons for train accidents in India using statistical methods. We have analyzed the different train accidents occurred between 2002 and 2016. And, we have taken different parameters for the same. In this manuscript, we have taken multivariate data analysis, principle components analysis, the least squares regression line, least square techniques based on linear equation, exponential equation, asymptotic equation, curvilinear equation and logarithmic equation [9–12].

3 Implementation Step-1. The first step of multivariate data analysis is to generate linear correlation coefficient from the original dataset. While computing we need to consider only the attributes, i.e., discarding the status column. Following values will come as correlation matrix (Table 1).

The Reasons for Rail Accident in India … Table 1 Correlation matrix

Table 2 Eigen value and percentage contribution

Table 3 Eigen vectors

11

1

−0.593

−0.593

1

−0.3603

0.1641

−0.3603

1

0.1641

Data attribute

Eigen value

A

0.188

Percentage contribution

B

1.028

34.26667

C

1.784

59.46667

6.266667

A

B

C

0.6256

−0.4946

0.6034

0.7362

0.1184

−0.6663

0.2581

0.861

0.4382

Step-2. The input of this step is the correlation matrix and we need to generate the Eigen value from it. Step-3. From the table, it has been observed that percentage contribution of the item E is very less as compared to other items therefore it has been ignored. The eigen vectors of the five items (A, B, C, D, F) have been furnished in Table 3. √ Step-4. The major factors have been calculated as per the formula ( eigen value × square (eigen vector)) using the selected eigen values as furnished in Table 2 and eigen vectors as furnished in Table 3. The major factors have been furnished in Table 4. Step-5. The cumulative effect value for all these data items have been calculated by adding the values row wise corresponding to each element of Table 4. The cumulative values for all items have been furnished in Table 5. Table 4 Contribution of eigen vector corresponding eigen value

Table 5 Cumulative effect value of items

Data attribute/Eigen value

3.1341

1.3713

0.7375

A

0.169696

0.24803

0.486304

B

0.235002

0.014213

0.592976

C

0.028884

0.751628

0.256473

Data attribute

Cumulative effect

A

0.949316

B

0.933902

C

0.982214

12

D. Singh et al.

Step-6. Now a relation has been formed by using the cumulative effect value of all the elements to produce total effect value. Total effect value = (0.949316) × A + (0.933902) × B + (0.982214) × C + using the Table 6, total effect has been found and the concept of least square-based linear, exponential, asymptotic, curvilinear and logarithmic have been applied to find the optimal equation for the optimal result. Furthermore, the result and percentage error have been furnished in Tables 7 and 8, respectively.

3.1 Analysis on Day, Month, and Year of Accident From the table, it has been observed that the least square-based linear equation has given the optimal result for the selected dataset. Therefore, least square-based linear equation will be used as inference model for the selected data set. Moreover, the concept of factor analysis and least square-based linear equation has been also applied on the train accident estimation of Indian Railways. The data of the aforesaid work has been taken from the Wikipedia and converted in excel sheet based on the day month and year of the accident. Out of 165 data elements, only 150 data has been taken as the training data and remaining 15 data elements will be used as the tested data to check the optimality if the estimation. The concept of the least square-based linear equation has been applied on the training data and percentage error is found as 17%. Tested data: Certain data that has not been processed earlier has been taken as the tested data and inference and prediction were made on the taken data. Furthermore, it has been observed that training data year started with 12 years ahead so that aforesaid value has been deducted at the time of estimation of result of tested data. The estimation and actual value has been furnished as follows (Table 9). The estimated percentage error is 10.1% and the percentage error has been added to the last row of the actual data of day month and year. Following result of day month and year furnished as follows. Day = 24 (actual) + 10 (Percentage) = 34−31 (no. of days in month is maximum 31 or 30) = 3, month = 11 + 10 = 21−12 = 8, but for the year, estimated value has been taken, Year = 2028−10 = 2018. (Because next year is 2017 and if we take it then it will became 2038.), So next probable date of the accident will be 3rd day of August 2018.

4 Conclusion It has been observed that people lost their life due to train accident and suffered from many problems. Therefore, here in this paper an effort has been made to estimate the causes of the railway accidents based on the data provided by the government site. (https://data.gov.in/catalog/number-persons-killed-and-injured-

In train accidents, failures of Railway equipment and misc. accidents

157

84

35

168

38

9

52

67

235

100

60

42

118

Year

2002–03

2003–04

2004–05

2005–06

2006–07

2007–08

2008–09

2009–10

2010–11

2011–12

2012–13

2013–14

2014–15

6

7

4

17

8

4

12

10

6

9

5

3

29

168

103

140

202

138

167

145

172

166

138

181

155

232

292

152

204

319

381

238

209

191

210

315

221

242

418

Total by factor 1

533

546

142

73

54

46

146

125

166

178

83

54

94

146

163

32

32

26

46

60

91

90

77

64

50

84

1657

3913

3674

3560

3471

3654

4428

3926

3713

3127

1842

2593

1210

In unusual occurrences caused by movement of railway vehicles exclusive of train accidents, etc.

Table 6 Number of people killed in three major parameters

2336

4622

3848

3665

3551

3746

4634

4142

3969

3382

1989

2697

1388

Total by factor 2

59

14

1

1

2

2

2

0

0

0

0

0

0

25

9

8

8

17

8

5

12

13

23

16

24

30

143

6

4

4

4

3

2

4

3

0

0

0

1

In unusual occurrences on railway premises not connected with the movement of railway vehicles

227

29

13

13

23

13

9

16

16

23

16

24

31

Total by factor 3

2680.562

4486.91

3798.115

3736.476

3698.749

3735.186

4532.577

4063.137

3919.697

3478.351

2082.027

2770.661

1722.808

Total Effect on all factors

The Reasons for Rail Accident in India … 13

14 Table 7 Estimated result using statistical methods on three parameters of rail accident

Table 8 Percentage error on statistical methods

D. Singh et al. Parameters three estimated result

Parameters two estimated result

Parameters one estimated result

−3.65902

−3.72527

2643.441

2.655216

2.703297

2766.263

8.969449

9.131868

2889.086

15.28368

15.56044

3011.909

21.59791

21.98901

3134.732

27.91215

28.41758

3257.554

34.22638

34.84615

3380.377

40.54061

41.27473

3503.2

46.85485

47.7033

3626.023

53.16908

54.13187

3748.846

59.48331

60.56044

3871.668

65.79754

66.98901

3994.491

72.11178

73.41758

4117.314

Least square method

Percentage error (%)

Linear

18.3

Exponential

19.65

Asymptotic

21.49

Curvilinear

18.82

Logarithmic

20.16

railway-related-accidents). The estimation of the cause will be helpful for the Government to take necessary action for the aforesaid causes to reduce the number of the accident in future. Moreover, train accident (https://en.wikipedia.org/wiki/List_ of_Indian_rail_accidents) date, month and year estimation were also done based on the aforesaid methods to provide the information to government to be careful for the particular month of the year. This effort will be able to minimize the chance of occurrence of accident due to different causes. The concept of factor analysis has been used to select the major factor that caused the railways accident in the future and the least square-based method has been used to predict the number of people who might be killed in the forthcoming year due to rail accident.

The Reasons for Rail Accident in India …

15

Table 9 Day, month, and year of train accident in India Actual data Day

Month

Estimated data Year

Day

Month

Estimated error Year

Day

Month

Year

29

9

2016

1

7

2022

28

2

6

20

11

2016

1

7

2023

19

4

7

6

12

2016

1

7

2023

5

5

7

6

12

2016

1

7

2023

5

5

7

28

12

2016

1

7

2024

27

5

8

28

12

2016

1

7

2024

27

5

8

21

1

2017

1

7

2025

20

6

8

7

3

2017

1

7

2025

6

4

8

30

3

2017

1

7

2025

29

4

8

9

4

2017

1

7

2026

8

3

9

15

4

2017

1

7

2026

14

3

9

19

8

2017

1

7

2026

18

1

9

23

8

2017

1

7

2027

22

1

10

24

11

2017

1

7

2027

23

4

10

24

11

2017

1

7

2028

23

4

11

References 1. Jian-hua Q, Lin-sheng L, Jing-gang Z (2008) Design of rail surface crack detecting system based on linear CCD sensor. In: IEEE international conference on networking sensing and control 2. Vijayakumar K, Wylie SR, Cullen JD, Wright CC, AIShamma’a AI (2009) Non invasive rail track detection system using Microwave sensor. J Appl Phys 3. Cacciola M, Megali G, Pellicanµo G, Calcagno S, Versaci M, Morabito FC (2010) Rotating electromagnetic field for crack detection in railway tracks. PIERS ONLINE 6(3) 4. Wojnarowski RJ, Welles IKB, Kornrumpf, WP. Electromagnetic system for railroad track crack detection and traction enhancement. US Patent 6,262,573. www.patentstorm.us/patents/ 6262573/description.html 5. Tranverse crack detection in rail head using low frequency eddy currents. Patent US 6,768,298. www.google.com/patents/US6768298 6. Polivka AL, Matheson WL (2014) Automatic train control system and method. US Patent 5,828,979, 27 October 2014 7. Palmer SB, Dixon S, Edwards RS, Jian X. Transverse and longitudinal crack detection in the head of rail tracks using Rayleigh wave-like wideband guided ultrasonic wave. Centre for Materials Science and Engineering The University of Edinburgh. www.cmse.ed.ac.uk/ AdvMat45/Rail-crack-detection.pdf 8. Lanza di Scalea F, Rizzo P, Coccia S, Bartoli I, Fateh M, Viola E, Pascale G. Non-contact ultrasonic inspection of rails 9. Ambegoda ALATD, Silva WTSD, Hemachandra KT, Samarasinghe TN, Samarasinghe ATLK (2013) Centralized traffic controlling system for Sri Lanka railways. In: 4th international conference on information and automation for sustainability (ICIAFS’08), Sri Lanka, pp 12–14 10. Mohaiminul MD, Khan I (2014) Automatic railway track switching system. Int J Adv Technol 54

16

D. Singh et al.

11. Boylestad FRL, Nashelsky L (2012) Railway crack detection using GPA technology, 9th edn. Prentice Hall, USA, pp 196–199 12. FRA Safety Statistics Data 1992–2002, FRA, U.S. Department of Transportation (2009)

Automatic Music Genre Detection Using Artificial Neural Networks Pratanu Mandal, Ira Nath, Nihal Gupta, Madhav Kumar Jha, Dev Gobind Ganguly and Souvik Pal

Abstract In this paper, we have explored the use of artificial neural networks (ANNs) for automatically detecting the genre of music. The challenge faced when hand-classifying music is that it is highly dependent on the accuracy and experience of the person classifying it. The main objective of this paper is to build an arrangement which will reduce the burden and increase the accuracy of classifying the genre of music. We use MFCC as feature vectors and use multilayer perceptrons (MLPs) to classify the data into the various genres. We have trained our model on a novel dataset that reflects current trends in music and addresses the problems faced with existing datasets. Keywords Genre · ANNs · FFT · Music · MFCC

P. Mandal · I. Nath (B) · N. Gupta · M. K. Jha · D. G. Ganguly Department of Computer Science & Engineering, JIS College of Engineering, Kalyani, India e-mail: [email protected] P. Mandal e-mail: [email protected] N. Gupta e-mail: [email protected] M. K. Jha e-mail: [email protected] D. G. Ganguly e-mail: [email protected] S. Pal Department of Computer Science & Engineering, Brainware University, Kolkata, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_3

17

18

P. Mandal et al.

1 Introduction With the advent of online music streaming services and ever-growing music industry, our music collection has been growing as well. This has presented us with the problem of having to classify music by genre. This when done manually is a highly tedious and error-prone task. In this paper, we propose the use of ANNs to automate the detection of the genre of music. The performance of the presented model has been evaluated on a novel dataset that is more up to date with modern music genres. The performance has also been compared with simpler machine learning approaches, for example, support vector machine (SVM) and k-nearest neighbour (KNN). The scope of the work is to build a system that can automatically analyze any previously unseen music file and predict its genre with reasonable accuracy.

2 Related Work In order to propose a methodology for classifying music by genre, we conducted the following background study where we explored previous works in this field and related fields such as voice and speech recognition. In 2001, Tzanetakis G, Essl G, Cook P presented a paper exploring algorithms for automatic genre classification. They used metal surface feature and rhythm feature to extract music characteristics [1]. In 2007, Annesi P, Basili R, Gitto R and Moschitti A presented a paper, where they utilized SVMs to plan an automatic classifier of music genres [2]. In 2011, Michael H, Yang H and Kenny K presented a paper in which they used KNN, kmeans, multi-class SVM and ANNs to classify music into classical, jazz, metal and pop genres. They used the GTZAN dataset from the MARSYAS website [3]. In 2015, Archit R and Margaux D presented a paper for classification of the genre of music using traditional machine learning approach. They used five features: MFCC, spectral centroid, zero crossing rate, chroma frequencies and spectral roll-off [4]. In 2016, Piotr K and Bartosz M developed a live music genre recognition system using convolutional recurrent neural networks. It provides a user-friendly visualization for the network’s current belief in real time [5].

3 Proposed Work In this paper, we propose the process illustrated by the flow chart in Fig. 1 for automatically classifying music by genre.

Automatic Music Genre Detection Using Artificial Neural Networks

19

Fig. 1 The proposed workflow

3.1 Dataset Formation A new dataset has been compiled consisting of the latest and the greatest songs from each genre with 100 for each genre, i.e. a total of 600 songs. Six genres were chosen, viz., Metal, Rock, Pop, Rap, Electronic and Jazz. The songs used for the dataset formation are of minimum 128 kbps bitrate in stereo mp3 format. Each song is first converted into 160 kbps bitrate mono wav format. Then each song has been split into tracks of 5 s. We obtained the Mel-Frequency Cepstral Coefficients (MFCC) for each 5 s segment followed by the feature vector extraction from the MFCC. We have utilized 80% of the dataset as training data and the rest as testing data.

3.2 Feature Extraction Feature extraction is the process of obtaining relevant data or information from a set of raw unprocessed data. Feature extraction is necessary because it is not feasible to train any model with an entire music track. Instead, we need to extract certain features from the music tracks on which we can train models. The aim is to extract relevant features such that the feature-patterns are similar intra-genre and as quite different inter-genre.

20

P. Mandal et al.

Table 1 Feature vector Sl. no.

Description

No. of features

1

Mean of MFCC

31

2

Standard deviation of MFCC

31

3

Covariance of MFCC

31

4

Genre transformed onto a numerical mapping (0–5)

Total no. of features

3.2.1

1 40

Mel-Frequency Cepstral Coefficients

Mel-Frequency Cepstrum (MFC) is a representation of a sound wave as the power cepstrum of the sound wave in the non-linear mel-scale. An MFC is jointly consisting of mel-frequency cepstral coefficients (MFCCs). Discrete cosine transform (DCT) is performed on the MFC as if it were a signal to obtain the MFCCs. In the MFC, the mel-scale is used to map the frequency bands such that they are equally spaced from each other. This provides a better representation of the human sense of hearing and is extremely useful for analyzing the audio in a manner that a human would perceive it.

3.2.2

Feature Vector

The feature vector contains the features that have been extracted in Sect. 3.2.1. Each music track consists of multiple feature vectors where each feature vector corresponds to a certain segment in the track. The feature vector used in our paper is formed from the MFCC of the 5 s tracks derived from each music file. The details of the feature vector are shown in Table 1.

3.3 Multilayer Perceptron (MLP) A multilayer perceptron (MLP) is a type of feedforward artificial neural network (ANN) capable of distinguishing non-linear data. An MLP consists of one input layer, one or more hidden layers and one output layer in that order. In our model, the layers are densely connected and certain drop out layers are introduced in between layers to prevent overfitting. We had experimented with various combinations of layers and weights. The following network was found to be near optimal in our research on our hardware environment which could be obtained in reasonable time (Fig. 2).

Automatic Music Genre Detection Using Artificial Neural Networks Input Layer Dense Layer

39 neurons 512 neurons

1024 neurons Dropout Layer

Dense Layer

Dropout Layer

Dropout Layer

relu 30%

512 neurons Dropout Layer

Output Layer (Dense)

relu

30% 1024 neurons

Dense Layer

relu 30%

2048 neurons

Dense Layer

relu 30%

Dropout Layer Dense Layer

21

relu 30%

6 neurons

softmax

Fig. 2 Multilayer perceptron layer used in our research

4 Experimental Results It is our recommendation to use a system with at least an Intel Core i5, with 4 GB RAM. The simulation work has been done using the Python 2.7 programming language. The project was developed on Linux, kernel 4.14. We are using Nvidia CUDA technology to leverage the power of the GPU to obtain massive speed-up.

4.1 Accuracy See Table 2.

22

P. Mandal et al.

4.2 Precision See Table 3.

4.3 Recall See Table 4. Table 2 Classification accuracy Classifier

Accuracy %

Random forest

65.16

Non-linear SVM

72.12

K-nearest neighbours

76.43

ANN

85.45

Table 3 Classification precision Classifier

Genre Metal

Rock

Pop

Rap

Electronic

Jazz

Random forest

0.630668

0.599315

0.523454

0.700083

0.669811

0.753045

Non-linear SVM

0.728877

0.675506

0.627789

0.758859

0.713789

0.795007

KNN

0.736842

0.758769

0.688247

0.728788

0.811671

0.859688

ANN

0.861086

0.833908

0.780488

0.900093

0.838675

0.901892

Metal

Rock

Pop

Rap

Electronic

Jazz

Random forest

0.676991

0.612316

0.465403

0.8125

0.504057

0.796137

Non-linear SVM

0.69469

0.677397

0.58673

0.844231

0.624746

0.865522

KNN

0.79292

0.741777

0.654976

0.925

0.62069

0.828326

ANN

0.828319

0.846746

0.788626

0.926923

0.796146

0.920601

Table 4 Classification recall Classifier

Genre

Automatic Music Genre Detection Using Artificial Neural Networks

23

Fig. 3 a Confusion matrix of ANN, b Confusion matrix of ANN showing misclassifications only

4.4 Confusion Matrix See Fig. 3.

5 Conclusion and Future Work In this paper, we have found that ANNs are quite effective in classifying music genres with an accuracy of 85.45% which is comparable to previous works in this field and better than most. The use of a novel dataset has helped to capture the latest trends in music and new genres like EDM. As we can observe in Fig. 3 b, the misclassification rate is quite low. There is some confusion between Metal and Rock and also between EDM and Pop due to the presence of common elements (like beats, tempo and instruments). The best classification is obtained for Jazz and Rap music genres as they are quite unique in their style, instruments and beats. In the future, our work can be extended to train recurrent neural networks, convolutional neural networks, and other deep learning models for even better results. We estimate that extending the dataset further in the future to include more samples and more genres and subgenres will produce even better results.

References 1. Tzanetakis G, Essl G, Cook P (2001) Automatic musical genre classification of audio signals. Princeton Universiy, Computer Science Department, Computer Science and Music Department 2. Annesi P, Basili R, Gitto R, Moschitti A (2007) Audio feature engineering for automatic music genre classification. Department of Computer Science, Systems and Production, University of Roma, Tor Vergata, Italy

24

P. Mandal et al.

3. Haggblade M, Hong Y, Kao K (2011) Music genre classification. Stanford University, Department of Computer Science 4. Rathore A, Dorido M (2015) Music genre classification. Indian Institute of Technology, Kanpur, Department of Computer Science and Engineering 5. Kozakowski P, Michalak B. Music genre recognition, posted on Wed 26 October 2016, at http://deepsound.io/music_genre_recognition.html, https://github.com/deepsound-project/ genre-recognition 6. Rosner A, Kostek B (2017) Automatic music genre classification based on musical instrument track separation. Institute of Computer Science, Silesian University of Technology, Gliwice, Poland, Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology, Gdansk, Poland 7. Paradzinets A, Harb H, Chen L (2009) Multiexpert system for automatic music genre classification. Ecole Centrale de Lyon, Departement Math Info 8. Anusuya MA, Katti SK (2011) Front end analysis of speech recognition: a review. Int J Speech Technol, Springer 14:99–145 9. Cataltepe Z, Sonmez A, Adali E (2005) Music classification using Kolmogorov distance. Representation in Music/Musical Representation Congress, Istanbul, Turkey 10. Saha G, Yadhunandan US (2017) Modified mel-frequency cepstral coefficient. Department of Electronics and Electrical Communication Engineering India Institute of Technology, Kharagpur, West Bengal, India 11. Benesty J, Sondhi MM, Huang Y (2008) Handbook of speech processing. Springer 12. Milner B, Shao X (2011) Speech reconstruction from mel-frequency cepstral coefficients using a source-filter model. University of East Anglia, UK 13. Washani N, Sharma S (2015) Speech recognition system: a review. Int J Comput Appl 115(18): (0975-8887) 14. Panagakis Y, Kotropoulos C (2013) Music classification by low-rank semantic mappings. EURASIP J Audio Speech Music Process 15. Saini P, Kaur P (2013) Automatic speech recognition: a review. Int J Eng Trends Technol. 4(2)

Role of Ad Hoc and Sensor Network for Effective Business Communication Ishu Varshney and Sunny Prakash

Abstract Ad hoc networking is a crucial aspect in fields of research as communicating technology. It is one of the appealing advanced technologies which is enabled in various applications, like rescue and tactical operations. Today, the enhancement and modernistic sequence in computing technology are positioned on the familiar insistence of ad hoc domain and for over two decades domain of ad hoc network is having distinct forms. This paper investigates two main essential attributes of an ad hoc technique for organizational transmission in the form of an effective communication network. First, the main attribute is mobile ad hoc network, which is advantageous to organizational employees for interacting with each other without making any changes in their network positions and this will help in reducing the time of communication from beginning to end which will be increased. Second main attribute wireless sensor network, which leads ‘ZigBee’ standard for organizational connection to oversee unremarkable activities of an employee. This will help employees to communicate inside or outside the organization effectively via ZigBee standard. WSNs are one of the witnessed noticeable revisions of facts technique which help in building the communication network to be more efficient and trustworthy. This paper highlights the role of networking in organizational communication and associates ad hoc research blueprint whichever adjoins the apperception, the conception, figuration, and relevance of networking encompasses by the effectiveness of an organization. Keywords MANET · WSN · ZigBee

I. Varshney (B) · S. Prakash G.L. Bajaj Institute of Technology and Management, Greater Noida, Uttar Pradesh, India e-mail: [email protected] S. Prakash e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_4

25

26

I. Varshney and S. Prakash

1 Introduction MANET: The modern clarification of MANET is solitarily the leading sense which equate our cosmos also than we consistently thinking attainable. Usually, abstraction of MANETs [1] is treated as credible range of view of such communication where the information and communication technologies are firmly linked with integral physical infrastructure; where the usage of networked embedded devices is accomplished through inventive systematic management review. In such transmit pragmatic measurement information is interrelated with refined dynamic system devices and restraint edification via ad hoc networking technology. Information resource is totally based on the prediction of people’s operative environments and benefited to end users through association of various communication networks. In this aspect, the new prototype of prevalent computing is formed due to advanced technologies. Further to advanced leading, equipment, architectures and code [2]. Moreover, computing and communicating power were attached to enduring home gadgets like digital cameras, refrigerator, etc., and it also enhances the core of computing surroundings. For acquisition of wireless nodes, a reciprocate network of MANET [1] is required which is positively or constantly formed without any requirement of any fixed infrastructure network which is benefitted to fruitful organizational advancement. With special appearance of MANET will result in advancement of mobile computing. During the last decade one of the most influential technology is Mobile Networking [3] which upholds prevalent computing and lying overture in both techniques like hardware and software. Commonly, for enabling wireless units for interaction with each other, then we are required to discuss only two main specific approaches: (1) Infrastructure: Consistently, cellular positioned on WSNs for good view and relied groundwork backing [1] and here admittance points are linked with fixed network groundwork in mobile device. Intermittently WLAN, UMTS and GSM are treated as a case WSNs. (2) Infrastructure-less: For maintaining a well balance on cellular system ‘MANET’ is only the reason which states the wide area of research and applications of ad hoc computing network. Well-organized and freely wireless mobile nodes are possible only through enact or integrate with MANETs. With the help of pre-existing communication infrastructure [1] in case of arbitrary and temporary ad hoc topologies, the people and devices are freely permissible to Internet work within range. This brings flourishing revolution in the advancement of MANET (Fig. 1). Aspect of Internet is only the hidden program afterwards the progression in telecommunication networks over the last few years which results in an adequate swift standard spreading of aggregate packet data traffic. All at once, distributed co-ordination is treated as pervasive development in wireless field which signifies the rising advancement of mobility and cutting edge for ad hoc networks. WSN: Considerable figure of sensor nodes help in organized wireless sensor network (WSN) where the sensor is attached to find out physical circumstances with

Role of Ad Hoc and Sensor Network …

27

Fig. 1 Communication via MANETs [3]

every exclusive node such as light, heat, pressure, etc. In application of wellness program, tillage, environment checks, etc., are possible only through Wireless Sensor Networks. Besides, WSNs are highly signalized by high diversity with different proprietary and non-proprietary results. New deployment and integration of extant sensor networks is an ample dimension of technologies. A network of nodes can be simultaneously characterized by ‘WSN’ through which the senses of environment can be monitored [4]. For joint consultation of distributed data processing emerged only with the help of a cross-layer design path [5]. Frequent outlines help in assisting the generation of WSNs which is due to low cost of sensor technology. For attaining virtual wide sensor networks WSN is treated as an extensive area of solutions [6]. For specific awakens regarding applications, the need for handling gateways is emerged and sensor networks are positioned on proprietary systems. Internet is only the medium through which WSN data is associated to other devices and there are no chances for accessible communication among convoluted app-specific and distinct criterion which results in restructuring are actualized in gateways or proxies (Fig. 2). This paper mainly archives the conceptual of ad hoc networking community for providing a better self-organized ad hoc network, wireless sensor network, ad hoc routing, ZigBee standards which prompts to make organizational communication network strong and efficient.

28

I. Varshney and S. Prakash

Fig. 2 Interworking among heterogeneous WSNs [6]

2 Analogues Upbringing Next-generation wireless functionalities are distributed through the effectiveness of ad hoc networking community and the upbringing backgrounds are as highlighted below: A. Extension of MANETs: • The University of Hawaii fabricate ALOHA NET is introduced in 1970. • DARPA was encouraged in 1972. • In 1980, DARPA introduced SURAN (Survivable Radio Networks) which helps in pointing the problem related to PRNET [5]. • The existence of IEFT was introduced in 1994. • DARPA Global Mobile Information System program commenced by Department of Defense (DoD) in 1994 [6]. • Ericsson introduced the ground existence of Bluetooth in 1995. B. ZIGBEE Standard: The design and framework of ZigBee wireless were elaborated by the ZigBee Alliance due to its inexpensive constituent data rate and contrary convention [7]. The four main layers of ZigBee Protocol is consisting of the application layer, media access control layer, network layer and physical layer.

Role of Ad Hoc and Sensor Network …

29

3 Efficient Communication Will Assist an Organization Literally, communication enacts an essential role in conceptualization, design and development of product (NDP), CRM, Staff management and in practical facet of business operations [8]. Employees are treated as key audience because they are in direct touch with other audiences. So, that is why communication channel may be effective. It helps in establishing a clear hope between employees and customers via efficient communication through ad hoc networking technique and it also conveys how the actual performance of employees will directly impact the company’s profitability through customer response or feedback. For building strong relationships effective communication is essential and it also helps in giving birth to trust and loyalty while individual needs meetings, fetching important business facts and figures and providing feedback. It also assists in building strong tie-ups with external audiences which highlights company culture and values through effective communication and it can be possible only through ad hoc networking. Ad hoc networking also helps in leading new ideas and innovations through several open channels of communication. For building strong feeling of teamwork among employees may be attained through organizational goal. Effective organizational communication at all levels of the organization is essentially establish through ad hoc network technique.

4 Values of Communication in the Organizational Structure Within an organization, command over communication is a tricky process unusually one with complicated levels and multiple issues. Overall productivity and workflow can improve only when all the segments of organizational communication can smoothly flow. Interconnection: For building strong connections between staff individuals and levels of workers in aspects of expert and social level, correspondence is necessary and it also helps in designing an open sound communication atmosphere which makes it safer for employees to express their ideas. For employees not to feeling isolate and feeling of teamwork is possible only through sound and dynamic communication system. When relationships are sound, it results in building trust among one another employees and work together more effectively. Transparency: In an organization if there is no clarity in communication channel then it results in ambiguity which creates adverse feelings and strained atmosphere. This helps in reducing misunderstandings via sound communication. Alliances: Effective and efficient communication helps employees to collaborate easily which results in a more fruitful team overall [9]. When several departments are

30

I. Varshney and S. Prakash

engaged on different in distinct appearance facets of identical project, communication is one of the effective techniques to obtain end results. In this situation, ad hoc network plays an essential role for legatee to the atmosphere of organizational structure.

5 Defended Transmissions in MANET Captions Persuasion Management: On the aspect of network, the worthiness of WSN is only the assessment system for the distributed networks for MANETs and also helps in analyzing the sensor network. For decentralization and initialization in WSN environment, the trust management system is mandated to design and prepare for reacting against issues. The node trust value is obtained based on cryptographic mechanism. If the node’s location is advised unsure and bypass in the routing process, then it means the computed trust value of node is less than threshold. Signature Schemes: As compared to other networks, security MANET is becoming a more implicate issue in transmitting information, which is why authentication is essential. For security purpose a digital signature is required that helps in detecting threats and several types of intrusion detection in WSN. For identifying attacked node several type signature schemes are used and also for the given secured data transmission through channels. Threshold Signature Scheme: For the attainment of confidential and authentication of packets in MANETs, it is essential to develop the trust-based security protocol based on cross-layer theme for detecting isolation of several malicious nodes which uses routing information. In this phase a trust-based packet is designed. Every packet is using trust value by maintaining a trust counter. The respected node is treated as malicious when trust value falls below the threshold. A multi-threshold signature scheme is used for designing a sound secure and effective communication channel.

6 Secure Transmissions in WSN Key Distribution Schemes: In comparison among operational requirement and WSN security, three main resultant models may be introduce. First one is the network keying model which is so simple to handle and it requires a lesser amount of resources and deriving advantageous over pairwise and group keying models. It helps in allowing the alliances among several nodes. Certificate Generation Method: Under certificate generation mode WSN subsists with several sensor nodes which is flourishing over medium-scaled network. Under this method [10], WSN consists of thousands of densely populated sensor nodes which are spreading over a medium-scaled network which results in identifying the area which segments several small clusters in network area.

Role of Ad Hoc and Sensor Network …

31

7 Conclusion This paper was investigating two main essential terminologies of ad hoc technique, i.e., MANET and WSN for organizational transmission in the form of an effective communication network. At the end stage of this research, we found out several new alternatives regarding designing and forms may be obtained in mobile communication. This research paper helps to designate a few definitive advance widths of experimentation proposition for ad hoc technology which is sure to simplify the upgrading and trigger of publicizing dispatch layer of the ad hoc application of precise observation.

References 1. Chitkara M, Ahmad MW (2014) Review on MANET: characteristics, challenges, imperatives and routing protocols. Int J Comput Sci Mob Comput 3:432–437 2. Varshney I, Ali S (2018) Impact of Ad-hoc networking on organizational communication. BIZCRAFT J Fac Manag Sci 12(1). ISSN-2231-0231 3. Verma S, Singh P (2014) Energy efficient routing in manet: a survey. Int J Eng Comput Sci 3:3971–3977 4. Sekhar Chandra P (2013) A survey on MANET securing challenges and routing protocols. 4(2):248–256. ISSN 2229-6093 5. Aftab MU, Nisar A, Asif D, Ashraf A, Gill B (2013) RBAC architecture design issues in institutions collaborative environment. Int J Comput Sci Issues 10:216–221 6. Bang Ankur O, Ramteke Prabhakar L (2013) MANET: history challenges and applications. Int J Appl Innov Eng Manag 2(9):249–251 7. Aarti SS (2013) Tyagi, study of MANET: characteristics, challenges, application and security attacks. Int J Adv Res Comput Sci Softw Eng 8. Odeh A, Fattah A, Alshowkan E (2012) Performance evaluation of AODV and DSR routing protocols in MANET networks 9. Varshney I, Ali S (2017) Study on MANET: concepts, features and applications. ELK’s Int J Comput Sci 2(3). ISSN 2394-0441 10. Varshney I, Ali S, Singh B (2018) Approach for detection of selfish nodes in MANET. J Eng Technol 7(4.39):206–209

Implementation of Integrated Security System by Using Biometric Function in ATM Machine Pushpa Choudhary, Ashish Tripathi, Arun Kumar Singh and Prem Chand Vashist

Abstract The paper proposes an efficient approach for authentication/identification of a valid person at ATM machine using some body parameters along with normal procedures which are currently used. Body parameters are taken through hidden camera installed in ATM and the extracted data is processed based on matching percentage with the stored data to identify/authenticate the valid person. If a person is found suspicious then appropriate action is triggered. For experimental work, two body parameters, i.e., fingerprint and face recognition have been taken to do the analysis for better result. Two databases, i.e., Olivetti Research Laboratory (ORL) face database and Extended Cohn-Kanade database (CK+) have been used in face recognition. For fingerprint analysis, stored data and test data have been compared to get the valid result. All experiments have been done in MATLAB and the result obtained by the proposed work is quite impressive. Keywords ATM · Fingerprint · Cornea and iris · Voice and face recognition

1 Introduction In today’s modern era, technology is providing us every facility at our doorstep but at the same time security threats are also rising exponentially and need to be dealt with following proper safety measures. Every country is preparing biometric database of its residents and all activities are being linked with a person’s biometric data. In P. Choudhary · A. Tripathi (B) · A. K. Singh · P. C. Vashist Department of IT, G. L. Bajaj Institute of Technology & Management, Greater Noida, India e-mail: [email protected] P. Choudhary e-mail: [email protected] A. K. Singh e-mail: [email protected] P. C. Vashist e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_5

33

34

P. Choudhary et al.

future, it is expected that there may be a provision to carry only one smartcard in place of carrying multiple cards to do all activities such as money transaction, bill payment, etc., and in different roles such as passport, driving licence, aadhar card, pan card, etc. Biometric data stored in the card will help to identify a person for every activity and its usage. Biometrics involve identifying objects (human being) by their behavioural characteristics which is captured from human behavioural structure. Behavioural characteristics are derived from human body parts. These characteristics are unique and measurable and can be used to automatically recognize an individual’s identity. Three modes of operations generally used in biometric system are given below: • Enrolment: Enrolment is the process to generate a template for identifying a person by sensing and measuring certain type of characteristics. • Verification: In this process one-to-one mapping is done for verification, when a person provides some data for verification then that person is verified by the stored data by matching the claimed identity. • Identification: In this process one-to-many identification is done from text data to stored data. In Figs. 1, 2 and 3 the process of enrolment, verification and identification of biometric system have been shown, respectively. In these figures, the image of user has been taken from Cohn-Kanade database. Rest of the paper is organized into three sections. Section 2 shows the proposed work that covers the body parameters used for feature extraction, authentication process used in ATM and working process of the proposed model. Section 3 depicts the performance analysis and results of experiment. The conclusion part is covered in Sect. 4.

Fig. 1 Enrolment process

Implementation of Integrated Security System …

35

Fig. 2 Feature verification (1:1)

Fig. 3 Feature identification (1:N)

2 Proposed Work In the proposed work, two body parameters have been used for feature extraction to validate the person. The detail of the same is as follows (Fig. 4): (a) Fingerprint: When a person feeds his secrete number through the keypad available on the ATM, his/her fingerprint is recorded in our database. Fingerprint can be an alternate method of authentication in case if a person loses his secrete number. Fingerprint authentication is rarely used but recording and recognition is done silently by a machine each time [1–8]. Figure 2 shows the ATM keypad. Here, there are two methods of feature extraction available which are used for classification: (a) Global features such as integral flow of ridges and (b) Local features which are minutiae such as ridge ending, delta and island [3]. Fingerprints of every individual are unique and the technology of fingerprint recognition is easy to use and it gives high recognition rates [4]. Fingerprint systems are constrained with following limitations:

36

P. Choudhary et al.

Fig. 4 ATM keypad

• Inability to enrol some users: Due to poor quality of fingerprints, identification of some persons is not possible, such as the elder persons and manual workers. For this reason one need to consider other biometrics solution. • Need to deploy specialized devices: Specialized devices are needed for capturing fingerprints. Thus, it requires fingerprint scanners. However, this system may not perform well if finger tissues are damaged due to age factor or manual work or other reason. Further, user interaction is always required in fingerprint systems. (b) Face Recognition: It is non-intrusive method, because in this method images are used to identify the person which can be captured without the awareness of person about the camera. No physical interaction is required and it is very hygienic. Facial characteristics can be videoed or captured easily without requiring any contact at all, even from a distance [9–12]. This is the only biometric that allows passive identification in a one-to-many environment. In the proposed work, an ATM machine has been used as a medium to create, update and compare the database of suspicious persons. The proposed system uses visual channel, audio channel and sensor based channel. By using these channels the system is operated on biometric functions. Camera, microphone and finger print reader have been used in ATM for data capturing and analysis to identify/authenticate a person. Feedback received from this exercise will help to improve the security level of a bank to stop fraudulent activity.

Implementation of Integrated Security System …

37

Fig. 5 ATM used for capturing data

2.1 Authentication Process in ATM In current scenario a customer is being authenticated by a secrete pin code which has to be remembered by the customer at the time of using ATM. This secrete pin can be cracked by other person. Nowadays so many cases have been reported that cards are being cloned and secrete pins are also being cracked using different techniques. In our proposal we are not concentrating on cloning of smart cards but we are trying to increase the level of authentication, assist the security agencies to nab the suspicious persons/criminals. The proposed system provides detailed database of a person which commit fraud/crime at any ATM and helps in investigation to reach the culprit. The sample image of ATM with data capturing devices is shown in Fig. 5.

2.2 Working Process of the Proposed System In this process ATM performs routine authentication of customer for transaction, at the same time ATM collects the data for fingerprint and face recognition. The data is continuously recorded and parallelly compared with existing database of black listed or suspicious persons who are absconding. If data is matched with existing database then machine slows its processing and triggers alarm to the concerned security agency. Collected data are body parameters, which is unique for each person. On the basis of the captured body parameters a suspicious person can be identified easily and triggered alarm generates information to the security concerns. If everything is ok then ATM collects all parameters for the new person and updates in database. If

38

P. Choudhary et al.

Fig. 6 Flow chart for data capturing

any wrong activity is reported at any ATM in future then on the basis of collected data the culprit can easily be identified and his/her data can be immediately incorporated in blacklist or list of suspicious persons. Below are the steps used in the proposed system (Fig. 6): • If it matches then triggers alarm to concern person. • If does not match, then record the data. • In future if any mishap is reported then data of suspected person can be easily found. • This data will be updated immediately in database of suspicious person.

3 Performance Evaluation and Result Analysis Figures 7 and 8 depicted the process of authentication/identification of fingerprint which is obtained from sensors used at keypad of ATM machine. This obtained fingerprint is processed and compared with existing database. For comparison, huge database of fingerprint has been collected. • • • •

Obtaining the test Fingerprint from ATM Machine. After that Test Fingerprint is preprocessed. Extraction is performed to collect the minutiae points from test fingerprints. Then Matching is done with test fingerprint and stored database.

Implementation of Integrated Security System …

39

Fig. 7 Comparison of sample fingerprint with existing fingerprint database

• Computation of test image and stored image score has been observed. • If matching score is greater than 90% then fingerprint is authenticated/identified else not verified. Comparison of fingerprint is performed in MATLAB with stored database and test data of fingerprint with the help of histogram, if both the data matches then the person can be identified otherwise data is stored to the database for further processing.

3.1 Face Recognition Result For the capturing of face image a small camera is installed in the ATM machine. Whenever a person is doing a transaction at ATM machine his face is captured by camera. And this captured image is compared with stored database. The proposed face recognition system is tested on the Olivetti Research Laboratory (ORL) face database and Extended Cohn-Kanade database (CK+). For the reduction of computational complexity and memory consumption, PGM format images of these databases have

40

P. Choudhary et al.

Fig. 8 Histogram of sample fingerprint and matched fingerprint

been resized from 112 × 92 to 56 × 46 bmp format images as shown in Fig. 9. Proposed system is trained for each person in the database. For each person there are ten images, in that five face image of the same person are used in training and the remaining five face image are used for testing. Recognition

Fig. 9 Three examples of ORL face database

Implementation of Integrated Security System …

41

Table 1 Recognition rate as compared to the number of training images Database

No. of training images 3

4

5

6

7

8

ORL

83.21

91.66

96.60

98.75

98.95

99.25

Cohn-Kanade (CK+)

90.47

94.44

95.43

95.66

97.22

100

rate of both the databases, i.e., ORL and CK+ are 96.5% and 95.33%, respectively. As numbers of training images increases, the recognition rate also increases, as shown in Table 1.

4 Conclusion In the current scenario security is a major issue in financial institutions specially in banking sector where rate of criminal activity is very high. Continuous observation and analysis of criminal activities is required to be monitored to find suspicious person at the spot. Humans are currently working on those work places where continuous observation and analysis is required but human errors cannot be denied. Thus, moving towards the computerized system is the necessity in the current scenario. In the proposed system, fingerprint and face recognition have been used to validate the person on ATM. Two databases such as ORL and CK+ have been used for the experimental work and recognition rate on both the databases are 96.60 and 95.43, respectively, which is quite impressive and near to 100%. The result is based on using five images for training and five images for testing out of ten images for the same person. In Table 1, ratio of number of training images and testing images vary as 3:7 (30:70), 4:6 (40:60), 5:5 (50:50), 6:4 (60:40), 7:3 (70:30) and 8:2 (80:20) out of ten images and respective recognition rate is also given. As we can see that the recognition rate increases as the number of training images increases and it reaches to 99.25 and 100 for ORL and CK+ databases, respectively, using eight training images out of ten images. But, recognition rate obtained on 50:50 ratio, i.e., half should be used for training and rest half should be used for testing among available images always considered as good. Thus, here 96.60 and 95.43 are considered as recognition rate for the proposed system.

References 1. Ravi J (2009) Fingerprint recognition using minutiae score calculation. Int J Eng Sci Technol 22:55–62 2. Cao K, Jain AK (2019) Automated latent fingerprint recognition. IEEE Trans Pattern Anal Mach Intell 41(4):788–800

42

P. Choudhary et al.

3. Lin C, Kumar A (2018) Contactless and partial 3D fingerprint recognition using multi-view deep representation. Pattern Recognit 83:314–327 4. Cao K, Jain AK (April 2019) Latent fingerprint recognition: role of texture template. In: 2018 IEEE 9th international conference on biometrics theory, applications and systems (BTAS). pp 1–9 5. Alsmirat MA, Al-Alem F, Al-Ayyoub M, Jararweh Y, Gupta B (2019) Impact of digital fingerprint image quality on the fingerprint recognition accuracy. Multimed Tools Appl 78(3):3649–3688 6. Reshma KV, Nair AT, Babu T, Shareef MP, Abraham N (2015) Identity of user thrashing and privacy protection of fingerprints. Procedia Comput Sci 46:652–659 7. Hoax: ATM security advice message: Enter PIN In Reverse to Call Police. http://www.hoaxslayer.com/reverse-pin-ATM.html 8. Toth B (2009) Liveness detection: iris. Encycl Biom 931–938 9. Tsai HH, Chang YC (2018) Facial expression recognition using a combination of multiple facial features and support vector machine. Soft Comput 22(13):4389–4405 10. Hsieh CC, Hsih MH, Jiang MK, Cheng YM Liang EH (2016) Effective semantic features for facial expressions recognition using SVM. Multimed Tools Appl 75(11):6663–6682 11. Zhen Q, Huang D, Wang Y, Chen L (2016) Muscular movement model-based automatic 3D/4D facial expression recognition. IEEE Trans Multimed 18(7):1438–1450 12. Farhan HR, Al-Muifraje MH Saeed TR (March 2017) A novel face recognition method based on one state of discrete Hidden Markov model. In: 2017 annual conference on new trends in information & communications technology applications (NTICT). pp 252–257. IEEE

DTSS and Clustering for Energy Conservation in Wireless Sensor Network Arpana Mishra, Shubham Shukla, Akhilesh Kumar Singh and Anika Gupta

Abstract In current scenario, wireless sensor networks (WSNs) have lives of many years. One very important application is environmental monitoring of wireless sensor networks (WSNs). The restriction to carrying energy within the charging backup of sensing point creates large problem to achieve a better network lifetime, which becomes a bottleneck in such applications of WSNs. The prime objective of the framework is to decrease the battery requirement and communication budget while confirming the data processing and data transmission. In this framework, data processing and gathering are achieved by using TDMA scheduling protocol and HEED clustering algorithm. For clustering, we use dual Wiener prediction scheme with optimized step size by reducing the mean-square derivation (MSD), in a way that the zonal supervisor can obtain a good approximation of the real data from the sensor nodes. A centralized principal component analysis (PCA) technique is utilized to perform the compression and recovery for the predicted data on the CHs and the sink, separately in order to save the communication cost and to eliminate the spatial redundancy of these used data about environment [1]. For all the nodes in every group, a deterministic weight (w) is extended depends on the instants period, it sense or achieve the message information from different neighboring source nodes. Each node of every group can start to work depending on the group of node’s weight. The weight of the node’s group is defined as a group of timing parts to every sensing point source in that group. That is why in a hierarchal flow diagram, there may be many node groups of individual importance. A group of largest importance (low timing part) can be assigned the preferable timing part. Here weighing importance is enhanced based on the message information collected to the nodding point from different nodding points and surroundings. All generated errors of same operation are finally examined conceptually, which come out to be controllable. Based on the conceptual analysis, designing can be done with number of algorithms for implementation. The real world data can simulate in cost effective manner to such as monitoring of environment based on clustered WSNs.

A. Mishra (B) · S. Shukla · A. K. Singh · A. Gupta G. L. Bajaj Institute of Technology and Management, Greater Noida, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_6

43

44

A. Mishra et al.

Keywords Time-division multiple access (TDMA) · Distributed TDMA slot scheduling (DTSS)

1 Introduction A developing sensor networking concept is arrowed toward a significant impact in today’s digital world. The major goal of WSN is to processing transmitting the smart information towards integrated and interconnected networks, making the internet even more ubiquitous. WSN is based on several enabling technologies including sensing aggregating, processing, transmitting, etc. With the help of wireless transreceiving and sensing point networking, sensing networks have many benefits in applications over other technological networks, according to the basis of standing quality, self-organization properties, and clustering for scalability [2–4]. On other hand, in same manner of regular observations, the many of messages transients at low frequency, which provides in a great messages uncertainty in time or space, subsequently high speed communications between sensing points will be a unwanted of stored energy. Conceptually, the more long life of sensing points will be proportional to the less communicated grouped messages. Using this approach, grouped messages decrement has become very important required remedy that is goaled to decrease the grouped messages sending and receiving [5–8].

Fig. 1 Basic structure of WSN

DTSS and Clustering for Energy Conservation in Wireless Sensor …

45

As shown in Fig. 1, the wireless small energy required sensing points are collectively making the wireless sensor network, and it is not a simple operation to regain the power scale of those sensing points for a life of long age. The sensing nodes are made-up with fixed radio communication and computing abilities. The given performance is an afford to decrease the battery required of nodding sensors, by objecting on the frequency spectrum, and there are four operational conditional positions at various timing moments. These sensing networks are constructed with small sensor nodes and are used for observing and sensing surrounding message information. Due to less size, charging is given with tiny battery. The time at which deploying these nodes in a “non-approachable area”, there will be no possibility of charging or recharging simply. The aim of every sensing point is to observe or compute the needed message information and transmit toward sink head point station. This head point or controlling may be in a non-approachable area (remote). Since WSN’s sensing node is tiny, its battery board should be as tiny and should be able to help to tasks without decreasing in the results. A transmitting receiving protocol can be used and it would be of less weighing importance. This protocol will not require high battery. To save battery power, we will transit condition of frequency spectrum in deactivated mode if no information is to be transmitted or received. The process of converting the communicating device to be in deactivated mode and transit it in working mode if single incident is abnormal is known as event-based strategy or on-demand strategy. Here, one different method of timing synchronization, i.e., on continuous instant’s slots, all the nodes will be deactivated mode or working phase [9]. This becomes a suitable synchronized method, whereas the overhead of making every sensing point in the proper synchronized way will not be easy. This may not be mandatory to make all the sensing points in working state on same time. These sensing networks may use a method of synchronized timing waveforms, correspondingly, at every moment, it can provide only few counting of sensing points in working condition. Thus, here the proposed task in this paper is basis of nonsynchronous timing pattern.

2 Protocol of Timing Schedule The protocol for scheduling is more suitable when tracks the activated sensing points at one instant of can time [2]. Thus, for a good synchronous timing waveforms instructions, the battery required should be less in designing aspect. In literature, many scheduling protocols are available for practicing. A narrowband modulation technology is used in scheduling protocol of WSNs [9]. Less transmission per time is sufficient for these sensing networks with no wire, in application of atmospheric observations. The sensing points will not be in working phase until their prescribed timing parts achieve TDMA (Time-Division Multiple Access) time

46

A. Mishra et al.

accessing rules. The minimum distance path will be selected with a designated link for working mode by TDMA [10]. A Medium Access Control (MAC) scheduling protocol in WSN should have the given properties: • • • • • • •

Capacity of throughput—good, Less messages per time transmission—moderate, Circuit simplicity—more, Modulation techniques—Narrowband, Time required in process—Low, Time spend in message transfer—low, and Low overhead—low.

A TDMA basis synchronized timing waveforms provide individual timings of every sensing point to use the channel for transmitting the observed message information or to send in go-ahead direction the processed information.

3 TDMA Scheduling Distributed TDMA slot scheduling (DTSS) instructions are used to design the scheduling protocol for WSNs. DTSS algorithm favors heterogeneous and homogeneous modes of transmission. The main goal of DTSS instruction is to stop the increment in the duration needed to run the instruction while negatively forcing the scheduling time to great mismatching in plots [11]. DTSS time accessing rules instruction is specified in the mean that the common instruction can be used to proper processing for timing parts of uncommon sensing points of communication, namely, broadcast, multicast, and unicast. In DTSS instructions, a sensing point is needed to know only the heading name of concerned collectors, at the point of every it’s two-hop nearby sensing point. Also, in DTSS instruction, the nearby sensing points can take uncommon timing parts regularly, if the compatible synchronized result is obtaining. Intrusion detection and weather monitoring are such applications where sensing points are most of the time no dynamic fixed. In given work, we also considered them to be static. We considered that, for any work in a practical point of view, every sensing point knows its information collecting sensing point. Initially, a task starts its execution, and the DTSS algorithm is executed to generate a TDMA schedule. After the task is covered, the TDMA schedule is removed. We considered that every fixed sensing point in the WSN has a specific identifier [2]. WSN’s every sensing point has some processing strategy along with a frequency spectrum to make able transmitting–receiving among them. Each sensing point uses the same frequency spectrum. Two types of communication ability are symmetric and two directional [3]. The phase of communication between any nearby sensing point is type of halfduplex; means, only single sensing point at one time can transmit or receive. Message

DTSS and Clustering for Energy Conservation in Wireless Sensor …

47

sent by a node is inherently received by all the nodes arranged in its transmission area defined as omnidirectional communication. Fixed size format is made by dividing the time line and each format is further subdivided into limited number of timing parts, called length of schedule. We already considered the synchronized nodes in time fashion with respect to timing part 0 and sensing point which are knowing of the timing part size and the length of schedule. Timing part 0 is defined by the sensing point which initiates the scheduling operation.

4 Wake-up/Sleep Mode Scheduling In nonsynchronous category of wake-up/sleep scheduling in regular time intervals, only specific sensing points have to be activated and transmitted or gathered. All the remaining sensing points will be in non-working mode [12]. While moving for a change from working (wake-up) to non-working (deactivated) phase, and then from non-working (deactivated) to working (sensing/wake-up) phase, the phase to be reviewed if the power it needed for the transit (phase change) will be less while analyzing with the power it needed in the working (sensing/wake-up) phase [13]. Energy consumed in switching should be low than energy saved. This protocol of scheduling is designed based on the radio of the node. Frequency spectrum in deactivated phase—rs. Frequency spectrum in sending phase—rt. Frequency spectrum in gathering phase—rr. Frequency spectrum in listening phase—rl. The energy required for the transit state can be represented by Esl (sleep to listen). Est (sleep to transmit). Esr (sleep to receive). For a time slot, say “t”, if the node is in sending state, it will be denoted by rt, t = 1. If it is not in sending state, it will be denoted by rt, t = 0. So at a particular instant say t, a node can be in any one of the four states. So we can state the condition like this rt, t + rr, t + rl, t + rs, t = 1. For the state switch from sleep state to listening state, the energy consumed can be calculated by t (rs, t + rl, t + 1) Esl. During every cycle, the radio goes to all the four states. When the radio is either sending or receiving or simply listening, it is said to be active. If it goes to sleep, it is said to be in sleep state. The energy wasted in simply switching between states E wasted in switching can be calculated. E wasted in switching = ts-a (p active + p sleep)/2. Designated representation. Radio in sleep mode—rs. Required energy while changing from deactivated phase to listening phase—Esl. Required energy while changing from deactivated phase to sending phase—Est.

48

A. Mishra et al.

Required energy while changing from deactivated phase to gathering phase—Esr. Sensing power needed in sensing phase—p deactivated power needed in deactivated phase rs. Required time for moving to sensing phase (rt, rr, rl) from deactivated phase—ts. Required time for moving to deactivated phase from sensing phase (rt, rr, rl)—ts. A moment when frequency spectrum going to do task according to time for transmitting and collecting sensing information—t. A moment when frequency spectrum finalizes to move to deactivated phase according to timing—ts [14]. Clustering Data aggregation is required when the counting of sensing points required in the practical perspectives is big [15]. When every node tries to transmit the observed/sensed message information to the sink head station point (BS), large battery is needed, which means large sensing points can waste their battery in short time. That is why data should be aggregated by gathering from a set of nodes and transmitted to the sink BS from that sensing point. For this, wireless sensor nodes are used in the form of tree in our task. Clustering is used to make the groups of nearby nodes in the form of different clusters [16] and it is a dominant process to improve in delaying the first node death; on the other hand, aggregation plays an important role in delaying the last node battery empty. In HEED (Hybrid Energy-Efficient Distributed Clustering), energy wasted in switching should be less than energy saved.

5 Conclusions In the process of scheduling, a sensor node is only initiated once to receive whole message information from its neighbors, that is why it can minimize the energy expenditure and time overhead in the mode/state exchange. Mainly, if the configuration is a tree, the nodes can only initiate twice in one scheduling period [17, 18]. We also propose distributed and centralized algorithms that use time axis partitions at most a fixed factor of the minimal. The practical conclusions corroborate the conceptual analysis and represent the efficiency of our algorithms in the form of the number of state changes, time slots assigned numbers, and delay of time. When comparing multi-hop and one-hop DTSS, it is very challenging multi-hop than single-hop instruction process since temporal reutilization of timing parts may not be impossible. In no conflicting parts of the network in centralized form, tradeoffs are existing as primary and nonprime. A primary confusion is generated when a sensing point sends and collects at the common timing instant or collects many communicating objecting points to it on the common timing instant. A no primary confusion is generated when a sensing point, a specific collector of a predicated communication, is also within the transmitting and receiving area of another communicating specific for different sensing points. In the principle of TDMA, generated issue is to decide the shortest path distance non-conflicted designation of timing parts

DTSS and Clustering for Energy Conservation in Wireless Sensor …

49

in which every communicated way or sensing point is working at least one time [4]. Initial process of instruction rules triggering every communicating way or sensing point at least once along with a TDMA format depends on the considerations that are different non-dependent end-to-end hierarchal transmission in the network. In no wired sensing networks message information are transmits from the sensing point to a non-head message envelope gathering points. In counting controlling [7], for example, the sensing points sense the roadway tracks of moving things at many freeway standing points or at a common of them and send the messages to the controlling point on the side of the freeway or common standing points. The envelopes are further sent to the controlling point over the location path hierarchy in many hops. The non-easy work, therefore, is to select the very short distance objection-free designation of timing instants at which the message envelopes created at every sensing point reach the processing point along with the hierarchal pathway [8].

6 Future Work The problem of scheduling process in many-hop networks is a use of a TDMA multiple access controlling instruction. It is difficult to get the shortest distance pathway non-conflicted designation of timing parts in which each connecting site or sensing point is processing at one in time period. This depends on the considerations that there are a lot of independent point-to-point processes in the network. In wireless sensor networks, where data are often transferred from the sensor nodes to a limited head data gathering point, the problem is to determine envelopes oriented at every sensing point their collecting point.

References 1. Ye W, Heidemann J, Estrin D (2002) An energy-efficient MAC protocol for wireless sensor networks. In: Proceedings of IEEE INFOCOM 2. Arikan E (1984) Some complexity results about packet radio networks. IEEE Trans Inf Theory 30(4):681–685 3. Ephremedis A, Truong T (1990) Scheduling broadcasts in multihop radio networks. IEEE Trans Commun 38(4):456–460 4. Ramanathan S, Lloyd EL (1993) Scheduling algorithms for multihop radio networks. IEEE/ACM Trans Netw 1(2):166–177 5. Tulone D, Madden S (2006) Paq: time series forecasting for approximate query answering in sensor networks. In: Proceedings of the 3rd European conference on wireless sensor networks (EWSN) pp 21–37 6. LeBorgne Y-A, Santini S, Bontempi G (2007) Adaptive model selection for time series prediction in wireless sensor networks. Signal Process 87(12):3010–3020 7. Tulone D, Madden S (2006) An energy-efficient querying framework in sensor networks for detecting node similarities. In: Proceedings of the 9th ACM international symposium on modeling analysis and simulation of wireless and mobile systems pp 191–300

50

A. Mishra et al.

8. Zhou H, Wu Y, Hu Y, Xie G (2010) A novel stable selection and reliable transmission protocol for clustered heterogeneous wireless sensor networks. Comput Commun 33(15):1843–1849 9. Dam T, Langendoen K (2003) An adaptive energy-efficient MAC protocol for wireless sensor networks. In: Proceedings of the first ACM conference on embedded networked sensor systems (SenSys) 10. Sun Y, Du S, Gurewitz O, Johnson DB (2008) DW-MAC: a low latency, energy efficient demand-wakeup MAC protocol for wireless sensor networks. In: Proceedings of ACM MobiHoc 11. Keshavarzian A, Lee H, Venkatraman L (2006) Wakeup scheduling in wireless sensor networks. In: Proceedings of ACM MobiHoc 12. Gandham S, Dawande M, Prakash R (2005) Link scheduling in sensor networks: distributed edge coloring revisited. In: Proceedings of IEEE INFOCOM 13. Jiang H, Jin S, Wang C (2011) Prediction or not? an energy-efficient frame work for clustering based data collection in wireless sensor networks. IEEE Trans Parallel Distrib Syst 22(6):1064– 1071 14. Li H, Lin K, Li K (2011) Energy-efficient and high-accuracy secure data aggregation in wireless sensor networks. Comput Commun 34(4):591–597 15. He Z, Lee BS, Wang XS (2008) Aggregation in sensor networks with a user-provided quality of service goal. Inf Sci 178(9):2128–2149 16. Quer Y, Masiero R, Rossi M, Zorzi M (2012) Sensing, compression and recovery for wireless sensor networks: monitoring frame work design. IEEE Trans Wirel Commun 11:3447–3461 17. Wang W, Wang Y, Li XY, Song WZ, Frieder O (2006) Efficient interference-aware TDMA link scheduling for static wireless networks. In: Proceedings of ACM MobiCom 18. Caione C, Brunelli D, Benini L (2012) Distributed compressive sampling for life time optimization in dense wireless sensor networks. IEEE Trans Ind Inf 8(1):30–40

Load Distribution Challenges with Virtual Computing Neha Tyagi, Ajay Rana and Vineet Kansal

Abstract Computing with cloud is the today’s era computational epitome. It is vastly usable figuring technology, i.e., promptly uniting the aforementioned as the upcoming of spread and popular computing. Virtual Figuring renders elastic, as-aservice resource abstraction relies on usage-based payment model, and is rising as an appealing computing epitome. Cloud computing uses the principle of virtualization and is becoming potent supporter of current Internet businesses. Cloud computing accomplishes the needs of immense data storage and majorly parallel computing. Keywords Cloud Computing · Task scheduling · System management

1 Introduction There emerged the necessity for subsidence security and outturn challenges for superior Computing; economical resolution for this was accurately known by SARA in Intel® LAN ten Gigabit Converged Network Adapters ten. There have been significant rise in DNA Sequencing through the usage of High Output Sequencing. Heavy Concurrent Sequencing rendering high throughput at lower costs is made possible through HTS. With regular upgradation in instructional and analysis field we have currently achieved the likelihood to sequence an entire human ordination with the assistance of one instrument in some day. Illumina® MiSeq1, Ion PGM™ (Personal Genome Machine) [1] and PacBio RS II [2] are minor and charge actual HTS platforms. Difference in these platforms lies in their N. Tyagi (B) G.L. Bajaj Institute of Technology & Management, Greater Noida, India e-mail: [email protected] N. Tyagi · A. Rana Amity University Noida, Noida, India e-mail: [email protected] V. Kansal IET, Lucknow, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_7

51

52

N. Tyagi et al.

protocols, technology used, throughput, and read length. Nowadays there exist those instruments which can have capability of rendering >90–100 Gbytes of reads in 1 day. Besides, the produced data is not free from noise. There is a significant computational challenge of combining heterogeneous and massive datasets in HTS data which can be recognized as Big Data. With the increasing need for analysis of HTS data and its processing, there emerged the requirement of cloud and grid computing with high performance. For solving the computational challenges of HTS, cloud and grid-based computing systems were studied thoroughly and then their application/role in real world were discussed like its role in personalized medicine.

2 Proposed Methodology The common and highly significant purpose is to have strong implementation of high throughput computing in medical field or any field. Cloud computing emerged as a groundbreaking computing capability which provides high reliability and availability. There always exists the needs to strengthen the technology in the way so that our present needs and future needs are handled in the most optimized manner. Hence, continual inventions and discoveries are crucial. As the cloud model for introduced users is losing its control over physical security, risk to data hacking is increasing day by day as all the data are available on cloud. Cloud computing is a boon for all IT Companies and other organizations. This computing depends on shared figuring resources instead of having native or servers or personal devices to switch submissions. Virtual computing has advanced in recent years and has covered the whole United States of America as a web. Similarly, cloud garage technology has extra demand. This virtual storage acts as a single storage although it includes many resources. It works on the idea of QoS confident. It is able to endure greater fault with the aid of redundancy. The IT organizations that have growing data generated by using these sectors are also developing, we cannot update our hardware often so we will adopt for cloud storage which is higher desire. Programs and sources hosted by others may be accessed via personal cloud garage (Figs. 1, 2, and 3).

3 Security and Throughput Challenges and Their Fixation in High Performance Computing Challenge – Provide a HPI between virtual machines to support Message-passing interface workloads in a cloud infrastructure, while meeting the rigorous safety ideals needed to protect worthful and delicate data.

Load Distribution Challenges with Virtual Computing

Fig. 1 Proposed process

Fig. 2 Proposed functionality

Fig. 3 Proposed encryption

53

54

N. Tyagi et al.

– Subdue security restrictions of basic InfiniBand within the cloud that impose vigorous quantity of TCP/IP over-head leading to reduction of output to unacceptable levels. Data Security Challenges in the Cloud and their solutions. Carrier providers can get entry to the facts from the cloud at any time so cloud computing has large privateness concerns. It is feasible that records can be circumstantially or intentionally deleted from the cloud which may be a massive trouble. Without a warrant, provider provides percentage statistics with the 1/3 birthday party which asks for user permission, which users should agree for the usage of cloud offerings. Customers can decide how information can be stored as in keeping with privacy coverage and law and can even save statistics in encrypted shape of records. Encryption can be a traumatic answer. Statistics encryption is basically a mathematical system of changing a field text to cypher text and the cypher text cannot be examined by means of everybody else than the patron or drug consumer keeping the encoding key. Tokenization can also be the solution as actual records reside domestically in an item statistics base which randomly generates keepsake which can be associated with statistics and are sent to the cloud, and the information can simply be examined through the custodian of the token database. Tokenization protects most effective against outside threat in view that all and sundry who can get admission to token database may want to easily get right of entry to open textual content records while encryption protects in opposition to the external and internal threats considering there’s clean segregation of duties between wherein the keys are controlled and wherein the encrypted data is saved. Tokenization calls for better capacity servers and databases to manage the essential token database that grows with increasing statistics mines whereas encryption calls for lighter weightiness stateless servers with no data memory.

4 Results and Discussions High Performance Computing Solutions A. Commodity Computing In Bioinformatics, artifact clusters serves expeditiously as they provide components at inexpensive and at larger scale to stay up with the user needs. This area unit like supercomputers with several processors as they represent of normal servers are connected to create a network. An everyday pc or servers has the chance of getting multiple cores for process. Apache Hadoop being the well-known open supply model uses the framework of parallel machine programming through Map scale back. Operating outline of Map scale back is: A Map scale back employee receives information from the main, methods it and sends back the generated output to the main. Map scale back is often used as an extract, load, and remodel tool that reads a collection of supply information, carries out data formatting or the transcription of steps, and delegates the results to the destination supply.

Load Distribution Challenges with Virtual Computing

55

B. GPU Computing Fashionable Processing Unit’s guide multiprocessing in order that they are extremely low-budget and rapid than consecutive imperative processing units. Normally cheap servers or workstations have 10’s of pc hardware cores whereas GPU card can have a thousand’s of cores. Vice exchange has brought on the worth of alternate items GPU’s to unendingly lower. Multicore Server or virtual laptop wishes Brobdingnagan vicinity and electricity than a GPU card. NVIDIA, a GPU card producer presents for the platform and model for programming that is causes unified device layout (CUDA) [3]. Endless CUDA-compatible gear are evolved inside the beyond for NGS processing and evaluation corresponding to Cushaw [4], spiny-finned fish [5], SOAP3 [6], CUDASW++ [7], SeqNFind [8], etc. The properly-arranged cluster of gear which might be used for sequence evaluation is SeqNFind. The analysis and visible image of carried out math facts comparable to (gene expression ranges) is feasible through ASCII text document R-surroundings. C. Cloud Computing Cloud computing renders a simplest way to access servers, storage, databases, and a broad set of application services over the web. It creates virtual process surroundings for the users so as to handle their process needs and storage. However, to set up and manage the central processing unit, memory and therefore the pipeline for information analysis need additional efforts and time of the user compared to business solutions. Collecting and maintenance of the system demands vital technical skills. Cloud can be categorized as Private cloud for the single organization, Public Cloud for the user over network in public use, hybrid cloud which can be mixture of two or more clouds. The biggest examples of participants in cloud provisions are Microsoft Azure, Google Genomics, Amazon Elastic Compute Cloud, etc.

5 Conclusion and Future Work The technology SARA is dependent on Intel local area network ten Gigabit Converged Network Adapters that integrate I/O virtualization with Intel. Virtual Computing for Connectivity (Intel VT-c) that aids in lowering mainframe usage and better-quality networking and I/O and the Information security for the virtualization is provided by PCI-SIG Single Root input–output Virtualization, a key element of Intel virtual technology that assist in reassuring the knowledge traffic at intervals where one VM cannot be accessed by alternative VMs, and alternative platform options that guarantee a reliable launch atmosphere. The distribution of load is one of the essential tasks in virtual computing environment to achieve the most optimum and maximized use of resources. In this broadside, we had simulated an algorithm which will make our load balancing more efficient and give brief insight of the cloud computing and the existing load balancing algorithms, and had an overview of the issues in them and their advantages and disadvantages.

56

N. Tyagi et al.

A detailed study of the existing algorithms was done and MapReduce was taken as the algorithm to work upon. The benefits of inculcating the functionalities of Big Data in cloud computing were studied and the most suitable virtual environment for implementing the project at a smaller level was chosen. Converging the functionalities of Big Data Analytics for proper operation of the cloud computing paradigm for dealing with the increasing users and resource requirements, an efficient algorithm was designed which could enhance the overall efficiency and elasticity of the Cloud.

References 1. https://www.thermofisher.com/unitedkingdom/en/domestic/life-technologicalknow-how/ sequencing/next-technology-sequencing/ion-torrent-subsequent-technology-sequencingworkflow/ion-torrent-subsequent-generation-sequencing-run-sequence/ion-pgm-device-forsubsequent-generation-sequencing.html 2. http://www.p.c.com/products-and-services/pacbio-structures/rsii/ 3. CUDA GPUs | NVIDIA Developer. https://developer.nvidia.com/cuda-gpus. Accessed 12 Sep 2016 4. Liu Y, Schmidt B, Maskell DL (2012) Cushaw: a cuda like-minded quick study aligner to massive genomes based at the Burrows-Wheeler remodel. Bioinformatics 28(14):1830–1837 5. Klus P, Lam S, Lyberg D, Cheung M et al (2012) BarraCUDA—a quick short read sequence aligner the use of photos processing units. BMC Res. Notes 5(1):27 6. Liu CM, Wong T, Wu E, Luo R, Yiu SM et al (2012) SOAP3: ultra-fast GPU-based parallel alignment tool for quick reads. Bioinformatics 28(6):878–879 7. Liu Y, Wirawan A, Schmidt B (2013) CUDASW++ 3.0: accelerating Smith-Waterman protein database search by using coupling CPU and GPU SIMD commands. BMC Bioinform 14(1):117 8. Carr DA, Paszko C, Kolva D (2011) SeqNFind®: a GPU multiplied series evaluation toolset allows Bioinformatics. Nat Methods Appl Notes 1–4

Mobile Ad Hoc Network and Wireless Sensor Network: A Study of Recent Research Trends in Worldwide Aspects Ishu Varshney

Abstract MANET is equipping every device that maintains the data required to route traffic. Wireless Sensor Network (WSN) is highly signalized by high diversity with different proprietary and nonproprietary results. New deployment and integration of extant sensor networks is an ample dimension of technologies. A network of nodes can be simultaneously characterized by “WSN” through which senses of the environment can be monitored. Basically, this paper analyzes the recent research trends in MANETs and WSNs in the categories of real estate, law and government, finance, electronics, computer field, etc., through web, image, news, and YouTube search. Keywords MANET · WSN · Wireless mobile nodes · Infrastructure network · Infrastructure-less network

1 Introduction MANET: It is specified as a network that has many autonomous nodes that often is composed of mobile devices. WSN is one of the witnessed noticeable revisions of facts and technique which helps in building the communication network more efficient and trustworthy. Today, the enhancement and modernistic sequence in computing technology are positioned on the familiar insistence of the ad hoc domain and for over two decades the domain of ad hoc network is having distinct forms [1]. For instance communication in an organization required interim network and it can be possible through only “MANET” which is advantageous to organizational employees for interacting with each other without making any changes in their network positions and this will help in reducing the time of communication and from beginning to end it will increased [2]. In the second main attribute, we are focusing on wireless sensor network, which leads “ZigBee” standard for organizational connection to oversee unremarkable activities I. Varshney (B) GL Bajaj Institute of Technology & Management, Greater Noida, Uttar Pradesh, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_8

57

58

I. Varshney

of an employee. This will help employees to communicate inside or outside the organization effectively via ZigBee standard. WSN: Figure of sensor nodes help in organizing a wireless sensor network (WSN) where the sensor is attached to find out physical circumstances with every exclusive node such as light, heat, pressure, etc. In application of wellness program, tillage, environment checks, etc., is possible only through Wireless Sensor Networks. Besides, WSNs are highly signalized by high diversity with different proprietary and nonproprietary results. New deployment and integration of extant sensor networks are an ample dimension of technologies. A network of nodes can be simultaneously characterized by “WSN” through which the senses of environment can be monitored [3]. For joint consultation of distributed data processing emerged only with the help of a cross-layer design path [4]. Frequent outlines help in assisting the generation of WSNs which is due to low cost of sensor technology. For attaining virtual wide sensor networks, WSN is treated as an extensive area of solutions [5] (Figs. 1 and 2).

Fig. 1 Scenario of MANET

Mobile Ad Hoc Network and Wireless Sensor Network …

59

Fig. 2 Interworking among WSNs

2 Challenges of Wireless Sensor Network Health Screening Programs, Automatic centralized control of building, Energy Saving, Flexibility in manufacturing process, Monitoring and Billing, Aging energy infrastructure, Integrated real-time environment monitoring, Virtualization of information infrastructure, Fleet Management system, Sedentary, and Merchant Center.

3 Enactment of Ad Hoc Network 1. Military battleground: Forthwith all the belongings of the Military were designed with the assistance of computer hardware [6]. For sustained information, ad hoc community helps in building grid among army personnel, vehicles, etc. 2. Profit making sector: At the time of calamity relief effort Ad hoc network is one of the trust building communication practice with rescue teams & its operations where communication are impermissible in case of any calamities. In these critical circumstances ad hoc networking technique enables the communication network sound.

60

I. Varshney

3. Provincial level: For conference or classroom participants, ad hoc community can freely build prompt association. Ad hoc network can also be tested for home network where other devices like taxicab, sports stadium, etc., can be attached. 4. Personal Area Network (PAN): Intercommunication interim MANET naturally is fruitful among mobile devices like PDA, a laptop and a cellular phone. Wireless connection can easily recoup with the lined cable and with the help of MANET.

4 Research Trends The research trends of MANETs and WSNs compared and investigated for the last 10 to 15 years in worldwide aspects, searching in the categories of real estate, law and government, finance, electronics, computer field, etc., through web, image, news, and YouTube search are shown in Figs. 3, 4, 5, and 6, respectively. Region-wise comparison of MANET & WSN in the categories of real estate, law and government, finance, electronics, computer field, etc., through web, image, news, and YouTube search are shown in Figs. 7, 8, 9, and 10, respectively [7].

Fig. 3 Comparing MANET and WSN through web search in worldwide aspects

Fig. 4 Comparing MANET and WSN through image search in worldwide aspects

Fig. 5 Comparing MANET and WSN in news search explore

Mobile Ad Hoc Network and Wireless Sensor Network …

Fig. 6 Comparing MANET and WSN in YouTube search explore

Fig. 7 Region-wise comparing of MANET and WSN in web search explore

Fig. 8 Region-wise comparing of MANET and WSN in image search explore

Fig. 9 Region-wise comparing of MANET and WSN in news search explore

61

62

I. Varshney

Fig. 10 Region-wise comparing of MANET and WSN in YouTube search explore

5 Conclusion This paper investigates the recent research trends of MANETs and WSNs in worldwide aspects in various categories such as web, image, news, and YouTube search region-wise. Many routing protocols are used for MANET but technically are still having challenges in security purpose and for the insistently of system bandwidth efficiency. WSN is one of the witnessed noticeable revisions of facts technique which helps in building the communication network more efficient and trustworthy. The continuous progression in the field of networking is driving a new alternative way and sensor field more in use.

References 1. Sekhar PC (2013) A survey on MANET securing challenges and routing protocols. 4(2):248– 256. ISSN: 2229-6093 2. Bang AO, Ramteke PL (2013) MANET: history, challenges and applications. Int J Appl Innov Eng Manag 2(9):249–251 3. Varshney I, Ali S (2018) Impact of Ad-hoc networking on organizational communication. BIZCRAFT J Fac Manag Sci 12(1). ISSN: 2231-0231 4. Aarti S, Tyagi S (2013) Study of MANET: characteristics, challenges, application and security attacks. Int J Adv Res Comput Sci Softw Eng 5. Varshney I, Ali S (2017) Study on MANET: concepts, features and applications. ELK’s Int J Comput Sci 2(3). ISSN Print: 2454-3047, ISSN Online: 2394-0441 6. Medetov S, Bakhouya M, Gaber J, Wack M (2013) Evaluation of an energy-efficient broadcast protocol in mobile Ad-hoc networks. In: 20th international conference on telecommunications 7. Varshney I, Ali S, Singh B (2018) Approach for detection of selfish nodes in MANET. Int J Eng Technol 7(4.39):206–209

Comparative Analysis of Clustering Algorithm for Wireless Sensor Networks Smriti Sachan, Mudita Vats, Arpana Mishra and Shilpa Choudhary

Abstract In today’s era, Wireless Sensor Networks (WSN) are associated with numerous technologies and has a variety of applications in different field like health care, phones, military and disaster management, etc. Sensor nodes are sometimes deployed in a large scale which works independently in rough environments, because of constraint resources, usually the scarce battery power, these wireless nodes are sorted into clusters for energy economical communication. Nowadays, cluster hierarchical schemes have achieved nice interest in minimizing energy consumption. This paper explains about the hierarchal cluster-based approaches. In cluster-based approaches, nodes are sorted into clusters, wherever an inventive detector node is nominative as a cluster head (CH). This paper highlights and discusses the planning challenges for cluster-based schemes, existing cluster-based techniques by evaluating it in account with certain parameters. Moreover, a close outline of those protocols is conferred with their benefits and disadvantages. Keywords WSN clustering · Hierarchal clustering · LEACH · CCWM

1 Introduction Wireless nodes that are combined together and spread over a particular area in various environments made a wireless sensor network. These nodes transfer the information to nearby nodes and to the base station by processing and sensing them. This small S. Sachan (B) · M. Vats · A. Mishra · S. Choudhary G. L. Bajaj Institute of Technology & Management, G.B. Nagar, Greater Noida, Uttar Pradesh, India e-mail: [email protected] M. Vats e-mail: [email protected] A. Mishra e-mail: [email protected] S. Choudhary e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_9

63

64

S. Sachan et al.

equipment has limited memory, low processing, low analyzing capabilities, and most significantly little emu (usually equipped with batteries) [1–4]. However, problems like optimizing energy potency need additional analysis. Moreover, topology construction is additionally important to distributing nodes uniformly within the cluster to create the network economical. In hierarchal techniques, nodes are collected together to form the clusters and by using some standard basis, a cluster head is chosen to process the routing. Hierarchal protocol proceeds toward two-layer technique in which the first layer is for environment sensing and another is for routing process. Sensing is done by nodes that consist of lower energy and higher energy nodes which are used for assembling, combining, and transferring information [5]. Clustering techniques are the most popular one to attain energy efficiency and efficient communication. Cluster-based hierarchal protocols efficiently utilize the data transfer, bandwidth consumption and gather the information also. This paper basically focuses on the most commonly used energy efficient hierarchal clustering techniques. The following paper is patterned as follows. In the first Section, clustering in WSN is explained and in Section second, taxonomy of hierarchal clustering is explained. Section three is regarding hierarchal clustering techniques, and numerous cluster-based approaches are explained thoroughly. The cluster-based techniques are summarized with strengths and weaknesses in Section four. The fifth section shows a comparative review of the discussed techniques followed by the conclusion in the sixth section.

1.1 WSN Clustering Due to lack of sources in WSN, direct communication of nodes with neighbor nodes and base station becomes a tedious task as energy utilization is very high which terminates the sensor nodes early. Structure of hierarchal system has a two-level hierarchy that includes cluster head into the upper part and member nodes at the lower part. Nodes at lower lever transfer the information to their respective CH. The information is combined by a cluster head and sent toward the base station. As CH is continuously transferring the information, so it consumes more energy compared to the member nodes [1]. After, once bound round, the chosen CH is not able to perform or die as it uses more energy, so as to make the load balancing, the CH periodically changes [3, 6].

Comparative Analysis of Clustering Algorithm …

65

BASE STATION

MN

MN

MN

MN

CH

CH MN

MN

MN

MN

CH MN

MN

MNNN MN

CH: Cluster Head MN: Member Node

2 Hierarchal Clustering Taxonomy WSN has various clustering techniques, but the limitation of this paper is that only four categories are explained (i) homogeneous and heterogeneous networks (ii) centralized or distributed algorithms (iii) static and dynamic clustering (iv) probabilistic and non-probabilistic algorithms.

2.1 Homogeneous and Heterogeneous Networks The techniques for both networks are depending upon the properties and performance of the sensor node. In homogeneous networks, all of the sensor nodes have similar processes and hardware specifications [7]. Moreover, it supported numerous parameters like remaining energy state and distance from the middle of a cluster. Each node is a CH, to attain energy potency and equalization, the role of CH is turned sporadically, whereas in heterogeneous networks, wherever there are sometimes two kinds of sensor nodes, nodes with higher hardware and process capabilities

66

S. Sachan et al.

are sometimes used as CH inside a cluster, operate as information collectors or perhaps is used as a backbone inside the network. Nodes having lower capabilities are common device nodes that sense the required field attributes [5, 8].

2.2 Centralized or Distributed Algorithms In centralized techniques, usually network partition and cluster formation is done by CH or base station. These algorithms are not appropriate for big scale networks, only appropriate for limited-area applications. In comparison to centralized one, the distributed techniques sensor nodes themselves chose the CH and do the cluster formation. It is more reliable, fast, and less time consumable [9].

2.3 Static and Dynamic Clustering Static and Dynamic clustering reckon on the applying necessities. In static, the formation of cluster and CH election are firm. When the formation of clusters is done then it’ll stay for a particular time. In most techniques, once the clusters are formed, they retain for a certain period; however, CHs are sporadically modified to achieve energy potency. Dynamic clustering shows the high energy potential as it selects the cluster head on a periodic basis and clusters are formed again. It is basically used for the environment where clusters combine efficiently to enhance the energy level [10].

2.4 Probabilistic and Non-probabilistic Approaches In probabilistic techniques, every node is assigned a previous likelihood to come to a decision whether or not the CHs or any random choice technique is employed [11, 12]. These probabilities given to the nodes work as a primary basis; however, there is a secondary basis as well to select the CH and formation of clusters to increase the energy uses and improve the network life span. This approach shows quick execution, less time, and decreases the number of exchanged messages. In nonprobabilistic approach, a defined standard is used to select the CH and formation of clusters. Basically, it works on the principle of receiving messages from one hop or multiple hops which results in a large number of messages exchange and makes it a complex system in terms of time as compared to probabilistic techniques. These techniques have been very commonly used for different applications in several environments to maintain energy potency and these techniques will modify the management of the nodes, cut back energy consumption, improve load reconciliation, increase strength, and improve information aggregation.

Comparative Analysis of Clustering Algorithm …

67

3 Hierarchical Clustering Techniques A lot of surveys are done in various routing techniques in WSN, but this paper basically focuses on various kinds of clustering hierarchal techniques. Clustering techniques are accustomed to change the node arrangement, to scale back energy consumption, and to enhance the load reconciliation and information aggregation. Nodes are combined together to make clusters. A node that’s called a cluster head (CH) is created accountable for gathering information from member nodes (MN), aggregates it, and transfers it to the base station directly or through some medium [13–15]. Rather than causing information of all device nodes in an exceedingly cluster, CH solely sends the mass information, that successively minimize the quantity of packets transmitted in an exceedingly network and minimize energy consumption.

4 Cluster-Based Techniques 4.1 Low Energy Adaptive Clustering Hierarchy (LEACH) LEACH is one among the primary energy economical routing protocols and continues to be used as a progressive protocol in WSN. The essential plan of LEACH was to pick out CH in between the various nodes by rotation system in a way that the energy dissipation from communication will be sent to all the other nodes. The process is split into two phases, the setup section and steady-state section. Within the setup section, every node decides themselves either or not to be a CH which depends on the CHs proportion advised and a variety of times a node has been CH. A random range is selected in between zero to 1; if the quantity is a smaller amount than threshold, the node becomes a cluster head [16, 17]. The second section is the steady-state section, in which nodes sense and send the information to its CH which is then combined and transfered to base station directly so as to avoid collisions, TDMA/CDMA MAC is employed and because of distributed approach, LEACH doesn’t need any global data.

4.2 Low Energy Adaptive Clustering Hierarchy Centralized In LEACH-C, the quantity of CHs in every spherical has equal rounds of associate to optimize the defined value. A centralized routing approach is one within which base station computes the typical energy of a network for a collection of device nodes having energy state higher than average. A CH is hand-picked from the cluster of nodes to make sure that nodes hand-picked ought to have enough energy level to become cluster head [18]. The network is divided into two sub cluster divisions and they are again

68

S. Sachan et al.

divided into a specified range of CHs. This way of distribution ensures that load distribution is equal. The base station finds the lower energy path and transfers the data of clusters and CH to any or all nodes within the network employing a minimum spanning tree approach. Additionally, each cluster can send request; so energy consumption is going to be high [19].

4.3 Cluster Chain Weighted Metrics (CCWM) CCWM [20] enhances the energy potency and increases network performance supported by weighted metrics. A collection of CHs is chosen based on these metrics. Member nodes use direct communication for transferring information toward their several CHs [21–23]. A routing chain of elective CHs is made for intercluster’s communication and every CH forwards information to its neighboring CH till it reaches to base station.

4.4 K-Means Rule The cluster head is chosen by K-means rule to increase the overall network lifetime [24]. Authors split the entire method into 3 phases. Initially, CH is selected by using LEACH protocol. The network is divided into K clusters on the basis of euclidian distance between the nodes joining the nearest CH. When the nodes are connected with the CH, every node is given an ID depending upon the space from the center. Node nearer to the centroid can have a smaller range. On the rotational basis, CH is elected by comparing the distance from the center and the node nearer to center is elected as a new CH. In comparison to other techniques k-means rule enhances the whole network lifespan however the periodically formed clusters results in more load to the network and large energy utilization [25, 26].

5 Comparison Summary In this section, the above-discussed cluster-based techniques are summarized. The benefits and drawbacks of the prevailing are highlighted to assist researchers to pick the technique as per their demand. These techniques were analyzed keeping in sight the cluster head choice approach to spot whether or not the technique is probabilistic or non-probabilistic. Moreover, the sort of cluster and CH choice is known to research whether or not to use centralized or distributed or may be hybrid. In Table 1, the cluster-based hierarchic protocols are summarized.

Comparative Analysis of Clustering Algorithm …

69

Table 1 Comparison of clustering-based approaches Sr. no.

Name of the technique

Approach for CH selection

Type

Merits

Demerits

1.

LEACH

Probabilistic

Distributed

(i) Load distributed to nodes are equal to some extent (ii) Unnecessary collisions are avoided by TDMA (iii) Excessive energy consumptions are avoided by allocated time slots

(i) Inter cluster communication is single hop (ii) Problems related energy, holes and coverage (iii) Without considering energy, CH selection is probabilistic (iv) Dynamic clustering has extra overheads

2.

LEACH-C

Probabilistic

Centralized

(i) Globally network view (ii) Evenly distributed load (iii) Effective Energy efficient routes

(i) Overhead Network (ii) CH selection is probabilistic (iii) Reselection process is very resource expensive

3.

CCWM

Non-probabilistic

Distributed

(i) Routing and network lifetime and routing are improved (ii) Reliable for small scale and static networks

(i) Election of CH is non optimized (ii) Enhances network overhead (iii) Not reliable for large scale networks

4.

K-means clustering

Probabilistic/centroid-based

Distributed

(i) Simple in approach (ii) Enhances lifetime of network

(i) Reformation of clusters are periodic (ii) Unevenly distributed load (iii) Reselection process is based on centroid distance

70

S. Sachan et al.

6 Conclusion Wireless Sensor Networks (WSN) is getting more focused because of their lesser price, small size, supercharged battery nodes, etc. Nowadays, these nodes are combined with other technologies like IoT, mobile phones, IEEE 802.11, and far additional, that makes WSN one among the efficient approaches of the twentyfirst era. As of the variety of environmental conditions and scarcity of battery sources, wireless sensor networks are a difficult space of analysis and research. This paper has shown a comparative analysis of existing cluster-based hierarchal schemes and energy economical cluster protocols with their respective significance and limitations.

References 1. Sabor N, Sasaki S, Abo-Zahhad M, Ahmed SM (2017) A comprehensive survey on hierarchicalbased routing protocols for mobile wireless sensor networks: review, taxonomy, and future directions. Wirel Commun Mob Comput 2017, Article ID 2818542, 23 2. Sharma KP, Sharma TP (2017) Energy-hole avoidance and lifetime enhancement of a WSN through load factor. Turk J Electr Eng Comput Sci 25(2):1375–1387 3. Lim JM-Y, Chang YC, Alias MY, Loo J (2016) Cognitive radio network in vehicular ad hoc network (VANET): a survey. Cogent Eng 3(1), Article ID 1191114 4. Wang F, Hu L, Hu J, Zhou J, Zhao K (2017) Recent advances in the internet of things: multiple perspectives. IETE Tech Rev 34(2):122–132 5. Gupta G, Younis M. Fault-tolerant clustering of wireless sensor networks. In: Proceedings of the IEEE wireless communications and networking conference (WCNC ‘03), pp 1579–1584, IEEE, New Orleans, La, USA, March 2003 6. Yu YQJ, Wang G, Guo Q, Gu X (2014) An energy-aware distributed unequal clustering protocol for wireless sensor networks. Int J Distrib Sens Netw 2014:8 7. Zin SM, Anuar NB, Kiah MLM, Pathan A-SK (2014) Routing protocol design for secure WSN: review and open research issues. J Netw Comput Appl 41(1):517–530 8. Mahalik NP (2007) Sensor networks and configuration. Springer 9. Enam RN, Qureshi R, Misbahuddin S (2014) A uniform clustering mechanism for wireless sensor networks. Int J Distrib Sens Netw 10(3), Article id 924012 10. Yu Y, Krishnamachari B, Prasanna VK (2004) Issues in designing middleware for wireless sensor networks. IEEE Netw 18(1):15–21 11. Mamalis B, Gavalas D, Konstantopoulos C, Pantziou G (2009) Clustering in wireless sensor networks. In: Zhang Y, Yang LT, Chen J (eds) RFID and sensor networks: architectures, protocols, security and integrations, pp 324–353 12. Huang H, Wu J. A probabilistic clustering algorithm in wireless sensor networks. In: Proceedings of the 62nd vehicular technology conference, pp 1796–1798, Sept 2005 13. Essa A, Al-Dubai AY, Romdhani I, Eshaftri MA (2017) A new dynamic weight-based energy efficient algorithm for sensor networks. In: Smart grid inspired future technologies: first international conference, SmartGIFT 2016, Liverpool, UK, May 19–20, 2016, Revised Selected Papers, pp. 195–203, Springer International Publishing, Cham, Switzerland 14. Mahajan S, Malhotra J, Sharma S (2014) An energy balanced QoS based cluster head selection strategy for WSN. Egypt Inf J 15(3):189–199, Article 101 15. Nayyar A, Gupta A (2014) A comprehensive review of cluster-based energy efficient routing protocols in wireless sensor networks. Int J Res Comput Commun Technol 3:104–110

Comparative Analysis of Clustering Algorithm …

71

16. Jia D, Zhu H, Zou S, Hu P (2016) Dynamic cluster head selection method for wireless sensor network. IEEE Sens J 16(8):2746–2754 17. Park GY, Kim H, Jeong HW, Youn HY. A novel cluster head selection method based on k-means algorithm for energy efficient wireless sensor network. In: Proceedings of the 27th international conference on advanced information networking and applications workshops (WAINA ‘13), pp 910–915, Barcelona, Spain, March 2013 18. Pantazis NA, Nikolidakis SA, Vergados DD (2013) Energy-efficient routing protocols in wireless sensor networks: a survey. IEEE Commun Surv Tutor 15(2):551–591 19. Xu D, Gao J (2011) Comparison study to hierarchical routing protocols in wireless sensor networks. Proced Environ Sci 10:595–600 20. Mamun Q (2012) A qualitative comparison of different logical topologies for wireless sensor networks. Sensors 12(11):14887–14913 21. Batra PK, Kant K (2016) LEACH-MAC: a new cluster head selection algorithm for Wireless Sensor Networks. Wirel Netw 22(1):49–60 22. Nayak P, Devulapalli A (2016) A fuzzy logic-based clustering algorithm for WSN to extend the network lifetime. IEEE Sens J 16(1):137–144 23. Yu J, Qi Y, Wang G, Gu X (2012) A cluster-based routing protocol for wireless sensor networks with nonuniform node distribution. AEU-Int J Electron Commun 66(1):54–61 24. Thonklin A, Suntiamorntut W. Load balanced and energy efficient cluster head election in wireless sensor networks. In: Proceedings of the 8th electrical engineering/electronics, computer, telecommunications and information technology (ECTI ‘11), pp 421–424, Khon Kaen, Thailand, May 2011 25. Youssef A, Younis M, Youssef M, Agrawala A. Wsn16-5: distributed formation of overlapping multi-hop clusters in wireless sensor networks. In: Proceedings of the IEEE global telecommunications conference (GLOBECOM ‘06), pp 1–6, Dec 2006 26. Soro S, Heinzelman WB. Prolonging the lifetime of wireless sensor networks via unequal clustering. In: Proceedings of the 19th IEEE international parallel and distributed processing symposium (IPDPS ‘05), pp 236–243, Washington, DC, USA, April 2005

Concept of Cancer Treatment by Heating Methodology of Microwave Awanish Kumar Kaushik, Smriti Sachan, Shradha Gupta and Shilpa Choudhary

Abstract In past few years Hyperthermia has become a very popular customary clinical procedure to treat Cancer. It uses the heat to deteriorate the cancerous tissue by irradiation of energy by using electromagnetic applicators. This paper outlines the effect of heat in microwave applications. The best treatment for retained tumors by using Hyperthermia, involve giant surface areas which may physically adapt to body contours, and regionally alter their power deposition patterns to regulate the nonuniform temperature caused by tissue in homogeneities and blood flow variations. The cancer killing ability is shown in the treatment done by combining both radiations and chemotherapy by the expert biologists in both vitro and in vivo. Biologists found that its tough to boost and to stay the temperature of tumors at therapeutic levels. This paper explains the methodology to treat the cancer in both ways i.e. conventionally and recent one. Keywords Microwave · Heating · Cancer treatment

1 Introduction In the course of the most recent decades number of individuals harrowed with cancer growth has seen a sensational increment. As indicated by the World Health Organization (WHO), cancer growth is the main reason for death overall today. In the year A. K. Kaushik (B) · S. Sachan · S. Choudhary Electronics and Communication Department, G. L. Bajaj Institute of Technology & Management, Greater Noida, India e-mail: [email protected] S. Sachan e-mail: [email protected] S. Choudhary e-mail: [email protected] S. Gupta Department of Applied Science (Physics), G. L. Bajaj Institute of Technology & Management, Greater Noida, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_10

73

74

A. K. Kaushik et al.

2012, the number of death brought about by malignant growth previously achieved 8.2 million. Among others, the most widely recognized reason for cancer death are disease of the lungs and liver, asserting 1.59 million and 745,000 death, separately [1]. The cooling mechanism used in superficial tissue circulation has created it troublesome to heat deep tissue by semi conductive heating. Microwaves ranges from 300 MHz to 300 GHz, hyperthermia uses 433, 915 and 2450 MHz, these frequencies are assigned for ISM band (industrial, scientific, and medical). Frequency range above than 2450 MHz doesn’t have any practical value so for hyperthermia frequency range below than microwave range is used. Various reports show that human and animal tumors are successfully treated by heating therapy. Also, there are several research on effective hyperthermia therapy, radiotherapy and chemotherapy. Hyperthermia has a small range of temperature varies from 42.5 to 44 °C. At lower temperatures, the impact of temperature is extremely minimum however at higher temperatures more than 44 °C, the basic cells are broken. Due to this variation in blood flow, as comparison to normal tissue tumor tissue are having high temperature. In addition, it’s believed that tumors are additional sensitive to heat. The temperature rise in tumors and tissues is computed by the energy deposited and also the physiological responses of the patient. HYPERTHERMIA is cancer treatment wherein raised temperature causes direct cytotoxicity or expands affectability of dangerous cell to different treatments, for example, radiation treatment and chemotherapy [2]. An outline of the reproduction apparatuses and procedures for hyperthermia treatment arranging was checked on and showed in [3]. And various author also provide the different concept like as local hyperthermia in recurrent breast cancer treatment [4, 5] were reported. For this point of view different microwave system also design using several investigations in recent years [6–8]. The energy deposition could be a complicated operation of frequency, intensity, geometry, and size of applicator when electromagnetic techniques are used. Typically the temperature isn’t solely depends upon the energy deposition however additionally on blood flow and thermal conductivity in tissues. Nowadays hyperthermia state analysis, thermal measurement and treatment coming up with microwave and RF waves that is way far from adequate. Additionally clinical applications cannot absolutely proceed without the prior knowledge or the necessary information of however heat is to incline in numerous clinical things. The event of advanced hyperthermia instrumentation and tec can permit eminent treatment of cancers that are immune to alternative ways of medical aid. Hyperthermia is the process to treat cancerous cell that increases the temperature of the particular body within the temperature range 42–45 °C. At this temperature it’s potential to provide thermal damages simply to the cancerous cell and not to the healthy tissue due to the variations between their vascular structures. Usually normal tissue incorporates a a lot of uniform and regular biological process compared with the tumor cell, thus in normal tissue the blood flow has a lot of prospects to act as cooler in order to that it’s potential to boost the temperature of tumor cell over healthy tissue. Figure 1 shows the variation in temperature within healthy and cancerous tissue. It is seen that in first half initially temperature increases linearly

Concept of Cancer Treatment by Heating Methodology of Microwave

75

Fig. 1 Behavior of temperature in healthy and tumor tissue with same heat condition

and has stabilization at higher values of temperature in tumor cell as compare to the healthy tissue.

2 Basic Cancer Treatment Methods Around 1946, a USA scientist named Dr. Farcy Spencer, ascertained that the an egg put in a microwave device not split all over so he analyze that microwaves can be used in different applications. The following heating strategies are used for cancer treatment in rehabilitation medication. Throughout the last 10 years, these techniques are changed and refined for heating tumors. Specific clinical applications are represented in future sections.

2.1 Microwave Illumination Set-up for Breast Imaging The Breast tissues are less clangorous than high water content tissue, and creating the microwave signal propagation apace possible over 10 cm of breast tissue. The microwave camera incorporates a mounted array of thirty two monopole antenna mounted to a positioning plate in circle at every distance. Every antenna operates in either transmit or receive mode. During this system every non active antenna is designated as a microwave sink so the transmitted signals are absorbed and not radiated. Once the antenna is within the non-active state, any coupled signal is transmitted employing a coax to the switch terminated while not being radiates. The antenna placed to the liquid and position within the circle close the breast is within fifteen cm diameter. Whereas the antennas are merged within the saline bath and associated electrical system are unbroken within the dry setting by a bellow at tank bottom that connect them from saline water. The info acquisition system is intended specified channel to channel separation is maintained on top of 120 db. An

76

A. K. Kaushik et al.

Fig. 2 Coagulation therapy treatment

easy pumping system from a reserve tank maintains the water label throughout the clinical session (Fig. 2).

2.2 Hypothermia Treatment The machine which is used to treat hyperthermia uses 430 MHz of microwave energy, the machine consists of surface cooler, thermometry unit and applicator. To heat the effected space, the temperature should be 42.5 °C. Temperature is measured by a thermocouple sensor having very thin layer of Teflon. After every 24 s power is switched off for 5 s. To measure the temperature. External heating of effected tissue is done by microwaves. If the position of the tumor is 2–3 cm from the surface of the skin, then it is heated by the surface applicator and to analyze it, the device is connected with a computer system (Fig. 3).

2.3 One-Channel Hyperthermia System In this type of treatment, the high frequency with respect to the ground to provide the cooling effect in the particular tissue. Figure 4, represent the basic structure of system. The component explanation is given in the below section. While doing the

Concept of Cancer Treatment by Heating Methodology of Microwave

77

Fig. 3 Hypothermia system

Fig. 4 One-channel hyperthermia system

heating process surface temperatures are controlled so that healthy tissue should not damage. Presently systems are under human supervision, after sometime this work will be done by supercomputers. (i)

Components Description RF power generator- works at 150MHz. Surface CoolingThe working of surface cooling is to retain the surface temperature at predefined value. It has a plastic bag between applicator and the skin circulating water. The circulation of water is done by peristaltic pump; this water is deionized to ignore heating. (ii) Temperature Measurement The temperature measuring device is associated with a multipoint probes, relay box, and a thermocouple amplifier connected to the printer. (iii) Microwave Coagulation Therapy This therapy is used for small tumors, Microwave coagulation therapy uses a thin layered microwave antenna that is placed inside the effected tissue. The

78

A. K. Kaushik et al.

Fig. 5 Microwave coagulated therapy

working frequency of antenna is 2450 MHz and power is in terms of tens of watt. Tumor is heated by microwaves (Fig. 5). The coagulated area made because of heating has about 3 cm of diameter. But in actual the coagulated area is much smaller. This type of issue is not solved by one antenna. The microwave antenna is placed into a catheter made of a PTFE that prevents adhering of the coagulated tissues to the antenna. Two such coaxial slot antennas are used. Applicator and antenna has a spacing of 10 mm array and it makes a 3 cm diameter coagulated area inside a tumor. By changing the number, width and position of the slots heating can be controlled along the axis of the antenna. (iv) Inductive Regional Heating Applicator System Firstly the irradiation issues are solved by the effective heating done by a ferrite core applicator system. From then onwards, different types of applicators work out and developed. In ferrite core a magnetic field is developed between a pair of poles. Regional heating works at low power so that the irradiation in applicator is decreased in comparison to the similar type of inductive applicator. The applicator system consists of cylindrical ferrite core of length 20 and 7 cm. Ferrite cores is adjusted according to the size of the breast. Usually magnetic cores are associated with a close loop magnetic path. As shown in the Fig. 6 various magnetic poles are developed due to RF current and a magnetic coupling is introduced in the pair of ferrite core. When the heat is introduced into ferrite core, the magnetic field developed an eddy current which is penetrated into the breast. Magnetic field is produced by conducive plates whose heating position is controlled vertically or horizontally. A conducive

Concept of Cancer Treatment by Heating Methodology of Microwave

79

Fig. 6 Construction of applicator system

plate is inserted in between a pair of ferrite cores wherever other two plates are associated with a ferrite core Due to the magnetic field induced in the system an eddy current is produced, this current flows at the lower level of conductive plate. So, the magnetic flux induced increases. This makes a strong coupling between the poles. (v) RF Hyperthermia for Deep-Seated Regions This technique uses an RF rectangular resonant cavity applicator. Different type of L-shape antennas are placed at specific places in the applicator. This methodology measures the frequency and temperature of the electrical quantity. The setup is shown in Fig. 7. This setup includes signal generator,

Fig. 7 RF hyperthermia for deep-seated regions setup

80

A. K. Kaushik et al.

3 Conclusion Hyperthermia is as yet not a created treatment for disease (cancer). There are heaps of enhancements require for the equivalent since cancer is an alternate and commonplace malady, it require a great deal of consideration. Inside the scope of research center types of gear to propel innovations the approach can upgrade its viability and effective working in this field. The further examination [9] demonstrates that the dispersion of vitality for the particular zone is a vital factor for the total reaction. The individual with the affected tumor access to the hyperthermia has an exceptionally decent impact of the treatment. In spite of the constraints of this framework, the root level encounters are encouraging to the point that further examination, innovative work will be a subject important to do. The framework in by and by additionally connected with an automated procedure for cooling the surface so the warming example ought to be limit. Fundamentally cancer growth treatment is our last objective and warming treatment utilizing microwave has potential of that level which is a large step towards cancer prevention.

References 1. World Cancer Report 2014, World Health Organization 2. Cavagnaro M et al (2011) A minimally invasive antenna for microwave ablation therapies: design, performances, and experimental assessment. IEEE Trans Biomed Eng 58(4):949–959 3. Kok H et al (2015) Current state of the art of regional hyperthermia treatment planning: a review. Radiat Oncol 10 4. Zee JVD et al (2010) Reirradiation combined with hyperthermia in breast cancer recurrences: overview of experience in Erasmus MC. Int J Hyperth 26 5. Zagar TM et al (2010) Hyperthermia for locally advanced breast cancer. Int J Hyperth 26 6. Iero DAM et al (2014) Thermal and microwave constrained focusing for patient-specific breast cancer hyperthermia: a robustness assessment. IEEE Trans Antennas Propagat 62:814–821 7. Stang J et al (2012) A preclinical system prototype for focused microwave thermal therapy of the breast. IEEE Trans Biomed Eng 59:2431–2438 8. Nguyen PT et al (2015) Microwave hyperthermia for breast cancer treatment using electromagnetic and thermal focusing tested on realistic breast models and antenna arrays. IEEE Trans Antennas Prop 63:4426–4434 9. Lubner M, Brace CL, Hinshaw JL, Lee FT (2010) Microwave tumor ablation: mechanism of action, clinical results and devices. J Vasc Interv Radiol 21:S192–S203

Novel Approach to Detect and Extract the Contents in a Picture or Image Awanish Kumar Kaushik, Shilpa Choudhary, Shashank Awasthi and Arun Pratap Srivastava

Abstract Unmistakable evidence of substance in shaded pictures of complex back ground is a noteworthy troublesome issue. This paper gives a count for perceiving content in pictures. Exploratory outcomes on indoor, outside, captcha and moving follows pictures demonstrate that this system can perceive content words unequivocally. The proposed estimation joins the upsides of two or three already philosophies for substance distinguishing proof, and usages a point of union of thought approach for substance finding. Our trial result on four diverse pictures display that the procedure subject to line edge discovery is reasonably superior to the present methodology. This count has 93.1% recall rate, and average time is 4.12 s, for English content. Keywords Contents detection · Contents localization · Contents extraction · Region-based extraction

1 Introduction Content discovery in video and picture has pulled in experts’ cogitation for quite a while. Thusly, innumerable extensive stretches of authentic chronicles are being secured and shared. Three sorts of substance in pictures are: indoor—outside substance which regularly happens in the field of perspective of the camera and subtitle/sensible/counterfeit substance which is misleadingly merge with the video and exuberance content which happens in the field of point of view of the web like captcha. Sometime the content extraction in a picture is difficult because of composite foundation, Unknown content character shading and distinctive stroke widths. A. K. Kaushik (B) · S. Choudhary · S. Awasthi · A. P. Srivastava G. L. Bajaj Institute of Technology & Management, G.B. Nagar, Greater Noida, Uttar Pradesh, India e-mail: [email protected] S. Choudhary e-mail: [email protected] S. Awasthi e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_11

81

82

A. K. Kaushik et al.

To choose spatial connection depend edge based and associated segment features of content characters, basically there are two type of methods are used. Firstly Edge based methodology [1] which is occupied with manage the have a high difference among content and foundation. In this manner, edges from letters are seen and joined. Related part technique [2] utilized a base up framework by iteratively combine sets of related picture elements. Different calculations dependent on inadequate decay has likewise been proposed for content extraction, where the content extraction is accomplished by accepting appropriate earlier on the content component [3]. There are additionally numerous different works dependent on histogram investigation, maximally stable outer district (MSER), and appearance [4–8]. The peruser is alluded to [9] for a decent review of content acknowledgment. Versatile dynamic form wind calculation is utilized in [10] for division of twisted content lines. Versatile thresholding, edge location administrators and scientific morphology are utilized in [11] for identification of the twisted content line in cam-time caught report images. Liang et al., Dhanya and Jayalakshmi [12] introduced a far reaching investigation of uses of binarization of camera caught record pictures (thick books, authentic compositions, message in scenes and so forth.). The extraction of content and mass from normal scene pictures and comic pictures is performed in [13, 14]. Our suggested procedure for picture content information extraction network (as shown in Fig. 1) removing a content segment from a picture can be extensively assembled into three fundamental advances: (1) discovery of the content area in the picture, (2) confinement of the locale, and (3) extracted the yield word picture. In the content recognition step, the region with a likelihood of content in the picture is perceived and the system of imprisonment incorporates furthermore enhancing the content areas by wiping out non-content segments. At long last in content extraction process delivers a yield picture with white content against a dull foundation.

Fig. 1 Algo

Novel Approach to Detect and Extract the Contents …

83

This paper characterized the substance extraction of four types of pictures using the line edge marker. The whole paper is organized properly with explanation of develop include the Experiment results with example. Portion IV gives closes independently. The algorithm is explain using following diagram (Fig. 1).

2 Proposed Algorithm Here the planning venture of the proposed content extraction is exhibited. Our point is to build speedy and amazing content identification framework which can manage still pictures with unpredictability foundation. It is clear from Fig. 1 that the suggested computation is predominantly executed by three phases, which will be portrayed underneath

2.1 Content Detection In our proposed framework, pictures are next convolved with directional channels at various introduction spread for line edge separating evidence in the level (0° or 180°), vertical (90° or 270°) headings [15]. So it will all in all be viewed as that the zone with higher edge quality in these ways is the going with locale. The line conspicuous verification spreads utilized are appeared in Fig. 2. Which upgrade perceived the edge of substance, by then procedure edge? In the event that the edge of the perceived edge set a sensible respect, the other perceived delicate edge can be separated. The whole process can be explained using some different stages• Stage-1—Make line revelation to characterize the edges at 0 or 180 and 90 or 270 introduction. Therefore we can get map with directional edge which characterizes the edge thickness and edge quality in vertical and level course. Figure 3b, characterize the normal edge picture. Figure 3a, demonstrates the edge picture in level and vertical course.

Fig. 2 Edge of substance

84

A. K. Kaushik et al.

Fig. 3 a Demonstrates the edge picture in level and vertical course b Characterize the normal edge picture

Fig. 4 Morphological result

• Stage-2—dependent on Otsu limit, Convert the edge image to binary image. • Stage-3—with the help of Otsu edge, the morphologically assignment apply in the picture. While morphological exercises as a general rule are performed on combined pictures. Figure 4 define the Morphological exercises.

2.2 Content Abstraction The content limitation approach is explained in this section • Stage-4—in this progression, the level and vertical projection profiles for the confident content segments are investigated. These projection profiles are essentially histograms where every canister is a check of the aggregate number of pixels present in every line or portion. The vertical and level projection profiles for the sharpened edge picture from Fig. 4, are showed up in Fig. 5a, b.

Novel Approach to Detect and Extract the Contents …

85

Fig. 5 a Vertical Projection profile b Level Projection Profile

• Stage-5—Procedure the level and vertical projection profiles of increased picture utilizing histogram with a proper edge regard and Create refined picture utilizing augmentation of parallel picture and focus expelling picture. • Stage-6—Get frail refined picture in 0 and 90 introduction utilizing morphological structure portion and make the last refined picture utilizing subtract the refined picture and weak refined picture. • Stage-7—In this progression, remove the long edge in last refined pictures with the help of related part checking director names the reduced vertical edges. Here, 4-neighbor related portion are used. After the related fragment naming, each edge is particularly named as a lone related section with its exceptional portion number. • Stage-8—In this movement, partition out non-content locale utilizing major to minor turn degree with assistance of Heuristic Filtering. Here, just those districts in the held picture which have a space more conspicuous than or relative to 1/20 of the most uncommon territory locale and clear those regions which have Width/Height NP VP (Here S stands for sentence and NP stands for noun phrase and VB stands for verb phrase). A set of defined rules and notations make a phrase structure as proposed by Chomsky, which is basically transitional grammar i.e. it is defined by the alpha movement of the auxiliary verb. Here in the English form the alpha movement of the verb is being done which doesn’t affect the machine translation and assigns priority to the lexicons in the sentence according to the following order: Noun phrase: these are the lexical units that come from fact that they can be replaced by it or this [17]. Verb phrase: these are the syntactic units which can be replaced by did(it). For e.g.: Blumen sind schön (written in German form)—flower is beautiful (English). Statistical Parts of Speech Tagging This approach is based on transition probabilities. This approach calculates the probability of the occurrence of a particular token in the given corpus. The popular techniques used under these approaches are Markov Model [1], conditional random fields and Hidden Markov Models. Under this approach, Hidden Markov Models are discussed. The morphological analyser for parts of speech tagging in Punjabi language will be including a statistical tagger along with syntactic word classes (rule based approach). A statistical parts of speech tagger is used for tagging the words accurately and removing ambiguity in the words. The most crucial function served by this parser would be eliminating the ambiguity issues and reconstructing the structure of the sentence if it is not according to rules as defined above in rule based approaches. Algorithm: • The first step of parsing is to extract rules for translation from SL (source language) to TL (target language) and then encoding utterances into machine interpreted representations. • The second step is extracting rules between natural language and machine language using phrase based translation and context-free grammar rules. • The third step is language modelling which is a weighted sum of the best translated candidate sentences in target language according to predefined grammar rules. Here annotated data is used which is tagged by a parts of speech tagger and its probability of occurrence is calculated: Bigram = Count(wi−1 , wi )/Count(wi−1 )

112

S. Kaur Jolly and R. Agrawal

This is the quintessential step where a bigram model is used to calculate the tag of the word given the tag and word count of previous and next words. P(ti /wi ) = P(wi /ti ) · P(ti /ti+1 ) In the above equation ti = tag of the current word. Wi = current word. The next crucial step in parsing is semantic role labelling which uses existing parts of speech tagger to assign roles to the lexicons. In the last step the parsing model and statistical model are combined to generate features out of the sentence. Hybrid Approach for Parts of Speech Tagging This approach is a combination of rule based parts of speech tagging and statistical based parts of speech tagging approach [18]. This approach is based on both rules of the language and stochastic models. In this approach, if more than one category is assigned to the token using rule based approach then statistical based models are used to compute the tag of the word entered. Hence it is an iterative approach that assigns tag on basis of the rules and Hidden Markov Model. This approach works as follows: • • • • •

Enter the sentence to be tagged. Tokenize the input into its constituent parts. If the token is plural convert it into singular form. Apply Rule based approach to the input text and compute the tags. if a single tag is found then display the result otherwise if multiple tags exist for a particular token then, • Apply Hidden Markov Model to the input text to compute the tags on basis of probabilities and display the Bigram Probabilities.

4 Analysis of Parts of Speech Tagging from English to Punjabi Language In order to analyse the accuracy of parts of speech tagging in translation from English to Punjabi language POS tagger tool is used developed by Punjabi University, Patiala. Morphology in Punjabi language plays an important role to augment its meaning, class and affix generation in machine translation. The challenges faced while doing rule based parts of speech tagging of a word is similarity of unknown words to words present in lexicon (dictionary) and to find out whether it is a lexical unit or not in the dictionary. So to overcome the above challenges already present, the lexicon was updated by adding known words using statistical parts of speech tagging approach.

Parts of Speech Tagging for Punjabi Language …

113

Punjabi is an Indo-Aryan language written in Gurumukhi script as shown in Fig. 2 and it follows subject object verb agreement so we discuss the inflectional morphology of English and Punjabi and how they are divergent from each other: Nouns: In Punjabi nouns can be singular or plural or both E.g. P: munde (munde) T: munde (boy) Hence rules are that boy—munda and girl—kuri, hence most of the gender translation occurs on the basis of rules above while there are no such rules in English language. • Adjectives: adjectives are also inflected on the basis of gender in Punjabi language. • Prepositions: the prepositions are actually post positions in Punjabi language that comes after the object in Punjabi language. • Pronouns: pronouns are also inflected on basis of singular and plural. • Verbs: verbs are the main part of Punjabi sentence and they are usually followed by auxiliary verbs like hai, haina, si, ham, hana. There are two types of phrases in Punjabi language nominal phrases and verb based phrases. Nominal phrases are made of modifier and noun e.g. Lambi kuri in Punjabi while verb phrases have the main verb phrase and supporting verb phrases. Hence the main difference existing in Punjabi morphology and English morphology is an arrangement of the verb phrases in the sentence: P: vakil apna niji kaam chal rahia hai T: lawyer is running his private firm Hence in the above translation the main difference is in the word order as translation is quite difficult directly hence we need an indirect approach based on n-gram based word model that uses dictionary approach to map the bigrams to bigrams and trigrams to trigrams and mapping of multi-words like mom–dad to ‘mappe’ in Punjabi. Hence using the POS tagger tool it was analysed that statistical parts of speech tagging were more accurate than rule based parts of speech tagging. The tool gave the following results: English Sentence: Police said a 23-year-old man was arrested on Monday after eight small vehicles were thrown in a small fight in Lowman Nagar area of the police station. :

114

S. Kaur Jolly and R. Agrawal

Fig. 2 Punjabi-English mapping

Rule Based Parts of Speech Tagging

Statistical Based Parts of Speech Tagging (Fig. 2)

.

5 Results The accuracy of the parts of speech tagger can be calculated by percentage of the words that are accurately tagged. It is defined below: ACCURACY = Total No of Words that are correctly tagged/Total No of Words Tagged In order to calculate the accuracy of tagger we evaluated 10,000 words divided into two parts from a Punjabi corpus and got the following results. The corpus that was evaluated was agriculture corpus having tags along with the Punjabi words as shown.1 1 The

text in () is phonetic translation of Punjabi words.

Parts of Speech Tagging for Punjabi Language …

115

Table 1 Accuracy of the POS tagger Test set

Size of corpus

Accuracy (%)

1

5000 words

77.8

2

5000 words

87.5

Table 2 Parts of speech tags for Punjabi language No.

Tag

Tag description

Example

English equivalent

1

N_NN

Common noun

Saharan

2

N_NNP

Proper noun

July

3

N_NST

Noun loc

Above

4

V_VM_VNF

Non-finite Vb

Sow

). ( ). The accuracy of the parts of speech tagger on the corpus is described in Table 1. The following tagsets are used for Punjabi parts of speech tagging from the agriculture corpus are described in Table 2. Hence it is observed that statistical parts of speech tagger solved the problem of complex sentences in Punjabi language. It is also observed that statistical parts of speech tagger solved the problem of unknown words in the corpus (out of the vocabulary words) and parts of speech tagging in the corpus.

6 Conclusion The information present in natural language is entirely dependent on the rhetorical structures (parse trees) generated by parsers and parts of speech tagger in machine translation systems. The crucial addendum in parts of speech tagging can be made by adding semantic principles to modify the algorithm further efficiently. Semantic addendum is needed for verb related movements in the sentence. Hence parts of speech tagging algorithm defined above justify the sentence formation and the structural effects. Hence a combined joint structure i.e. hybrid approach for parts of speech

116

S. Kaur Jolly and R. Agrawal

tagging needs to be defined in future to address different aspects of morphologically rich languages like Hindi and Punjabi.

References 1. Bandyopadhyay S, Ekbal A (2006) HMM based POS tagger and rule-based chunker for Bengali. In: Proceedings of the 6th international conference on advances on pattern recognition (ICAPR 2007), 2–4 Jan 2006, ISI, Kolkata, India, pp 384–390, World Scientific Press (Singapore) 2. Rama T, Gali K(2009) Modeling machine transliteration as a phrase based statistical machine translation problem. In: Proceedings of the 2009 named entities workshop: shared task on transliteration. Association for Computational Linguistics, pp 124–127 3. Goyal V, Lehal, GS (2008) Hindi morphological analyzer and generator. IEEE Computer Society Press, Los Angeles, CA, pp 1156–1159 4. Gill MS, Lehal GS (2008) A grammar checking system for Punjabi. In: 22nd International Conference on on Computational Linguistics: Demonstration Papers. Association for Computational Linguistics, August, pp 149–152 5. Cutting D, Kupiec J, Pedersen J, Sibun P (1992) A practical part-of-speech tagger. In Third conference on applied natural language processing (ANLP-92), pp 133–140 6. Chomsky N (1981) Introduction to government and binding theory 7. Chomsky N (1981) On the representation of form and function. Linguist Rev 1(1): 3–40 8. Patil VF (2010) Designing POS tagset for Kannada languages, organized by central institute of Indian Languages, Department of higher education, MHRD 9. Jagarlamudi J, Kumaran A (2008) Cross-lingual information retrieval system for Indian languages. In: Advances in multilingual and multimodal information retrieval. Springer, Berlin, Heidelberg, pp 80–87 10. Brill E (1994) Some advances in transformation-based part-of-speech tagging. In Proceedings of the 12th national conference on artificial intelligence (AAAI-94), vol 1, pp 722–727, Seattle, WA 11. Madhavi G, Mini B, Balakrishnan N, Raj R (2005) Om: one tool for many (Indian) languages. J Zhejiang Univ Sci A (Zhejiang University Press) 6(11):1348–1353 (2005) 12. Mishra N, Mishra A (2011) Part of speech tagging for Hindi corpus. In: 2011 International Conference on Communication Systems and Network Technologies. IEEE, June, pp 554–558 13. Ali H (2010) An unsupervised parts-of-speech tagger for the bangla language. Department of Computer Science, University of British Columbia, 20, 1–8 14. Singh S, Gupta K, Shrivastava M, Bhattacharyya P (2006) Morphological richness offsets resource demand-experiences in constructing a POS tagger for Hindi. In: Proceedings of Coling/ACL 2006, Sydney, Australia, July, pp 779–786 15. Gill MS, Lehal GS, Joshi SS (2008) Part of speech taggset for grammer checking of Punjabi. Apeejay J Manage Technol 3(2):146–152 16. Mahar JA, Memon GQ (2010) Rule based part of speech tagging of Sindhi language. In: ICSAP10, International conference on IEEE Signal acquisition and processing, 2010, pp 101– 106 17. Goyal V, Lehal GS (2008) Hindi morphological analyzer and generator. In: First international conference on emerging trends in engineering and technology, USA, pp 1156–1159 18. Mohnot K, Bansal N, Singh SP, Kumar A (2014) Hybrid approach for part of speech tagger for Hindi language. Int J Comput Technol Elec Eng 4:25–30 19. Gupta R, Goyal P, Diwakar S (2010) Transliteration among Indian Languages using WX notation. In: Semantic approaches in natural language processing 20. Janarthanam SC, Sethuramalingam S, Nallasamy U (2008) Named entity transliteration for cross-language information retrieval using compressed word format mapping algorithm. In: Proceedings of the 2nd ACM workshop on improving non english web searching. ACM, pp 33–38

Development of Decision Support System by Smart Monitoring of Micro Grid Laxmi Kant Sagar and D. Bhagwan Das

Abstract This paper presents a brief overview of the remote real time monitoring system of the smart micro grid at Dayalbagh Educational Institute. Various components of smart micro grid i.e., battery, inverter etc., are smartly monitored by continuous data collection and real time graphs. Potential harmful conditions of the solar PV system and the parameters of these grid components that effects efficiency of smart micro grid are taken into account and timely preventive/corrective control action is taken thereby contributing toward the decision support system. Also comparison is done with existing methodology and design of micro grid to prove the effectiveness of micro grid at DEI. Keywords Smart micro grid · Real time monitoring · Reliable · Economy

1 Introduction Smart micro grid is basically an integration of conventional grid along with renewable energy generation resources and communication technologies including internet which provides two way communication between utility and customer. Remote Monitoring, Analysis, and Control are the various factors which make smart grid better than traditional grid. Increased population growth, carbon emissions, and global warming are the various factors responsible for the extensive use of renewable energy and their integration with existing power systems lead to the growth of smart grids [1]. Smart micro grid provides electricity to customers at reduced price, high reliability, and low pollution [2]. Nowadays with an increasing use of renewal sources for energy generation smart micro grid is becoming very popular, as using micro L. K. Sagar (B) Department of Computer Science & Engineering, G.L. Bajaj Institute of Technology & Management, Greater Noida, India e-mail: [email protected] D. Bhagwan Das Department of Electrical Energy, Dayalbagh Educational Institute, Agra, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_15

117

118

L. K. Sagar and D. Bhagwan Das

grid integration of electricity with renewal energy can be done. Beside this there are many other advantages of using grid are: • • • •

increased security more effective way of energy transmission reduced operational cost quicker restoration of energy after disturbances.

Since the generation depends on the weather, which is stochastic in nature and also the load is dynamic, efficient and reliable operation of the plants is one of the main challenges for smart grid. Renewable Energy Smart Grid should be able to provide quality and reliable electricity with decision-making capabilities. This paper highlights some of the salient features of the smart micro grid developed in DEI Agra support by DST’s Solar Energy Research Initiative (SERI) in 2014.

2 Literature Survey Till date a lot of work is done and still continuing to make smart micro grid more efficient and reliable. Koykul [3] discusses about advantages and disadvantages of using smart grid in distributed renewal energy generation. Gungor et al. [4] focus on the use of Wireless sensor networks for improving efficiency and reliability of smart grid. Kohsri and Plangklang [5] emphasise on monitoring and data collection from generation to load consumption for decision support system. Pipattanasomporn et al. [6] discuss about providing intelligence to smart grid through multi agent system. Metke and Ekl [7] discuss about grid security and propose PKI (Public Key Infrastructures) to be the best among various available security techniques. Logenthiran et al. [8] propose a generalized technique for based on load shifting. Maharjan et al. [9] propose distributed algorithm—Stackelberg Game which will solve Demand Response Management (key component of smart grid) problem and will take care of both companies and consumers. Mwasilu et al. [10] discuss about future aspects of smart micro grids and discuss about integrating smart grid with Electric Vehicles. Reference [11] focus on crucial issues related to integrating of smart grid with building which includes use of smart meters, interoperability, and demand response capabilities. Keshtkar et al. [12] propose use of fuzzy logic for load reduction with thermal comfort considerations. Bahrami and Sheikhi [13] propose integrated dynamic Response program to reduce dependency on customers in the existing Dynamic Response program. Alam et al. [14] propose cognitive based communication for smart grid. Zame et al. [15] discuss about RD and D policies for energy storage to enhance operational experience and reduce operational cost (Table 1).

Development of Decision Support System by Smart Monitoring …

119

Table 1 Summary of work done Year/Author

Methodology

Results

2001/Kian A. [16]

Study factors that impact renewable energy

Idea for renewable energy pricing

2002/Filho. J. C. R. [17]

Integrating renewable energy with smart grid

Design price tariffs for consumers

2010/Ayompe L. M. [18]

Real time monitoring for future

Model for predicting PV system AC output power

2011/Kohsri [5]

Testing system to choose proper strategy

Real time monitoring for decision making

2011/Koykul W. [3]

Discuss smart grid applications

Survey and summarize smart grid applications

2012/Geviano A. [19]

Discuss smart grid applications

Integrating new devices with existing equipments

2012/Phaungpornpitak N.

Develop algo for optimal locating and sizing of components

Methods to find out optimal location for DG

2012/Phaungpornpitak N.

Challenges to integrate renewable energy in smart grid

Financial and sizing issues for adding renewal energy

2019/This study

Real time monitoring of smart micro grid

Smart micro grid toward decision support

2.1 Smart Micro Grid at Dayalbagh Education Institute In order to harness the goodness of renewable energy, whole DEI campus has 9 rooftop Solar PV power Plants having a total capacity of 558.2 kWp with battery backups. The institute is self-sufficient to provide electricity to the whole campus thereby reducing noise and sound pollution and contributing to National Solar Mission. Ministry of Renewable Energy, Government of India has declared the Dayalbagh Town area as a Green Campus. Besides many advantages, there are many challenges posed by SPV power plants that includes: • • • • •

Monitoring and Fault diagnosis of inverters Monitoring of loads Fault diagnosis of SPV panels Overloading of inverters Inefficient operation of inverters due to under loading.

To deal with the above-mentioned challenges and ensure reliable efficient, and economic operation of various distributed energy resources, a need based R & D project was started in DEI with an objective of designing a smart micro grid that will handle all the above-mentioned challenges. The project was raised in support with DST’s Solar Energy Research Initiative (SERI) (Table 2).

120 Table 2 Distributed SPV power plant capacities

L. K. Sagar and D. Bhagwan Das S. no

Site

SPV capacity (kWp)

1

Faculty of Engineering

147.8

2

Faculty of Science

148.32

3

Faculty of Arts

40.8

4

Faculty of Education

40.8

5

Faculty of Social Science

40.8

6

USIC Complex

94.68

7

Boys Hostel

5

8

Seminar Hall Complex

20

9

New Girls Hostel

20

3 DST SERI Project: Design The Dayalbagh renewable energy smart micro grid is designed with an objective of providing electricity at reduced cost and high reliability. It is a modern, small scale electricity system which integrated with distributed renewable energy sources and a grid from Torrent Agra. In this study we are mainly focusing on two aspects of smart grid i.e., monitoring and controlling as two different modules.

3.1 Methodology and Features—Monitoring A Real Time Monitoring system (software) has been designed in Dayalbagh under SERI project where real time monitoring, analysis, and decision based on the analysis are made to increase the efficiency of micro grid. Some of the salient features of automated system include: • It has been designed in .Net framework using C#. • Each plant is provided with a unique static IP address which provides uniqueness to all plants. • The Data form inverter and meter are collected and stored in Cloud and can be monitored globally. • The data are refreshed at the rate of 3 s. • All the data are stored in MySQL server at Dayalbagh. • Data from all inverters are integrated and displayed for control action. • Data from smart energy meters are also integrated and collected for load demand evaluation. The automated software is graphical user interface based and continuously monitors all the solar plants and monitors various parameters of Inverter i.e., volts, Amps, Freq etc., and also stores them at local server.

Development of Decision Support System by Smart Monitoring …

121

GUI OF AUTOMATED SOFTWARE OF SMART MICRO GRID AT DEI

Fig. 1 GUI for arts department

• A special feature has been embedded in software where some constraint have been imposed on some parameters (example voltage) with some threshold value and if value of that parameter falls below the assigned threshold a timely alert i.e., mail and sms is sent to the concerned person thereby contributing to the decision support system. Figure 1 shows the GUI for Arts department and same GUI’s run for seven other departments.

3.1.1

Real Time Monitoring

A dashboard in tableau has been designed which provides real time monitoring of all the plants. The plotted graph between time versus solar Kw provide health status of all the eight plants, thereby ensuring good reliability and good economy. Various important parameters are monitored through graphs. Some of these parameters include Solar_kW—monitors Solar Generation, load_Kw—monitors site load, Src_kW—monitors Grid Supply i.e., electricity from torrent power. Figures 2, 3, and 4 showing all three monitoring.

Data Collection Data from inverter i.e., voltage, current, power etc., and meter data i.e., load currents, voltages are displayed at front end and are collected and stored at local MySQL

122

Fig. 2 Solar generation monitoring

Fig. 3 Site load monitoring

Fig. 4 Grid supply monitoring

L. K. Sagar and D. Bhagwan Das

Development of Decision Support System by Smart Monitoring …

123

Fig. 5 Schema of data stored

server. These various parameters help in controlling behavior of inverter by selecting different modes of operations. An automated system has been designed which automatically takes backup of whole database at 11.59 p.m. and stores it at pre-defined location so as to recover data in case of any mishandling (Fig. 5).

3.2 Methodology and Features—Controlling Automated system has been designed which monitors the load and as per the requirement switches from solar to grid or vice versa. The software is embedded with microprocessor controlled switchgear and switchovers which handles the switching (Fig. 6).

Fig. 6 GUI of energy meter data

124

L. K. Sagar and D. Bhagwan Das

Fig. 7 Microcontroller based remote controlled change-over switches

3.2.1

Remotely Controlled Switchgear and Switchovers

Selection of sources i.e., solar and grid during off-grid and on grid hours is done automatically as per the threshold value of selected parameters through software which reduces man labor and thereby increasing reliability and economy (Fig. 7).

4 Decision Support System at DEI The Automated System (Software) provides a basis of decision support system which helps in making decision thereby ensuring increased efficiency of the smart micro grid (Fig. 8). The data collected are very useful for monitoring the health of SPV power plants as well as providing a good source of data for research purposes for UG and PG projects. On the basis of the collected data, various machines learning algorithms such as for Solar generation forecasting and prediction, load demand analysis etc., are being implemented in R and Python environment which will further enhance the efficiency of grid.

Development of Decision Support System by Smart Monitoring …

125

Fig. 8 Decision support system

5 Conclusion The smart micro grid at DEI has proved to be very beneficial many times. On 30 and 31st July 12 when almost half the population of India was paralyzed due to blackout, the smart micro grid at Dayalbagh proved its effectiveness by providing uninterrupted power supply to the whole campus. The smart micro grid at Dayalbagh fully satisfies the objective of smart micro grid for which it has been designed. It provides uninterrupted power supply to the whole campus even at off-grid hours.

References 1. Vaccaro A, Velotto G, Zobaa A (2011) A decentralized and cooperative architecture for optimal voltage regulation in smart grids. IEEE Trans Ind Electron 58(10):4593–4602 2. IEEE. Smart grid: reinventing the electric power system. IEEE power and energy magazine for electric power professionals. IEEE Power and Energy Society, USA (2011) 3. Koykul W (2011) Current status and action plan of utility sector on smart grid and smart community in Thailand. Thailand—Japan workshop on smart community in Thailand: Provincial Electricity Authority, Thailand 4. Gungor VC, Member, IEEE, Lu B, Senior Member, IEEE, Hancke GP, Senior Member, IEEE (2010) Opportunities and challenges of wireless sensor networks in smart grid. IEEE Trans Ind Electron 57(10) 5. Kohsri S, Plangklang B (2011) Energy management and control system for smart renewable energy remote power generation. Ener Procedia 9:198–206 6. Pipattanasomporn M, Feroze H, Rahman S (2009) Multi-agent systems in a distributed smart grid: design and implementation. In: IEEE PES 2009 power systems conference and exposition. Seattle, Washington, USA, pp 1–8

126

L. K. Sagar and D. Bhagwan Das

7. Metke AR, Ekl RL (2010) Security technology for smart grid networks. IEEE Trans Smart Grid 1(1) 8. Logenthiran T, Srinivasan D, Shun TZ (2012) Demand side management in smart grid using heuristic optimization. IEEE Trans Smart Grid 3(3) 9. Maharjan S, Zhu Q, Zhang Y, Gjessing S, Ba¸sar T (2013) Dependable demand response management in the smart grid: a Stackelberg game approach. IEEE Trans Smart Grid 4(1) 10. Mwasilu F, Justo JJ, Kim E-K, Do TD, Jung J-W (2014) Electric vehicles and smart grid interaction: a review on vehicle to grid and renewable energy sources integration. Renew Sustain Ener Rev 34:501–516 (Elsevier) 11. Kolokotsa D. The role of smart grids in the building sector, vol 116, pp 703–708, 15 March 2016. Elsevier 12. Keshtkar A, Arzanpour S, Keshtkar F, Ahmadi P (2015) Smart residential load reduction via fuzzy logic, wireless sensors, and smart grid incentives. Ener Build 104:165–180 (Elsevier) 13. Bahrami S, Sheikhi A (2016) From demand response in smart grid toward integrated demand response in smart energy hub. IEEE Trans Smart Grid 7(2) 14. Alam S, Farhan Sohail M, Ghaurib SA, Qureshib IM, Aqdas N (2017) Cognitive radio based smart grid communication network. Renew Sustain Ener Rev 72:535–548 (Elsevier) 15. Zame KK, Brehm CA, Nitica AT, Richard CL, Schweitzer III GD (2018) Smart grid and energy storage: policy recommendations. Renew Sustain Energy Rev 82(1):1646–1654 (Elsevier) 16. Kian A, Keyhani A (2001) Stochastic price modeling of electricity in deregulated energy markets. In: 34th Hawaii international conference on system sciences. Hawaii, USA 17. Filho JCR, Affonso CM, Oliveira RCL (2002) Pricing analysis in the Brazilian energy market: a decision tree approach. In: IEEE PowerTech conference. Bucharest, Romania 18. Ayompe LM, Duffy A, McCormack SJ, Conlon M (2010) Validated real-time energy models for small-scale grid-connected PVsystems. Energy 35(10):4086–4091 19. Gaviano A, Weber K, Dirmeier C (2012) Challenges and integration of PV and wind energy facilities from a smart grid point of view. Energy Procedia 25:118–125

Gender Recognition from Real-Life Images Apoorva Balyan, Shivani Suman, Najme Zehra Naqvi and Khyati Ahlawat

Abstract Pattern Recognition from images has always been a prime research area when machine learning is talked about. Gender recognition from real-life images is a similar area which poses many challenges since humans are efficient in identifying gender but it is not easy to accomplish the same task with computers. Most of the social interactions depend upon gender perception. In this paper, a machine learning based approach has been discussed to recognize gender from real-life images. Various classifiers have been trained on 17,500 images from the Adience benchmark with multiple train and test data splits. The popular machine learning technique Principal component analysis is used to represent the face image in a low dimensional space with the help of a feature vector. The experimental results have shown that the maximum accuracy achieved is 90% from support vector machine using a radial basis function on a 70–30% split of training and testing data. Keywords Gender recognition · Support vector machine · Radial basis function · Principal component analysis · Linear kernel · Linear discriminant analysis · Feature extraction · Logistic regression

A. Balyan (B) · S. Suman · N. Z. Naqvi · K. Ahlawat Indira Gandhi Delhi Technical University for Women (IGDTUW), Delhi, India e-mail: [email protected] S. Suman e-mail: [email protected] N. Z. Naqvi e-mail: [email protected] K. Ahlawat e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_16

127

128

A. Balyan et al.

1 Introduction Gender recognition is an important problem in machine learning in computer vision society. The study of state-of-art methods suggests that no perfect solution exists to this problem which makes it even more interesting. As technology has emerged, gender recognition is required in numerous areas of applications comprising of human– computer interaction (HCI), biometrics, surveillance, and security, etc. This results in the generation of an efficient model that recognizes gender with the same accuracy as humans. A neural network known as Sexnet [1] was trained to discriminate sex in human faces and performed on 90 images. The average error rate was 8.1% when compared to a human whose average was 11.6%. This was the first attempt in the field of gender recognition. Moghaddam B and Yang M-H [2] investigated support vector machine on 1700 faces with low resolution from FERET dataset. Various classifiers like linear discriminant analysis, support vector machine with Gaussian and polynomial kernel, k-nearest neighbors, etc., were trained. The support vector machine showed the lowest error rate as it was not affected by the resolution of the images. Calfa and others [3] identified the gender by applying linear discriminant analysis (LDA) and principal component analysis (PCA). Linear discriminant analysis performs similar to support vector machine if the resources are scarce. Multiple datasets were used like FERET, PAL, UCN, etc. Lemley J et al. [4] did a comparative analysis of state-of-art methods that are used for detecting gender. FERET and Adience datasets were used for the comparison. Adience consisted of images in the natural environment as compared to FERET. Edinger et al. [5] provided a distinctive dataset of facial images labeled for age and gender that were captured from mobile devices such as iPhone. This made the dataset challenging as compared to other facial image datasets. The dropout support vector machine was used for facial feature estimation along with the cross-dataset testing using Gallagher and Adience. A novel methodology was nominated by Basha and Jahangeer [6] to classify gender into male and female using ORL dataset (400 images). Continuous wavelet transform (CWT) was used for feature selection. Selected features were given as input for classification using support vector machine (SVM) with a linear kernel. A hybrid classification model was proposed by Gutta and Weschler [7] for gender and ethnic classification of human faces trained on 3006 facial images. It was a combination of inductive decision tree and radial basis function. Due to its hybrid nature, it was robust and flexible. Jain et al. [8] did the classification of images using independent Component analysis and support vector machine. Independent component analysis was used to convert input image into a feature vector within a low dimensional space. Experiment on varying sizes of training data was done to see the effect on the accuracy of the classifier. Graf et al. [9] studied gender from human faces using discriminatory experiments. The dimensionality of a set of facial image was reduced using principal component analysis and various classifiers were trained on this reduced representation like support vector machine, relevance vector machine, prototype classifier. The support vector machine outperformed all the other classifiers.

Gender Recognition from Real-Life Images

129

2 Dataset Description The dataset used is Adience Benchmark [5]. This benchmark consists of images taken in real-world conditions (or natural conditions). It includes all the effects of lighting, noise in the environment, the pose of the person, etc. This dataset captures humans in their natural settings or ‘in the wild’. Features of Adience benchmark: 1. 2. 3. 4. 5. 6.

17,492 face photos of males and females are used. The total human subjects are 2,284. The age is divided into 8 groups: (0–2, 8–13, 15–20, 25–3, 38–43, 48–53, 60+). Each photo is labeled by gender. All the images are taken in natural conditions. Flickr albums are the source of the dataset. The images are uploaded with mobile phones like the iPhone. All the images are public using the Creative Commons (CC) license.

It is clear from the above photos that they have been captured in a non-artificial environment. They have a different background, movements, and different facial expressions like a smiley face, serious face, with glasses and without glasses.

3 Preprocessing and Classification Methods Principal component analysis (PCA) [10] is a dimensionality reduction and feature extraction algorithm that extracts all the features from an image. It finds ‘k’ n-dimensional orthogonal vectors when there are n attributes or dimensions. These orthogonal vectors represent the data. Also the value of ‘k’ is less than or equal to n. The original data is projected toward lower dimensional space which leads to dimension reduction. Practically, PCA creates a matrix of n features into a new dataset of less than n features i.e., ‘k’. Principal component analysis of two-dimensional data—Normalization of input data is done such that all the attributes are in the same range. Computation of k orthogonal vectors—The input data is a linear combination of principal components. The components are arranged in decreasing order of variance. These components serve as a redefined set of axes for the data, providing information about the variance. The data can be scaled by eliminating those components which have low variance. Support vector machine (SVM) [11] is a classification method that transforms the data through a nonlinear mapping into high dimensional space. This is done by searching a maximal hyperplane that separates one class from another. It finds the hyperplane using support vectors which are significant data points that lie close to the decision boundary. There can be two cases, when the data is linearly separable and when it is linearly inseparable. In the latter case, SVM uses ‘Kernel trick’ to find the hyperplane in higher dimension. There are various kernel functions like Polynomial kernel, Gaussian radial basis function (RBF), Sigmoid Function etc. [12].

130

A. Balyan et al.

4 Proposed Methodology The classification of gender from the real-life images is done using Python version 3.7 on Anaconda software version 2018.12. The proposed methodology is depicted in Fig. 1. We will divide the datasets into two parts, training dataset to train the model and testing dataset to check the performance of the model. Adience dataset— It contains around 17,500 images that are as true as the real-world scenarios. This problem lies in the supervised learning domain (Classification) of machine learning as the classes are already defined i.e., MALE and FEMALE [13]. Steps followed are: Step 1: Data preparation—The Adience dataset can be downloaded from: “https:// talhassner.github.io/home/projects/Adience-data.html”. The data will be stored in a python list. All the images need to be resized and converted to grayscale. The features will be stored in a CSV file along with their gender values. Step 2: Feature extraction—The feature vector of the image is created with the help of principal component analysis (PCA). Before applying principal component analysis, scaling of the data is done using a min–max scalar of sklearn preprocessing library. Different principal components value like 75, 85, 100, 150, 500, etc. are tried on data to understand the effect on various models. Step 3: Splitting the dataset into training and testing sets—The split is done in three ways. 70–30% with testing data being 30% of the overall dataset, 75–25% with testing data being 25% of the overall dataset, 80–20% with testing data being 20% of the overall dataset. Step 4: Classification—Various classification algorithms can be used to recognize gender. All of them will be imported from the scikit-learn python library [13].

Fig. 1 Proposed methodology

Gender Recognition from Real-Life Images

131

Being a classification problem, the algorithms that can be used are, Support vector machines (SVM) using different kernel functions such as linear, radial basis function, Naive bayes [14], Logistic regression [15], Decision tree [16], K-nearest neighbor (KNN) [17], Linear discriminant analysis (LDA) [18], Quadratic discriminant analysis (QDA) [19].

5 Experimentation and Results The main outcome is to classify the gender in the images with maximum accuracy i.e., the classifier used should be a reliable classifier that predicts the gender of all images provided in the test dataset correctly. After training models using various algorithms of classification, the code was run five times to maximize accuracy with different train and test split. These splits lead to different results which are summarized below: When the training data is 70% and testing data is 30% of the overall data— Radial basis support vector machine (SVM) outperforms all the other classifiers with its average accuracy of 90% although support vector machine takes time in comparison to others. The next best performer is a support vector machine with linear kernel with an average accuracy of 89%. K-nearest neighbor gives a medium average accuracy of 87% after SVM linear kernel. Decision tree performed the worst on the testing data as they are very prone to overfitting. The accuracy is 85% which is the least average accuracy achieved on training the model in every iteration. The accuracy is shown in Table 1. When the training data is 75% and testing data is 25% of the overall data— Radial basis support vector machine and linear kernel support vector machine (SVM) outperforms all the other classifiers with their average accuracy of 89.4%. The next best performers are naive bayes, logistic regression, and linear discriminant analysis with their average accuracy of 89%. K-nearest neighbor and logistic regression gives medium average accuracy of 87%. Decision tree performs the worst in this split with an accuracy of 85% due to the major drawback of overfitting. The accuracy is shown in Table 1. When the training data is 80% and testing data is 20% of the overall data— Radial basis support vector machine and linear kernel support vector machine (SVM) outperforms all the other classifiers with their average accuracy of 89.3%. The next best performers are naive bayes, logistic regression, and linear discriminant analysis with their average accuracy of 88%. K-nearest neighbor and quadratic discriminant analysis gives medium average accuracy of 87%. Decision tree performs the worst in this split with an accuracy of 85% due to the major drawback of overfitting. The accuracy is shown in Table 1.

132

A. Balyan et al.

Table 1 Accuracy for various splits on data Algorithm

Split (%)

1st run

2nd run

3rd run

4th run

5th run

Average

RBF support vector classifier

70–30 75–25 80–20

0.9080 0.8945 0.8960

0.9001 0.8904 0.8900

0.8993 0.8980 0.8970

0.8993 0.8925 0.8930

0.8993 0.8982 0.8920

0.9002 0.8940 0.8930

Linear support vector classifier

70–30 75–25 80–20

0.9028 0.8945 0.8960

0.8932 0.8904 0.8900

0.8927 0.8980 0.8970

0.8917 0.8925 0.8930

0.8991 0.8982 0.8982

0.8954 0.8940 0.8930

Linear discriminant analysis

70–30 75–25 80–20

0.8917 0.8830 0.8910

0.8871 0.8920 0.8920

0.8864 0.8870 0.8900

0.8841 0.8970 0.8850

0.8868 0.8940 0.8800

0.8872 0.8900 0.8860

Naive Bayes

70–30 75–25 80–20

0.8885 0.8890 0.8910

0.8854 0.8840 0.8830

0.8839 0.8940 0.8910

0.8864 0.8910 0.8870

0.8862 0.8920 0.8830

0.8854 0.8900 0.8870

Logistic regression

70–30 75–25 80–20

0.8801 0.8810 0.8850

0.8828 0.8740 0.8800

0.8742 0.8820 0.8840

0.8801 0.8810 0.8740

0.8786 0.8780 0.8770

0.8791 0.8750 0.8800

Quadratic discriminant analysis

70–30 75–25 80–20

0.8807 0.8770 0.8770

0.8750 0.8750 0.8770

0.8790 0.8850 0.8810

0.8792 0.8800 0.8780

0.8837 0.8880 0.8740

0.8782 0.8900 0.8770

KNN

70–30 75–25 80–20

0.8830 0.8792 0.8740

0.8700 0.8744 0.8710

0.8742 0.8680 0.8780

0.8723 0.8792 0.8740

0.8770 0.8797 0.8750

0.8753 0.8750 0.8740

Decision tree

70–30 75–25 80–20

0.8597 0.8545 0.8530

0.8637 0.8535 0.8490

0.8603 0.8598 0.8550

0.8586 0.8573 0.8570

0.8447 0.8628 0.8590

0.8547 0.8560 0.8550

6 Conclusion and Future Scope The characteristics of distinction between male and female can be understood by the gender of the person. Gender classification is a challenge that provides the computer with the facility of categorizing gender information. This classification is fairly complex as it is very easy for humans to understand the gender of a person as compared to a computer. The important fields where gender classification is used are Human– computer interaction (HCI), Artificial intelligence (AI), Social media marketing by business. Since this problem does not have any perfect and unique solution therefore, various machine learning approaches are employed for classifying gender and their results are compared. The highest accuracy in the state-of-art methods is 98% on less training data that is 400 images and on Adience benchmark highest accuracy is 77.4% by applying machine learning. The accuracy achieved in this research is 90% by applying the support vector machine, which is maximum as compared to all the other classifiers when trained on Adience benchmark. Due to the immense importance of recognizing gender correctly, accuracy can be improved. This can be done with the help of neural networks. The neural networks take a lot of training time but

Gender Recognition from Real-Life Images

133

their performance is better than machine learning algorithms because of the propagation and fitting of nonlinear models. Prediction of age can be done by using face image data of the person. Furthermore, efficient algorithms can be derived for feature extraction to achieve better accuracy of the model like extracting the most relevant features from the image which distinguish gender. Prediction of more characteristics from the face images adds more value to the computer. Enhancement of the systems can be done by adding the prediction of age, emotion, ethnicity. Each of the above explains more about a person helping the computer to improve the interaction with the user.

References 1. Golomb BA, Lawrence DT, Sejnowski TJ (1991) SEXNET: a neural network identifies sex from human faces. In: Advances in neural information processing systems, pp 572–577 2. Moghaddam B, Yang M-H (2002) Learning gender with support faces. IEEE Trans Pattern Anal Mach Intell 24(5):707–711 3. Bekios-Calfa J, Buenaposada JM, Baumela L (2011) Revisiting linear discriminant techniques in gender recognition. IEEE Trans Pattern Anal Mach Intell 33(4):858–864 4. Lemley J, et al (2016) Comparison of recent machine learning techniques for gender recognition from facial images. MAICS 97–102 5. Eidinger E, Enbar R, Hassner T (2014) Age and gender estimation of unfiltered faces. Trans Inf Forensics Secur (IEEE-TIFS), special issue on facial biometrics in the wild 9(12):2170–2179 6. Basha AF, Jahangeer GSB (2012) Face gender image classification using various wavelet transform and support vector machine with various kernels. IJCSI Int J Comput Sci Issues 9(6, 2) 7. Gutta S, Wechsler H, Phillips PJ (1998) Gender and ethnic classification. In: Proceedings of the IEEE international automatic face and gesture recognition, pp 194–199 8. Jain A, Huang J, Fang S (2005) Gender identification using frontal facial images. In: IEEE international conference on multimedia and expo. IEEE, 4 pp 9. Wichmann FA, Graf ABA, Simoncelli EP, Butlthoff HH, Scholkopf B (2005) Machine learning applied to perception: decision-images for gender classification. In: 18th annual conference on neural information processing systems (NIPS 2004). Canada 10. Han J, Kamber M Dimensionality reduction: principal component analysis. In: Data mining: concepts and techniques, 2nd ed. Morgan Kaufmann Publishers, pp 79–80 11. Han J, Kamber M Classification and prediction: support vector machines. In: Data mining: concepts and techniques, 2nd ed. Morgan Kaufmann Publishers, pp 337–342 12. Liu L, Shen B, Wang X (2013) Research on kernel function of support vector machine. School of Electronic and Information Engineering, Beijing Jiaotong University. Received 12 May 2013; Revised 23 September 2013; Accepted 20 October 2013 13. Sathya R, Abraham A (2013) Comparison of supervised and unsupervised learning algorithms for pattern classification. (IJARAI) Int J Adv Res Artif Intell 2(2) 14. Taheri S, Mammadov M (2013) Learning the Naive Bayes classifier with optimisation models. Int J Appl Math Comput Sci 23(4):787–795 15. Peng CYJ, Lee K, Ingersoll GM (2002) An introduction to logistic regression analysis and reporting. J Educ Res 96(1) 16. Patel BN, Prajapati SG, Lakhtaria KI (2012) Efficient classification of data using decision tree. Bonfring Int J Data Min 2(1) 17. Guo G, Wang H, Bell D, Bi Y, Greer K (2003) KNN model-based approach in classification. School Comput Math

134

A. Balyan et al.

18. Tharwat A, Gaber T, Ibrahim A, Hassanien AE Linear discriminant analysis: a detailed tutorial. AI Commun 30(2):1–22. IOS Press. https://doi.org/10.3233/aic-170729 19. Srivastava S, Gupta MR, Frigyik BA (2007) Quadratic discriminant analysis. J Mach Learn Res 8:1277–1305

Applications of Raspberry Pi and Arduino to Monitor Water Quality Using Fuzzy Logic Padmalaya Nayak, Chintakindi Praneeth Reddy and Devakishan Adla

Abstract Water is one of the most important natural resources that must be monitored, analyzed continuously for safety and survival of human life. The traditional method relies on collecting water samples, testing, and analyzing the water samples in specific laboratories which is not only cost effective but also causes access latency, and delay in disseminating the information among the end users. The huge growth of wireless technology and VLSI design has brought a tremendous change in developing small microsensors that are being utilized for various monitoring applications since the last decade. In this paper, an effort has been made to measure the drinking water quality with less cost in a hardware platform with the help of some water-related Sensors, Raspberry Pi, and Arduino Microcontroller. The proposed method utilizes the Fuzzy Logic algorithm and the experimental result shows that the proposed method has many more advantages over traditional systems. It is also observed that the proposed system works effectively in a real-time environment with immediate response and less cost. Keywords WSN · Raspberry Pi · Arduino · Fuzzy logic

1 Introduction The WSNs applications provide many challenges even though these sensor nodes are very tiny, battery operated and can be deployed randomly or deterministically to monitor the environmental parameters. The applications are huge that ranges from military, civil, health care, agriculture, disaster hit areas, water quality, and many P. Nayak (B) · C. P. Reddy CSE Department, Gokaraju Rangaraju Institute of Engineering and Technology, Hyderabad 500009, Telangana, India e-mail: [email protected] C. P. Reddy e-mail: [email protected] D. Adla CSE Department, Vidya Jyoti Institute of Technology, Hyderabad 500009, Telangana, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_17

135

136

P. Nayak et al.

more applications [1]. It is quite amazing that the tiny sensor nodes are becoming the heart of the system design for sensing and monitoring the activities to bring Internet of Things (IoTs) into practice. Further, Raspberry Pi is a small device capable of communicating and processing having its own advantages with less cost and compact size [2] (Fig. 1). It can operate like desktop computer that starts from surfing from the Internet, playing high-definition video games, making spreadsheets, and more over huge support from online activities [3]. Figure 2 depicts the screenshot of Raspberry Pi that consists of a processor and graphics chip, memory, various connectors and interfaces for external devices. Raspberry Pi works like a standard PC, uses a keyboard and mouse (used for command entry), a power supply, and a display unit. It can also work like nonstandard version media server or smart TV. Ethernet cable is used for Internet connectivity. USB dongle can be used for Wi-Fi connectivity [2, 3]. Linux is a compatible software for Raspberry Pi and is used as an operating system for Raspberry Pi. There are several versions of Linux ported to the Raspberry Pi’s BCM2835 chip, that includes Debian, Fedora Remix, and Arch Linux [4]. An operating system called Raspbian is used for Raspberry Pi. There are also many non-Linux OS operating systems available [5]. Those can be used with Raspberry Pi. One of the biggest advantages of Raspberry Pi over personal computer is that it contains some GPIO ports that can be easily connected to external Fig. 1 Basic architecture of WSN

Fig. 2 Raspberry Pi

Applications of Raspberry Pi and Arduino to Monitor Water …

137

Fig. 3 Arduino UNO microcontroller

devices. In this work, Ubuntu MATE has been used for Raspberry Pi. Arduino Micro controller is another example of open source hardware on which each sensor node can be built upon with the help of the individual sensor. There are few versions of Arduino microcontroller available such as Arduino UNO, Arduino LEONARD, Arduino MEGA, and Arduino DUE, etc. In the proposed work, Arduino UNO r3 has been used. The screenshot of Arduino is shown in Fig. 3. In the proposed model, pH sensor, Conductivity sensor, and Temperature sensor has been used. Then the microcontroller is programmed to get the sensor value. Conductivity sensor model conductivity probe k 1.0 (Atlas Scientific) is used. PT-1000 temperature kit and pH probe (ENV-40-pH) is used in our experiments. The rest of the paper is organized as follows. Section 2 presents the related work and Sect. 3 discusses about the Proposed Algorithm and Fuzzy Inference Modules. Section 4 discusses experimental results followed by the conclusion in Sect. 5.

2 Related Work The applications of Raspberry Pi are proven in various fields. This section presents few applications. Raspberry Pi is a ticket size computer with an integrated RAM, a CPU and on chip graphical processing unit (GPU). In [6], a Raspberry Pi reads the inputs from sensors, store the sensed values in a database for historical trending and when the sensed value crosses a threshold point the output goes to the off state. In [7], the author describes that wireless sensors and a Raspberry Pi can do all possibilities in this world. In [8], the author discusses that how the small, powerful, and inexpensive sensor nodes can be used with Raspberry Pi. In [9], Raspberry Pi is used for healthcare monitoring system such as to measure the Blood Pressure and Sugar. Further, water quality measurement in an IoT environment is discussed in [10]. But no paper discusses about fuzzy logic for monitoring water quality. Motivated by this fact, we made an attempt to experiment with different sensors, Arduino and

138

P. Nayak et al.

Raspberry Pi for some applications like water quality and environmental monitoring. The main motto of this research paper is to develop a mobile application through which water quality can be monitored by pH and conductivity value of water at different temperature using fuzzy logic model.

3 Proposed Model In this section, we have discussed the Proposed Model, Proposed Algorithms, and Experimental Set up in detail. A. System Assumption In our experiment, three sensor nodes are deployed randomly to monitor the water quality continuously. The sensors are pH sensor, Conductivity sensor, and Temperature sensor. 1. All the sensor nodes are considered to be mobile. 2. The sensors are configured into sensor nodes through Arduino microcontroller which is a programmable device and the same data can be transmitted to the Raspberry Pi. 3. Raspberry Pi has been used as a processor which can process the data and send to the cloud. 4. Temperature sensor is used to measure the environment sensor as each time conductivity changes with respect to increase or decrease in temperature. B. System Model In the proposed model, Fuzzy Logic (FL) model is used viewing that FL model can solve uncertainties involved in higher level exist in the complex real problems. As the quality of water changes with respect to various parameters, we have considered few parameters to apply the well-known Mamdani’s Rule and checked the water quality. The basic Fuzzy Logic model is presented in Section D. The proposed Fuzzy Logic model is depicted in Fig. 4. C. Proposed Experimental Approach /* for monitoring the drinking water quality*/

Fig. 4 Example of fuzzy logic model

Applications of Raspberry Pi and Arduino to Monitor Water …

139

1. Select some parameter to measure the water quality (Here pH value and Conductivity has been selected) 2. Use some sensors based on the parameters of your Interest (Here pH sensor, Conductivity sensor, and Temperature sensor is selected) 3. Calculate a Threshold Value 4. Water is drinkable within the threshold value 5. Apply fuzzy if-then-else rules to measure the water quality /* for Noptimal Sensors */ 6. All the Sensors are capable of sensing the water-related parameters 7. Send the sensed data to Raspberry Pi through Arduino Microcontroller /* end of for */ 8. Raspberry Pi processes the data and sends the data to the Cloud 9. Fetch the data from the cloud in your Mobile Phone /* end of applications */ D. Fuzzy Logic Model Figure 4 shows the Inference techniques and the fuzzy logic model used in our proposed model. FL model comprises four components: A fuzzifier, fuzzy inference engine, fuzzy rules, and a defuzzifier. The following procedure is required to complete the process. 1. Fuzzifier: Inputs/crisp values are transferred to fuzzy set. 2. Rule Evaluation: If-Then-Else Rule is applied here. 3. Fuzzy Inference Engine: Both the crisp inputs are fed, and fuzzy if-then-else rule is applied and finally, a fuzzy inference is produced. 4. Defuzzification: Fuzzy set values can be converted into crisp value using a defuzzifier. In this approach, pH value and conductivity of the water is considered as two fuzzy inputs. Each input variable has individual membership function. The fuzzy set representing pH value and Conductivity is shown in Fig. 5a, b correspondingly. The linguistic variables for pH represent more, medium and less. Triangular membership is assumed for more, medium and less. For conductivity, the linguistic variables are more, moderate and less. Membership functions of all the input variables is provided

Fig. 5 Membership function plots. a pH value, b conductivity

140

P. Nayak et al.

Table 1 MF for input variable

PH value

Conductivity

Low

Less

Medium

Moderate

More

More

Fig. 6 Fuzzy set for output variable acceptance factor

in Table 1. For sake of demonstration, the grading of the membership function is attached with a numerical value. Rule Base and Inference Engine: In proposed system 9 rules are taken in Fuzzy Inference Technique. Here X represents pH value, Y represents conductivity and C represents the Acceptance Factor (AF). The o/p AF is taken as 5 membership functions i.e. Reject, Unsafe, Nearly Safe, Safe and Safest. Figure 6 shows the acceptance factor. We have derived the rules from the formula as used in Eq. (1). AF = Avg

m 

pH + Avg

0

m 

Conductivity

(1)

0

where, m varies from 0 to 5. For the output variable AF membership function is shown in Table 2 and the Acceptance Factor (AF) value is shown in Table 3 based on Fuzzy Rules. Table 2 MF for output variable AF

Membership functions Acceptance factor Reject (0), Unsafe (1), Nearly safe (2), Safe (3), Safest (4)

Applications of Raspberry Pi and Arduino to Monitor Water … Table 3 Fuzzy rules and value of AF

pH Value

Conductivity

141 AF for drinking water

More (2)

More (2)

Safest (4)

Medium (1)

More (2)

Safe (3)

Low (0)

More (2)

Nearly safe (2)

More (2)

Moderate (1)

Unsafe (1)

Medium (1)

Moderate (1)

Nearly safe (2)

Low (0)

Moderate (1)

Unsafe (1)

More (2)

Less (0)

Nearly safe (2)

Medium (1)

Less (0)

Unsafe (1)

Low (0)

Less (0)

Reject (0)

4 Results and Discussion A. Experimental Set Up In our experiments, we have used one Raspberry Pi, Arduino and some sensors like pH sensor, Conductivity sensor and Temperature sensor. One bread board is used for multi-point input/output connections. The pH and Conductivity sensors are used to sense the water quality and send the data to Raspberry Pi through Arduino. The complete experimental set up is shown in Fig. 7a, b. Figure 7a shows the experimental set up for drinking water whereas 7b shows the experiment set up for mud/ground water. We have implemented Fuzzy Logic Algorithm and the code has been dumped into the Raspberry Pi to measure the water quality. The output is monitored in a monitor. The same value can be uploaded to the remote cloud and it can be accessed through the Mobile Phone when the system is not in use. To avoid the connectivity between sensors and breadboards, high quality wireless senor motes can be used. B. Results and Discussions The quality of drinking water is verified through Mamdani’s Fuzzified Rule as discussed in Table 3. Two inputs such as pH value and conductivity are given as the

Fig. 7 a Experimental set up for drinking water, b for mud/ground water

142

P. Nayak et al.

Fig. 8 Temperature versus conductivity

fuzzy inputs. We have verified the AF for drinking water with 9 conditions using Mamdani’s rule through MATLAB. The same information can be fetched from the remote cloud to our mobile phone. Figure 8 shows the relationship between conductivity and environmental temperature. When temperature increases, conductivity increases proportional to the temperature. We have taken two samples of water such as mud water and drinking water. When the temperature reaches at temperature 250, the conductivity increases to 7.8. Whenever the temperature reaches below 130, conductivity decreases, and the water is not drinkable because the optimal value of conductivity for drinking water lies in between 6.7 and 7.8.

5 Conclusion The summary of the paper concludes that an efficient Wireless Sensor Network application has been developed with the help of some water-related sensors, Raspberry Pi and Arduino. Drinking Water quality can be monitored on our mobile phone by fetching the information from the remote cloud with less cost and less overhead. Two important parameters of water such as pH value and electrical conductivity has been considered and Mamdani’s rule has been applied to check the drinking water quality. The technical novelty of the paper emphasizes on Wireless Sensor Network applications that utilizes the principle of fuzzy logic control to monitor the drinking water quality. It has been proven that fuzzy logic control handles the real-time problems more accurately and effectively than any other traditional model. Acknowledgements This project is carried out under the UGC grant Ref. F. No: 4-4/2015-2016 (MRP/UGC-SERO), Oct 2016. We are thankful to the funding agency for providing support to carry out the research work at GRIET, Hyderabad.

Applications of Raspberry Pi and Arduino to Monitor Water …

143

References 1. Heinzelman WR, Chandrakasan A, Balakrishnan H (2000) Energy-efficient communication protocol for wireless micro sensor networks. In: IEEE computer society proceedings of the thirty third Hawaii international conference on system sciences (HICSS ‘00), Washington, DC. IEEE, pp 1–10 2. Schmidt M (2012) Raspberry Pi—a quick start guide. The Pragmatic Bookshelf 3. Richardson M, Wallace S (2013) Getting started with Raspberry Pi. O’Reilly, USA 4. Horan B (2013) Practical Raspberry Pi. Apress, New York 5. Raspberry Pi getting started guide. RS Components, Version 1.0, 2012 6. Home monitoring projects for Raspberry Pi. http://www.projects.privateeyepi.com/home. Accessed 22 Dec 2013 7. Bell CA (2013) Beginning sensor networks with Arduino and Raspberry Pi. Press Media 8. Ferdoush SM (2014) A low-cost wireless sensor network system using Raspberry Pi and Arduino for environmental monitoring applications. Master of Science thesis, University of North Texas 9. Nayak P, Sowmya M, Pranay J Integration of sensor node with Raspberry-Pi for health care monitoring application. In: Proceedings of international conference on advances in electronics and computer system (IAECS-2015), pp ECS-048–ECS-052. ISSRD, Bangalore, India 10. Vijaykumar N, Ramya R (2015) The real time monitoring of water quality in IoT environment. In: Proceedings of international conference on circuit, power, and computing technologies (ICCPCT), Tamil Nadu

Developing a Smart and Sustainable Transportation Plan for a Large-Sized City: An Approach to Smart City Modeling Sushobhan Majumdar and Bikramjit Sarkar

Abstract Rapid and haphazard city growth in developing countries like India leads to urban sprawl, traffic congestion, pollution, etc., which increases transportation externalities of the city. In case of a smart city, the uses of information technology are high than the old cities or any other newly developed cities. Govt. of India, has taken the decision to develop or redevelop those cities in a sustainable way. Kolkata is one of the largest cities in the eastern part of India. The demographic growth rate of Kolkata city is relatively higher than the other cities in India. The objective of this paper is to scrutinize the transport and communication scenarios of Kolkata city and what needs to be done to make Kolkata city more sustainable and smart in transport. This study reveals that most of the people prefer private taxi and bus services during their trips. Kolkata city is now facing various transportation problems like traffic congestion, illegal trafficking, unscientific construction of roads, blocked footpath by street vendors, etc. This study suggests the needs of Information and Communication Technology System (ICT), Special Traffic Management System (STMS), Intelligent Transport Management System (ITMS), etc. This type of study will throw new light on urban sustainability studies, smart city studies, etc. The probable remedies used in this study to develop a smart and sustainable transportation plan will help the Govt. officials, planners, and decision-makers to take decisions for the sustainable planning and management of the area. Keywords Transportation externalities · Smart city · Urban sustainability · Sustainability · Planning

S. Majumdar (B) Department of Geography, Jadavpur University, Kolkata, India B. Sarkar Department of Computer Science and Engineering, JIS College of Engineering, Kalyani, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_18

145

146

S. Majumdar and B. Sarkar

1 Introduction Cities are playing an important role in societal development by increasing GDP and purchasing power of a region. Cities are complex systems. Rapid and unplanned urban growth in the developing countries leads to urban sprawl, traffic congestion, pollution, etc., which affects demographic, economical, societal, and physical environmental condition of the area [1]. Smart city is relatively a new term which is related to the uses of ICT (Information and Communication Technology) and IoT (Internet of Things) [2]. In a smart city the uses of information technology are high than the old cities. Various authors like [3–5] have already discussed various aspects of smart city. Caragliu et al. [2] have also discussed about different issues related to smart city like infrastructural development, social inclusion, urban development, environmental sustainability, etc. [4] has mentioned some characteristics of smart city like political efficiency, socio-cultural development, formation of social capital, and planned development for future. Lombardi et al. [6] proposed a framework based on the concept of the Triple Helix’s concept regarding smart city development. The concept of urban sustainability and urban growth is related to the various aspects of issues like residential, industrial, transportation, commercial, or residential [7– 9]. Urban development of an area must be related to environmental sustainability in case of the smart city [10]. Kolkata city is one of the metro cities in the eastern part of India. The demographic growth rate of Kolkata city is relatively higher than the other cities in India. The objective of this paper is to scrutinize the transport and communication condition of Kolkata city and what needs to be done to make Kolkata city more sustainable and smart in transport.

1.1 Study Area Kolkata city and its surroundings area are named Kolkata Metropolitan Area. This is under the jurisdiction of Kolkata Metropolitan Development Authority (KMDA). It covers six districts in West Bengal i.e., Kolkata, Howrah, Hooghly, North 24 Parganas, South 24 Parganas, and Nadia. According to the Census 2011, the total population of KMA is 1 crore and 58 lakhs comprising an area of 1841.47 km2 . Between 2001 and 2011, the growth rate population in this area is 10.30% (Figs. 1 and 2; Table 1).

1.2 Database and Methodology This work is based on primary data and secondary data. The map of Kolkata Metropolitan Area has been collected from the Kolkata Metropolitan Development

Developing a Smart and Sustainable Transportation Plan …

147

Fig. 1 Study area

Authority office. Other than these maps transport and communication maps have been collected from the ILGUS Bhawan, KMC office, National Library etc.

2 Analysis and Discussion Kolkata city is situated on the east bank of the river Hooghly (Hugli) on the lower Gangetic delta which is part of the Bengal Basin. It stands on a thick alluvial deposit of the great antiquity. Geologists agree that Kolkata is a part of submerged sundry trees found beneath the city [11]. Figure 3 explains the growth of the population in Kolkata Metropolitan Area. Figure 4 explains various types of road transport in Kolkata city. From this, it has been seen that most of the people prefer private cars for their transportation purposes. Only few people prefer mini bus, tram, cab services, etc. From the various modes of transport in Kolkata city, it has been found that most of the people prefer bus services (45%). Because of various bus transport authorities like Calcutta State Transport Corporation (CSTC), South Bengal State Transport Corporation (SBSTC) etc. bus service is very good. Most of the authorities related to bus services have fixed

148

S. Majumdar and B. Sarkar

Fig. 2 Transport and communication in KMA

POPULATION IN MILLION

Table 1 Present composition of KMA

Area (km2 )

Units

No’s of units

Municipal corporations

3

272.32

Municipalities

39

667.28

Census towns

155

364.53

Out growth’s

6

Rural areas

394

Total KMA

–

4.68 532.66 1841.47

20 10 0 1921

1931

1941

1951

1961

1971

YEAR

Fig. 3 Growth of population in Kolkata metropolitan area

1981

1991

2001

2011

Developing a Smart and Sustainable Transportation Plan … Fig. 4 Road transport in Kolkata city

5%

4%

149 Private Cars and Taxi

6%

Bus Mini Bus

44%

12%

Tram

Auto Rickshaw

22%

Heavy Vehicles

3%

Others

4%

Two Whellers

the interval of bus services as maximum 5 min during peak hours. Except during peak hour, the interval of bus service is nearly 10–15 min. For this reason, most of the people prefer bus services. Other than these few people prefer auto-rickshaw services, suburban railway during their travel purposes (Figs. 5 and 6). Other than these problems, traffic congestion is another problem in Kolkata city. The main reason is the high growth rate of transport. For this reason during peak hours this area is totally congested as a result the movement of vehicles and passengers is slow during peak hours. Another problem in Kolkata city is that most of the footpath in Kolkata city is totally blocked by the dwellers. For this reason in some of the congested areas (like Burrabazar, Park Circus area, etc.) pedestrians have to use road while walking. Other problems related to transport in Kolkata city —illegal parking, unscientific construction of roads, footpath blocked by street vendors, and unplanned

2%

11%

Bus

2% 2%

Autorickshaw/Taxi Suburban Railway

45% 16%

Personalized Vehicle 23%

Metro Pedestrian Ferry

Fig. 5 Transportation modes in Kolkata city

Fig. 6 Average speed of vehicles in crossing point

PICNIC GARDEN SCIENCE CITY GARDEN REACH KHIDDERP ORE

BBD BAG

MOULALI

MG ROAD

SHYAMBAZ AR ESPLANAD E

BALLYGUNJ E TOLLYGUNJ E ULTODANG A MANIKTOL A

HAZRA

BEHALA

GARIA

RASHBEHA RI

PARK CIRCUS

GARIAHAT

JADAVPUR

16 12 8 4 0

150 Fig. 7 Accidents in major roads

S. Majumdar and B. Sarkar E M BYPASS BBT ROAD GARIAHAT ROAD CHITTARANJAN AVENUE JAWAHAR LAL NEHERU ROAD A.J.C. BOSE ROAD 0

Table 2 Types of traffic congestion in Kolkata

20

40

60

Types of traffic congestion in Kolkata Simple congestion Congestion due to multiple roads Bottle neck condition of old Kolkata Multiple modes of transport Unplanned traffic related congestion Unscientific construction of roads

construction of roads are the major causes which effects the transportation system of Kolkata city (Fig. 7; Table 2).

3 Recommendations There is no land use and transportation plan for Kolkata city and also in the case of Kolkata Metropolitan Area (KMA) or Kolkata Metropolitan Region (KMR). So the decision-makers and policy planners should take immediate measures to make an interlinked with transport and land use of this city to reduce the negative externalities of Kolkata city (Figs. 8 and 9). In case of transportation rules and regulations, it must be equal for all citizens. In some cases, new city models can be followed. Decision-makers and planners should formulate a plan which is focused mainly on the three aspects of sustainability i.e., physio-environmental sustainability, economical, and societal sustainability. During the formulation of future transportation plan, it must be kept in mind that quality of life should be maintained. In case of few areas, various new city models can be followed. Govt. officials should increase the tax on private vehicles and subsidy should be allotted for public vehicles. At the same time services of the public vehicles should be increased. Future transport in this megacity should focus on the electric operated and battery operated system which is eco-friendly. Incentives should be given on the commercial and housing development along the transportation routes to increase public transport. Govt. officials should focus on the non-motorized modes of transport. In case of paratransit modes of transport (Such as auto-rickshaw) Liquid

Developing a Smart and Sustainable Transportation Plan …

151

Fig. 8 Proposed model to linked up land use with transport

Fig. 9 Criteria’s for smart transport in Kolkata

Petroleum Gas (like LPG) operated auto must be used. Older roads must be merged with the major roads. Except these few measures uses of Information and Communication Technology (ICT) in transport and communication System must be followed. During peak hour or rush hours of time (8 A.M.–12 P.M. and 6 P.M.–9 P.M.), Special Traffic Management System (STMS) has to be followed. Intelligent Transport Management System (ITMS) should be developed like proper and fast auto signaling system, Passenger Information System (PIS), good parking place, integrated ticketing system (already has been launched), and air-conditioned bus stops (already started in various parts of the salt lake area) should be constructed.

4 Conclusion Kolkata is a very old city with unplanned growth. Recently in Kolkata city, the services of metro railway services have been extended and officials have sanctioned a project called ‘East West Metro’. The construction work for this project has already been completed. The major objective of this project is to connect the metro railway

152

S. Majumdar and B. Sarkar

with all the congested parts in the city which also helps by reducing traffic congestion. In some areas (basically in the peripheral areas) the services of ‘Toto’ (battery operated public transport in which maximum four persons can travel at a time) has been started. This city is recently developed toward the eastern part of from the CBD area. One new town (namely Rajarhat) has been established. Another major noticeable fact is that decentralized growth has been started in this area. Unplanned and haphazard growth is high in the peripheral areas from the city area. Most of the development in this area is mainly transport corridor oriented. The growth of the city is mainly toward the major transportation routes from the city. The proposed measures mentioned in this study will help the concerned authorities to take immediate measures. Except these Govt. officials, planners, and decision-makers develop a smart and sustainable transportation plan then it will minimize negative externalities of transport. This type of study will throw new light in the urban sustainability studies regarding the transport condition of this area which is helpful for the better planning and management of the area.

References 1. Neirotti P, De Marco A, Cagliano AC, Mangano G, Scoranno F (2014) The current trends in smart city initiatives: some stylised facts. Cities 38:25–36 2. Caragliu A, Del Bo C, Nijkamp P (2011) Smart cities in Europe. J Urban Technol 18:65–82 3. Giffinger R, Fertner C, Kramar H, Kalasek R, Pichler-Milancovic N, Meijers E (2007) Smart cities ranking of European medium sized cities. Final Report, Centre of Regional Science 4. Albino V, Berardi U, Dangelico RM (2015) Smart cities: definitions, dimensions, performance, and initiatives. J Urban Technol 22:3–21 5. Chourabi H, Nam T, Walker S, Gil-Garcia JR, Mellouli S, Nahon K, Pardo TA, Scholl HJ (2012) Understanding smart cities: an integrative framework. In: Hawaii international conference on system science (HICSS), pp 2289–2297 6. Lombardi P, Giordano S, Farouh H, Yousef W (2012) Modelling the smart city performance. Innov Eur J Soc Sci Res 25:137–149 7. Goonetilleke A, Yigitcanlar T, Ayoko G, Egodawatta P (2014) Sustainable urban water environment: climate, pollution and adaptation. Edward Elgar, Cheltenham 8. Pietrosemoli L, Monroy CR (2013) The impact of sustainable construction and knowledge management on sustainability goals. A review of the Venezuelan renewable energy sector. Renew Sustain Energy Rev 27:683–691 9. Solanki VK, Makkar S, Kumar R, Chatterjee JM (2019) Theoretical analysis of big data for smart scenarios. In: Balas V, Solanki V, Kumar R, Khari M (eds) Internet of things and big data analytics for smart generation, vol 154. Intelligent Systems Reference Library 10. Dizdaroglu D, Yigitcanlar T (2016) Integrating urban ecosystem sustainability assessment into policy-making: insights from the gold Coast City. J Environ Planning Manage 59(11):1982– 2006 11. Nair PT Calcutta in the 19th century, vol 1. Firma KLM, Calcutta

Improved Data Dissemination Protocol for VANET Using Whale Optimization Algorithm Bhoopendra Dwivedy and Anoop Kumar Bhola

Abstract In VANET, the process of data dissemination used to improve the quality of travel to avoid unwanted accidents. Many existing protocols use such type of message activities to ensure fair road safety without concentration on network congestion. As the control overhead data packet increases, the congestion on node increases; therefore the paper proposes an optimal adaptive data dissemination protocol (OADDP). This uses control overhead reduction algorithms and optimal clustering mechanism to improve overall data dissemination quality and to minimize network congestion. The Whale optimization Algorithm (WOA) is used for clustering and cluster head (CH) selection. Here, the CH node acts as a candidate (relay) node to forward data from source to destination. To diminish overhead on nodes, we use the predictor-based decision-making (PDM) algorithm. The proposed OADDP is simulated and compared with ADDP in variable traffic flow. Simulation results exhibit better performance with higher data dissemination efficiency and lower collision rate. Keywords VANET · Data dissemination · Cluster head (CH) · Candidate node · Network simulator

1 Introduction VANET is a mobile ad hoc network (MANET) but some of the characteristics of VANET may differ from MANET such as higher node and restricted rode topology [1]. VANET is similar to MANET data propagation in terms of exchanging messages between different nodes [2]. Hence, propagating data to the intended node or region of interest is a unique problem in VANETs and requires incorporating effective B. Dwivedy (B) Department of CSE, G. L. Bajaj Institute of Technology & Management, Greater Noida, India e-mail: [email protected] A. K. Bhola Department of Computer Science, Banasthali Vidyapith, Rajasthan, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_19

153

154

B. Dwivedy and A. K. Bhola

techniques to disseminate data. A vehicle can broadcast data to the other nodes but it cannot exceed the transmission range [3]. One issue in data dissemination occurs when the propagation of data is beyond the sender’s transmission range. To handle this problem flooding method is helpful [4, 5]. For effective broadcasting, each vehicle would make a decision to rebroadcast the message even in both cases: either the node is in communication range or beyond the communication range [6]. Many approaches have been proposed for data dissemination, especially for commercial advertisement dissemination and it must be provided the secure and practical incentive scheme, a secure and practical incentive scheme for commercial advertisement dissemination [7] designed by the decoding algorithm called Reed–Solomon codes (RS-codes). The number of communication between vehicles is minimized to one single broadcast communication, and no interaction between vehicles is needed. The content-based dissemination approach [8] considers the relevance of the data. An efficient and competitive broadcasting algorithm (BODYF) [9] is specifically designed for dealing with highly fluctuant networks, which reduces the communication overhead. The paper is organized as follows. The literature review is discussed in Sect. 2. The network model and methodology of OADDP method are given in Sect. 3. Construction of the proposed OADDP method is presented in Sect. 4. Results are discussed in Sect. 5 and the conclusion is in Sect. 6.

2 Literature Review The aim of any routing protocol is to search the optimum path between vehicles for communication and data dissemination purposes. There are mainly two types of approaches as divided by their nature: one is proactive and the second is reactive. Since reactive protocols are used in VANET, it depends on routing protocols and the performance of these protocols depends on parameters used in these protocols for performance evaluation [6, 8]. Different techniques are used to improve communication and data dissemination in the VANET environment. Liu et al. proposed two 1-D lattice methods using network coding [10]. One technique like cloud-helped security message scattering plan (CMDS) is used by Liu et al. [10]. VMaSC-LTE hybrid architecture using the 3GPP/LTE system is proposed by Ucar et al. [11]. Vehicles situated in a particular communication range may form a cluster and in this cluster, CH are identified with a heuristic approach so some paper considers the prefetchbased data dissemination method using RSU [12]. Under the assumption that the routes and data request information of vehicle subscribers are readily available, and two algorithms are devised to determine how to pre-fetch a set of data from a data center to roadside wireless access points. Lin et al. [13] have proposed a mutually enhanced company-wise spatial time division multiple access (STDMA) scheduling plan. The two key classes of message streams are used to describe the capacity of inside substance streams. The dispersal of basic occasion driven message streams must meet stringent postpone limitations.

Improved Data Dissemination Protocol for VANET …

155

3 Methodology with Network Model 3.1 Methodology Oliveria et al. [14] have proposed an adaptive data dissemination protocol which is a multi-hop broadcasting protocol. This method enables the candidate selection mechanism from neighboring nodes. The vehicles are predicted to facilitate consumers with infotainment-rich, protected, dependable, driving knowledge through the intelligent transportation system. Moreover, the increasing amount of messages is a considerable issue that reduces the performance of any constrained systems resources. Furthermore, the possibility of obtaining the negative impacts of exchanging messages frequently in the high-density network. Our contributions are as follows: 1. Cluster head (CH) selection is based on parameters like energy consumption, delay, cooperation rate, and congestion rate. 2. Responsibility of CH is to communicate data between vehicles. 3. The frequently exchange control OH messages (beacon) are reduced by the PDM algorithm. Finally, simulations were done to predict the effectiveness of OADDP mechanism with ADDP [15].

3.2 Network Model of OADDP Mechanism We assume that vehicles are equipped with sensors, onboard units (OBUs), location of the node and roadside units (RSUs). The network system is incorporated of vehicles outfitted with OBUs and RSUs as Fig. 1 shows. Here, the short information about RSU is given that RSU is acted as an intermediate node that is utilized for transferring data between the vehicles for effective communication. Nodes in cluster forward traffic information to CH through intra-routing manner and it forwards to BS using neighboring CHs.

4 Clustering Using Whale Optimization Algorithm (WOA) Consider various particles (vehicles) are arbitrarily set into motion through the space and note that the fitness value of own and their neighbors thus emulate effective neighbors by attracting them. Clustering of particles uses semiautonomous herds or every one of the particles can belong to a single global flock. Clustering begins with gathering arbitrary particles. Let Initial best solution = Pbest And Global best = gbest So any particle in population is represented as

156

B. Dwivedy and A. K. Bhola

Fig. 1 Network model

   If fitness( pk ) < fitness pbestik

(1)

pbestk , otherwise     pk, If fitness( pk ) < fitness gbestik

(2)

 pbestk = gbestk =

pk,

gbestk , otherwise

4.1 CH Selection Step 1: Initialization Here the weight value represents the random value wi (i = 1, 2, 3, …, n) where n shows the quantity of weight value. And also initialize the coefficient vectors of whale such as a, A, and C. Step 2: Fitness Calculation Evaluate the fitness based on the condition (2) Step 3: Encircling prey

Improved Data Dissemination Protocol for VANET …

157

Best search agent is calculated, the updated method is indicated by the accompanying conditions:     U = C · wbest (t) − w(t)

(4)

w(t + 1) = wbest (t) − A · U

(5)

where t denotes a current iteration, A and C denotes a coefficient vector, wbest denotes a position vector for best solution, w denotes a current position vector, and | | denotes an absolute value. The vectors A and C are computed as follows: A = 2 a · r − a

(6)

C = 2 · r

(7)

where a is directly reduced from 2 to 0 throughout emphases (in both investigation and exploitation stages) and r is an irregular vector in (0, 1). Step 4: Repeat until better fitness values are achieved.

4.2 Control OH Reduction Using Predictor-Based Decision-Making (PDM) Algorithm At first, every vehicle that enters any group coordination zone sets its status as vehicle naturally. At that point, it should sit tight for τ second, on the off chance that it does not get vehicles ad (CHA) message; at that point it changes the status to vehicles and begins intermittently to forward CHA message. On the off chance that the CH gets the CHA message, it remains as vehicle and answers with just a single HELLO message. At the point when the CH gets all answers from the vehicles inside its related lifetime, the BS is competent to figure the following best CH before leaving the group. Consider network scenarios shown in Fig. 2, for exchanging the VHELLO message; initially, when vehicle-1 enters the bunch zone, at that point it ought to send VHELLO message; and when the vehicle-3 leaves the group zone, it sends another VHELLO message. When the vehicles 2, 4, 5, and 6 are in the group zone then do not require sending any HELLO message. Here, we confront an issue as far as CH change because of CHA misfortune. Therefore, we utilize the indicator based basic leadership (PDM) calculation to anticipate CH change before it happening. PDM calculation is used to process the LT edge of CH vehicles in light of measurements, for example, VS , position and speed. The cluster metrics are separately collected

158

B. Dwivedy and A. K. Bhola

Fig. 2 HELLO message when enters and leaves the cluster in existing method

from all CHs in the order of position, velocity, and Vs . Then average values of these parameters will be calculated. Algorithm 1: OADDP Input: Number of population, control variables, and termination level Output: CH selection and OH message reduction 1 for iteration =1 2 compute best and worst solution 3 compute boundary condition 4 if f(xr)< f(xi) 5 direction = xr-xi 6 else 7 direction = xmax-xi 8 end if 9 end for 10 for each vehicle 11 compute vehicle strength (Vs) 12 Compute CH = max (Vs1, Vs2…) 13 end for 14 for each CH vehicle 15 initialize position, velocity, and Vs 16 compute relative difference 17 end for 18 if relative difference >> threshold 19 pass trigger signal 20 else 21 no action 22 end if

Improved Data Dissemination Protocol for VANET … Table 1 Parameters

159

Parameters

Values

Vehicles

100, 200, 300, 400, 500, and 600

Average speed

100 kmph

Area

3000 × 3000 m2

Transmission range urban environment

300 m

Transmission power urban environment

0.98 mW

Transmission range highway environment

256 m

Packet size

2048 byte

5 Results and Setup Using Network simulator (NS-2), Our proposed algorithm (OADDP) uses 5 × 5 network scenarios. The number of vehicles is varied by 100, 200, and 400 and vehicle speed ranges from 10 to 50 kmph. The simulation parameters are given in Table 1. The effectiveness of the proposed OADDP mechanism is compared with performance metrics such as success rate, number of control messages, propagation distance, end-to-end delay, and data dissemination efficiency.

5.1 Experiments on Performance Analysis in Variable Traffic Flow In the following plot, we will do performance analysis with new and old methods in various traffic flow conditions. Figure 3 shows the number of collisions. Figure 4 shows the comparative data dissemination efficiency with existing and proposed method.

6 Conclusion In this paper, we have proposed an optimal adaptive data dissemination protocol (OADDP) for VANET road safety. The proposed algorithm uses the Whale Optimization technique for clustering and predictor-based decision-making (PDM) algorithm for control overhead messages reduction. The selected cluster head (CH) node acts as a candidate node and is able to forward the message between source nodes to the destination node in the network. The simulation result shows that our algorithm

160

B. Dwivedy and A. K. Bhola

Fig. 3 Collision rate versus traffic flows

Fig. 4 Dissemination efficiency in variable traffic flows

reduces the collision rate and improves the data dissemination efficiency in different traffic flow and vehicle density situations.

References 1. Schoch E, Kargl F, Weber M (2008) Communication patterns in VANETs. IEEE Commun Mag 46(11):119–125 2. Lin X, McKinley P, Ni L (1995) The message flow model for routing in wormhole-routed networks. IEEE Trans Parallel Distrib Syst 6(7):755–760

Improved Data Dissemination Protocol for VANET …

161

3. Hu C, Chen M (2005) Adaptive multichannel data dissemination: support of dynamic traffic awareness and push-pull time balance. IEEE Trans Veh Technol 54(2):673–686 4. Misener J (2007) EDITORIAL: intelligent transportation systems and safety: innovation and directions. J Intell Transp Syst 11(3):105–106 5. Wisitpongphan (2007) Broadcast storm mitigation techniques in vehicular ad hoc networks. IEEE Wirel Commun 14(6):84–94 6. Vinel A, Campolo C, Petit J, Koucheryavy Y (2011) Trustworthy broadcasting in IEEE 802.11p/wave vehicular networks: delay analysis. IEEE Commun Lett 15(9):1010–1012 7. Tseng F, Liu Y, Hwu J, Chen R (2011) A secure reed-solomon code incentive scheme for commercial Ad dissemination over VANETs. IEEE Trans Veh Technol 60(9):4598–4608 8. Cenerario N, Delot T, Ilarri S (2011) A content-based dissemination protocol for VANETs: exploiting the encounter probability. IEEE Trans Intell Transp Syst 12(3):771–782 9. Ruiz P, Dorronsoro B, Bouvry P, Tardón L (2012) Information dissemination in VANETs based upon a tree topology. Ad Hoc Netw 10(1):111–127 10. Liu B, Jia D, Wang J, Lu K, Wu L (2017) Cloud-assisted safety message dissemination in VANET–cellular heterogeneous wireless network. IEEE Syst J 11(1):128–139 11. Ucar S, Ergen S, Ozkasap O (2016) Multihop-cluster-based IEEE 802.11p and LTE hybrid architecture for VANET safety message dissemination. IEEE Trans Veh Technol 65(4):2621– 2636 12. Kim R, Lim H, Krishnamachari B (2016) Prefetching-based data dissemination in vehicular cloud systems. IEEE Trans Veh Technol 65(1):292–306 13. Lin Y, Rubin I (2017) Integrated message dissemination and traffic regulation for autonomous VANETs. IEEE Trans Veh Technol 66(10):8644–8658 14. Oliveira R, Montez C, Boukerche A, Wangham M (2017) Reliable data dissemination protocol for VANET traffic safety applications. Ad Hoc Netw 63:30–44 15. Liu F, Chen Z, Xia B (2016) Data dissemination with network coding in two-way vehicle-tovehicle networks. IEEE Trans Veh Technol 65(4):2445–2456

A Novel IoT-Based Approach Towards Diabetes Prediction Using Big Data Riya Biswas, Souvik Pal, Nguyen Ha Huy Cuong and Arindam Chakrabarty

Abstract Big data is a modern teamster of today’s economical world. Data are being digitalized in today’s world as imperative judgment is taken by Big data analytics. In our manuscript, we have discussed about Big data analytics in IoT ecosystems and its implications in healthcare. Healthcare is concerned now a days and big data is holding all the supportive hands in IoT-based healthcare systems. In healthcare, we have discussed about Diabetes Mellitus which is a non-communicable disease. This paper deals with the proposed system of diagnosis of diabetes. Hence it is assertive that we do some surveys on how we can manage to handle large data files, technologies are defined and also predictions of diabetes through IOT sensor and management have been discussed. Keywords Big data · Hadoop · Map reduce · HDFS · Pig · HIVE · HBase · IoT

1 Introduction In current surroundings data is generating from multiple origin. These data is of multiple diversity. This bulk of enormous data is considered as Big Data. Big data and IoT are the buzz words now days. It is used to express cumbrous bulk of structured and unstructured data. Some characteristics of Big data being discussed [1, 2]. R. Biswas (B) Department of Computer Science & Engineering, JIS College of Engineering, Kalyani, India e-mail: [email protected] S. Pal Department of Computer Science & Engineering, Brainware University, Kolkata, India e-mail: [email protected] N. H. H. Cuong College of Information Technology, The University of Da Nang, Da Nang, Vietnam e-mail: [email protected] A. Chakrabarty Department of Management, Rajiv Gandhi University, Itanagar, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_20

163

164

R. Biswas et al.

The amount of data can be blunder, malleable and scalable data which are processed through technologies used called Hadoop, Map Reduce, HIVE, Pig. The production of volume of data is rapidly enlarging every year due to occurrence of newly technologies, accessory and communication [3]. In the modern milestone of smart phones and wearable devices, endless sum of health data folder of patient from different challenges continue featured by healthcare industry [4]. Mostly confusion begin where system process through heterogeneous data sets [5, 6]. In healthcare sector a leading non-communicable disease (NCD) is Diabetes Mellitus. There are basically three stages of type 1 diabetes, type 2 Diabetes, Gestational diabetes. Diabetes is a scheme of metabolic diseases consist of high blood sugar levels concluded lengthy season a numerous operation supported on Internet Of Things been developed for management of diabetes and it composite of physical objects [7, 8]. IOT is mostly a model for interconnecting sensor which does tracking, sensing, processing and diagnosing, coming up with a enclosed device and detector which can link up and also exchange content beyond the internet [9, 10]. In this paper, we are going to discuss literature survey of the related work in the Sect. 2. Section 3 deals with the architecture of the proposed diabetes diagnosis system, proposed algorithm, and sequence diagram of the algorithm.

2 Literature Survey In this section, we have discussed on literature survey of the background study. Chavan et al. [9] have discussed about Big data is a word which defines massive and convoluted set of data. Some technologies like Hadoop, HDFS, Map Reduce, Pig, Hive, HBase being used. Khan et al. [11] have expressed a proposed data life cycle which utilize employ the technologies and nomenclature of Big data management, investigating and scarceness. Nizam and Hassan [12] have discussed that it is tough to operate with Big data resolving management traditional dataset. Chen et al. [13] have discussed that Initially generic background of Big data is inform then study about the connected technologies i.e. cloud computing, Internet Of Things, data centers and Hadoop. Archenaa and Mary Anita [14] have deliberate about the approach of how we expose newly expose surplus value from the data autogenic by healthcare and government. Prasad et al. [15] have discussed that diabetes is one of the leading non-communicable disease Mellitus. This system will prophesy searching algorithm in Hadoop/Map Reduce. Huzooree et al. [16] has explains that Diabetes Mellitus (DM). The goal of this paper is to ecumenical review centering on recent glucose projection model is declared depending on the rating to performing data analytics in wireless body area in network system. Kumar and Pranavi [17] has discussed that the important function is providing dilution healthcare by modern application such as Big data and cloud. A ecumenical survey is made on diabetes dataset with random forest (RF), SVM, k-NN, CART and LDA algorithms. Joaheer and Nagowah [18] have explained Telemedicine, Electronic Health Records (EHR) and social media. This paper also describes the repung of Big data and also it proposed architecture for

A Novel IoT-Based Approach Towards Diabetes Prediction …

165

diabetes Mellitus to predict patient with chronic disease in maturius. Saravana kumar et al. [19] have discussed that the unstructured nature of lifecycle from healthcare of Big data This paper analyzing algorithm in Hadoop/Map Reduce is used for prediction of diabetes type, hindrance. Al-Taee et al. [20] has discussed that the self-management of diabetes by IOT based podium. A completely practical model system is created, achieved, point-to-point function is approved successful.

3 Architecture of Proposed Diabetes Diagnosis System This section describes the architecture of diabetes diagnosis system that analyzes the various Data and initially it accumulated data from numerous devices and it is initially stored in an unstructured or semi-structured format. Initially data should be digitized to stem EHR as well as data are smart devices, research and development SNM data repository which is begin captured by existent technologies and used to redirect those data to centralized database for anatomy. The data are gathered for processing in Hadoop data system then data will be accumulated by apache flume. Apache flume is used here which is a item of hadoop ecosystem. Then the data will be pushed to Hbase by agents for further processing (Fig. 1). The outputted data moves to HIVE it is a business application running in SQL queries against a hadoop cluster. It uses then map reduce. Map reduce has two tasks

Fig. 1 Architecture for our algorithm

166

R. Biswas et al.

data is splitted and passed into mapping function for produce output values. Next data can use spark which is a frame work that is same as hadoop which provides opensource platforms and can be used by anyone. Then the outputted data are aggregated in hadoop data manager where YARN is used i.e. it is a centralized platform. Then implementing various machine learning algorithms such as RF, LDA, CART and K-NN for prediction it also learn specific data and the test absolute model which will be back to hadoop and hadoop data system will completely send it back to end user tool. Here it initially identify the course of people tolerate from diabetes registry by working healthcare analytics to big data technology for identifying the diabetes.

3.1 Proposed Algorithm This section deals with the proposed algorithm of the diabetes diagnosis system. Step 1: Step 2:

Step 3:

Step 4:

Step 5: Step 6: Step 7:

Step 8: Step 9:

Initially data assembles from various sources like EHR, smart device and sensors devices and research and development and SNM data respitory. Hadoop is a framework which permits for distributed processing of enormous data set. It is a framework which has a capability for stocking and considering data which are prevailing in various machines. It also service map reduce which permits for diving the query into limited chunk and achieve them in co-ordinately. Initialized data’s from various sources need to be delivered to hadoop data system to process the data where the data’s are accumulated by using apache flume, then data moves to HIVE which run SQL queries then data’s are place down to map Reduce for summarization. After that spark is used which furnish a open-source platform. The processed data’s aggregated from hadoop data system need to be managed, so to manage the data are implemented in hadoop data manager where YRAN is used to add new features to the hadoop. It is a centralized platform used for Resource Management. The outputted values of managed data are a switched back to hadoop data system. The outputted data should be evaluated so Hadoop data system will generate machine learning algorithm. Hadoop ensures the appropriate algorithm for the data does evaluated according to their category, from the set of algorithm to determine the appropriate data pattern and lining the data for earning prediction. Outputted data tested by Hadoop and draws one specialized model of algorithm and learned the specific model. Outputted data are for specific algorithm switch back to hadoop data system.

A Novel IoT-Based Approach Towards Diabetes Prediction …

167

Fig. 2 Sequence diagram for our proposed algorithm

Step 10: Once, the above steps are done hadoop data system will switch over to end user tools.

3.2 Sequence Diagram Initially data are accumulated from multiple sources. Data are being processed by hadoop data system and managed by hadoop data manager and outputted managed data are swapped back to hadoop data system. The hadoop data system will generate machine learning algorithm. According to the kind of data it will choose specific model and the outputted model will reversed back to hadoop data system. After completion of above steps hadoop will swap back to end user tools. In this section, Fig. 2 is expressing that the data sources which are being assembled from various devices are processed and managed. It also analyzes various algorithm and choose appropriate algorithm to learned specific model to predict diabetes.

4 Analysis In our work, we have built the need of predicting techniques to measures the diabetes unlike the traditional models which doesn’t provide enough efficiency, accuracy and fastest delivery. This technique possesses several data from EHR, R&D and other sources like smart devices. By using existing technologies, it is possible to capture and send to a centralized database for analysis. Also, unlike most of the other proposed works, we used to gathered data from various devices and processed the data in hadoop data system and then processed data are being managed by hadoop data management and additionally it also applied machine learning algorithm such as RF, LDA, CART and K-NN provoked by hadoop. The machine learning algorithm has main benefits over the most other techniques as it provides more accurate throughput to user and it gains the performance rate of the model.

168

R. Biswas et al.

Fig. 3 Compares the traditional approach with the proposed approach shows accuracy will increase as data sources with increase

In our proposed work, considering all factors, we can say that the efficiency, accuracy and fastest delivery medical care at lower cost. This can be compared with the traditional approach by depicting them in the form of graphs for both the traditional approach and our proposed approach. It is depicted that data that we are getting from multiple sources are used to predict the diabetes. So, here accuracy perform a efficient role. Figure 3 compares the data sources with the accuracy as the data sources will increase accuracy throughput will also increase in proposed approach and decrease in traditional approach. Figure 4 is expressing the comparison of cost of traditional approach as compared to proposed approach the cost will decline in proposed approach.

5 Conclusion Peoples are engaged in today’s world in the feverish slots and not pickings any care of their own health, starring to difficulties of continuing disease such as diabetes. In this paper, a recent framework is suggested that utilize. This framework will analyze and predict diabetes Mellitus and providing way to improve healthcare complexity and delivering earliest potential working. As well as this framework is operate for self-treatment and also in future providing faster medical care within a chip costs. In this paper, it also provides many various machine learning algorithms such as RF, SVM, CART, LDA and K-NN to predicting data patterns. The frame work is working currently under development.

A Novel IoT-Based Approach Towards Diabetes Prediction …

169

Fig. 4 Compares the traditional approach cost with the proposed approach and display the increase in cost in traditional approach and decrease in proposed approach

References 1. Joshitta S (2016) Applications of big data analytics for diagnosing diabetic mellitus: issues and challenges. Int J Recent Trends Eng Res 02:454–461 2. Emani CK, Cullot N, Nicolle C (2015) Understandable big data: a survey. LE2I UMR6306, CNRS, ENSAM, Univ. Bourgogne Franche-Comté, F-21000 Dijon, vol 17, pp 70–81 3. Honest N, Patel A (2016) A survey of big data analytics. Int J Inf Sci Tech (IJIST) 6(1/2) 4. Pooja M, Das D (2016) Comparative analysis of IoT based healthcare architectures. Int J Appl Eng Res 11(20):10216–10221. ISSN 0973-4562 5. Gaitanou P, Garoufallou E, Balatsoukas P (2014) The effectiveness of big data in health care: a systematic review. In: Closs S et al (Eds) MTSR 2014, CCIS 478, pp 141–153 6. Thara DK, Premasudha BG, Ram VR, Suma R (2016) Impact of big data in healthcare: a survey. In: 2016 2nd international conference on contemporary computing and informatics (IC3I), Noida, pp 729–735 7. Azzawi MA, Hassan R, Abu Bakar KA (2016) A review on internet of things (IoT) in healthcare. Int J Appl Eng Res 11(20):10216–10221. ISSN 0973-4562 8. Deshkar S, Thanseeh RA, Menon VG (2017) A review on IoT based m-health systems for diabetes. Int J Comput Sci Telecommun 8(1) 9. Chavan V, Rajesh, Phursule N (2014) Survey paper on big data. Int J Comput Sci Inf Technol (IJCSIT) 5(6):7932–7939 10. Alelyani S, Ibrahim A (2018) Internet-of-things in telemedicine for diabetes management. In: 2018 15th learning and technology conference (L&T), Jeddah, pp 20–23 11. Khan N, Yaqoob I, Hashem IAT, Inayat Z, Kamaleldin Mahmoud Ali W, Alam M, Shiraz M, Gani A (2014) Big data: survey, technologies, opportunities, and challenges. Sci World J (Hindawi Publishing Corporatione) 2014:18 pages 12. Nizam T, Hassan SI (2017) Big data: a survey paper on big data innovation and its technology. Int J Adv Res Comput Sci 8(5):2173–2177 13. Chen M, Mao S, Liu Y (2014) Big data: a survey. Mobile Netw Appl 19:171–209

170

R. Biswas et al.

14. Archenaa J, Mary Anita EA (2015) A survey of big data analytics in healthcare and government Proc Comput Sci 50:408–413 15. Prasad ST, Sangavi S, Deepa A, Sairabanu F, Ragasudha R (2017) Diabetic data analysis in big data with predictive method. In: 2017 international conference on algorithms, methodology, models and applications in emerging technologies (ICAMMAET), Chennai, pp 1–4 16. Huzooree G, Khedo KK, Joonas N (2017) Glucose prediction data analytics for diabetic patients monitoring. In: 2017 1st international conference on next generation computing applications (NextComp), Mauritius, pp 188–195 17. Kumar PS, Pranavi S (2017) Performance analysis of machine learning algorithms on diabetes dataset using big data analytics. In: 2017 international conference on Infocom technologies and unmanned systems (trends and future directions) (ICTUS), Dubai, pp 508–513 18. Joaheer R, Nagowah SD (2017) A big data framework for diabetes in Mauritius. In: 2017 international conference on Infocom technologies and unmanned systems (trends and future directions) (ICTUS), Dubai, pp 126–132 19. Saravana kumar NM, Eswari T, Sampath P, Lavanya S (2015) Predictive methodology for diabetic data analysis in big data. Published by Elsevier B.V. 50:203–208 20. Al-Taee MA, Al-Nuaimy W, Al-Ataby A, Muhsin ZJ, Abood SN (2015) Mobile health platform for diabetes management based on the Internet-of-Things. In: 2015 IEEE Jordan conference on applied electrical engineering and computing technologies (AEECT), Amman, pp 1–5

Technical Solutions to Build Technology Infrastructure for Applications in Smart Agricultural Models Nguyen Ha Huy Cuong, Souvik Pal, Sonali Bhattacharyya, Nguyen Thi Thuy Dien and Doan Van Thang

Abstract The 4.0 revolution has played an increasingly important role in boosting the economy at the macro and micro levels in Vietnam. Among them is the development of smart agriculture (high-tech agriculture) and sustainable tourism development which is receiving much attention from the community. The pervasive and very timely nature of the 4.0 revolution has occupied a high proportion of conference forums from the government and scientists to the farmers in the remote areas to the key priority areas in economic development. The composition of industry 4.0 can be found, including 4 basic pillars: AI (artificial intelligence), Big Data (big data), IoT (Internet of Things), and Network and Cloud Infrastructure (Infrastructure Network and Cloud Computing). This paper focuses on providing technical solutions to Cloud Computing, specifically the technical solutions related to the cost of operating the entire smart agricultural economy 4.0. Cloud Computing, with a system of physical servers arranged on a global scale and interconnected through a telecommunications system, plays an increasingly important role in operating, delivering, and declaring resource extraction and sharing. By surveying existing and developing farms, it is observed that smart agricultural models of households need to apply high technology; providing resources for these models need technical solutions. Ensuring that it is accurate and timely, it is required to avoid overlapping and conflicting requirements N. H. H. Cuong (B) College of Information Technology, The University of Danang, Da Nang, Vietnam e-mail: [email protected] S. Pal · S. Bhattacharyya JIS College of Engineering, Kalyani, India e-mail: [email protected] S. Bhattacharyya e-mail: [email protected] N. T. T. Dien QuangNam College of Economics and Technology, Tam K`y, Vietnam e-mail: [email protected] D. Van Thang Industrial University of Ho Chi Minh, Ho Chi Minh City, Vietnam e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_21

171

172

N. H. H. Cuong et al.

from users, especially distant resources. It has been observed that a strong, reliable solution, as the basis for the development of control systems within the cloud, is not yet available. Keywords Cloud Computing · Cluster · Cluster focus · Control systems

1 Introduction In today’s world, network infrastructure and Cloud Computing are indispensable factors in the operation and development of IoT applications. In 2017, according to the assessment of the Asian Cloud Computing Association (Cloud Asia), resource provision, an important service in Cloud Computing, is becoming a major demand in scientific applications. Most of the existing Cloud Computing infrastructures in Vietnam have been providing services with many problems and challenges as stated below: 1. Firstly, the cost of translation includes services provided through remote data centers and is built on virtual servers rented from abroad, so the price is very high. 2. The availability of human resources to ensure the operation of the entire system is insufficient and unsecured. 3. Cloud Computing resources often appear as the only access point for all Cloud Computing servers, so there is no guarantee for users about consistent time and configuration. 4. Cloud centers have not built service provision infrastructures that must have superior features of flexibility in scalability, security recovery, and network congestion. In recent years, with the attention of the government, the whole country, from big cities to remote areas of the country and in almost every home, everyone is interested in the revolution of technology 4.0. Through the actual survey, we can see that with the advantage of the country as an agricultural economy associated with services, the localities have also made resolutions to promote smart agriculture, agriculture 4.0. Some plant varieties were also planted in farms from medium to several tens of hectares, mainly Pomelo trees. With the advantages of discussion, it is very suitable for Pomelos that has high growth and productivity after several years of planting and caring. However, the problem faced by the cooperative members (when planting Pomelo and other fruit trees) is to apply high technology in cultivation such as automatic irrigation model. In this paper, we offer solutions to support virtual server services and provide effective solutions in providing resources for remote needs (farmers) and virtual server systems. The solution meets the following specific requirements: – Technical solution to ensure anti-conflict resources – Technical solution to avoid congestion

Technical Solutions to Build Technology Infrastructure …

173

– Technical solution to ensure reduced price for users – Technical solution for private cloud – Architecture for models that provide heterogeneous resources. This article is organized as follows: Section 2 presents some related work. Our proposed algorithm is presented in Sect. 3. Section 4 in turn presents the experimental setup on and compares the experimental results. Section 5 concludes the paper.

2 Related Work The Information system, indistinguishable relation current solutions, often using the input mechanism for virtual machines that require resources such as Walsh’s [1] study and colleagues suggest two classes using utility functions, through context. Automatic and autonomous scenes; while author Song [2] and colleagues use adaptive controllers to automatically adjust the use of virtual resources to meet customer satisfaction. Song [2] proposed an algorithm to provide adaptive resources with a policy of giving priority to previous resource request registration processes. Meanwhile, Qiang [3] and colleagues proposed a solution of automatic multidimensional resource provisioning and a solution to reduce the responsibility for the agent nodes (NA) with this model that solved the limitations of scattered nodes or clusters. However, the author [3] also points out the drawbacks of the inability to coordinate actions together for the whole system to operate efficiently, which is a combination that cannot be installed. In the Mark Stillwell study [4] and the publication of research on providing infrastructure services based on open source Cloud Computing system, proposed optimal scheduling algorithm, and consideration of the ability to allocate spirit Activation allows maximum use of physical server resources. Meanwhile, Xiao [3] and his colleagues proposed a load balancing policy with a smart resource management mechanism. The author [5] argues that existing resource delivery systems are processor resources that depend on cores and resource management processes. The solution in [5], proposed through these strategies, focuses on resource-based processes that rely on process management to be flexible and responsive to resource requirements effectively. It can be seen that there are many different approaches in the direction of resource supply research, the proposed solutions have advantages but still problems and difficulties exist. Therefore, there is no solution that best meets the quality of service for virtualized resource users. In the Cloud Computing environment with multiple Data Centers (Data Centers), the Data Centers are distributed across all geographic surfaces. These virtual server centers are pooled from physical servers connected through a networked environment built on a distributed and mixed hardware platform. The study of technical solutions to provide resources based on the virtual server system on the heterogeneous dispersion platform is also interested in research by domestic and world researchers.

174

N. H. H. Cuong et al.

3 Proposed Resource Allocation Algorithm In this section, we have discussed our Resource Allocation Algorithm. Proposed Resource Allocation Algorithm Input: W[ci ; mi ; ni ; di ;] Output: Resource Allocation ti 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

Do Job.Insert (t i ) k ← 0; max_k = input While (Read.isfree() = true do Ready.Insert (Job.delete(t i )) J-cur ← Readly.delete(t i ) J-cur.Status = Assigned for each j ∈ Ri do; If(usedbyi [j] = 0) send request (t i ) to pj end for Else if response = Success Return result Else wait (2k++ ) if (Timeout Or k = max_k) Return fail While (True)

4 Experimental Results of Resource Allocation Algorithm In this section, we have discussed experimental results in Table 1. We require creating 6 virtual machines with the results obtained as shown. The Table 2 depicts the result of successfully creating 7 virtual machines required through Cloudlet, represented by VM ID 1 to VM ID 7. Starting time is 0.1 ms and ending 85 ms for virtual machine VM ID 1 and 175 ms for VM ID 7 virtual machines with many different scenarios such as: we increase the number of virtual machines required, as well as allocate physical servers on multiple D data centers different materials. In Fig. 1, in test case 2 after 4 virtual machines were created in the second center, next to the testing center 3 created 2 virtual machines. In the third test center 2 also created 4, then the center 3 created 2, the center 4 created 1. With the limited Table 1 Required time in between starting and destroying VM Required time Available CPU p (%)

10

20

30

Non create VM and No Wait/For

23.95

45.65

160

Non create VM and used Wait/For

14.05

28.40

78.46

Create VM

15.21

23.27

60.10

Technical Solutions to Build Technology Infrastructure …

175

Table 2 Required time in between starting and destroying VM Required time Available CPU p (%)

10

20

30

Non create VM and No Wait/For (%)

0.15

0.37

10.7

Non create VM and used Wait/For (%)

0.46

1.26

6.09

Create VM (%)

5.46

7.40

55.9

Fig. 1 Time contract graph with the ability of each CPU to provide virtual server creation resources

capacity of the laptop we tested, it can create up to 8 Virtual machines with 4 data centers.

5 Conclusion Whenever there is an interaction between agents, the active agent checks to make sure the system does not perform a transition to a bad state. When the request is sent to an active agent, it predicts its next state. There is a predictable state, the active agents check in the list, to find out whether the state is forecasted in the subsists. Currently in the world data centers use virtual server technology to increase both in research and deployment of Cloud Computing services application. So research and development of algorithms to provide resources in the virtualization platform is one of the most prominent issues. In subsequent studies, we will delve into research applications and demonstrate the accuracy of this algorithm to suggest additional preventive measures and prevent deadlock from occurring. The next direction of the paper is to conduct research on solutions based on available algorithms that propose solutions to prove the effectiveness of the solution in modeling language.

176

N. H. H. Cuong et al.

References 1. Walsh WE, Tesauro G, Kephart JO, Das R (2010) Utility functions in autonomic systems. In: Proceeding international conference on autonomic computing, 17–18 May 2004, pp 70–77 2. Song Y, Sun Y, Shi W (2013) A two tiered on demand resource allocation mechanism for VM–based data centers. Int J IEEE Trans Serv Comput 6(1):116–129 3. Xiao Z, Song W, Chen Q (2013) Dynamic resource allocation using virtual machines for Cloud Computing environment. Int J IEEE Trans Parallel Distrib Syst 24(6):1107–1117 4. Knapp E (1987) Deadlock detection in distributed database systems. ACM Comput Surv 19(4):303–327 5. Singhal M (1989) Deadlock detection in distributed systems. IEEE Comput 22:37–48 6. Darabant AS, Databant L (2011) Clustering methods in data fragmentation. Rom J Inf Sci Technol 14(1):81–97 7. Cuong NHH, Solanki VK, Van Thang D, Thuy NT (2017) Resource allocation for heterogeneous Cloud Computing. Netw Protoc Algorithms 9(1–2):71–84 8. Still M, Vivien F, Casanova H (2012) Virtual machine resource allocation for service hosting on heterogeneous distributed platforms. In: Proceeding of the 2012 IEEE 26th international parallel and distributed proceeding symposium (IPDPS’12), Washington, DC, USA. IEEE Computer Society, pp 786–797 9. Stillwell M, Schanzenbach D, Vivien F, Casanova H (2013) Resource allocation algorithms for virtualized service hosting platforms. Int J Parallel Distrib Comput (JPDC) 70(9):962–974 10. Li Q, Hao Q, Xiao L, Li Z (2009) Adaptive management of virtualized resources in Cloud Computing using feedback control. In: Proceedings of the 1st information science and engineering (ICISE), 26–28 Dec 2009. IEEE, pp 99–102

Edge Detection Through Dynamic Programming in Ultrasound Gray Scale Digital Images Anju Mishra, Ramashankar Yadav and Lalan Kumar

Abstract Ultrasound medical pictures area unit are widely utilized since ultrasound might be a non-obtrusive and non-ionizing nosology system. Edge detection technique is generally utilized for any segmentation or extra estimations of components within the image. Edges speak to high frequency segments of an image. Initially we will convert ultrasound images into digital images through images processing technique. Then we propose dynamic programming procedure for locally discovering edges in gray level of ultrasound digital images. The DP algorithms method the for ultrasound images to urge the minimum accumulative value matrix to tracing a world best edge. Dynamic programming methodology and a threshold inclination administrator procedure are being utilized for edge discovery on each initial and noisy form of this gray level digital ultrasound image. Keywords Ultrasonography · B-Mode · Doppler · Robert’s operator

1 Introduction Edge detection is that the characteristic focuses in an exceedingly digital image at that the image brightness will converted pointedly, termed as edges. Edges represent high frequency segments of an image. They’re wide utilized since ultrasound may be a non-invasive and non-ionizing medically diagnostics method. Perception of human body structures including ligaments, joints, vessels, muscles and inward organs has been done through ultrasound strategy. Ultrasound are often used for imaging, detection, measure and cleaning. Most of the edge detection algorithms are optimized with classic techniques i.e., (1) Bacterial A. Mishra (B) · R. Yadav · L. Kumar G.L. Bajaj Institute of Technology and Management, Greater Noida, India e-mail: [email protected] R. Yadav e-mail: [email protected] L. Kumar e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_22

177

178

A. Mishra et al.

Foraging Algorithm commonly known as BFA, (2) Sobel Edge Detection and (3) SUSAN Edge Detection. Bacterial Foraging works on swarm intelligence and Sobel utilizes Image Gradient for Edge detection.

2 Ultra Sound Images Ultrasonography might be a demonstrative imaging method that is utilized for visualizing figure structures together with ligaments, vessels, joints, muscles and inside organs. Ultrasound could likewise be a cyclic quick power for every unit area wave with a recurrence bigger than the furthest reaches of the human hearing shift. Ultrasound is along these lines not isolated from “normal” (capable of being heard) sound upheld varieties in physical properties, exclusively the very actuality that people can’t hear it. Ultrasound devices work with frequencies from twenty rate up to a several giga cycle. Limiting the depth of the penetration by using higher frequencies will increase the resolution of the image. For conversion of ultrasound images we have to go through different phases of ultrasound image acquisition Phase of Image Acquisition.

3 Digital Conversion of Ultrasound Images Ultrasound pictures will contain a lot of noise content—particularly speckle noise. During the stages of Image acquisition noise can be added in any respect. • The noise are often introduced throughout the beam forming method and additionally throughout the signal procedure stage. • Even throughout the Scan conversion, there may well be loss of information because of the interpolation. The fine details and edge definition get degraded due to speckle noise and limits the refinement goals by making it troublesome to watch little and low distinction lesions in body. Distinctive hub and horizontal channels are connected on the signals at totally distinguished stages to boost the image quality. The handled signals would then be lean to scan conversion module as output lines, which will can genuinely create the image from the sweep lines by completing a geometrical mapping. Echo signals are amplified by ultrasound scanners and make amends for attenuation losses. Mistreatment filtering that’s connected to the centre frequency provide noise reduction and alternative advantages such as information measure of the transmitted pulse. The Noise reduction procedures in Ultrasound pictures will be generally isolated to 2—During obtaining and after securing. The speckle noise will be diminished by wavelet-change, multi-look preparing, spatial sifting or homomorphic separating and so forth. Speckle decrease by spatial

Edge Detection Through Dynamic Programming …

179

and homomorphic separating is performed on image when its obtaining. Ultrasound image quality can even be enhanced throughout the procurement by • Using high quality transducers • mistreatment gel to reach and to maintain a strategic distance from air holes among body and electrical gadget • mistreatment pertinent hub and sidelong channels while doing the flag procedure • mistreatment adjustive addition methods though doing the scan conversion. B-mode ultrasound frameworks utilize digital scan conversion (DSC) to show learning as an exceedingly organize framework non heritable in an exceedingly coordinate group.

4 Edge Detection Edges are defined as any points wherever image brightness changes pointed square measure generally sorted out into a collection of sinuous line segments. Edge detection alludes to the method of distinguishing and finding sharp discontinuities in a picture. Three stages in edge detection are: • Image smoothing: for noise decrease these means include sifting the picture for enhancing the execution of edge locator. • Detection: separating all edge focuses that are square measure potential possibility to wind up edge point. • Edge confinement: choosing from the hopeful edge focuses solely the focuses that are genuine individuals from set of focuses containing an edge. Edge detection is tough in howling pictures, since both the noise and the edges incorporate high-frequency quintessence. Endeavors to scale back the noise outcome square measure obscured and mutilated edges.

4.1 Gradient Based Edge Detection Detection of perimeters by searching down the most and least within the calculation of the image done by the gradient technique. If the gradient is higher than the edge, that means there’s an object within the image. The popular edge detection administrators square measure Roberts, Sobel, Prewitt.

180

A. Mishra et al.

4.2 Sobel’s Operator A convolution mask is utilized is once in a while inexhaustible humbler than the genuine picture. The model underneath demonstrates the spread being fell over the most noteworthy surprising astounding left part of the input image tended to by the green format. The I and J esteems are won’t to move the record pointer therefore you can multiply, for instance, pixel (A22) by the comparing veil esteem (M22). It is necessary to note of that pixels in the first and last lines, just as the first and last segments can’t be controlled by a 3 × 3 cover. Input Image

+ Psuedo-Convolution Mask

A11

A12

A13

…

A1k

A21

A22

A23

…

A2k

A31

A32

A33

…

A3k

:

:

:

:

:

=

M11

M12

M13

M21

M22

M23

M31

M32

M33

Output Image

B11 B12 B13 …

B1k

B21 B22 B23 …

B2k

B31 B32 B33 …

B3k

:

:

:

:

:

The above figure outlines the operating of sobel pseudo convolution masks once it’s connected to the information given information image. The sobel pseudo mask plays out the accompanying activity to compute the point B22 of the picture: B22 = (A11 ∗ M11) + (A12 ∗ M12) + (A13 ∗ M13) + (A21 ∗ M21) + (A22 ∗ M22) + (A23 ∗ M23) + (A31 ∗ M31) + (A32 ∗ M32) + (A33 ∗ M33) The veils is connected one by one to the input image, to make separate estimations of the incline segment in each presentation (call these Gx and Gy). The gradient magnitude is given as: |G| =



(Gx2 + Gy2 )

(1)

Typically, an approximate magnitude is registered utilizing |G| = |Gx| + |Gy|

(2)

Edge Detection Through Dynamic Programming …

181

The angle of orientation of the edge offering ascend to the spatial angle is given by  = arctan(Gx/Gy)

(3)

For this situation, orientation 1 is interpreted as meaning that the bearing of most extreme contrast from dark to white keeps running from left to directly on the image, and different edges are estimated anticlockwise from this.

4.3 Robert’s Operator 2-D spatial gradient estimation on a image has been performed by Roberts Cross operator. The administrator involves of a strive of 2 × 2 convolution kernels. One kernel is basically the other turned by 90°. This can be terribly almost like the Sobel operator. The kernels are often applied on an individual basis to the input image, to create distinct estimations of the gradient component in every orientation (Gx and Gy). The gradient magnitude is given as: |G| =



(Gx2 + Gy2 )

(4)

Typically, associate degree inexact extent is registered utilizing |G| = |Gx| + |Gy|

(5)

The angle of orientation of the sting offering ascends to the abstraction gradient is given by:  = arctan(Gx/Gy) − 3π/4

(6)

4.4 Canny’s Edge Detection Canny operator is relies on three criteria. The fundamental set up uses a Gaussian operate to sleek image first off. As it were, the two with sensational adjustment of gray- scale (solid edge) and focuses with slight modification of grayscale (weak edges) compare to the second spinoff zero-intersection point. Therefore these 2 thresholds square measure accustomed notice sturdy edges and weak edges. Canny edge detector utilizes Gaussian convolution, to sleek the image, which unfold the Gaussian and controls the level of smoothing.

182

A. Mishra et al.

g(m, n) = G(m, n) ∗ f(m, n)

(7)

    Gσ = 1/ 2π σ 2 exp −(m2 + n2 )/2σ2

(8)

The maxima and minima of the primary spinoff gradient square measure constant because the zero crossings of the second directional spinoff. M(m, n) =

 gm2 (m, n) + gn2 (m, n)

(9)

Only the maxima crossings square measure of interest as a result of these pixels define the territories of the most keen force changes within the image. Ridge pixels has been gone through, a two-threshold technique or physical phenomenon, to work out the ultimate arrangement of edges.

5 Proposed Edge Detection Technique Dynamic programming technique has been opted for classifying the characteristic functions, that is planned from that the most effective “edge function” for every window. The windows are chosen by initial decaying the whole picture into equivalent square areas and after that dividing into four equivalent sub squares any locale where the standard fit of the foremost effective edge perform (as estimated by the optimality foundation) falls beneath a tough and fast resilience, etc. The methodology is outlined in a theoretical calculation, and the strategy is initially represented by application to a gray scale ultrasound digital images of a counterfeit structure. Second, for examination, the dynamic programming strategy and a split-and-union methodology are the unit utilized for image improvement for a rambunctious variation of a muscle cell culture image.

6 Dynamic Programming for Images To detail the edge following technique as dynamic programming, one must define an evaluation function that embodies the notion of the “best edges”. Then one conceivable basis for a “decent edge” is a weighted aggregate of high combined edge quality and low total shape; that’s, for an acquainted degree n-segment bend, h(x1 , . . . , xn ) =

n  k=1

S(xk ) + α

n−1 

q(xk , xk+1 )

k=1

where the implicit constraint is that consecutive x k ’s must be grid neighbor:

(10)

Edge Detection Through Dynamic Programming …

xk − xk+1  ≤

183

√

2

  q(xk , xk+1 ) = diff ∅(xk ), ∅(xk+1 )

(11)

where α is negative. The function g we take to be edge strength, i.e., g(x) = s(x). Notice that this evaluation function is in the form of and can be optimized in the stages: h(·) = h 1 (x1 , x2 ) + h 2 (x2 , x3 ) + h 3 (x3 , x4 )

(12)

f 0 (x1 ) ≡ 0

(13)

f 1 (x2 ) = max[s(x1 ) + αq(x1 + x2 ) + f 0 (x1 )] x1

f k (xk+1 ) = max[s(xk ) + αq(xk + xk+1 ) + f k−1 (xk )] x2

(14)

These equations can be put into the steps: Algorithm: Dynamic Programming for edge detection 1. Set k = 1 2. Consider only x such that s(x) ≥ T. For each of these x, define low curvature pixel “in front of “ the contour direction. 3. Each of these pixels may have a curve emanating from it. For k = 1, the curve is one pixel in length. Join the curve to x that optimizes the left hand side of the recursion equation. 4. If k = N, pick the best fN − 1 and end. Else, set k = k + 1 and move to step a pair of. This algorithm can be generalized to the case of picking a curve emanating from x (that we have already generated. Find the end of that curve, and join the best of three curves emanating from the end of that curve. Figure 1 shows this process. The equations for general case are f 0 (x1 ) ≡ 0  fl (xk+1 ) = max s(xk ) + αq(xk , t (xk+1 )) + fl−1 (xk )] xk

(15)

where the curve length n is related to α by building sequence n(l) such that n(1) = 1, n(L) = N , and n(1) − n(l − 1)

(16)

is a member of {n(k)|k = 1, . . . , l − 1}. Likewise, t (X K ) is a function that extricates the tail pixel of the curve headed by X K .

184

A. Mishra et al.

Fig. 1 Dynamic programming optimization for edge selection

7 Lower Resolution Evaluation Function In the Dynamic Programming formulation, the components g(X K ) and q(X K , X k+1 ) in the evaluation function is limited; the variable x for successive s and q are in fact constrained to be grid neighbors. This need to be the case: The x can be exceptionally removed from one another without adjusting the fundamental system. Furthermore, the function g and q need not be local gradient and absolute curvature, respectively, yet can be any capacities characterized on allowable x. Template functions, signified by g(x), measure how well a part of model matches a part of image at the point x. Relational functions, denoted by qkj (x, y), measure how well the position of the matches of the kth part at (x) agrees with the position of the match of the jth part at (y). T (x) ≡ Template centered at x computed as an aggregate of a set of chest radiographs g(xk ) =

 T (x − xk ) f (x) |T || f | x

θ xk , x j = expected angular orientation of xk from x j







q xk , x j = θ xk , x j



yk − y j − arctan xk − x j



This method was formalized using lower resolution objective function.

(18)

Edge Detection Through Dynamic Programming …

185

8 Generalization to Higher—Dimensional Image Data The generalization to higher-dimensional spaces is straight forward. In these images the notion of the gradient is the same (a vector depicting the greatest gay level alter and its comparing course), however the natural elucidation of the relating edge component might be troublesome. In three dimensions, edge elements are primitive surface elements, isolating volumes of contrasting gray scales. The objective of contour following is to link together neighboring surface elements with high gradient modulus esteems and similar orientations into bigger limits.

9 Conclusion In this paper, we proposed the dynamic programming approach for edge detection in ultrasound images. The ability of the proposed edge detector was incontestable in a very dynamically ever-changing atmosphere manufactured from a group of digital grayscale ultrasound images. First we convert the ultrasound images into digital images through digital scan converter (DSC). After that digitally converted ultrasound sample images has applied dynamic programming approach for edge detection. The windows are disintegrating the whole picture into equivalent square locales and after that breaking into four equivalent sub squares any district where the standard of work of the best edge work. The calculation reacted to the progressions by producing sub square examples as indicated by the dispersion of the recently made edges. It conjointly established to be hearty since even a BFA, Sobel, SUSAN and watchful of a littler size could recognize the edges, despite the fact that the amount of identified edge pixels was decreased and expectedly quicker than these edge detection procedure.

Impact of Heterogeneous IoT Devices for Indoor Localization Using RSSI Bhagwan Sahay Meena, Sujoy Deb and K. Hemachandran

Abstract Internet of Things (IoT)-based indoor localization technology is making a profit in various appearances and standards. A lot of works are being done and are going on indoor localization by many students and research scholars from various parts of the world and no adequate procedure has been proven for indoor localization, unlike the outdoor localization that uses the GPS satellite technology. In this following paper, we used a combination of smartphones and BLE devices for indoor localization using the RSSI values of the BLE and smartphones Bluetooth. We have considered three cases, in the first case, we only placed the BLEs in a room at random locations and retrieve the RSSI values from each BLE. Similarly, in the second case, we only placed the smartphones at random locations of the room, and in the third case, we placed both the BLEs and smartphones at random locations in the same room and retrieve their RSSI values. The simulation is done for two scenarios, without noise (no user interference/empty room) and with noise (with user interference). The row data collected from each device are stored in a database and the average RSSI value of each device has been considered in this implementation. Using this average RSSI value, we calculate the distance of each device and passed this distance matrix to MDS algorithm to calculate coordinates of each device, and finally applied transformation to find the coordinates of each device in three-dimensional form and compare it with the original coordinates manually taken during the placement of the devices. Keywords Internet of things · Bluetooth low energy · Smartphone · MDS · Received signal strength indicator

B. S. Meena (B) · S. Deb · K. Hemachandran Assam University Silchar, Assam, India e-mail: [email protected] S. Deb e-mail: [email protected] K. Hemachandran e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_23

187

188

B. S. Meena et al.

1 Introduction In the past current years, IoT-based smart environment system is introduced to replace the manual system of various domains. The system consists of various sensors, RFID, PIR, BLE, etc. [1]. The idea of scattered devices devoted to a single motive application has been substituted with machine-to-machine communication and Internet of Things (IoT) for general-purpose applications [2]. One of the predominant petitions of IoT is localization, i.e., the technique for determining the position of a device/object connected to the IoT network. The widely used satellite technology like GPS is nearly purposeless in indoor localization and numerous contrasting methods for indoor localization have been initiated in the past years [3, 4]. The data emerging from various IoT devices also known as smart data are enormous in nature. Therefore, real-time processing and capturing of these data are challenging tasks. The linked stream Big Data processing mechanisms play a vital role in real-time data processing and capturing of IoT smart data [5]. Semantic interoperability is used to exchange the smart data between Iot devices in a significant way [6]. Indoor localization is a network that can be used to determine the position of a device/object placed inside a building, shopping mall, a room, etc. where GPS is not appropriate for the task. GPS is a straightforward procedure for outdoor localization network. Although for GPS system, a huge amount of energy is required and the implementation cost is very high for a large network. The line of sight between GPS satellite and receiver become fails for indoor localization fails, the maximum accuracy provided by the GPS system is only up to 5 m [7]. The concept of IoT-based indoor localization is extending at a fast rate, chiefly due to the improvement in the smartphone industry and the establishment of large number of smartphone features and latest IoT devices like BLE, ZigBee, Wi-Fi, etc. [8]. Localization algorithm is cleaved into two categories, proximity based and distance based. Proximity-based localization algorithm only contemplates the information apropos the proximity to distinct nodes in the network, whereas the distance-based localization algorithm contemplates distance between the nodes. To calculate the distance between the nodes, various ranging techniques can be applied such as Angle of Arrival (AoA), Received Signal Strength Indicator (RSSI), Time of Arrival (ToA), and Time Difference of Arrival (TDoA) [9–14]. The relative distance can be used in methods like triangulation, trilateration, ISO-MAP, MDS-MAP, etc. to determine the relative coordinates and finally applying Helmert transformation technique to map the relative coordinates with the original coordinates [15, 16] out of which RSSI is the most reconnoitered technique since no extra hardware is required. MDS-based indoor localization can be used to determine the coordinates of the nodes in 3D which gives better accuracy [17]. In [18], a comparative study between BLE and Wi-Fi technology has been practically experimented in different conditions. The experiment was performed using trilateration for both indoor and outdoor environments including both LoS and nonLoS conditions. Experiment results compare the propagation characteristics between BLE and Wi-Fi in order to determine which technology provides better accuracy for

Impact of Heterogeneous IoT Devices …

189

a localization system. From the experimental results, it was clear that the RSSI value measured from BLE provides better localization accuracy than Wi-Fi technology. Similarly in [19], RSSI-based trilateration technique was implemented to compare between ISM868 and ZigBee for localization. Both the technologies end up resulting in poor accuracy for localization. In [20], a comparison between RFID and BLE is demonstrated in an outdoor environment for locating an object. In this paper, RSSI along with the path loss model is used for implementing trilateration for localization of an object. Results concluded that BLE provides better localization accuracy than RFID. In large-scale user tracking can be carried out with a considerable number of people using the GPS system in a scenario [21] or by using the sensors and applications installed in smart devices or smartphones of a user [22]. GPS can be used for outdoor localization of a user. Accessibility has made possible for tracking a user for indoor localization. Bluetooth Low-Energy (BLE) beacons-based indoor localization is an optimistic procedure especially in position-based services (PbS) applications. It is suitable for mobile devices for wide range and it is cost-effective. In [23], an optimized way for determining the location of BLE and RFM have been discussed. In [24], sophisticated adaptive propagation model is proposed, in which the parameters of the path loss model with a particle filter are modeled. In [25], a combination of channel-separate polynomial regression model is proposed for determining the distances between the receiver and the beacons, and the results show better accuracy as compared to the simple model. The RSSI-based localization is mostly implemented through simulations, where RSSI influenced with noise is modeled with a conventional allotment in the simulation scenarios [26]. This paper aims to demonstrate the localization accuracy for a hybrid model where both BLEs and smartphones are randomly placed in a room. RSSI measurements are collected for two scenarios with user and without user interference. The simulation is done using well-known localization algorithm MDS [27] and hence applying the transformation technique to map the simulation output coordinates with the original coordinates to determine the accuracy of the mixed devices for indoor localization. The results also compare the accuracy of BLEs and smartphones for both scenarios. It also compares the accuracy for all three cases considered, that is, both only BLEs, only Smartphones and both BLEs and Smartphones as the path loss value is changed for both scenarios. The paper is further divided into following sections: Sect. 2 gives brief description about the problem definition. Section 3 describes the detailed setup and implementation of experiment. Section 4 gives the simulation results obtained. Section 5 concludes the paper and Sect. 6 throws light for the future works.

190

B. S. Meena et al.

2 Problem Definition The following paper explains the impact of indoor localization using RSSI values of BLE and smartphones. First, the BLEs and smartphones are randomly placed in a room and the RSSI values are collected and kept. Using this RSSI values, the distance of each device is calculated and is passed to a well-known localization algorithm like MDS. MDS takes pairwise distance while making the distance matrix and if the distance is larger than the RSSI value, it may contain some error. In this research work, real-time simulation was done under two scenarios with user interference and without user interference. The simulation was done once with only BLEs, once with only smartphones, and lastly taking with combination of both BLEs and smartphones. The simulation results show the difference in the localization error as the value of the path loss n is changed for each case. The value of n varies from 2 to 4.

3 Experimental Setup and Implementation 3.1 Experimental Setup The hardware and software used for the experiment are mentioned below: 1. Hardware – BLE devices, – Smartphones, and – Raspberry Pi 3. 2. Software – Matlab 2017a. Figure 1 shows the experiment environment where the simulation was done and Fig. 2 shows the hardware used.

3.2 Implementation The experiment was performed inside a classroom of size 10 m × 7 m in two scenarios without user and with user interference. First, the Raspberry Pi was placed in the middle of the room. The Raspberry Pi acts as the centralized system running a Linux-based OS where the RSSI values from different nodes are received by executing a command. BLEs and smartphones were placed in specific locations of the room as shown in Fig. 3. First, RSSI values were collected from all the nodes without

Impact of Heterogeneous IoT Devices … Fig. 1 Experimental environment

Fig. 2 Hardware used

Fig. 3 Node placement in the room

191

192

B. S. Meena et al.

user interference. Similarly, the RSSI values from smartphones were collected by switching OFF the BLE devices and the same was done for the BLEs by switching the smartphone’s Bluetooth OFF. The same procedure was followed for collecting the RSSI values for each case in the second scenario by making two persons walk around the room, thus making an interference. A total of 600 RSSI measurements are collected 300 for each scenario and store the row RSSI values in a database.

3.3 RSSI to Distance Translation Localization algorithms are based on the distance between each pair of devices. Each RSSI measurement collected needs to be converted into the distance and it is the most exacting part of the localization algorithm. The RSSI values are measured in decibel milliwatt (dBm). The relationship between power in decibel milliwatt and power in watt is given as follows: RSSI = 10 log10

Pw [dBm] 1 mW

(1)

Therefore, the relation between the distance and power is given as RD = −c − 10n log10 D D = 10

−c−R D 10n

(2) (3)

where c is the power measured from 1 m, n is a constant for path loss which ranges from 2 to 4 (2 for free space), and R D is the RSSI value measured at distance D. Using Eqn. 3, the distance between each pair of devices can be calculated to form the distance matrix D, which will be the input to the MDS algorithm.

3.4 MDS Algorithm – The classical MDS algorithm involve some linear algebra. The algorithm is based on the fact that the coordinate matrix X can be derived by eigenvalue decomposition of the scalar product matrix A XX . The issue of deriving A from the proximity matrix D is solved by multiplying the squared proximities with the matrix K = I − n−1 11 . The procedure is known as double centering. The classical MDS algorithm consists of the following steps: 1. Compute the squared proximity matrix D2 . 2. Apply double centering: A = − 21 KD2 K using matrix K = I − n−1 11 , where I is the identity matrix and n is the number of devices.

Impact of Heterogeneous IoT Devices …

193

3. Extract the first three largest positive eigenvalues d1 , d2 , d3 of matrix A and the corresponding eigenvectors v1 , v2 , v3 . 4. Finally, the coordinate matrix X is derived by multiplying eigenvalues and eigenvectors.

3.5 Transformation Technique Transformation technique is required to map the local coordinates with the global coordinates. Helmert transformation is used by the localization algorithms and it consists of seven parameters [16]. – Three rotation parameters: It is used to align the local coordinate axis with the global coordinate axis. – Three translation parameters: It is used to map the origins of the two coordinate axis by moving the points along three coordinate axis. – One scaling parameter: It is used to scale the local coordinates with their global coordinates. Let Xl,i be the local coordinate of a device i and Let Xg,i be the global coordinate of a device i. A minimum of four anchor nodes are required for Helmert transformation. By increasing the number of anchor nodes, the transformation accuracy increases [28]. The following equation is used to transform the local coordinate to global coordinate: (4) C  = T + RoS X where X is the local coordinate of a device and C  is global coordinate of a device, Ro represents the rotation parameter, S represents the scaling parameter, and T represents the translation parameter. The transformation algorithm consists of the following steps: 1. Calculate the centroid nodes of local and global coordinates. Let X l and X g be the centroid nodes of the local and global coordinates. Therefore, X l and X g are calculated as follows: Xl =

n 1 X l,i n i=1

(5)

Xg =

n 1 X g,i n i=1

(6)

2. Calculate the new coordinates for the devices. The following equations are used to find the new coordinates using the centroid nodes as follows:

194

B. S. Meena et al.  X l,i = X l,i − X l

(7)

 = X g,i − X g X g,i

(8)

3. Calculate the rotation matrix. Let Rom be a matrix and is defined as follows: Rom =

n    T (X g,i )(X l,i )

(9)

i=1

Let v1 , v2 , v3 and e1 , e2 , e3 be the eigendecomposition derived from RomT ∗ Rom , therefore the rotation matrix is calculated using the following equation:  Ro = Rom ∗

1 1 1 √ e1 e1T + √ e2 e2T + √ e3 e3T v1 v2 v3



4. Calculate the scaling factor. The scaling factor is calculated using the following equation:  n  2 i=1 ||X gb,i || S = n  2 i=1 ||X loc,i ||

(10)

(11)

5. Calculating translation matrix T and global coordinate matrix C  . Translation matrix t is calculated using the following equation: T = X g − S RoX l

(12)

Finally, for each device point X in the local coordinate the corresponding global coordinate C  is computed using the equation: C  = T + RoS X

(13)

4 Simulation Results Figure 4a, b shows the average RMSE error for each case in both the scenarios. In case of the smartphone, the RMSE error gets increased as the value of n increases in the first scenario (without user interference) but in the second scenario (with user interference) the RMSE error remains constant and is less than that of the BLEs and the combination of BLEs and smartphones. In case of the combination of BLEs and smartphones, the RMSE is slightly less than in case of only BLEs and the value in constant as the value n is increased in the first scenario, but in the second scenario the error is more in case of only BLEs as compared to the combination of both devices.

Impact of Heterogeneous IoT Devices …

(a) Average RMSE error without user interference.

195

(b) Average RMSE error with user interference.

Fig. 4 Average RMSE error

(a) Mean error without user interference.

(b) Mean error with user interference.

Fig. 5 Mean error

Figure 5a, b shows a comparison of the mean error for both the scenarios. Here, in the first scenario, the mean error is less in case of only smartphones compared to the other two cases only BLEs and the combination of both devices. The mean error is more and is constant in case of only BLEs but less than the mean error for the combination of both BLEs and smartphones which is also constant as the value of n is increased. In the second scenario, the mean error for smartphones remains constant as the value of n is increased but less than the other two cases. The mean error gets increasing as the value of n increases in case of only BLEs and the combination of both the devices. Figure 6a, b shows the difference in the values of variance and Fig. 7a, b shows the difference in the values of standard deviation of the mean error in both the scenarios

196

B. S. Meena et al.

(a) Variance without user interference.

(b) Variance with user interference.

Fig. 6 Variance

(a) Standard deviation without user interference.

(b) Standard deviation with user interference.

Fig. 7 Standard deviation

and the values of variance and standard deviation are more for the second scenario when compared to the first scenario as the value of n is increased.

5 Conclusion As per the simulation results discussed above, the localization error is more in case of BLEs and the error is less in case of smartphones for both the scenarios. This is because there are more fluctuations reason being the multipath propagation due to object (walls) in the RSSI values of BLE as compared to that of the smartphone RSSI values. While comparing both the scenario, the error is less if there are no user interferences (idle mode) and the error is more with user interference (presence

Impact of Heterogeneous IoT Devices …

197

of two users) as the signal changes its pathbreaking the line of sight giving huge differences in the RSSI values as compared to the first scenario. In some cases, the error is increased or decreased as the value of the path loss n is increased.

6 Future Works In this paper, RSSI-based indoor localization has been implemented using IoT devices such as BLE and smartphone and also a hybrid approach of abovementioned IoT devices has been taken into consideration and their result parts are also evaluated. After, the result evaluation, we have concluded that still some more IoT devices and their hybrid-combination-based indoor localization are big areas of research.

References 1. Meena BS, Hemachandran K (2019) An experimental approach for understanding internet of things (IOT) for corporate model and various platforms. Int J Innov Technol Explor Eng 8 2. Mainetti L, Patrono L, Vilei A (2011) Evolution of wireless sensor networks towards the internet of things: a survey. In: 2011 19th international conference on software, telecommunications and computer networks (SoftCOM), pp 1–6 3. Gu Y, Lo A, Niemegeers I (2009) A survey of indoor positioning systems for wireless personal networks. IEEE Commun Surv Tutorials 11(1):13–32, First Quarter 2009. https://doi.org/10. 1109/SURV.2009.090103 4. Zahid F, Rosdiadee F, Mahamod I (2013) Recent advances in wireless indoor localization techniques and system. J Comput Netw Commun 2013:1–12 5. Sivadi I, Solanki V, Vinit G, Thirumaram M (2019) Performance analysis of linked stream big data processing mechanisms for unifying IoT smart data. https://doi.org/10.1007/978-981-138461-5_78 6. Sivadi B, Solanki B, Vinit G, Thirumaram M (2019) A survey on semantic approaches for IoT data integration in smart cities. https://doi.org/10.1007/978-981-13-8461-5_94 7. Zafari F, Papapanagiotou I, Devetsikiotis M, Hacker TJ (2017) An iBeacon based proximity and indoor localization system. arXiv:1703.07876 8. Stojkoska BR, Kosovi´c IN, Jagušt T (2017) How much can we trust RSSI for the IoT indoor location-based services? In: 2017 25th international conference on software, telecommunications and computer networks (SoftCOM), Split, pp 1–6. https://doi.org/10.23919/SOFTCOM. 2017.8115543 9. Bahl P, Padmanabhan VN (2000) RADAR: an in-building RF-based user location and tracking system. In: Proceedings IEEE INFOCOM 2000. Conference on computer communications. Nineteenth annual joint conference of the IEEE computer and communications societies (Cat. No. 00CH37064), Tel Aviv, Israel, vol 2, pp 775–784. https://doi.org/10.1109/INFCOM.2000. 832252 10. Cheng X, Thaeler A, Xue G, Chen D (2004) TPS: a time-based positioning scheme for outdoor wireless sensor networks. In: IEEE INFOCOM 2004, Hong Kong, vol 4, pp 2685–2696. https:// doi.org/10.1109/INFCOM.2004.1354687

198

B. S. Meena et al.

11. Karalar TC, Rabaey J (2006) An RF ToF based ranging implementation for sensor networks. In: IEEE international conference on communications, June 2006, vol 7, pp 3347–3352 12. Kułakowski P, Vales-Alonso J, Egea-Lopez E, Ludwin W, Garcia-Haro J (2010) Angle-ofarrival localization based on antenna arrays for wireless sensor networks. Comput Electr Eng 36(6):1181–1186 13. Kumar P, Reddy L (2009) Distance measurement and error estimation scheme for RSSI based localization in wireless sensor networks. In: Fifth IEEE conference on wireless communication and sensor networks (WCSN), Dec 2009 14. Niculescu D, Nath B (2003) Ad hoc positioning system (APS) using AOA. In: IEEE INFOCOM 2003. Twenty-second annual joint conference of the IEEE computer and communications societies (IEEE Cat. No. 03CH37428), San Francisco, CA, vol 3, pp 1734–1743. https://doi. org/10.1109/INFCOM.2003.1209196 15. Chaurasiya VK, Jain N, Nandi GC (2014) A novel distance estimation approach for 3D localization in wireless sensor network using multi dimensional scaling. Inf Fusion 15:5–18 16. Watson GA (2006) Computing helmert transformations. J Comput Appl Math 197(2):387–394 17. Meena BS, Hemachandran K, Laskar SH (2016) A study on effects of multi-dimension reduction on three-dimensional wireless sensor network localization. In: Information and communication technology for competitive strategies, March 2016 18. Zhao X, Xiao Z, Markham A, Trigoni N, Ren Y (2014) Does BTLE measure up against WiFi? A comparison of indoor location performance. In: Proceedings of the 20th European wireless conference, May 2014, pp 1–6 19. Touvat F, Poujaud J, Noury N (2014) Indoor localization with wearable RF devices in 868 MHz and 2.4 GHz bands. In: Proceedings of the IEEE 16th international conference e-health networking, application and services (Healthcom), Oct 2014, pp 136–139 20. Fernandez-Llorca D et al (2016) Comparison between UHF RFID and BLE for stereo-based tag association in outdoor scenarios. In: Proceedings of the 6th international conference on IT convergence and security (ICITCS), Sept 2016, pp 1–5 21. Zeng J, Li M, Cai Y (2015) A tracking system supporting large-scale users based on GPS and G-sensor. Int J Distrib Sens Netw, First Published 27 May 2015. https://doi.org/10.1155/2015/ 862184 22. Tsitsigkos A, Entezami F, Ramrekha T, Christos P, Panaousis EA (2012) A case study of internet of things using wireless sensor networks and smartphones 23. Stojkoska BR (2016) A taxonomy of localization techniques based on multidimensional scaling. In: 2016 39th international convention on information and communication technology, electronics and microelectronics (MIPRO). IEEE, pp 649–654 24. Zuo Z, Liu L, Zhang L, Fang Y (2018) Indoor positioning based on bluetooth low-energy beacons adopting graph optimization. In: MDPI, 4 Nov 2018 25. Cinefra N (2013) An adaptive indoor positioning system based on bluetooth low energy RSSI. PhD thesis, Politecnico di Milano, Milan, Italy, January 2013 26. Zhuang YY, Li J, Qi Y, El-Sheimy L (2016) Smartphone-based indoor localization with bluetooth low energy beacons. Sensors 16 27. Wickelmaier L (2003) An introduction to MDS. In: Sound quality research unit, Aalborg University, Denmark 28. Deakin RE (1996) 3D coordinate transformations. Surv Land Inf Syst 58:223–234

Indoor Localization-Based Office Automation System Using IOT Devices Bhagwan Sahay Meena, Ramin Uddin Laskar and K. Hemachandran

Abstract In the Internet of Things, each device communicates with each other through the Internet, in other words it is a communication between devices over a network to perform a particular task for which it is programmed. Such systems are also called automation systems. The Automation systems are those systems which perform a task without any human efforts. In existing automation systems, they are only able to detect the presence of human in a room and accordingly control the appliances. Such systems are quite good for a single user. But not reliable and moreover not suitable for multi-users. Here, the system cannot localize a person therefore it can switch ON appliances that are not required resulting again in electricity wastage. To overcome this problem there must be a system that can track a person in room and control each appliances independently. In this paper, the proposed system will first localize a person and then control the appliances. To localize a person an Indoor Localization Technique is used. This technique is used to find the position or area of a person inside a room. To implement such a technique some PIR sensors and RFID tags can be used by recognizing their patterns. Keywords Home-office automation system · Indoor localization · Radio-frequency identification (RFID) · Passive infrared (PIR) sensor · Global positioning system (GPS)

B. S. Meena (B) · R. U. Laskar · K. Hemachandran Assam University Silchar, Assam, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_24

199

200

B. S. Meena et al.

1 Introduction The IoT is responsible for extending Internet connectivity beyond standard devices such as smartphones, desktop, laptop, and tablet to any range of non-Internet-enabled mechanical devices. The IoT makes an interface between Internet-enabled devices and non-Internet-enable devices (i.e., sensors, motors, etc.). In paper [1], they proposed a system that is based on door permission technique with monitoring and controlling of home appliances. In paper [2], they implemented an automation system based on Wi-Fi technology when there is a technology like remote control or radio connection are used, devices are need to be within a particular range. In paper [3], they used Wi-Fi for communication and Raspberry Pi as server system. In this system user can interact with system using web-based interface over Internet and can control home appliances through smartphones. In paper [4], the proposed system is cloud-based system to store data coming from different sensors and controlling appliances depending on the patterns. In paper [5], the proposed system concentrated on ventilating, heating, door-access, lightning, illuminating, and reconfiguration are designed to save energy and promote sustainable development. In paper [6], the proposed system provides the study on how it is possible to ensure the indoor office comfort (thermal, visual, air quality, acoustic, etc.) and how each of them could be analyzed. In paper [7], the proposed system analyzes the existing works multimodal data fusion techniques for smart commercial buildings and occupancy monitoring. In paper [8], the proposed system is able to monitor and manage the energy consumption of existing public buildings. The system was used in different building construction. In paper [9], the proposed system provides a distributed solution that ensures a consensus is attained among all denizen, irrespective of their ideal temperature preferences being in conflicting or coherence. In paper [10], the proposed system provides environment simulation and daylighting control strategy to gain energy-efficient lighting while it also provides appropriate lighting levels at the aiming points. The system uses self-tuning multivariable controller for lighting strategy which improves the energy consumption due to artificial lighting. In paper [11], the proposed system mainly focuses on security operations, where it uses finger print identification module which allows only enrolled employees. The system used ARM (Advanced RISC Machines) 11 controller which processes the data and gives output. In paper [12], they proposed a general methods and algorithms to interconnect residential smart buildings with smart grids. In paper [13], they proposed a smart meeting room usually refers to working environment, which provides meeting with other important information acquisition and exchange space in order to improve the working and decision making efficiency. In papers [14, 15], both proposed how Internet can be used as a interface between devices to communicate in order to perform a task. In paper [16], researchers proposed an automation system where a Raspberry Pi as a controller and actions are coordinated by the home agent running in PC. In paper [17], the proposed system uses an automation logic in order to optimize the

Indoor Localization-Based Office Automation System …

201

power consumption costs throughout the day by observing the energy production and consumption, are cheap and considering the specific requirements of each appliance at home (battery charger, refrigerator, and oven). In paper [18], the proposed system is for improving the automation in industrial plants with a massive deployment of RFID tags associated with the production parts. In paper [19], researchers proposed a novel mobility simulation framework based on behavior pattern for office environments. In paper [20], the system presents the outcomes of a study that examines human interactions and mobility patterns in indoor spaces such as office environments. In papers [21, 22], researchers proposed Bluetooth based home automation system where Bluetooth Low Energy devices are used as communication medium for devices. In paper [23] researchers proposed a Voice recognition-based home automation where devices are controlled by commands that can be given through mic. In paper [24, 25], researchers proposed a ZigBee-Based Wireless Home Automation System. The system uses ZigBee protocol to transfer information among them. In papers [26, 27], researchers proposed Global System for Mobile communication (GSM)-based home automation system, where Short Message Service (SMS) features are also included. In papers [28, 29], researchers proposed a newly developing energy harvesting technology used in transportation, building, and home automation systems called EnOcean technology. In paper [26], the proposed system makes use of a PIC16F887 microcontroller for home appliances to control which also uses GSM for SMS features. In paper [30], the proposed system as an M2M system. It uses GSM for communication. GSM offers options for M2M which include DualTone Multi-Frequency (DTMF), SMS, General Packet Radio Service (GPRS). In paper [31], researchers proposed phone-based home automation system. It provides a system for a smart home that includes facilities like system controller, house-wide wiring, and a common interface. The proposed home automation system is server based. Here, the system will first localize a person that means find exact location of a person in room and then control the required appliances. To implement this system a Raspberry Pi is used as a server. The Pi is the central controller of entire system, all other devices are controlled by Raspberry Pi. The components that are used in our system is given in Table 1. Table 1 Parameters of devices used in the proposed system PIR sensors

Range 0–3 m

Node MCU ESP8266

Storage 4 MB, Range 0–300 m, 802.11n, Rx Power = 56 mA × Power = 120 mA, 65 mbps, PA +25 dbm, 2.4 GHz, Packet 1024 bytes

Wi-Fi

2.4 GHz, 150 feet (indoor range), 300 feet (outdoor range)

Room size

14 × 14 feet

Cabin size

6 × 6 feet (each)

202

B. S. Meena et al.

2 Proposed Problem This system will help to make home/office or other workspaces more portable, easy to handle, maintainable, electricity conservation, and cost-effective which leads to sustainable development. For example, consider a scenario where one forgot to switch OFF light or fan or other appliances of his/her room, so there may be a possibility of electricity wastage. Sometimes there are only a few people inside a room and still, all appliances are unnecessarily left ON. Hence, in such case, the proposed system will keep track of each person as well as there current location in the room (through indoor localization) and accordingly control the appliances and manages the electrical resources effectively.

3 Proposed Architecture The system is divided into three layers as shown in Fig. 1. The first layer is a physical layer that consists of the devices which are to be controlled by the system (fan, light, air condition, etc.). The second layer is for backend operation which includes Raspberry Pi (central controller), Server, Wireless Router, and Database. And the last layer is an input layer, which consists of RFID reader and tags, PIR sensors, and

Fig. 1 System architecture

Indoor Localization-Based Office Automation System …

203

Node MCU. In this system architecture generally, the input layer (3rd layer) sends data to the backend layer (2nd layer) through a Wi-Fi network. In the backend layer, Raspberry Pi compares the data with the data stored in the server’s database and if a match found than Raspberry Pi requests Relay module to control the appliances. The following assumption(s) are considered for implementation of the system: – The Office building consists of a Wireless network with a router capable of access point isolation. – Every employee of the office is equipped with an RFID-based tag. Detail connections and working of each layer are given in the below sections.

3.1 Experimental Setup Devices that are included in this layer are the Relay module and electrical appliances that are to be controlled like fans, lights, etc. – Relay module: This board requires 5 V power supply and each channel needs a 15.2 mA driver current. Relay modules are quite handy since it has a standard interface that can be controlled directly by the microcontroller. In this system, it is connected with Raspberry Pi and controls the appliances once command received from the Pi. – Raspberry Pi: A cheap computer used as a microcontroller in this system which includes a python script to control whole system. The set of GPIO (General Purpose Input/Output) pins are connected with Relay module to control it for physical computing. Since it is being an easily programmable device comes handy while performing tasks that also include multiple programming languages and interfaces. This microcontroller also includes database to store the data coming from different sensors. – Node MCU: A Wi-Fi module used to connect over the network. In the system NodeMCU is connected with PIR sensors, it receives data from PIR sensor and sends it to the database of Raspberry Pi through Wi-Fi. Here, NodeMCUs and Raspberry Pi are connected with common Wi-Fi router. – PIRs: These electronic sensors that measured infrared (IR) light radiating from objects in its field of view are used in the system to detect the motion of a person. These sensors are cheap and have quite an efficient accuracy the reason why it is used for indoor localization. – RFID: Radio-Frequency Identification module, a technology whereby digital data encoded in RFID tags are captured by a reader via radio waves are used in the system to identify a person uniquely that helps in localizing a person more accurately.

204

B. S. Meena et al.

4 System Working The system is installed in a building where multiple rooms are there. In each room, a PIR sensor is installed at the entrance door that represents the room uniquely. Each room contains two cabins with one PIR sensor installed in each. An RFID reader is installed at the entrance door of the building shown in Fig. 2. An employee is only allowed to enter into the building only if he/she has a valid RFID tag. The person will should have a valid RFID tag to enter into the building. Once he/she swipes his/her valid tag in RFID reader the reader will send the data to the server for verification through NodeMCU. Once the tag is verified a person is allowed to enter into the office building. After that he/she will enter into a room. While entering into a room, the PIR sensor installed at the entrance door of that room will detect motion and send it to the server. Next, he/she will sit into his cabin, where another PIR sensor is installed, that will detect his/her movement and send it to the server. During this whole process the server will concatenate all the data received from the RFID reader and PIR sensors in an order they are received to form a pattern of strings. Now, the resultant pattern will be matched with existing pattern stored in the database, and if match is found the Raspberry Pi will command the Relay module to control the appliances of that particular cabin. Similarly, for exit the pattern will be reversed in order. Here, the system will wait for given delay time and then switch OFF all the appliances of that particular cabin. All possible patterns are given in

Fig. 2 Components used in the proposed system

Indoor Localization-Based Office Automation System …

205

Table 2 List of all existing pattern for a single person Direction

Pattern

Description

Entry

RFID → R PIR → C PIR

Person entered into the particular room in office

Entry

RFID → R PIR → C PIR1

Person entered into cabin 1

Exit

R PIR → RFID

Person exited from office without entering into his cabin

Exit

C PIR → R PIR → RFID

Person exited from his/her cabin and then from office

Exit

C PIR → R PIR

Person exited from his room but he is inside the office

Table 2. The pattern formed by server in order to find the location of a person is a technique of indoor localization. Detail explanation of this technique is given in below section.

4.1 Localization The localization technique is actually used to localize a person in a particular room of the office. To find the location of a person in an office, we use some pairs of PIR sensors installed in the different areas of the office. Each PIR sensor has its unique ID and they represent a particular area of the room. They are named as: – R PIRs: These PIRs are installed at the entry door of each room. – C PIRs: These PIRs are installed at each cabin in the room. So, for each room there are exactly one R PIR and two C PIRs (as shown in Fig. 3). Each PIR sensors are connected with NodeMCU (ESP8266 Wi-Fi module) in order to send the reading of sensors through Wi-Fi. Each NodeMCU is connected with a common Wi-Fi router to send data to server. The algorithm used to localize the person and accordingly controlling the appliances is given in Algorithm 1.

206

Fig. 3 System design

B. S. Meena et al.

Indoor Localization-Based Office Automation System …

207

5 Result The simulation is done in a room of size 14 × 14 feet. The room contains two cabins, cabin 1 and cabin 2. There are three PIR sensors installed in a room. One at the entrance door of the room and the other two at each cabin. Each cabin is 6 × 6 feet in size. The gap between these two cabins is 3.9 feet. The RFID reader is installed outside the room (at corridor). So whenever event is detected sensor sends the data through NodeMCU to server database. The data that are entered in database are raw data; therefore in order to filter the data so that we can get Entry/Exit pattern we need to execute a python script which takes the tuples with identical timestamps and parses them as an event. The below section explains each graph in detail.

208

B. S. Meena et al.

5.1 Graphs All graphs indicate the number of triggers in sensors and RFID per two hours. Figure 4 represents raw data and filtered data for entry event in both cabin 1 and 2. In Graph (a), it indicated raw data for number of triggers in R PIR sensor installed at entrance door of the room and raw data for number of triggers in C PIR01 sensor which is installed inside the cabin 1. At the end graph shows filtered data that is entry event where triggering in R PIR is followed by C PIR01 within 10 s delay time. Similarly, for entry event in cabin 2, the filtered data is shown in graph (b). Figure 5 represents raw data and filtered data for exit event in both cabin 1 and 2. Here in graph (a), it represents raw data for number of triggers in R PIR and C PIR01 and at the end filtered data for exit event from cabin 1 that is Triggering in C PIR01 followed by R PIR. Similarly, in graph (b), for exit event from cabin 2, the filtered data is shown. Figure 6 represents raw data and filtered data for entry event in both cabin 1 and 2 including RFID. In previous graphs, the filtered data shown are without RFID. In this case raw data for RFID is also included along with raw data of R PIR and C

(a) Entry Event in cabin 1

(b) Entry Event in cabin 2

Fig. 4 Entry events without RFID in the proposed system

(a) Exit Event in cabin 1 Fig. 5 Exit event without RFID in the proposed system

(b) Exit Event in cabin 2

Indoor Localization-Based Office Automation System …

(a) Entry Event in cabin 1

209

(b) Entry Event in cabin 2

Fig. 6 Entry event with RFID in the proposed system

(a) Exit Event in cabin 1

(b) Exit Event in cabin 2

Fig. 7 Exit event with RFID in the proposed system

PIRs. Here, Graph (a) represents entry event into the cabin 1 where the pattern is triggering in RFID followed by R PIR followed by C PIR01. Similarly graph (b) represents entry event for cabin 2. Figure 7 represents raw data and filtered data for exit event in both cabin 1 and 2 including RFID. That means a person will exit from room as well as from building. Here, Graph (a) represents exit event from the cabin 1 where pattern is triggering in C PIR01 followed by R PIR followed by RFID. Similarly graph (b) represents exit event for cabin 2.

6 Conclusion The use of RFID and PIR sensors with pattern matching technique base localization implementation can reduce the processing cost while maintaining the accuracy of the system to a great extent. It shows that first of all the user has been localized as per the event detection pattern. Then as per the user’s localization the electronic devices

210

B. S. Meena et al.

Fig. 8 Energy consumption graph with and without system

have been controlled. Other technologies such as inbuilt GPS system, Bluetooth Low Energy devices, cameras, etc., can be used for localization. As far as the cost and accuracy has been concerned such systems increase the cost and accuracy is also not up to the mark. But in this proposed system, it is based on a low-range PIR sensor and RFID which reduces the cost and the result shows the accuracy is also higher. Energy consumption before and after the system was installed is recorded and when comparing the energy consumption of both cases, it is found that energy consumption after the system was implemented is relatively less. Energy consumption of both cases is shown in Fig. 8. The application of the proposed system is huge. It can be used in any below to average crowded places like: home, offices, library, museum, small shops, cyber cafe, etc.

7 Scope for Future Works The proposed system works well where the crowd is very less. If crowd increased according to that the required hardware and database pattern may increased the cost of the implementation. So, the researchers may give some focus while considering much crowded scenario and explore more research work in this field using the combination of IoT and other devices.

References 1. Kumar P, Pati UC (2016) IoT used monitoring and control of appliances for smart home. In: IEEE international conference on recent trends in electronics information communication technology, 20–21 May 2016, India

Indoor Localization-Based Office Automation System …

211

2. El Shafee A, Hamed KA, (2012) Design and implementation of a Wi-Fi based home automation system. Int J Comput Electr Autom Control Inf Eng 6(8) 3. Pavithra D, Balkrishnan R (2015) IoT based monitoring and control system for home automation. In: Proceedings of 2015 global conference on communication technologies (GCCT’15) 4. Gunge VS, Yalagi PS (2016) Smart home automation: a literature review. Int J Comput Appl (0975-8887) 5. Li H (2014) A novel design for a comprehensive smart automation system for the office environment. In: Emerging technology and factory automation (ETFA). IEEE 6. Bujdei C, Moraru SA (2011) Ensuring comfort in office buildings: designing a KNX monitoring and control system. In: 2011 7th international conference on intelligent environments (IE). IEEE 7. Akkaya K et al (2015) IoT-based occupancy monitoring techniques for energy-efficient smart buildings. IEEE 8. Pellegrino A, Lo Verso VRM, Blaso L (2016) Lighting control and monitoring for energy efficiency: a case study focused on the interoperability of building management systems. IEEE Trans Ind Appl 52(3) 9. Gupta SK, Mishra S, Wen JT (2015) Collaborative energy and thermal comfort management through distributed consensus algorithms. IEEE Trans Autom Sci Eng 12(4) 10. Gaspare B, Mehrdad M (2015) Day lighting control and simulation for LED-based energyefficient lighting systems. IEEE Trans Ind Inform 12(1) 11. Renuka B, Saniya A (2016) Design and implementation of smart office automation system. Int J Comput Appl 151(3). ISSN: 0975-88887 12. Donatella S, Nacci AA (2014) On how to design smart energy-efficient buildings. In: 2014 12th IEEE international conference on embedded and ubiquitous computing (EUC) 13. Deng T, Feng L, Suo Y, Chen Y (2010) Spontaneous interoperation of information appliances in a smart meeting room. In: 2010 2nd international workshop on intelligent system and applications (ISA) 14. Palaniappan S, Hariharan N, Kesh NT, Vidyalakshmi S, Angel Deborah SM (2015) Home automation systems: a study. Int J Comput Appl 116(11) 15. Tiwari S, Gedem S (2016) A review paper on home automation system based on internet of things technology. Int Res J Eng Technol 03(05) 16. Patel MM, Mehul AJ, Dixita BV (2015) Home automation using Raspberry Pi. Int J Innov Emerg Res Eng 2(3) 17. Buckl C, Sommer S, Scholz A, Knoll A, Kemper A, Heuer J, Schmitt A (2009) Services to the field: an approach for resource constrained sensor/actor networks. In: Proceedings of WAINA’09, Bradford, United Kingdom, May 2009 18. Spiess P, Karnouskos S, Guinard D, Savio D, Baecker O, Souza L, Trifa V (2009) SOA-based integration of the internet of things in enterprise services. In: Proceedings of IEEE ICWS 2009, Los Angeles, CA, USA, July 2009 19. Shiquin S, Nanjing C, Guihai C (2009) A behavior pattern based mobility simulation framework for office environments. In: 2009 Wireless communications and networking conference, WCNC 2009. IEEE 20. Gluhak A, Martelli F, Verdone R (2013) Measuring and understanding opportunistic copresence pattern in smart office spaces. In: Green computing and communications (GreenCom), 2013 IEEE and internet of things (iThings/CPSCom), IEEE international conference on and IEEE cyber. Physical and Social Computing 21. Piyare R, Tazil R (2011) Bluetooth based home automation system using cell phone. In: 2011 IEEE 15th international symposium on consumer electronics (ISCE), Singapore, pp 192–195 22. Lee KY, Choi JW (2003) Home automation system via bluetooth home network. In: SICE annual conference, Fukui, vol 3, pp 2824–2829 23. Sen S, Chakrabarty S, Toshniwal R, Bhaumik A (2015) Design of an intelligent voice controlled home automation system. Int J Comput Appl 121(15):39–42 24. Al Shu’eili H, Gupta GS, Mukhopadhyay S (2011) Voice recognition based wireless home automation system. In: 2011 4th international conference on mechatronics (ICOM), Kuala Lumpur, pp 1–6

212

B. S. Meena et al.

25. Baviskar J, Mulla A, Upadhye M, Desai J, Bhovad A (2015) Performance analysis of ZigBee based real time Home Automation system. In: 2015 international conference on communication, information & computing technology (ICCICT), Mumbai, pp 1–6 26. Teymourzadeh R, Ahmed SA, Chan KW, Hoong MV (2013) Smart GSM based home automation system. In: 2013 IEEE conference on systems, process & control (ICSPC), Kuala Lumpur, pp 306–309 27. Pramanik AA, Rishikesh, Nagar V, Dwivedi S, Choudhury B (2016) GSM based smart home and digital notice board. In: 2016 international conference on computational techniques in information and communication technologies (ICCTICT), New Delhi, pp 41–46 28. Kuzlu M, Pipattanasomporn M, Rahman S (2015) Review of communication technologies for smart homes/building applications. In: 2015 IEEE innovative on smart grid technologies – Asia (ISGT ASIA), Bangkok, pp 1–6 29. Sharma H, Sharma S (2014) A review of sensor networks: technologies and applications. In: 2014 recent advances in engineering and computational sciences (RAECS), Chandigarh, pp 1–4 30. Alheraish A (2004) Design and implementation of home automation system. IEEE Trans Consum Electr 50(4):1087–1092 31. Brooke Stauffer H (1991) Smart enabling system for home automation. IEEE Trans Consum Electron 37(2):29–35

Ensemble Based Approach for Intrusion Detection Using Extra Tree Classifier Bhoopesh Singh Bhati and C. S. Rai

Abstract With the swift growth of Internet technology, various types of attacks and intrusions are taking place over the Internet. Intrusion Detection Systems (IDS) are widely used to detect attacks. Enormous research has been done in the area of IDS but due to new attacks it is still an open area to researchers. In this paper, an ensemble based scheme is proposed using extra tree classifier for intrusion detection. The proposed scheme has four major steps namely Data Collection, Preprocessing of data, ensemble based training and testing and results, respectively. The basic idea of ensemble based approach is to make separate—separate classifier and train these classifiers. Combining the decision of different classifier is done to obtain the strong decision. In implementation of proposed scheme, KDDcup99 and NSL-KDD datasets are used. These datasets are benchmarks for intrusion detection. The simulated results show that our proposed scheme is very effective for intrusion detection. Python programming environment is used in implementation. The proposed scheme achieved 99.97% accuracy on KDDcup99 dataset and 99.32% on NSL-KDD datasets. Keywords Information security · Ensemble · Machine learning · Intrusion detection

1 Introduction Due to the rapid growth of the Internet, various transactions are taking place over the networks. Since the past decade, people have started showing high dependency on the Internet or networking. However different types of attacks are also coming in to picture during exchange of the information through Internet [1]. It is a big challenge to researchers who belong to network security. Nowadays, it is essential to protect our useful and private information. Intrusion detection system is used to B. S. Bhati (B) · C. S. Rai USIC&T, Guru Gobind Singh Indraprastha University Dwarka, New Delhi 110078, India e-mail: [email protected] C. S. Rai e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_25

213

214

B. S. Bhati and C. S. Rai

protect useful information from different types of attacks. The main motive of IDS is to detect various attacks that are occurring in the network. Intrusion detection is a process to monitor the various states occurring in the networks and investigating the occurring states to detect malicious states [2]. Broadly the IDS techniques can be classified as below: Anomaly Based Technique: In this type of IDS, investigation on the occurring states is done based on deviation in behavior of occurring event. Suppose a deviation is found in the behavior then a particular state is detected as intrusion otherwise detected as normal state [3]. Signature Based Technique: In this type of IDS, investigation on the occurring states is done based on signatures stored in knowledge- base of IDS. The signatures of malicious activities are already stored in knowledge-base of IDS. If signature of occurring state matches with existing signatures, that particular state is detected as intrusion. It is also referred to as knowledge-base or misuse detection [4].

2 Related Work Performance of a model can be improved using ensemble modeling which combines various machine learning techniques into a single predictive model. The majority of ensemble methods falls under homogeneous learners. On the other hand, some methods use learners of different types, called heterogeneous learners. Bagging, Boosting, and Staking are some of the commonly used ensemble learning techniques [5]. Perdisci et al. [6] proposed a new approach for network anomaly detection by using unsupervised or unlabeled learning approaches. They used a clustering algorithm to solve text classification problems. Wang et al. [7] proposed a new scheme using random forest algorithm, which is useful for online learning classification. It is based on a greedy search. Results showed good accuracy for both online learning and object tracking. Mkuzangwe et al. [8] proposed a scheme for intrusion detection system based on an ensemble of classifiers. It helped in estimating the performance of the ensemble classifier even before its implementation. NSL-KDD datasets were used in different proportions for observational study. Giorgio et al. [9] proposed an anomaly based IDS for Network using Modular Multiple Classifier System (MCS). Giorgio et al. used KDDCup99 datasets and achieved high detection rate. Folino et al. [10] proposed an IDS using Genetic Programming Ensemble. This proposed IDS was used in a distributed environment. Chebrolu et al. [11] proposed an IDS based on general Bayesian network (BN) classifier and Classification and Regression Trees (CART). Ensemble were used for complementing base classifiers. Performance measures were also examined for Distributed IDS. Hence, a distributed model based on a hybrid architecture was proposed by them with the evidence of increased accuracy in intrusion detection on DARPA benchmark.

Ensemble Based Approach for Intrusion Detection …

215

Sindhu et al. [12] presented improved multi-class categorizing IDS using three different viewpoints such as preprocessing and cleaning input traffic pattern followed by a feature selection algorithm. They used neurotree model for increased classification and detection rate which is superior to NN* and extended C4.5 algorithms. Aburomman and Reaz [13] observed that if opinions from multiple experts can be combined into one it can help in demonstrating improved classification accuracy using ensemble approach. It was observed that Particle Swarm optimization obtained best results with an average improved accuracy of 0.756% indicating relatively less time and optimized weights which produced best possible accuracy. It was concluded that these techniques were good enough for binary classification. Mukkamala et al. [14] proposed an ensemble approach and also compared it with other techniques. The performance of ANNs, SVMs and MARs were compared with an ensemble method. It was concluded that the ensemble method was superior in accuracy of classification with the possibility of gaining 100% classification accuracy with correct intelligent paradigms. Among these techniques SVMs outperformed MARs and ANNs in terms of training time and prediction accuracy. Resilient back propagation performed the best with accuracy of 97.04% and training of 67 epochs. Hence, demonstrating the importance of ensemble in different learning paradigms.

3 Proposed Scheme Proposed scheme is divided into the following major steps namely Data collection, Preprocessing of datasets, Training and Testing using Extra tree classifier, and Results. The outline of proposed scheme is given in Fig. 1.

Fig. 1 Proposed scheme

216

B. S. Bhati and C. S. Rai

3.1 Data Collection Dataset is required as a benchmark for any proposed technique for intrusion detection. Here, most popular intrusion datasets are used. Both KDDcup99 and NSL-KDD datasets are used which are publically available for researcher of network security community [15]. These datasets have many normal records and malicious records. In the implementation all the records are categorized as Normal, Denial of Services (DOS), Probe, R2L and U2R [16]. In implementation KDDCup99 and NSL-KDD datasets are used.

3.2 Preprocessing When we collect the datasets from any source, the datasets is not normalized. In preprocessing some operation are performed on the datasets before the training and testing. The main objective of preprocessing is to eliminate unwanted data from the datasets. In simple way we can say preprocessing is a process to transform the data from one form to desirable form [17]. A preprocessed dataset behave in a good manner during training and testing phase and the outcome of model will be accurate. In this implementation, Python environment is used to preprocessing the both datasets KDDcup99 and NSL-KDD datasets.

3.3 Training and Testing Using Extra Tree Classifier Training and testing is a very important phase, in this phase model is built through training and then model is tested. Here, the Extra tree Classifier ensemble method is used. The graphical representation of the above method is given in Fig. 2. Extra tree classification is the modification of bagging where samples of the training dataset are used to construct the random trees [18]. Extra tree classifier is also known as extremely randomized trees. The working of extra tree classifier is given below: Step 1: Train the dataset. Step 2: Random selection is done in this step. Random selection (K) is used to determine the best split. Step 3: Multiple decision tree are build using random vector. Step 4: All tree generated by random vector are combine into a single decision tree.

Ensemble Based Approach for Intrusion Detection …

217

Fig. 2 Graphical representation of extra tree classifier

3.4 Results This phase is the outcome of the above phases. In this phase, the model takes the decisions whether the occurring event is a malicious activity or normal activity. The proposed scheme gives 99. 97% detection accuracy on KDDcup99 dataset and 99.32% on NSL-KDD datasets.

4 Analysis and Discussion Tables 1 and 2 show the summary of results obtained using proposed scheme. With the help of precision, recall, f1 score the results have been analyzed. Confusion Matrix for both the datasets is given in Tables 3 and 4. The terminology used at analysis stage can be explained as following: Table 1 Summary of results of proposed scheme on KDDcup99 datasets Category

Precision

DOS

1.00

Normal

1.00

Probe

1.00

R2L U2R

Recall

f1-score

Support

1.00

1.00

78,282

1.00

1.00

19,494

0.99

0.99

799

0.99

0.97

0.98

221

0.86

0.75

0.80

8

218

B. S. Bhati and C. S. Rai

Table 2 Summary of results of proposed scheme on NSL-KDD datasets Category

Precision

Recall

f1-score

Support

DOS

1.00

1.00

1.00

1857

Normal

0.99

1.00

0.99

2670

Probe

0.99

0.97

0.98

464

R2L

0.97

0.88

0.92

40

U2R

1.00

1.00

1.00

1

Table 3 Confusion matrix for KDD 10% using proposed scheme DOS DOS

Normal

Probe

R2L

U2R

78,282

0

0

0

0

Normal

6

19,482

3

2

1

Probe

0

6

793

0

0

R2L

1

6

0

214

0

U2R

0

2

0

0

6

Table 4 Confusion matrix for NSL 20% using proposed scheme DOS DOS

Normal

Probe

R2L

U2R

1850

7

0

0

0

Normal

0

2664

5

1

0

Probe

1

15

448

0

0

R2L

0

5

0

35

0

U2R

0

0

0

0

1

Precision: Precision may be defined as the ratio of True Positive (TP) to the sum of True Positive (TP) and False Positive (FP). It measures the model’s accuracy. Precision =

TP T P + FP

(1)

Recall: Recall is defined as the ratio of TP to the sum of TP and False Negative (FN). It measures the model’s completeness [19]. Recall =

TP T P + FN

(2)

Accuracy: Accuracy may be defined as the ratio of how correctly predict the observation to the total observation. Where TN refers to true negative.

Ensemble Based Approach for Intrusion Detection …

Accuracy =

TP +TN T P + FP + FN + T N

219

(3)

F-beta (β) score: F-beta score is defined as the average weight of Precision and Recall. When the value of beta in F-beta score is 1 then it is termed as F1 score. F-beta score gives best value when its value reaches 1 and at 0 value it gives the worst score [20].  1 + β2 P R F-score (β) = β2 P + R 

(4)

where, R, refers to Recall, P refers to Precision. F-beta (β) score can be computed as:   1 + β2 T P  (5) F-score (β) =  1 + β2 T P + β2 F P + F N

5 Conclusion In this paper, several ensemble based approach for intrusion detection have been reviewed. Here, a new ensemble based approach has been proposed and analyzed. Proposed scheme has four phases namely Data collection, Preprocessing of datasets, Training and Testing using Extra tree classifier, and Results. The proposed scheme has been applied on both the intrusion detection dataset i.e., KDDcup99 and NSL-KDD. For implementation, a Python programming environment has been used. Accuracy has been observed for both the datasets. Results have been analyzed based on accuracy and confusion matrix, which shows the proposed scheme achieved 99.97% accuracy on KDDcup99 dataset and 99.32% on NSL-KDD datasets. In future, the proposed scheme may be applied in real-time datasets and optimization technique may be used.

References 1. Lazarevic A (2005) Managing cyber threats: issues, approaches, and challenges. Springer Science+Business Media, Incorporated 2. Bace R, Mell P (2001) NIST special publication on intrusion detection systems 3. Bhati BS, Rai CS (2016) Intrusion detection systems and techniques: a review. Int J Crit Comput Based Syst 6(3):173–190 4. Liao HJ, Lin CHR, Lin YC, Tung KY (2013) Intrusion detection system: a comprehensive review. J Netw Comput Appl 36(1):16–24

220

B. S. Bhati and C. S. Rai

5. Basics of ensemble learning. https://www.analyticsvidhya.com/blog/2015/08/introductionensemble-learning. Accessed on 14 Feb 2019 6. Perdisci R, Gu G, Lee W (2006) Using an ensemble of one-class SVM classifiers to harden payload-based anomaly detection systems. In: ICDM , vol 6, pp 488–498 7. Wang A, Wan G, Cheng Z, Li S (2009) An incremental extremely random forest classifier for online learning and tracking. In: 2009 16th IEEE international conference on image processing (ICIP), pp 1449–1452. IEEE 8. Mkuzangwe NN, Nelwamondo F (2017) Ensemble of classifiers-based network intrusion detection system performance bound. In: 2017 4th international conference on systems and informatics (ICSAI), pp 970–974. IEEE 9. Giacinto G, Perdisci R, Del Rio M, Roli F (2008) Intrusion detection in computer networks by a modular ensemble of one-class classifiers. Inf Fus 9(1):69–82 10. Folino G, Pizzuti C, Spezzano G (2005) GP ensemble for distributed intrusion detection systems. In: International conference on pattern recognition and image analysis. Springer, Berlin, Heidelberg, pp 54–62 11. Chebrolu S, Abraham A, Thomas JP (2005) Feature deduction and ensemble design of intrusion detection systems. Comput Secur 24(4):295–307 12. Sindhu SSS, Geetha S, Kannan A (2012) Decision tree based light weight intrusion detection using a wrapper approach. Expert Syst Appl 39(1):129–141 13. Aburomman AA, Reaz MBI (2016) A novel SVM-kNN-PSO ensemble method for intrusion detection system. Appl Soft Comput 38:360–372 14. Mukkamala S, Sung AH, Abraham A (2005) Intrusion detection using an ensemble of intelligent paradigms. J Netw Comput Appl 15. Dataset Canadian Institute for Cybersecurity. http://nsl.cs.unb.ca/NSL-KDD/. Accessed on 14 Feb 2017 16. Tavallaee M, Bagheri E, Lu W, Ghorbani AA (2009) A detailed analysis of the KDD CUP 99 data set. In: 2009 IEEE symposium on computational intelligence for security and defense applications. IEEE, pp 1–6 17. Aldwairi M, Khamayseh Y, Al-Masri M (2015) Application of artificial bee colony for intrusion detection systems. Secur Commun Netw 8(16):2730–2740 18. Ensemble machine learning algorithms in python with scikit-learn. https:// machinelearningmastery.com/ensemble-machine-learning-algorithms-python-scikit-learn/. Accessed on 16 Feb 2019 19. Scikit learning. https://scikit-learn.org/stable/modules/generated/sklearn.metrics.precision_ recall_fscore_support.html. Accessed on 18 Feb 2019 20. Accuracy, Precision, Recall & F1 Score: Interpretation of Performance Measures, https://blog. exsilio.com/all/accuracy-precision-recall-f1-score-interpretation-of-performance-measures/ accessed on 26 February 2019

Fourth Industrial Revolution: Progression, Scope and Preparedness in India—Intervention of MSMEs Arindam Chakrabarty, Tenzing Norbu and Manmohan Mall

Abstract MSME (Micro, Small and Medium Enterprise) sector constitutes more than 99% of private firms operating in India which generate crores of jobs across the country. In fact, the MSME firms aim to support the large companies either in the form of outsourcing partners for supplying raw materials, W-I-P or adding value to one or few processes as ancillary to the big establishments. However, in the growing competition and the market complexity, the MSMEs have to compete with the large firms. The world is emerging toward Fourth Industrial Revolution (4IR) which not only prescribes for automation, speed, and prompt delivery mechanism but also it attempts to duplicate Human Intelligence in the form of Machine Learning or Artificial Intelligence (AI). In the dynamics of rapid changes across the Industrial Ecosystem, it is emergent for the MSMEs to re-module its business directions. The threshold level technology needs to be transferred, absorbed, and adopted by the MSME firms so that the can play a meaningful role in today’s knowledge economies. This paper has explored the Scope and Preparedness for the sector and has prescribed desired Policy Reforms to make the transition smooth, value-adding and resourceful. Keywords 4IR · MSMEs · Artificial intelligence · Threshold level technology · India

A. Chakrabarty (B) Department of Management, Rajiv Gandhi University (Central University), Itanagar 791112, Arunachal Pradesh, India T. Norbu · M. Mall North Eastern Regional Institute of Science & Technology (NERIST), Centre for Management Studies (CMS), Nirjuli 791110, Arunachal Pradesh, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_26

221

222

A. Chakrabarty et al.

1 Introduction 1.1 MSMEs in the World The contribution of MSMEs across the globe has been found highly significant particularly as a change agent for rapid socio-economic development [1, 2]. Research indicates that the MSME sector has a positive association with the economic growth and developmental indicators equitably both in developing and developed nations of the world [3, 4]. The sector has been instrumental to absorb a large pool of manpower directly or indirectly worldwide.

1.2 Indian MSMEs: Status, Scope, and Achievements Indian MSMEs have the capacity of absorbing around 40% of the total workforce that contributes almost 45% of manufacturing outputs worth of around 6% of manufacturing GDP and reserves the share of 40% of total exports of the country. It is observed that around 94% of the firms belonging to the MSME sector are not registered even though the growth of this sector has been recorded at around 11% per annum which is more than the average GDP of the country in recent years [5, 6]. However, with the implementation of Goods and Service Tax (GST) in India, the unregistered MSMEs are compelled to enroll as a part of legal bindings. The MSME firms have been largely facing a series of problems and inadequacy which are mostly in terms of lack of availability of resources and opportunities leading to high-end inefficiencies. However, India witnesses a minuscule of MSME firms that are performing at par with the big corporates while the larger section of Indian MSMEs acutely suffer from Industrial Sickness or pro-sickness. The financial package extended to such sick firms would not be able to address the root causes rather offering some other set of benefits which might result in a favorable outcome [7, 8]. Most importantly, the industrial development is a function of the ease, access, and successful use of technological development. The MSME sector essentially needs the constant support for skilling of its manpower and technology led transformational business practice. Barring a few Medium and high performing firms, it is difficult for Micro or Small firms to afford continuous investment for technological upgradation. It is indeed a great challenge for the policymakers and the promoters these firms to have a full-proof solution for its survival, growth and sustainability.

Fourth Industrial Revolution: Progression, Scope …

223

2 Review of Literature MSMEs are deemed to be an accelerator of economic growth across the world [1, 2, 9]. There is a positive relationship between the growth of MSMEs and the growth of the economy in many developed and developing countries [3, 4, 10]. MSMEs are the backbone of the Indian economy as they play a pivotal role by making a substantial contribution to the economy. They contribute around 40% of gross industrial value, 45% of the export and are considered to be the second largest employment generator in the country [9, 11]. Therefore, MSMEs are a necessity for the nation as they ensure innovation, revenue generation, and employment generation, etc. [12]. MSMEs, notwithstanding, face several challenges in India such as lack of tangible resources [13], HRM related issues [14], issues related to power, raw material procurement [15], lack of adequate financial assistance from Banks, absence of sophisticated technologies, scarcity of resources, and lack of skilled manpower leading to ineffective marketing [16].

3 Objectives of the Study The present study endeavors to: i. Study and understand the progression of the Industrial Revolution with a special focus on the Fourth Industrial Revolution (4IR). ii. Explore the scope and emergence of 4IR in India with special reference to MSME sector.

4 Research Methodology The paper has been conceptualized responding to the call of the hour about the emergence of the 4IR in the world and specifically in the Indian subcontinent. The paper has attempted to understand how the 4IR has progressed over a period of time which has been presented with the use of relevant and reliable secondary information. Since the country has been growing as one of the fastest economies of the world and also aspiring to optimize its demographic dividend, it has become imperative to understand the scope and emergence of 4IR and how the Indian MSMEs can play a responsible role as it caters to almost the entire Indian industries barring a few hundreds of larger firms.

224

A. Chakrabarty et al.

5 Analysis and Interpretation Analysis I The Industrial Revolution (IR) began with a view to meet the demands of growing population in terms of supply of products. The primary sector has its limitation and highly concentrated on agricultural produces, handlooms, and handicrafts products. With the landmark invention of Steam based Power and Low Productive Machines and other tools, the First IR took place in the mid of the eighteenth century which continued up to the first half of nineteenth Century. With the pace of technological development in consonance with the growing demand for higher productivity, the Second IR started from the second half of the nineteenth century to the mid of the twentieth century where the sector emphasized on assembly line and mass production techniques to cater to the need of the population. In the later part of twentieth century, the industry was dominated by growing advent of electronics, instrumentation, and computational devices, IT and automation, it introduced a new age of Industrial Revolution popularly known as Third IR. This period of IR focused on higher volume of production with superior quality, precision, and Quality Function Deployment (QFD). In fact, Total Quality Management (TQM) appears to be the prime focus among most of the learning and large firms. The emphasis of Product Development has been shifted to achieving excellence in Process Development since the outcome of the lead processes essentially improvise creating higher quality of product. We are now in the era of 4IR where not only the Production Operation Systems are being modified through continuous improvement process as prescribed by Deming’s P-DC-A Cycle but the sector intends to replicate Human Intelligence par excellence into the devised mechanism in practice. The flowchart of the progression of IR consistent with the timeline has been enveloped in Fig. 1. Analysis II The new regime of IR has been propelled by outstanding advancement of satellite and wireless technology and its successful adaption among the population like Mobile Telephony, access to Mobile led Internet, use of high configuration platform like androids, etc. The wave of such advancement was highly appreciated and absorbed by the Indians particularly the Youth population of the country. The tremendous growth of mobile and internet services have revolutionized the economic growth trajectory of the world and, of course, this is going to impact the Indian economy in the coming decades. The vivid penetration of Smartphone tremendously enhances the use of Mobile Internet as compared to Fixed Line Portals. This has made a growing propensity and user friendly internet access platform for the Indian users that resulted in greater participation in mobile led e-commerce activities across the country. This has motivated the Indian users to prefer new business model [17]. Internet economy of India is projected to double from the existing (April, 2017) 125 billion USD to 250 billion USD by 2020 at the behest of phenomenal growth in e-commerce/m-commerce of which the value of transactions would reach around 100 billion USD through digital platform. The ambitious project of Digital India

Fourth Industrial Revolution: Progression, Scope …

225

Developed by the Authors

Fig. 1 Industrial revolution: journey ahead

Campaign intends to create online economy worth trillion USD by the year 2025 [18]. According to Worldometers—real-time world statistics, the present global population is around 7.6 billion of which it is estimated that almost half of the population are Internet users and surprisingly, around 50% of global internet users reside in Asia. Around 24% Internet users from Asia belong to India [19]. It is projected that the market potential for IoT devices in India would reach up to 9 billion USD by 2020 [20] as the country is poised to execute large scale IoT intervention projects to cater to its diversified reform policy [21]. All these above figures and observations signify that there is a growing market opportunity in creating, manufacturing and servicing IoT led devices in India which can be shouldered by the Indian MSMEs either as a support hub of large scale enterprises or the independent providers in the segment. This would depend based on the Competency Mapping of the MSMEs firms in terms of firm’s core expertise, experiences, conformity of other value chain and the extent of absorbing and adopting new age technology within the least possible transition of time.

226

A. Chakrabarty et al.

6 Recommendations Based on the present study, it is imperative to understand the development of IoT led and other forms of digital ecosystem in the era of 4IR has been emerging as mammoth business opportunities where the MSMEs can play a leading role along with the larger firms. This transition needs certain policy reforms from the state as well as higher commitments by the enterprises operating in India since India is emerging as the fifth largest economy in the world and the second largest continent using part excellence technology. The sector can achieve enormous export opportunities in various countries of Asia, Africa and Latin America even a small segment of gulf nations since India enjoys a competitive edge over others in terms of its positioning as IT superpower, superior quality of human skill, competitive labor cost, and sustainable competitive advantage on the related domain trade. To achieve these agendas, the following recommendations may be incorporated: 1. The government should sponsor and organize massive skill development programs highlighting the necessary augmentations for creating devices related to IoT and other forms of digital and interactive ecosystem. 2. To reinforce the confidence among the smaller firms, the state may formulate time-bound financial incentives either in the form of tax exemption or extending case-specific subsidy so that financial aspect can be considerably supported. 3. The bank or Financial Institutions may be directed to promote firms for venturing into 4IR by allocating targeted budgetary provisions that should be disbursed in a time-bound manner. 4. The state must encourage Higher Educational Institutions like Universities, Colleges, and Research Institutes to take up innovative and need-based projects and applied research in the broader domain of IoT enabled devices so that the outcome of the research can move from lab to market. The present research has identified that people of India are keen to get IoT augmented high quality healthcare sector which can be prioritized along with other emerging areas. 5. Massive investment in the sector is essential in order to strengthen the infrastructure for delivering public utility services like health, education, Public Health Engineering (PHE), environmental protection etc., both in rural and urban areas of India in order to expedite rapid socio-economic transformation as prescribed by the United Nations’ Sustainable Development Goals (SDGs).

7 Conclusion The present paper is exploratory in nature which has been grounded by the latest dataset and information collated from most recent and reliable sources. The paper has attempted to showcase how 4IR has arrived and is knocking at the door. If we miss or delay to welcome, perhaps we would be compelled to invite ourselves to the

Fourth Industrial Revolution: Progression, Scope …

227

catastrophic consequence and would fail to get into the growth trajectory in the new millennium. India is about to encash its growing demographic dividends where it is inevitable to imbibe the youth with this new generation business model otherwise it would be detrimental to achieve the goals as doctrines by UN SDGs.

References 1. Chong MY, Chin JF, Loh WP (2013) Lean incipience spiral model for small and medium enterprises. Int J Ind Eng 20 2. Sun Y, Cheng J (2002) Hydrolysis of lignocellulosic materials for ethanol production: a review. Biores Technol 83(1):1–11 3. Demirbag M, Tatoglu E, Tekinkus M, Zaim S (2006) An analysis of the relationship between TQM implementation and organizational performance: evidence from Turkish SMEs. J Manuf Technol Manag 17(6):829–847 4. Ogunyomi P, Bruning NS (2016) Human resource management and organizational performance of small and medium enterprises (SMEs) in Nigeria. Int J Human Resour Manag 27(6):612–634 5. Accessed from https://economictimes.indiatimes.com/small-biz/policy-trends/smes-employclose-to-40-of-indias-workforce-but-contribute-only-17-to-gdp/articleshow/20496337.cms on 28 April 2019 6. Accessed from https://evoma.com/business-centre/sme-sector-in-india-statistics-trendsreports/ on 28 April 2019 7. Burpitt WJ, Rondinelli DA (2000) Small firms’ motivations for exporting: to earn and learn? J Small Bus Manag 38(4):1 8. Jani HJ, Joshi YC, Pandya FH (2015) Impact of fiscal incentives on MSMEs’ performance in Gujarat. J Entrep Manag 4(4):1–20 9. Singh B, Narain R, Yadav RC (2012) Identifying critical barriers in the growth of Indian micro, small and medium enterprises (MSMEs). Int J Bus Compet Growth 2(1):84–105 10. Bhuiyan N, Alam N (2004) ISO 9001: 2000 implementation—the North American experience. Int J Prod Perform Manag 53(1):10–17 11. Sharma RK, Kharub M (2015) Qualitative and quantitative evaluation of barriers hindering the growth of MSMEs. Int J Bus Excell 8(6):724–747 12. Boso N, Cadogan JW, Story VM (2013) Entrepreneurial orientation and market orientation as drivers of product innovation success: a study of exporters from a developing economy. Int Small Bus J 31(1):57–81 13. Singh RK, Garg SK, Deshmukh SG (2008) Challenges and strategies for competitiveness of SMEs: a case study in the Indian context. Int J Serv Oper Manag 4(2):181–200 14. Gakhar K, Kour H (2012) Issues and innovations of HRM in MSMES-a study of the developing states of India. Global J Arts Manag 2(1):83–86 15. Srinivas KT (2013) Role of micro, small and medium enterprises in inclusive growth. Int J Eng Manag Res (IJEMR) 3(4):57–61 16. Aruna N (2015) Problems faced by micro, small and medium enterprises—a special reference to small entrepreneurs in Visakhapatnam. IOSR J Bus Manag (IOSR-JBM) 43–49 17. Chakrabarty A (2019) Is India poised for M-commerce in the cashless Milieu? In: Duhan P, Singh A (eds) M-commerce: experiencing the phygital retail. Apple Academic Press. Hard ISBN: 9781771887144, E-Book ISBN: 9780429487736. https://doi.org/10.1201/ 9780429487736 18. Accessed from https://www.ibef.org/download/Role-of-Manufacturing-in-EmploymentGeneration-in-India.pdf on 28 April 2019 19. Internet World Stats, Internet usage statistics, Dec 2017. http://www.internetworldstats.com/ stats.htm

228

A. Chakrabarty et al.

20. Deloitte (2017) Internet of Things (IoT) to be the next big thing for operators—TMT India Predictions, Feb 2017. https://www2.deloitte.com/in/en/pages/technology-media-andtelecommunications/articles/tmt-india-predictions2017-press-release.html 21. Tata Communications (2018) India IoT report—emergence of a new civic OS. Accessed from https://www.tatacommunications.com/wp-content/uploads/2018/02/IoT-Report.pdf on 28 April 2019

Call Admission Control in Mobile Multimedia Network Using Grey Wolf Optimization Sanjeev Kumar and Madhu Sharma Gaur

Abstract Wireless mobile networks still need reliable traffic performance, link connectivity, and consistent terminal mobility with Call Admission Control (CAC) to obtain best services of mobile communication and data transmission. Generally, mobile networks contain base stations (BTS), mobile hosts (MH), links, etc. that are often vulnerable to failure. The main objective of this research is to choose the least figure of potential positions to deploy base stations with proper channel allocation so that it should be covered maximum population density. According to do this, a multi-objective function is proposed to enhance the Quality of Services (QoS) in terms of continuous service availability. Particularly, a channel allocation scheme is presented to minimize the call dropping probabilities. This multi-objective function is combined with a Grey Wolf optimizer (GWO) to address this problem. In this GWO based scheme, an efficient fitness function and GWO operators are used to represent the systematic solution of the channel allocation problem. This model is tested with two different mobility scenarios, one is random mobility, where each position is represented by the grid cross position and second is based on uniform mobility where the potential position is deployed uniformly. This model efficiently manage and uses the reserved radio resources to control the call failure rate or call dropping probability. The experimental results compared with some existing methods to illustrate the efficiency of the proposed scheme. Keywords Call admission control · Quality of service · Grey wolf optimizer · Resource management

S. Kumar (B) · M. S. Gaur G. L. Bajaj Institute of Technology and Management, Greater Noida, India e-mail: [email protected] M. S. Gaur e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_27

229

230

S. Kumar and M. S. Gaur

1 Introduction In the highly connected world of cellular or mobile networks, Wireless mobile networks, to achieve reliable mobile communication, data transmission, better performance, link connectivity, and consistent terminal mobility still need heavy traffic management with Call Admission Control (CAC). Generally, mobile networks contain base stations (BTS), mobile hosts (MH), links, etc., that are often vulnerable to failure. The main objective of this research is to choose the least figure of potential positions to deploy base stations with proper channel allocation so that it should be covered maximum population density. According to do this, a multi-objective function is proposed to enhance the QoS for continuous service availability. Particularly, a channel allocation scheme is presented to minimize the call dropping probabilities. This multi-objective function is combined with Grey Wolf optimizer (GWO) to solve this problem. In this GWO based scheme, an efficient fitness function and GWO operators are used to represent the systematic solution of the channel allocation problem. This model is tested with two different mobility scenarios, one is random mobility, where each position is represented by the grid cross position and second is based on uniform mobility where the potential position is deployed uniformly. This model efficiently manages and uses the reserved radio resources to control the call failure rate or call dropping probability. Rapidly growing mobile devices and increasing multimedia services are constantly adding resource management challenges and yet unsolved problems for wireless cellular networks. Although in real-time multimedia networks it is difficult to resolve the traffic congestion problem but by addressing QoS requirements expected by users of that traffic can be used to diminish the resource availability concerns.

2 Related Work Background: In recent years, mobile computing has received great interest of the mobile user for its various mobile/wireless applications. In order to provide communication for non-real-time (Static) and real-time (Dynamic) users, mobile computing makes use of various mobile/wireless communication network protocols. In mobile communications environments, systematic assignment of radio resources and coverage problem play important roles to increase the quality of the cellular system. The mobile networks are distributed over the geographical area in the form of hexagonal shape, called cells. Each cell is equipped with at least one base station (BTS) that provides the network coverage capacity to cells, which can be used for the transmission of voice, data, and types of services in the mobile networks. One of the most favorable properties of the BTS placement is to manage the design conflicts of cells because the cell typically uses the distinct set of radio frequencies from adjacent cells in order to avoid interference and provides quality of services (QoS) within each cell [1]. In mobile communication system, channel allocation schemes

Call Admission Control in Mobile Multimedia Network …

231

play a very important role to provide the best services to the mobile users (MU) but it may require more base stations to increase the coverage area which may also increase the system cost and connectivity problem. Literature Review: In the past year, a lot of models have been developed in the field of mobile computing to solve various mobile services problems by using metaheuristic approaches [2–13]. Here the author presents two types of problem solution methods, one method used the objective function to increase the coverage area of the BTS by using interference constraints. In order to make the most effective use of radio signal frequency, networks used parallel random probability distribution optimization techniques. In addition to this, an extra temporal phase offset variable is used to minimize the signal distributions [2]. An integrated approach for Frequency Division Multiple Access (FDMA) systems, however, is not considered in this work, as well as a precise formulation in terms of mathematical programming. Genetic algorithms and greedy heuristics are applied by the researchers to increase the coverage area and proper channel utilization [3–6] according to the potential position. In [10] author developed a dynamic channel allocation (DCA) scheme for fixed broadband wireless networks using Genetic Algorithms (GA). This GA based scheme was compared with two wireless communication models based on interference and channel isolation, which shows that GA based scheme obtains better SNR [7]. In [8] Kassotakis et al., developed a channel reuse model for heterogeneous mobile networks with the combination of GA and local search algorithms in order to provide reliability and accuracy of GA and hill-climbing methods, respectively. In heavy network traffic, the simulation results of HGA model were compared with the performance of graph coloring algorithm (GCA) [9], and with the comparison of these two methods we find that the performance of the HGA based model is better than the performance of CGA based model. Khanbary and Vidyarthi [10], developed a fault-tolerance based dynamic channel allocation (DCA) model using GA for the reassignment of available channel or reserved channels to maximize channel utilization and minimize handoff call dropping probability. Further, they [11] developed a GA based reliable mobile communication and channel allocation system to find the mobility of the mobile user. A communication network can be more reliable if it makes efficient usages of the channel and offers the quality of services to the mobile user anywhere and anytime. Sanjeev Kumar et al. [12], have developed a hybrid approach for dynamic channel allocation service the mobile users based on their handoff calls or mobility in mobile networks. The proposed model applied two famous models of neural network, error back-propagation neural network, and Hopfield neural network (HNN). Error backpropagation neural network finds the mobility based on the previous history of the mobile user and based on the traffic mobility the HNN model allocates the channels which not only increase resource utilization and QoS of the system but also decrease the blocking and dropping probabilities significantly as shown in simulation results. Further in [13] the author developed a new energy function using HNN that not only identifies the mobility of the user but also makes the assignment of the channels for new calls and handoff calls. This approach examines and compares the performance

232

S. Kumar and M. S. Gaur

of the traffic mobility, which increases overall QoS in terms of signals, traffic load, inter-cell handoff, and continuous service availability to mobile users. The performance of the system is compared with some existing models of mobility and channel allocation schemes. In [14], a location based CAC and resource allocation scheme have been developed using error back-propagation model that randomly forecast the exact future mobile location of the mobile user based on the previous mobility history of the user. In [15] the author provides a wide survey of call admission control and resource allocation by using soft computing techniques.

3 System Models and Problem Formulations Network Model: BTS can cover the maximum number of population or mobile terminals within its frequency range. As similar to LEACH [16–19], the data collecting operation is separated into rounds. In every round, all the base stations cover the maximum populations (mobile terminals) with proper channel allocation that are within its communication range. This MAC protocol is used to handle MAC layer communication [20]. Problem Description: The motivation behind this research is to develop the optimal base station setup in a given pool of configurations in Mobile Computing. Generally, mobile networks contain BTS, MH, links, etc., needs special attention and care with respect to service failure. Thus, there is a need to design an attractive trustworthy network communication network for reliable data transmission, in terms of continuous service of base stations and the communication channels of the mobile networks. The parameters are used in this proposed model 1. 2. 3. 4. 5.

U = Set of mobile users. P = Set of potential mobility position. BTS = Represents the communication range of the base stations. disti j = Represents the interval between u i and s j . Cov(u i ) = Defines a BTS set within the communication area of u i , in other words, u i covered by Cov(u i )     Cov(u i ) = s j |dist u i , s j ≤ BT Scomm  ∀ j , 1 ≤ j ≤ M

(1)

6. T Cov(u i ) represents the set of mobile users that are covered by base station S j covers all T Cov(u i ) mobile user points.     T Cov(u i ) = u j |dist u j , si ≤ BT Scomm  ∀ j, 1 ≤ j ≤ N

(2)

The optimal BTS placement with maximize coverage problem can be specified as follows. Let bi j and qi j be the Boolean variables defined as follows:

Call Admission Control in Mobile Multimedia Network …



1, if user u i covered by s j 0, Otherwise

(3)

1, if a potential position pi for BTS placement 0, Otherwise

(4)

bi j =  qi j =

233

Subsequently, the Linear Programming Problem (LPP) can be formulated as follows Minimize Y =

K 

qi

(5)

i=1

Subject to

M 

bi j ≥ K , ∀i , 1 ≤ i ≤ N

(6)

j=1

The constraint (5) ensures that every mobile user u i , 1 < i < N , is protected by at least K number of BTSs. Hence it satisfies maximum coverage of each user with minimum no required channels.

4 Proposed GWO Based Algorithm The problem of the optimal BTS section is reformulated as a combinatorial optimization algorithm in [21]. We propose a new fitness function based on minimum potential position, maximum population coverage and proper channel allocation. To optimize fitness function, we use the Grey wolf optimizer (GWO).

4.1 Solution or Agent Representation In Fig. 1, a string of zeros and ones represents the solution or agents and the length of every agent represents the number of possible positions. In this we use an encoding scheme for base station placement, if the value of the ith position is 1 then a base station is placed at ith positions, otherwise no base station at ith positions. In addition, for channel allocation, we initialize a separate encoding scheme. For each selected potential positions, we define an array of 14 elements, as shown in Fig. 1, where a number of blocks represented by the very first position of the array, the number of channels are represented by second position of the array, the channel lending information of its six neighbors is represented by next six positions and the last six positions collect channel borrowing information form six neighbor cells.

234

S. Kumar and M. S. Gaur

Fig. 1 Mobile network with 6 users and 5 potential positions Solution or agent structure

Proposed Algorithm

Derivation of Fitness Function. On the basis of the objective of this approach the capability of the agents and solution can be measured in terms of its quality level. Chosen least Number of Potential Position. In order to deploy the base station we have to choose at least M potential location out of K potential locations.

Call Admission Control in Mobile Multimedia Network …

 ConCost(u i ) =

k, if |Cov(u i ) ≥ k k − |Cov (u i )|, Otherwise

235

(8)

Hence, the second objective function is defined as follows: Objective 2: Maximum O F2 =

N  1 ConCost(u i ) N × K i=1

(9)

For the channel allocation problem, we use an objective function as follows: Each cell O F3 =

M 

block host + reserved channels + prime channels

(10)

i=1

where, M are the selected BTS. On the basis of the above description, it is found that the objectives vary from each other. In order to handle these conflicting objectives we developed a multiobjective fitness function by using the weight sum approach (WSA) [22]. WSA, is a traditional approach that can solve the multi-objective optimization problem. Here each objective Oi is multiplied by a weight value w j . At last, the sum of all objective function is converted into a single scalar objective function as follows Fitness = w1 + O F1 + w2 × (1 − O F2 ) + w3 × O F3

(11)

In our work we take w1 + w2 + w3 = 1 to minimize the Fitness value.

5 Simulation Result and Discussion The parameters of the proposed system are set as follows: the number of search agents is set to be 100, the number of iterations is equal to be 500. Experimental results related to Minimum BTS selection For the experimental study we indicate the starting population of 100 agents to implement the proposed algorithm. Figs. 2 and 3 represent that as the number of potential locations increases, the number of chosen location decreases. The parameter description is presented in Table 1. In Figs. 4 and 5 it is clearly represented that our proposed method achieves better performance than existing ones.

236

S. Kumar and M. S. Gaur

Fig. 2 Comparison in terms of selected BTS nodes for random order users

Fig. 3 Comparison in terms of selected BTS nodes for non-random order users

Call Admission Control in Mobile Multimedia Network … Table 1 Simulation parameter

237

Parameter

Data

Number of potential position

100–600

Number of mobile users

150–250

Communication range

150 m

Initial population size

150

Iteration

600

Cells

40

Channels

50, 100

Reserved channels

16, 30

Fig. 4 Comparison in terms of number of selected potential positions in random

6 Conclusion In this paper, a Grey Wolf Optimizer Algorithm (GWO) inspired algorithm was proposed to manage CAC based on handoff prioritization to solve resource allocation problems in wireless networks. In this GWO is used to search the least figure of selected potential position for deployment of BTS in order to obtain the maximum coverage of BTS nodes in mobile networks. This method is required when there is a need to supervise the maximum population targeted area at the same time by the BTS nodes. First of all, the linear programming formulation of the proposed problem is presented and after that GWO based method is applied to show the proper solution of the proposed problem by using fitness function derivation and updating operations. In addition to this, this function has been repeated by various potential

238

S. Kumar and M. S. Gaur

Fig. 5 Comparison in terms of number of selected potential positions in non-random

locations to maximize the population locations by applying different schemes of mobile networks. As a number of possible locations are selected, our proposed model shows the performance of the system in terms of decreasing call rate failure than the existing method.

References 1. Marappan, R, Sethumadhavan G, Harimoorthy U (2016) Solving channel allocation problem using new genetic operators—an experimental approach. Perspect Sci 2. Beutler R (1998) Digital single frequency networks: improved optimization strategies by parallel computing. Frequenz 52(5–6):90–95 3. Fallot-Josselin S (1998) Automatic radio network planning in the context of 3rd generation mobile systems. In: Cost 259, Duisburg, Germany, Sept 1998 4. Lieska K, Laitinen E, Lähteenmäki J (1998) Radio coverage optimization with genetic algorithms. In: Proceedings of the 9th IEEE international symposium on personal indoor and mobile radio communications, PIMRC’98, pp 318–321 5. Monlin A, Athanasiadou G, Nix A (1999) The automatic location of base stations for optimized cellular coverage: a new combinatorial approach. In: Proceedings vehicular technology conference ’99, Houston, Texas 6. Yu C, Subramanian S, Jain N (1998) CDMA cell site optimization using a set-covering algorithm. In: Proceedings of the 8th international telecommunication network planning symposium, Sorrento, Italy, Oct 1998 7. Wong SH, Wassell I (2002) Dynamic channel allocation using a genetic algorithm for a TDD broadband fixed wireless access network. Laboratory Communications, University of Cambridge, Cambridge, UK

Call Admission Control in Mobile Multimedia Network …

239

8. Kassotakis IE, Markaki ME, Vasilakos AV (2000) A hybrid genetic approach for channel reuse in multiple access telecommunication networks. IEEE J Sel Areas Commun 18(2):234–243 9. Huang NF, Liu HI (1998) A study of isochronous channel reuse in DQDB metropolitan area networks. IEEE/ACM Trans Netw 6(4):475–484 10. Khanbary LMO, Vidyarthi DP (2008) A GA-based effective fault-tolerant model for channel allocation in mobile computing. IEEE Trans Veh Technol 57(3) 11. Khanbary LMO, Vidyarthi DP (2009) Reliability-based channel allocation using genetic algorithm in mobile computing. IEEE Trans Veh Technol 58(8) 12. Kumar S, Kumar K, Pandey AK (2016) Dynamic channel allocation in mobile multimedia networks using error back propagation and Hopfield neural network. Proc Comput Sci 89:107– 116 13. Kumar S, Kumar K, Pandey AK (2016) Hopfield neural network based dynamic call admission control for quality of service provisioning in mobile multimedia networks. IJIP 10(3):88–97. ISSN: 0973-8215 14. Kumar S, Kumar K, Kumar P (2015) Mobility based call admission control and resource estimation in mobile multimedia networks using artificial neural network. In: 1st international conference on next generation computing technologies (NGCT-2015), 4–5 Sept 2015, pp 852– 857. ISBN: 9781467368070 15. Kumar S, Kumar K, Pandey AK (2014) A comparative study of call admission control in mobile multimedia networks using soft computing. Int J Comput Appl 107(16). ISSN: 0975-8887 16. Luo H, Shyu ML (2011) Quality of service provision in mobile multimedia- a survey. Springer J Human Centric Comput Inf Sci 1(1):1–15 17. Gaur MS, Pant B (2015) Impact of signal-strength on trusted and secure clustering in mobile pervasive environment. Elsevier Proc Comput Sci 57:178–188 18. Gaur MS, Pant B (2014) A bio-inspired trusted clustering for mobile pervasive environment. In: Proceedings of the third international conference on soft computing for problem solving (SocPros2013). Published by Springer series advances in intelligent systems and computing AISC, vol 259, pp 553–564. ISSN: 2194-5357 19. Xia Z, Hao W, Yen I-L, Li P (2005) A distributed admission control model for QoS assurance in large-scale media delivery systems. IEEE Trans Parallel Distrib Syst 16(12):1143–1153 20. Gaur MS, Pant B (2015) Trusted and secure clustering in mobile pervasive environment. Springer Open J Human-Centric Comput Inf Sci (HCIS). https://doi.org/10.1186/s13673-0150050-1 21. Yang J, Jiang Q, Manivannan D, Singhal M (2005) A fault-tolerant distributed channel allocation scheme for cellular networks. IEEE Trans Comput 54(5):616–629 22. Mirjalili S, Mirjalili SM, Lewis A (2014) Grey Wolf optimizer. Adv Eng Softw 69:46–61

Feature Classification and Analysis of Acute and Chronic Pancreatitis Using Supervised Machine Learning Algorithm R. Balakrishna and R. Anandan

Abstract Supervised Machine learning algorithms play a vital role in prediction of diseases, and the death rate comparatively reduced in twenty-first century; one of the most predominant diseases is acute and chronic Pancreatitis in which only 7% of patients were alive after treatment since the cancerous cells spread to all other parts of the pancreas gland. This marks the importance of diagnosing cancer at an incipient stage and as a result in this work more than 3000 CT scan tumor images from 82 patients around the world are considered. In preprocessing stage the speckle noise is removed using wiener’s filter and then the normalized images are partitioned by enhanced region-based active contour (ERBAC) to find the region of interest (ROI). The features are extracted from the segmented images by gray Llevel co-occurrence matrix (GLCM) and the extracted features are classified by using KNN and SVM classification algorithm. Based upon the results it is found that KNN produces 97.2% of accuracy that the patients are diagnosed at incipient stage. This feature classification may be further improved using artificial neuro-fuzzy inference system (ANFIS) to predict with 99% accuracy. Keywords Image processing · Enhanced region-based active contour · Region of interest · Feature extraction · KNN classification and preprocessing

R. Balakrishna (B) · R. Anandan Department of Computer Science and Engineering, School of Engineering, Vels Institute of Science Technology and Advanced Studies (VISTAS), Chennai, India e-mail: [email protected] R. Anandan e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_28

241

242

R. Balakrishna and R. Anandan

1 Introduction 1.1 Division of Pancreas Pancreas is a vital organ which plays an essential role in converting the food we have into liquid for the proper functioning of our body cells. Pancreas provides two major functions called endocrine and exocrine, in which the former function is used to improve the sugar level of the blood and the latter function will support our digestion system [1]. The structure of pancreas is shown in Fig. 1 which highlights the different portions of pancreas.

Fig. 1 Structure of pancreas

Feature Classification and Analysis of Acute and Chronic …

243

1.2 Pancreas Location and Pancreatic Cancer The pancreas is generally located at the back of the stomach, in the top left portion of the abdomen surrounded by liver, spleen, and small intestine. As discussed above, the head part is located at the junction of stomach meeting the small intestine where the food are converted into fluids; the body or neck lies in the mid-region of pancreas and finally the thin end called tail is at the left-most end of pancreas. Tumors in the pancreas’ cells are called pancreatic cancer [2]. Generally there are two types of tumors in pancreas such as endocrine tumors which occurs 10%, and exocrine tumors which occurs in the tissues of exocrine which accounts 95% of pancreatic cancer, this type of tumors is also called as pancreatic adenocarcinoma, also called pancreatic ductal adenocarcinoma. Apart from pancreatic adenocarcinoma other exocrine tumors are mucinous cystadenocarcinoma, adenosquamous carcinoma and acinar cell carcinoma [3].

1.3 Symptoms and Diagnosis of Pancreatic Cancer The initial stage of pancreatic cancer is usually quiet and the symptoms become visible or observed only when the tumors spread to other parts and affect the functions of pancreas. Generally, the symptoms for tumors present in head and body occurs due to the squeezing of neighbor parts like mesenteric nerves, duodenum, pancreatic duct, bile duct, etc. Magnetic resonance imaging (MRI) and computed tomography (CT) are the two most common diagnostic tools, which are used for diagnosing the tumors in pancreas. The main function of CT scan is to analyze and determine the location of tumors in pancreas and also to identify the ratio of tumors spread to the liver [4]. The architecture of current work is shown in Fig. 2, which illustrates the overall workflow of the research.

Fig. 2 Architecture of the system

244

R. Balakrishna and R. Anandan

2 Materials and Methods 2.1 Preprocessing and Enhancement of Images Preprocessing of intensity images is the next common step to improve or enhance the quality of images by processing data with low level of reflection and produce the desired results. In this work, we considered wiener filter for removing unwanted noises or distortions and this can be done by blurring the images using low pass filter and it is resorted by inverse filtering and at the same time it removes additive noise of the images and the result is optimal in terms of mean square error (MSE). The Fourier domain for wiener’s filter is W F(F1, F2) =

B ∗ (F1, F2)NKK (F1, F2) |K (F1, F2)|2 NKK (F1, F2) + NXX (F1, F2)

(1)

where B ∗ (F1, F2) is used to remove the blur and NKK (F1, F2), NXX (F1, F2) is the power spectra and addictive noise of the images. Image enhancement is the next process in preprocessing, where the images are processed to improve or enhance the contrast, sharpening of images for better analysis. There are several image enhancement techniques available and in that we considered histogram equalization, where the contrast of the images is enhanced by altering the intensities of pixels. Let us consider an image I and n x by n y matrix be pixel integer values ranging from 0 to N − 1 where N is the possible intensity values, usually 256. Let q be the normalized histogram of I, qn =

No. of pixel with intensity n Total number of pixels

(2)

where n = 0, 1, 2, . . . . . . .N − 1 Let H be the histogram equalized images and it is defined as Hn,m = floor(N − 1)

In,m 

qn

I =0

where floor () is used to round down to closest integer.

(3)

Feature Classification and Analysis of Acute and Chronic …

245

2.2 Enhanced Region-Based Active Contour (ERBAC) Segmentation Segmentation is one of the crucial processes in image processing [5] which divides or partitions images into segments having similar properties. There are several segmentation techniques available in image processing and we considered region-based segmentation for our work called modified region-based active contour, where partition constraints and energy forces of the pixels are used for further processing and analysis. The term active contour is referred as active model for the segmentation process and contour is used to denote the barrier defined for the region of interest of an image [6]. The main objective of enhanced region-based active contour method is to define sharpness of the images and develop sealed contour for the location.

2.3 Feature Extraction Using GLCM Feature extraction is one of the types of data reduction techniques and its main objective is to extract the useful features or information from the images. In order to distinguish the different features of images, some local features of the images are utilized and these features are classified on different key elements of image data like color intensity, textures data, shape, or edges of the images [7]. In GLCM, a matrix is created where RGB values of an image are converted into the number of rows and columns which is similar to the number of gray levels G of an image. Figure 3 illustrates the transformation of grayscale image into gray level co-occurrence matrix (GLCM).

(a)

(b) 10

9

8

7

6

5

4

3

2

1

6

6

6

6

6

6

6

6

6

6

3

3

3

3

3

3

3

3

3

3

1

2

3

4

5

6

7

8

9

10

Noted Pixel Fig. 3 Sample input image and gray level values

Acquaintance Pixel

246

R. Balakrishna and R. Anandan

2.4 Image Classification Using K-Nearest Neighbor Algorithm There are several classifiers in image processing for classification and we considered k-nearest neighbor (K-NN) for our work. In K-NN algorithm, K refers to total number of nearest neighbor and the core deciding factors is the number of nearest neighbors available. Therefore, higher the K values will give better smoothing effects which make the classifiers more resistant to outliers. The following illustrates the K-NN algorithm, Start,   Formulate the K pivotal points pp1 , pp2 , . . . . . . . . . . . . , ppn , For each image I, test,   Step 1: search for k1 pivotalpoints pps1 , pps2 , . . . . . . . . . . .. ., ppsk1 . In accordance to categories ps1 , ps2 , . . . . . . . . . . . . . . . , pski1 . Step 2: calculate coincidence images r,   for all the images in categories ps1 , ps2 , . . . . . . . . . . . . . . . , pski1 . Step 3: Constraint of KNN is used for labeling.

3 Results and Discussion 3.1 Preprocessing of Images The acquiesced images are preprocessed to remove unwanted noises or data by using different filtering methods. By considering the three parameters like Peak signal-tonoise ratio (PSNR), signal-to-noise ratio (SNR), and mean square error (MSE), we considered three filtering techniques: Gaussian, Median, and wiener’s filter which gives maximum results for noise removal and the output is shown in Fig. 4a, b.

(a)

(b) 200

MSE PSNR

0 I1 I4 I7 I10 I13

SNR

Fig. 4 Wiener filtered Image and Wiener filter graph where MSE, PSNR, SNR are referred as mean square error, Peak signal-to-noise ratio and signal-to-noise ratio

Feature Classification and Analysis of Acute and Chronic …

247

Fig. 5 Segmented image

3.2 Enhanced Region-Based Active Contour (ERBAC) Segmentation The filtered images are now segmented using modified region-based active contour method to extract the exact location of tumors in pancreas for feature extraction and the segmented image is shown in Fig. 5.

3.3 Feature Extraction Using GLCM From the segmented images we use Gray level co-occurrence matrix (GLCM) to extract the texture features like contrast, energy, homogeneity, etc., from the pancreatic tumor and non-tumor cells and Fig. 6 illustrates the scatter plot diagram of GLCM values. (a)

(b)

250

Img1

200

Img2 Img3

150 100 50 0

Img4 Img5 Img6 Img7 Img8

Fig. 6 a Scatter plot of GLCM values and b scatter plot of energy and entropy

248

(a)

R. Balakrishna and R. Anandan

(b)

Fig. 7 a Parallel coordinates and b confusion matrix

3.4 Classification of Images K-NN classifier is used to classify the tumor and non-tumor cells and the performance measures for scatter plot, parallel coordinates, confusion matrix are shown in Figs. 6 and 7.

3.5 Performance Measure The accuracy of the proposed system is calculated for KNN classifier and the value is 97.2%. The processing speed of the developed CAD system is 6000 obs/s and the training time is calculated as 3.7012 s.

4 Conclusion and Future Enhancement In this work, it is found that the KNN classifier is more predominant than SVM when the region of interest is closer to the tumor cells and it shows 97.2% of accuracy during prediction. The test data is compared with clinical data and it is found to be acceptable at the incipient stage. In 3000 samples of CT scan images, 1750 images are defected with nine associated features and remaining 1250 images are assumed to be normal. In this 1750 images, 649 images are found to be in worst case, 550 images are found to be defected 60% with cancer and 551 images are found to be defected with 80% of cancer. This work may be further extended to human computer interaction using artificial neuro-fuzzy classifier to increase efficiency.

Feature Classification and Analysis of Acute and Chronic …

249

References 1. Makkar S, Solanki VK, Big data and machine learning progression with industry adoption. A handbook of IoT and Big Data. CRC Press, Taylor and Francis 2. Balakrishna R, Anandan R (2018) Early diagnosis of chronic and acute pancreatitis using modern soft computing techniques. Int J Recent Technol Eng (IJRTE) 7(4S2). ISSN: 2277-3878 3. Devi BA, Rajasekaran MP (1921) Performance evaluation of MRI pancreas image classification using Artificial Neural Network (ANN) 671–681 4. Zhu M, Xu C, Yu J, Wu Y, Li C, Zhang M, Jin Z, Li Z (2013) Differentiation of pancreatic cancer and chronic pancreatitis using computer-aided diagnosis of endoscopic ultrasound (EUS) images: a diagnostic test. PLoS ONE 8(5):63820 5. Kamavisdar P, Saluja S, Agrawal S (2013) A survey on image classification approaches and techniques 2(1):1005–1009 6. Padmasheela Keshvan, Anandan R, Balakrishna R (2017) Segmentation of pancreatic tumor using region based active contour. J Adv Res Dyn Control Syst (JARDCS) Spl Issue (04) 7. Srivastava D, Wadhvani R, Gyanchandani M (2015) A review : color feature extraction methods for content based image retrieval 18(3):9–13

RUDRA—A Novel Re-concurrent Unified Classifier for the Detection of Different Attacks in Wireless Sensor Networks S. Sridevi and R. Anandan

Abstract Wireless Sensor networks find its applications in most fields such as health care, automotive, consumer electronics and most importantly Industry 4.0 automations. Nowadays integration of Internet of things (IoT), Wireless sensor networks (WSN) has become the most prominent and ubiquitous in day to day life, but these systems still suffer from the different security attacks, which makes the connected devices under immense pressure for an efficient and secured data transfer. To overcome this issue, intrusion detection system using hybrid artificial intelligence algorithm has been proposed for the better detection and classification. The paper proposes the most intelligent attack detection system (IADS) RUDRA which works on the principle of recon current LSTM networks (Long Short-Term memory) along with extreme learning machines (ELM) which is then used for detection of the different DoS attacks such as Sybil, wormhole, black hole, sinkhole and selective forwarding attacks. The proposed tool works on three different phases, such as feature decomposition, hybrid learning and decision phase. The real-time datasets were collected on the test bed which consists of RISC architecture as main CPU interfaced with CC2540 transceivers. Also, the proposed tool integrated with the hybrid classifier has been compared with other existing algorithms such as RNN-LSTM, ELM and SVM in which the accuracy of 98.4% is obtained for the proposed classifier. Keywords LSTM · RUDRA · Ubiquitous · IoT (Internet of things) · RNN · ELM · Industry 4.0 · DoS (Denial-of-Services)

S. Sridevi (B) · R. Anandan Department of Computer Science and Engineering, School of Engineering, Vels Institute of Science Technology and Advanced Studies (VISTAS), Chennai 600117, India e-mail: [email protected] R. Anandan e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_29

251

252

S. Sridevi and R. Anandan

1 Introduction Wireless sensor networks gather a brighter light of research due to its characteristics of implementing in all fields of applications. Wireless sensor network normally consists of the autonomous sensors interfaced with main CPU and transceivers. These networks can collect the information from the remote place and transfer to the WSN gateways which are primarily known as sink. By the end of 2020, it will be 14.4 billion connected devices [1], in which the wireless sensor networks play a very important and pivotal role for the data collection, transmission and reception. WSN’s are mostly vulnerable to security attacks due to their open and distributed nature of the sensor nodes [2]. Also, when the WSN broadcasts every time, it will be easier for the attackers to affect the networks. The attackers can affect the networks by injecting the fault messages, alter the integrity of the networks, eavesdrops the messages and waste the network resources [3]. Denial-of-Service (DoS) attacks are considered to be more important in WSN since it cause serious threats to the networks. Since the hardware resources for the sensor nodes remain to be very tiny and resource constraint, implementing the attack identification and detection system remains to be more challenging among the researchers [4]. Besides, non-availability of WSN datasets for integrating the intelligent algorithms for the detection adds to the challenges. Considering the above challenges, we design the intelligent attack detection systems (IADS) by integrating the strong deep learning and machine learning algorithms [5] to obtain high accuracy in identification of attacks [6].

2 Re-concurrent Unified Classifier for the Detection of Different Attacks in Wireless Sensor Networks [RUDRA] 2.1 Working Architecture of the RUDRA Figure 1 shows the working architecture of the proposed tool. The working mechanism has been detailed as three different stages such as feature extraction and decomposition, hybrid classification and decision stages.

2.2 Feature Extraction and Decomposition The various parameters were measured experimentally or by changing the different layers of transceivers using embedded C programming on the main CPU. The features which are obtained from the experimental setup are as follows : Node_ID (NID), RSSI (R): The RSSI (Received Signal strength Indicator) [7], Distance (D), Initial Energy

RUDRA—A Novel Re-concurrent Unified Classifier …

253

Where D= feature separator/decomposer Fig. 1 Overall architecture for the proposed RUDRA TOOLS for Intelligent attack detection System

(I.E): The initial Energy (I.E), Consumed Energy (C.E), Residual Energy (R.E), Cluster Head ID, Cluster head distance (dCH ), Data Sent from Base station to Nodes (Nbs ), Data Sent from Nodes to Base station (Nts ), Data sent from the Cluster head to Nodes (NcS ), Data sent from the Cluster head to base station (Nbc), Throughput (T).

2.3 RUDRA’s Classifier Mechanism The proposed tool integrates the re-concurrent network with LSTM for the energy prediction thresholds and other features for the extreme learning machines for classification of the different attacks.

2.4 Hybrid Integration Mechanism The inputs to the different learning algorithms have to be decomposed from the features which are extracted from the test beds. The Energy related inputs such as the remaining energy, residual energy and consumed energy along with the different distance at different iterations are given as Inputs to LSTM [8] for the Energy prediction (Ep) outputs. These outputs along with the other features are given as the inputs to the ELM for the better classification and detection of the different attacks [9].

254

S. Sridevi and R. Anandan

Consider that, for the ith input Pl in the training dataset is L  P l ; y l |P l ∈ R l , y l ∈ R l=1 . The practical outputs of the LSTM are Y s (Pl ) and ELM are Y e (Pl ). Then the training dataset for linear classification is given  L  as ys (P l ), ye (P l ); y l l=1 , [10]. Suppose the classification model for the hybrid model is expressed as 

yˆ ( p) = C0 + C1 Ys (P) + C1 Ye (P) the newly generated training dataset We expect that



ys (P l ), ye (P l ); y l

(1)

 L l=1

,.

C0 + C1 Ys (P l ) + C1 Ye (P l ) = y l l = 1, 2, 3 . . . . . . .L

(2)

Then the above equations can be changed as Ac = y ⎤ 1 Ys (P 1 ) Ye (P 1 ) ⎢ 1 Ys (P 2 ) Ye (P 2 ) ⎥

T ⎥ ⎢ A=⎢. ⎥C = [C0 C1 C2 ]T Y = y1 y2 yL .. .. ⎦ ⎣ .. . . 1 Ys (P L ) Ye (P L ) ⎡

(3)

Finally, hybrid model parameters can be written as 

C = A+

(4)

where A+ → Moore-Penrose Pseudo inverse Matrix of A.

3 RUDRA-Dataset Collection Table 1 represents the construction of the datasets collected from the real-time scenario for first iteration which consists of 15 nodes iterated for 15 h of time duration. For construction of the whole datasets, we have 20 iterations and collected nearly 1000 datasets which comprise of both the normal data and attack data.

4 Experimental Setup Figure 2 shows the experimental setup for the construction of the datasets and evaluating the proposed protocol. The hardware setup consists of the Ardiuno/Node MCu as the main CPU which operated at 3.3 V at 500 ma current which is then interfaced with TI CC2540 transceivers which also operates in the 3.3 V [11]. The

100

56

43

18

19

78

11

17

76

10

100

100

09

100

100

08

16

100

07

15

100

06

100

100

05

14

100

204

74

100

03

100

100

02

13

100

01

12

Throughput

Node ID

D

2

3

4

3

3.2

4

2.6

2.7

2.8

2.5

2.6

2.7

2

3

4

3

3.2

5

4

RSSI

−67

−68

−67

−67

−68

−69

−67

−67

−68

−68

−68

−67

−67

−68

−67

−67

−68

−90

−98 0.12

0.45

0.12

0.12

0.12

0.12

0.12

0.10

0.19

0.20

0.12

0.12

0.12

0.12

0.12

0.12

0.12

0.12

0.12

C.E

0.33

0.00

0.33

0.33

0.33

0.33

0.33

0.30

0.27

0.33

0.33

0.33

0.33

0.33

0.33

0.33

0.33

0.33

0.33

R.E

03

02

03

03

03

03

03

12

12

03

03

03

03

03

03

03

03

03

03

CHD

Table 1 Construction of datasets for evaluating the RUDRA IADS tool

3

3

2

2

2

3

2

4

4

2

2

2

2

2

2

2

2

3

2

dCH

02

00

08

08

08

08

08

03

04

04

08

08

08

08

08

08

08

08

08

NBS

03

00

08

08

08

08

09

01

02

04

08

08

08

08

08

08

08

08

09

Nts

02

00

09

09

09

09

08

00

00

03

09

09

09

09

09

09

09

09

08

Ncs

01

00

08

08

08

08

08

01

00

03

08

08

08

08

08

08

08

08

08

Nbc

0.33

0.45

0.33

0.33

0.33

0.33

0.33

0.24

02.21

0.23

0.33

0.33

0.33

0.33

0.33

0.33

0.33

0.33

0.33

Ep

Sinkhole

Selective forwarding

Normal

Normal

Normal

Normal

Normal

Sybil

Wormhole

Blackhole

Normal

Normal

Normal

Normal

Normal

Normal

Normal

Normal

Normal

C

RUDRA—A Novel Re-concurrent Unified Classifier … 255

256

S. Sridevi and R. Anandan

Fig. 2 Experimental setup

initial energy at startup phase is considered as 0.45 mJ. Fifteen nodes are considered for the iterations and distance of maximum 5–6 m is considered as the radius of circumference.

5 Results and Discussion The following parameters were evaluated for proving the significance of the proposed RUDRA tool and it is compared with the other existing machine learning algorithm such as the extreme learning machines (ELM), support vector regressors (SVR) and finally LSTM [5]. Detected Results × 100 Total Number of Iterations

(5)

Sensitivity =

TP (True Positive) × 100 TP (True Positive) + TN (True Negative)

(6)

Specificity =

TN (True Negative) × 100 TP (True Positive) + TN (True Negative)

(7)

Accuracy =

RUDRA—A Novel Re-concurrent Unified Classifier …

257

Fig. 3 Comparative analysis of accuracy detection

Figure 3, comparative analysis of proposed RUDRA tool along with the existing algorithm for the detection of different attacks. It shows the accuracy of detection of different attacks by the proposed RUDRA classifier along with existing algorithms. It clearly shows that the accuracy of RUDRA is 99.5% which outperforms other algorithms which have 82, 76 and 70% accuracy of other existing algorithms. After evaluating the accuracy of detection, sensitivity and selectivity for different attack detection for the proposed RUDRA classifier and comparative analysis between the other algorithms are evaluated and presented in Table 2.

6 Conclusion The proposed RUDRA algorithm has been designed based on the ensemble of LSTM and ELM with 50 neurons for the detection of different DoS attacks. The performance of the proposed RUDRA algorithm has better accuracy in detection, better sensitivity and selectivity than the existing algorithm. The proposed classifier seems to be more intelligent in detection of the attacks and finds its application even when WSN amalgamated with IoT. The algorithm can be improvised by testing with more number of attacks since the paper deals with only five different types of DoS attacks. Even 100% accuracy of detection is obtained by tuning the hidden layers of the proposed RUDRA algorithm.

Selective Forwarding attacks (%)

80

76

70

ELM

LSTM

SVR

99

73

73.6

81.5

98.5

71

77.4

82

98.4

72.4

72

82.4

98

70.3

75

83.5

98.4

73

78

80

99

74

77.3

82

98.5

Worm Hole attacks (%)

Black Hole attacks (%)

SinkHole attacks (%)

Sensitivity analysis (%) Sybil attacks (%)

Black Hole attacks (%)

Worm Hole attacks (%)

Selectivity analysis (%)

RUDRA

Algorithms used

Table 2 Comparative Analysis of different algorithms for different attacks

73

78

84

98

Sybil attacks (%)

72.4

74

83

97

SinkHole attacks (%)

70.3

75

83.5

98

Selective Forwarding attacks (%)

258 S. Sridevi and R. Anandan

RUDRA—A Novel Re-concurrent Unified Classifier …

259

References 1. Tian B, Yao Y, Shi L, Shao S, Liu Z, Xu C (2013) A novel Sybil attack detection scheme for wireless sensor network. In: IEEE international conference on broadband network & multimedia technology, pp 294–297 2. Bijalwan A, Minch A, Gamo Gofa E, Solanki VK, Pilli ES, Botnet forensic: issues, challenges and good practices 10(02):28–51 3. Chatterjee JM, Kumar R, Pattnaik PK, Solanki VK, Zaman N (2018) Privacy preservation in data intensive environment. Tour Manag Stud 14(2):72–79 4. Patil DS, Patil SC (2017) A novel algorithm for detecting node clone attack in wireless sensor networks. In: International conference on computing, communication, control and automation (ICCUBEA), pp 1–4 5. Harpal E, Tejpal G, Sharma S (2017) Machine learning techniques for wormhole attack detection techniques in wireless sensor networks. Int J Mech Eng Technol (IJMET) 8(9):337–348 6. Sheela D, Priyadarshini, Mahadevan G (2011) Efficient approach to detect clone attacks in wireless sensor networks. In: International conference on electronics computer technology, Vol 5, pp 194–198 7. Anwar RW, Bakhtiari M, Zainal A, Abdullah AH, Qureshi KN (2015) Enhanced trust aware routing against wormhole attacks in wireless sensor networks. In: International conference on smart sensors and application (ICSSA), pp 56–59 8. Shimpi B, Shrivastava S (2016) A modified algorithm and protocol for replication attack and prevention for wireless sensor networks. In: International conference on ICTBIG, pp 1–5 9. Chen D, Zhang Q, Wang N, Wan J (2018) An attack-resistant RSS-based localization algorithm with L1 regularization for wireless sensor networks. In: IEEE advanced information management, communicates, electronic and automation control conference (IMCEC), pp 1048–1051 10. Kumar S (2014) Improving WSN routing and security with an artificial intelligence approach. In: DWAI@AI*IA 11. Marano S, Matta V, Tong L (2006) Distributed detection in the presence of Byzantine attack in large wireless sensor networks. In: IEEE military communications conference, pp 1–4

Health-Care Paradigm and Classification in IoT Ecosystem Using Big Data Analytics: An Analytical Survey Riya Biswas, Souvik Pal, Bikramjit Sarkar and Arindam Chakrabarty

Abstract The Indian healthcare system is in a dilapidated state. Healthcare is important to society because people get ill. Healthcare is defined as the diagnosis, treatment, prevention and management of disease, illness and preservation of physical and mental well-being in humans. In our paper we have done healthcare surveys to analyze the aspects. In this paper some aspects of IoT healthcare and big data analytics are discussed Big data can be used for better health planning. It’s methodology can be used for healthcare data analytics which helps in better decision making. IoT is the fast developing wireless and web technologies sensors are used to predict the disease supported on IoT are used to develop the healthcare sector. Hence it is assertive that we do, various classifications in IOT are discussed. Hence it is assertive that we do initial surveys on the concept of Big Data, on Healthcare aspects and IoT ecosystem as how we can manage to handle large data files. Keywords Big data · IoT · Healthcare device · IoT ecosystems

1 Introduction Cumbrous amount of structured and unstructured data it is used to describe by Big data which is a buzzword. Some characteristics of Big data [1, 2]. As healthcare sector is expanding extremely. The volume of produced data is rapidly expanding every year R. Biswas (B) · B. Sarkar Department of Computer Science & Engineering, JIS College of Engineering, Kalyani, India e-mail: [email protected] B. Sarkar e-mail: [email protected] S. Pal Department of Computer Science & Engineering, Brainware University, Kolkata, India e-mail: [email protected] A. Chakrabarty Department of Management, Rajiv Gandhi University, Itanagar, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_30

261

262

R. Biswas et al.

due to existence of unique technologies, appliance and transmission [3]. In the current analysis of smart phones and wearable devices, endless figure of health data folder of patient from various challenges continue featured by healthcare industry [4]. Mostly complication arise where system proceed through divergent data sets [5, 6]. Various aspects of IoT is being discussed in this paper i.e Cancer which is a unchecked expansion unusual unit in any place in a body. Heart disease which is a dominating to conditions to that action area that influenced our heart and affect diminish or clog blood vessels. HIV/AIDS is a germ that mark and alter exempt system. Asthma is one of the most common chronic diseases that has a intelligent impact on people’s well-being and in our society. Diabetes is a scheme of metabolic diseases consist of high blood sugar levels concluded lengthy season. For tracking the disease IoT are implemented. IOT basically a model for interconnecting sensor which track, sensing, process and diagnosis [7]. IOT basically composite of physical objects and domain where enclosed device content across the internet [8, 9]. IoT assist self management of disease. Over internet areas like health, Logistics, industry, security, agriculture and environment etc are basically empowered by the IOT appliances [10]. In this paper, we are going to discuss literature survey and classification in the Sect. 2. Section 3 deals with the analytical survey, table and policy design

1.1 Motivation Big data is not only data it has turn into a entire subject, which involves different device, approach and scheme. Big data is transformative attempt in day-day-life. As in present day there is immense bulk of data, examining these acceptable sets which encompass of structure and unstructured data of various type and size; big data analytics grant the user to evaluate the impractical data to generate a faster and superior judgment. It is establish that big data is calculated to expanding rapidly in healthcare than in other sectors like manufacturing, financial services or media. Big Data and Analytics as with the Internet of Things (IoT). The term big data is one of burning technology. The big data analytics in healthcare covers assimilation and investigation of huge amount of data of complicated heterogeneous data. since Big Data can be advantage to consider user data and the prescribed assistance. It will trying to design program that will allow health care to reach those area where access to hospital was somewhat limited. IoT refers to the computerized intelligent curb and direction of connected associated devices over boundless regions via sensors and other computing capacity.

2 Classification and Analytical Survey In this section, we will discuss the IoT-based healthcare paradigm and its classification. We have also discussed the analytical survey in this related field.

Health-Care Paradigm and Classification in IoT Ecosystem …

263

Fig. 1 Classification of IoT aspects

Figure 1 describes the different aspects of IoT-based healthcare. Chavan et al. [9] discussed for creating, acquiring, comparing some technologies like Hadoop, HDFS, Map Reduce, Pig, Hive, HBase are used here. Khan et al. [11] It describes proposed data life cycle which utilize the technologies and nomenclature of Big data management, investigating and scarceness. Nizam et al. [12] discussed that Big data is a type of dataset which is very massive and complicated that get difficultly to computing them exploiting traditional data processing applications. Chen et al. [13] discussed Big data then study about the connected technologies i.e. cloud computing, Internet Of Things, data centers and Hadoop. Archenaa et al. [14] deliver about the perception of how we expose newly expose surplus. Prasad et al. [15] author discussed that diabetes is one of the leading non-communicable disease. This system will prophesy exploring algorithm in Hadoop/Map Reduce environment. Huzooree et al. [16] author explains that Diabetes Mellitus (DM) is one of the starring health hindrance about the world initiating national economical concern. King et al. [17] author discussed that the asthma is characterized by hyper-responsiveness and can be avert by convenient benefit of remedial assistant to conduct asthma charge. Alpert et al. [18] author discussed that the heart failure is an developing public health complication with huge morbidity and probity. Stewart et al. [19] author describes the cardiovascular disease is a compelling and constantly-developing complication. Simon et al. [20] author describes that the HIV-1 pandemic is a complex blend of distinct contiguous. This paper brings on epidemiology. Bhatti et al. [21] author describes that the exploration of the human immunodeficiency virus (HIV) as the original organism of captured immunodeficiency disorder (AIDS). Constatine et al. [10] the author described that breast cancer is one of the biggest fatal disease of world. This paper proposed machine learning algorithm that for Big data analysis leading of an map reduce and mahout. Priya et al. [22] author describes that in first phase, min max normalization algorithm is enforced. Second phase by need of pso character choice. In fourth phase the efficiency will be determined accepting root men square value. Nahar et al. [23] initial forecast of liver disease is very crucial to deliver life and holding appropriate step to curb the disease. Shandilya et al. [24] in the current age automation medical field has develop into one of the favored affair of researcher and cancer. This paper generate survey of such current research study that cause usage of online and offline data for cancer classification. Alharam et al. [25] author describes. The main aim of this paper is for conserving healthcare industry from attack of cyber. Kumar et al. [26] author describes that traditional health center based

264

R. Biswas et al.

approach healthcare is identified with the arrival of large precision sensors and IOT. Kumbi et al. [27] author describes that the IOT is the leading network infrastructure of shipment of connectivity, transportation Technology which is proposed healthcare by the IOT.

3 Analysis of IoT Devices for Healthcare See Table 1.

4 Policy Design and Constraints in Implementing in India India seriously needs for reforms in their policy mix particularly in the field of health sector. The world is preparing them to welcome and grab the opportunity matrix that is emerging through the incorporation of 4th Industrial revolution. This is the high time to prepare for optimal participation of Indian firms. The following policy interventions may be exercised. (i)

The 4th Industrial revolution has brought gigantic opportunities in the field of IoT, RFID led ecosystem primarily in the domain of health care sector. The larger enterprises should concentrate in the new business domain as the potential market opportunities are increasing day by day. The big firms or the consortium of large firms may invest on R & D in collaboration with the premier research organizations of the country. The govt. should encourage this mission by offering some lucrative package like tax holiday or relief for the firm for next three years. The firms may be incentivized by promoting SEZ or providing subsidy. (ii) The Indian MSMEs must initiate to this call of the hour. The firms may introduce their activities in the healthcare domain. As of now, there are various attempts are made to develop innovative branch of research augmenting various forms of technology with the healthcare domain. The branch of biomedical engineering, nano technology and electronic devices is being frequently used in the modern health services. The MSMEs can identify a niche market specializing any of the innovative techno-oriented direction and cutting-edge research which can be converged in modern medical system. (iii) The IoT based infrastructure can be conjugated with creating healthcare alarming devices. The psycho social behavioral pattern of a set of patients may be studied and the common patterns may be digitally incepted in the IoT led instruments in the line of censory device like e-nose, RFID censors for detecting aroma or pigments and even the unnatural body movements to detect the cases that comes under broader domain of ergonomics.

Health-Care Paradigm and Classification in IoT Ecosystem …

265

Table 1 Features and benefits of IoT devices For healthcare Sl. No.

Disease

IoT devices

Features of IoT device

Benefits

1

Cancer

1. Electronic-nose 2. Biosensor

1. Smart device, authentic, flexibility, quality control 2. Quick, authenticate detection, decent, observing of angiogenesis, cancer metastasis

1. The give off breath of patients with lung cancer characteristics that can be with a computerized nose 2. Biosensors can catch whether a tumor is exist, whether it is favorable or cancerous

2

Heart disease

1. Smartphone 2. Heart beat sensor

1. Rapid analysis, flat cost, familiar 2. Low-cost

1. The sensor associate to a module in the smart phone over the audio jack 2. Heart attack disclosure using Heart Beat Sensor effort on Photoplethysmography (PPG) art.

3

Diabetes

1. Insulin pump 2. Gluco track

1. Flexibility, predictable, reducing wide fluctuations in blood glucose 2. Pain free, Reading history data, user friendly, easy to read data

1. It is a small, automated device that device that bear insulin continuously all over the day 2. A glucose monitoring home device sensor is used to measure the concentration blood.

4

Liver disease

1. MRI 2. e-nose

1. Images come, approach organ morphology, physiology, functions contrast 2. Alluring, ancient and marginally faucal aroma of the emit breath

1. MRI evaluate liver function, usually expressed via the Child-Pugh score 2. e-nose could be a authentic non-invasive apparatus for characterizing CLD

(continued)

(iv) In the era of big data analytics, the predictive analogy in the healthcare sector is emergent to prevent it mammoth outcome on human civilization. The growing concerns of environmental pollution have been challenging the very existence of our civilization. The IoT based ecosystem may be applied in a less expensive manner to identify whether the region are crossing the vulnerable and critical

266

R. Biswas et al.

Table 1 (continued) Sl. No.

Disease

IoT devices

Features of IoT device

Benefits

5

HIV

1. Photonic crystal (PC) biosensors 2. Novel BioNanoSensor

1. Rapid, sensitive, 100% efficiency, label-free, 2. Inexpensive, portable, simple, sense gases

1. Biosensors optical detection method for bimolecular, cells, and viruses 2. BNS device that employ automation to identify the existence of the HIV

6

Asthma

1. Bracelet 2. HET wristband

1. Authentic measuring and measure 2. Controlling volatile organic compounds, circulatory humidity and temperature

1. It benefit wearers anticipate a looming asthma attack 2. It is a wearable system that could record framework to forecast asthma attacks

level of pollutants so that appropriate measures can be prescribed. The device may also identify the root causes or epicenter of such pollutants so that the multiple stakeholders can intervene and address the issue. To implement all such modern techniques, India should progress and contribute in the age of Fourth Industrial Revolution. There are several inherent constraints for its implementation particularly in Indian context. 1. The achievement in this new era of technology needs holistic, inclusive and comprehensive growth in the field of technology, its availability, ease of accessibility, technical knowhow for its use, capacity of investment and overall dynamics of its adaptability in real life practice. 2. The IoT technology is the platform to facilitate the healthcare support but the country desperately requires a basic infrastructural facility for healthcare services. The issues of malnutrition, vaccination, basic sanitation and the most importantly the awareness of people in general etc have been creating the stumbling block to achieve success in the cause of humanity.

5 Conclusion In this literature survey, big data and its various concepts are included. The words big data has been coined to depict this newness. This paper also defines the characteristics of big data. With the advance of big data, we could answer questions that were beyond research in the past, extract knowledge and insight from data. it is understood that

Health-Care Paradigm and Classification in IoT Ecosystem …

267

every big data platform has its individual focus. Big data analytics in healthcare is germinating into a promising field for affording acumen from very large data sets and enhancing conclusion while compressing amount. Big Data today carry a lot of promise for the healthcare sector. So, implementing healthcare analytics with expeditious organization, and evolution of big data will make rapid and exact diagnosis which will decrease blunder and bring convenient treatment. using. IoT devices to handle their health requirements. To afford relevant cure to the patient, symptoms are determined from the excessive number of data. Aspects of IoT have been presented. IoT can detect, determine, and accessed by devices like actuators, sensors or other smart devices. In this paper a review on big data, healthcare, IoT usage in healthcare has been presented.

References 1. Joshitta S (2016) Applications of big data analytics for diagnosing diabetic mellitus: issues and challenges. Int J Recent Trends Eng Res 2:454–461 2. Emani CK, Cullot N, Nicolle C (2015) Understandable big data: a survey. In: LE2I UMR6306, CNRS, ENSAM, Univ. Bourgogne Franche-Comté, F-21000 Dijon, Vol 17, Aug 2015, pp 70–81 3. Honest N, Patel A, Patel CM (2016) A survey of big data analytics. Int J Inf Sci Tech (IJIST) 6(1/2) 4. Pooja M, D Das (2016) Comparative analysis of IoT based healthcare architectures. Int J Appl Eng Res 11(20):10216–10221. ISSN 0973-4562 5. Gaitanou P, Garoufallou E, Balatsoukas P (2014) The effectiveness of big data in health care: a systematic review. In: Closs S et al (eds) MTSR 2014, CCIS 478, pp 141–153 6. Thara DK, Premasudha BG, Ram VR, Suma R (729–735) Impact of big data in healthcare: a survey. In: 2016 2nd international conference on contemporary computing and informatics (IC3I), Noida, pp 729–735 7. Alelyani S, Ibrahim A (2018) Internet-of-things in telemedicine for diabetes management. In: 2018 15th learning and technology conference (L&T), Jeddah, pp 20–23 8. Azzawi MA, Hassan R, Bakar KAA (2016) A review on Internet of Things (IoT) in healthcare. Int J Appl Eng Res 11(20):10216–10221. ISSN 0973-4562 9. Chavan V, Phursule RN et al (2014) Survey paper on Big Data. Int J Comput Sci Inf Technol (IJCSIT) 5(6):7932–7939 10. Constantine R, Batouche M (2015) Drug discovery for breast cancer based on big data analytics techniques. In: 2015 5th international conference on information & communication technology and accessibility (ICTA), Marrakech, pp 1–6 11. Khan N, Yaqoob I, Hashem IAT, Inayat Z, Kamaleldin W, Ali M, Alam M, Shiraz M, Gani1 A (2014) Big data: survey, technologies, opportunities, and challenges. Hindawi Publishing Corporatione Sci World J 18 12. Nizam T, Hassan SI (2017) Big data: a survey paper on big data innovation and its technology. Int J Adv Res Comput Sci 8(5):2173–2177 13. Chen M, Mao S, Liu Y (2014) Big data: a survey. Mob Netw Appl 19:171–209 14. Archenaa J, Anita EAM (2015) A survey of big data analytics in healthcare and government. Procedia Comput Sci 50:408–413 15. Prasad ST, Sangavi S, Deepa A, Sairabanu F, Ragasudha R (2017) Diabetic data analysis in big data with predictive method. In: 2017 international conference on algorithms, methodology, models and applications in emerging technologies (ICAMMAET), Chennai, pp 1–4

268

R. Biswas et al.

16. Huzooree G, Khedo KK, Joonas N (2017) Glucose prediction data analytics for diabetic patients monitoring. In: 2017 1st international conference on next generation computing applications (NextComp), Mauritius, pp 188–195 17. King TE Jr (1999) A new look at the pathophysiology of asthma. J Natl Med Assoc 91(8 Suppl):9S–15S 18. Norman R. Alpert • David M. Warshaw”, Human Heart Failure: Dilated Versus Familial Hypertrophic Cardiomyopathy” 2003 19. Stewart J, Manmathan G, Wilkinson P (2017) Primary prevention of cardiovascular disease: a review of contemporary guidance and literature 20. Simon V et al (2006) HIV/AIDS epidemiology, pathogenesis, prevention, and treatment. Lancet (London, England) 368:9534 21. Bhatti AB et al (2016) Current scenario of HIV/AIDS, treatment options, and major challenges with compliance to antiretroviral therapy. Cureus 8(3):e515 22. Banu Priya M, Laura Juliet P, Tamilselvi PR (2018) Performance analysis of liver disease prediction using machine learning algorithms 05(01) 23. Nahar N, Ara F (2018) Liver disease prediction by using different decision tree techniques. Int J Data Min Knowl Manag Process (IJDKP) 8(2) 24. Shandilya S, Chandankhede C (2017) Survey on recent cancer classification systems for cancer diagnosis. In: 2017 International conference on wireless communications, signal processing and networking (WiSPNET), Chennai, pp 2590–2594 25. Alharam AK, El-madany W (2017) Complexity of cyber security architecture for IoT healthcare industry: a comparative study. In: The proceedings of 2017 5th international conference on future internet of things and cloud workshops 26. Kumar N (2017) IoT architecture and system design for healthcare systems. In: The proceedings of 2017 international conference on smart technology for smart nation 27. Kumbi A, Naik P, Katti KC, Kotin K (2017) A survey paper on internet of things based healthcare system

Human Activity Recognition from Video Clip Rajiv Kumar, Laxmi Kant Sagar and Shashank Awasthi

Abstract The recognition of human activities from video files has been a challenging issue due to background clutter, viewpoint, appearance, and illumination. The human actions are characterized by various actions like walking, standing, and running. The surveillance system lets us to classify a human action either as malicious or non-malicious activities. This paper presents a real-time action recognition system where the camera captured frames are classified using KNN algorithm. The performance of the system is tested against the standard KTH dataset. Keywords Activity recognition · KNN · KTH dataset · Surveillance system

1 Introduction In today’s world, the increasing theft has increased the demand for surveillance system. Current existing system contains camera but needs more computing power to analyze the system. Therefore, Human Action Recognition [1, 2] (HAR) is an important research area in computer vision. As there are numerous cameras (CCTV) installed in various locations of different cities, all of them require proper monitoring; although, we can hire that much of human resource as employees to monitor but as we know humans cannot continuously monitor camera feeds and there are lot more chances of him to miss some events occurring in front of any CCTV. So, from there the need comes of human activity recognition systems. The main goal of HAR system is to recognize different types of human activities in real life. Accurate recognizing activities are a challenging task because of its diverse and highly complex nature. Human action recognition broadly classified into Template matching [3, 4], Flowbased techniques [5], Shape based [6, 7], and Internet of Video Things [8]. The feature extraction is a very important step in action recognition. Large feature vector takes more processing time and results more cost whereas low feature vector results R. Kumar (B) · L. K. Sagar · S. Awasthi G L Bajaj Institute of Technology & Management, Greater Noida, Uttar Pradesh, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_31

269

270

R. Kumar et al. Human Activity Recognition

Template matching

Flow Based

Shape Based

IoVT

Fig. 1 HAR systems

less processing time and hence low cost. The human action recognition system is shown in Fig. 1.

2 Related Work Machine learning plays an important role in Human Action Recognition (HAR). Work in this area has been initiated in early days but still there is lack of system which properly recognizes action. In order to achieve this Li and Hua [3] proposed transform and template matching based human action recognition. The technique has low computational complexity. It utilizes a string matching scheme to analyze various human actions. Sahoo and Ari [4] proposed Local Maxima of Difference Images (LMDI) techniques to detect interest region point for human action recognition. The difference images are computed using continuous frames difference techniques and further three-dimensional value of peak is used. Histograms of local features are calculated for each point of interest. Barnachon et al. [5] proposed a novel framework using Motion Capture (MoCap) data. The techniques are based on histogram of poses, according to Hausdorff distance. It also compares the result with Bhattacharyya distance. Moussa [6] proposed Scale Invarient Feature Transform (SIFT) feature to detect interest points from video frames. Further, normalized values from Bag of Video words are used to classify with Support Vector Machines. The experiment is conducted on KTH and Weizmann datasets. Azim and Hemayed [7] proposed trajectories-based approach to capture temporal-related information. Cuboid features are used to extract features from video frames. Finally, these features are classified by using SVM. Akula et al. [9] proposed Human Action Recognition based on Convolution Neural Network. The techniques are useful for persons with disabilities. It recognizes six action classes—sitting, standing, falling, walking, sitting with chair and desk, lying on grounds. Itano et al. [8] proposed Internet of Video Things (IoVT) architecture to detect various activities of humans. In order to improve robustness, the hyper parameters values obtained from Neural Network are optimized by using Genetic Algorithm. The progress in this area is summarized in Table 1.

Human Activity Recognition from Video Clip

271

Table 1 Progress in human action recognition Sl. No.

Methodology

Dataset used

Result

Li and Hua [3]

Transform and template matching

Action

NA

Sahoo and Ari [4]

Local maxima of difference images (LMDI)

J-HMDB and UCF101 datasets

NA

Barnachon et al. [5]

Motion capture

(MoCap) data

NA

Moussa [6]

Scale invariant feature transform

KTH and Weizmann datasets

97.89% for KTH and 96.66% for Weizmann

Azim and Hemayed [7]

Trajectory-based representation, cuboid features

KTH and Weizmann datasets

NA

Akula et al. [9]

Convolution neural network

6 action classes

83.7%

Itano et al. [8]

Internet of Video Things (IoVT)

Action recognition

NA

3 Methodology The experiments are conducted on KTH dataset. The KTH dataset consists of six different video clips of human actions Hand Clapping video clips, Boxing video clips, Hand Waving video clips, Jogging video clips, Running video clips, Walking video clips, Cycling video clips, and Surfing video clips as shown in Fig. 2.

Handclapping video clip

Boxing video clip

Running video clip

Handwaving movie clip

Joging Movie Clipg

Walking Movie Clipg

Fig. 2 KTH dataset samples for six different actions

272

R. Kumar et al.

It contains 100 movie clips for each action. It is further divided into 50 training dataset and 50 testing dataset. Dataset is stored in video format (.avi). The data frame images are obtained from movie clip. Since there are many frames available in movie clip, further 50 frames are extracted from each clip. These image frames are preprocessed by using low pass filter in order to smooth the images. From these preprocessed images, features are extracted. For calculating feature edges, images are extracted along with interest point location such that it captures the required action. These feature vectors are stored in 40 feature points which are used for classification. For classification, KNN techniques are applied. The steps are summarized in Fig. 3. The experiments are conducted in MATLAB 2015 a in Windows 10 environment. The dataset is divided into 50 training and 50 test dataset. The features are calculated and stored in a vector form. The feature vector for various actions is shown in Fig. 4. The feature extraction graph shows that features are distinguishable for different actions. The datasets are labeled. The feature vector obtained from datasets is further classified by using supervised learning KNN approach. The average recognition accuracy is calculated as 83% as shown in Table 2.

Movie

Frame

•Movie Clip

Pre-

ClassificaƟ •KNN

•FilteraƟon •Edge, Intrest point locaƟon

•Frame ExtracƟon

RecogniƟo •HAR Confusion Matrix

Fig. 3 Steps used for human action recognition

6

105

5 4 3 2 1 0 1

1.5

2

2.5

3

Fig. 4 Feature vector for six different actions

3.5

4

4.5

5

5.5

6

0

3

1

2

2

1

Hand waving (3)

Jogging (4)

Running (5)

Walking (6)

Cycling (7)

Surfing (8)

Average accuracy

1

41

Hand clapping (2)

Boxing (1)

1

Table 2 Confusion matrix for HAR

0

2

0

1

0

3

44

0

2 1 5

1

0

0

0

0

42

3

1

0

3

5

41

0

0

6

4 2

3

3

0

3

5

3

39

5

2

0

42

4

3

2

0

0

6 0

0

0

0

0

0

0

41

7

42

0

0

0

0

0

0

0

8 82

83

84

82

84

78

82

84

88

Human Activity Recognition from Video Clip 273

274

R. Kumar et al.

4 Conclusion This paper presents a model to detect human activities from a video clip. The algorithm is tested on KTH datasets. The KTH datasets contain six different human actions video clip. On training dataset, it provides best result (100%). Whereas on test dataset it provides 83% recognition accuracy. KNN classifier works better with training dataset as it preserves the prototype of samples. Still there is a gap in feature vector and classification scheme which can be improved by other machine learning algorithms. This can be used as a base to make a whole new surveillance system which can work independently and report to related authority about any abnormal activity.

References 1. Poppe R (2010) A survey on vision-based human action recognition. Image Vis Comput 28(6):976–990. ISSN 0262-8856 2. Roh M-C, Shin H-K, Lee S-W (2010) View-independent human action recognition with volume motion template on single stereo camera. Pattern Recognit Lett 31(7):639–647. ISSN 0167-8655 3. Li C, Hua T (2011) Human action recognition based on template matching. Procedia Eng 15:2824–2830. ISSN 1877-7058 4. Sahoo SP, Ari S (2019) On an algorithm for human action recognition. Expert Syst Appl 115:524– 534. ISSN 0957-4174 5. Barnachon M, Bouakaz S, Boufama B, Guillou E (2014) Ongoing human action recognition with motion capture. Pattern Recognit 47(1):238–247. ISSN 0031-3203 6. Moussa MM, Hamayed E, Fayek MB, El Nemr HA (2015) An enhanced method for human action recognition. J Adv Res 6(2):163–169. ISSN 2090-1232 7. Abdul-Azim HA, Hemayed EE (2015) Human action recognition using trajectory-based representation. Egypt Inform J 16(2):187–198. ISSN 1110-8665 8. Itano F, Pires R, de Sousa MADA, Del-Moral-Hernandez E (2019) Human actions recognition in video scenes from multiple camera viewpoints. Cogn Syst Res 56:223–232. ISSN 1389-0417 9. Akula A, Shah AK, Ghosh R (2018) Deep learning approach for human action recognition in infrared images. Cogn Syst Res 50:146–154. ISSN 1389-0417

A Framework for Enhancing the Security of Motorbike Riders in Real Time Yash Khandelwal, Sajid Anwar, Samarth Agarwal, Vikas Tripathi and Priyank Pandey

Abstract With the rise in the population these days the requirement of vehicles especially two-wheelers is increasing which is also leading to an increase in the number of accidents; therefore, we require a robust system to enforce the safety guidelines and ensure that people are following the traffic rules that are made to maintain their security; therefore, we have suggested an approach to detect dodgers using an automated system which detects bike-riders with and without a helmet. To achieve this goal, we have used Faster-RCNN inception v2 model, and have procured 96.08% accuracy in detecting bike-riders without a helmet and an accuracy of 86.7% in detecting bike-riders with helmet and the overall accuracy of 93.37%. Keywords Helmet detection · Faster-RCNN · Region proposed network

1 Introduction There are various modes of transportations available like cars, trains, airplanes, bikes, etc. But for short distances we prefer two-wheelers and therefore it is very popular, but because of less protection, they are associated with a high risk. Therefore, a helmet is a must for a bike-rider. If we talk about India alone, more than 1.4 lakhs people died in road accidents in the year 2017, according to the status given by [1] and are increasing continuously. So, following traffic rules play a vital role to drop down these accidents, but a lot of people do not follow traffic rules, which is a punishable act. There are various strategies that are adopted by the government but most of them are manual like detecting bike-riders without a helmet; detecting bike-riders manually becomes a tedious task in congested traffic and there are many chances of faults, therefore there is a need of an automated system to detect defaulters. Many people have proposed their model for this task. But the problem with their work is that they require some human intervention to make them work, due to this their efficiency will decrease in future as human efficiency decreases with time which will result in a decrease in Y. Khandelwal (B) · S. Anwar · S. Agarwal · V. Tripathi · P. Pandey Graphic Era Deemed to Be University, Dehradun, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_32

275

276

Y. Khandelwal et al.

efficiency of the model. Therefore, there is a strong need for a fully automated realtime system which will effectively detect the helmetless bike-rider. Since the number of road accidents is increasing day by day, many countries are implementing an automated system for detecting helmetless riders; hence, our approach will be useful in achieving the goal. Detecting dodgers in a real-time environment is a tedious task because of various challenges. Some of the challenges are 1. There can be multiple defaulters in a single frame, hence the system should be capable of detecting them; also there can be people who are just carrying a helmet in hand and not riding a bike, and if there are two people on the bike then according to traffic rules both of them should wear a helmet; hence, the model should be able to differentiate between all the above cases. 2. In the live stream video from CCTV, viewpoints play an important role, since we have to track the bike-rider head and so, the position of the head should be at an angle where the model can track it. But in real time it is not possible all the time. 3. The video quality is also a major issue since many CCTV cameras have low video resolution power, so the frames are not clear, which affects the efficiency of the model. 4. Since there are no predefined rules for the design of the helmet, some helmets are in the shape of cap or turban, etc. which looks similar to the helmet which makes them difficult to track. But a real-time system should overcome all the above challenges and give accurate, fast and cost-effective results. Observing these challenges, we suggest a model which detects dodgers in real-time and produces fast and accurate results using Faster-RCNN inception v2 model. As stated in [2] to find out the region of interest, selective searching is used by both RCNN and Fast-RCNN which results in a time-intensive and slow process. As we are working on the live stream and it requires fast analysis for predicting and tracking, we have used Faster-RCNN which in turn uses region proposal network (RPN) which helps in predicting image score and bounding boxes faster. The further paper is systematized as follows. Section 2 consists of a literature review which includes previous works that are done for detecting the defaulters, Sect. 3 includes a proposed approach for detecting helmetless bike-riders, Sect. 4 includes results and discussion, and finally, we have concluded our approach in Sect. 5.

2 Literature Review In the last few years, a lot of work has been done on the automatic detection of bike-riders with or without a helmet. Dahiya et al. [3] proposed a model in which they have used background subtraction and applied object detection by extracting features using HOG, SIFT, LBP and used SVM for detecting defaulters, and achieved an accuracy of 93.80% but in the public traffic; the used feature extraction technique struggles to extract visual feature methodically from the small portion of the frame which consists of the defaulters. Chang and Chen [4] proposed an intelligent helmet to identify the number plate of large approaching buses/trucks. Their helmet includes infrared transceivers, camera, ECP and battery. The accuracy to recognize the number plate at daytime is 75% and 70% at night. Talaulikar et al. [5] proposed a method

A Framework for Enhancing the Security of Motorbike …

277

to detect the bike-riders without helmets using background subtraction and features extraction. To implement background subtraction, they use aspect ratio and standard deviation of hue and for features extraction they use HOG. To develop a more accurate model they implement principal component analysis (PCA) on derived features. Wu and Zhao [6] designed a model to detect whether workers of the factory were wearing a helmet or not. Their model has three major parts, first KNN to detect moving objects, second to identify workers they use HOG descriptor and SVM classifier, third a colour-based hybrid descriptor to differentiate the colour of helmets. Mistry et al. [7] proposed a system to detect a helmet using CNN. In their model, there are two stages of YOLOv2 to increase accuracy, one to detect person trained on COCO dataset and another one is trained on helmet images dataset. Han et al. [8] proposed a double deep Q-learning network for detecting an object. They trained their double DQN with priority-based experience replay and predicted action at each step and checked the efficiency on the active vision dataset. Kulkarni et al. [9] proposed a method based on background subtraction and classifier. First, they used background subtraction to a detected moving object and then classified it as bike-rider with or without a helmet and then characters on number plate are extracted for defaulters. They have trained a convolution neural network using transfer learning. Doungmala and Klubsuwan [10] proposed a method in which they first detect face using Haar descriptor and then detect the helmet using circular Hough transform. Daimary et al. [11] proposed a helmet system which includes MQ3 alcohol sensor and a sensor for detecting helmet, which will enable ignition of two-wheelers after sensing alcohol and helmet, if it senses alcohol then ignition does not start hence protecting the rider from riding a two-wheeler. Their proposed helmet is rechargeable and powered by a lithium-ion battery. Some researchers proposed data mining approach to classify accident occurrence identification [12] and severity of accident identification [13]. Sharma et al. [14] proposed real-time approach to detect defaulters using motion analysis. After reviewing the above work, we found out that there is still scope for improvement in detecting the helmetless rider, therefore we have proposed an approach for enhancing the security of motorbike riders in real time by using FasterRCNN.

3 Methodology The foremost intention of this framework is to predict helmetless rider precisely and with less oversight, which will work in a real-time environment fulfilling all-natural conditions. The proposed framework uses Faster-RCNN algorithm to predict and distinguish the defaulters; the framework works in two stages as depicted in Fig. 1, the first stage includes conversion of video into frames then pass them to a trained model which will detect bike-riders and discard all those frames which do not have bike-riders in it; since, we are passing the whole frame for detecting the defaulters which will help in a condition where there are multiple bike-riders without helmet

278

Y. Khandelwal et al.

Fig. 1 A proposed framework for detecting bike-riders with or without a helmet. Source https:// www.iith.ac.in/vigil/resources.html [3]

in a single frame. The second stage includes the detection of bike-riders with and without a helmet and saving the result for future uses. As shown in Fig. 2, Faster-RCNN consists of region proposal network which helps in finding the region of interest and detecting object using this network. FasterRCNN includes anchors, which is basically a hyperparameter helpful in detecting an object with distinct aspect ratios and by default there are nine anchors at a position of an image. Faster-RCNN also includes RPN (region proposed network) and its output is a proposal which in next step send to classifier and regressor; classifier finds the probability of the proposal in the image and regressor predicts the position of the proposal, also softmax function is shown in Eq. 1. which is used to calculate final probability of the proposals; the main advantage of this is the output range of probability which lies between zero to one and the sum of all probability is equal to

Fig. 2 Working of faster R-CNN

A Framework for Enhancing the Security of Motorbike …

279

one,;also, RPN helps in finding the region of interest in the input frame, which helps in the faster prediction of bounding boxes and makes Faster-RCNN comparatively faster than RCNN and Fast-RCNN. F(Ψk ) = n

ek

m=0

em

k = 0, 1, 2, 3, . . . n

(1)

4 Results and Discussion Our machine configuration includes Windows machine with Intel(R) Core (TM) i56200U CPU @ 2.30 GHz × 4 with 6 GB RAM. We have used the dataset which is used by [3] and trained our model with the first one-hour video and tested on the second-hour video, which has a frame size of 640 × 480; the dataset includes various types of bike-riders, some are wearing cap like helmets, traditional helmets, some are wearing a cap instead of helmet, in some frames rider was going away from camera hence only head of the rider is visible; all these factors were making it apt for training and testing of the model. Our model was trained using Faster-RCNN inception v2 model as feature extractor and first stage features stride was kept 16, our first objective was to track bike-riders in the entire frame and then pass them to the second model which will track their head and check for helmet; we have trained the first model up to 20000 iterations, and the second model up to 40000 iterations (Fig. 3). Table 1 includes true positive, true negative, false negative, false positive values for both the classes; as we can see the false positive score is 71, which means it has detected frames which include riders without-helmet but it has detected them as with-helmet this is because there is no predefined design of helmet and since the frames were blurry and people with cap were resembling as with-helmet. But the overall accuracy for detecting defaulters is 96.08% and detecting riders with the helmet is 86.7%. Table 2 includes proportional scrutiny in the form of recall,

Fig. 3 Sample frames from dataset [3]

280

Y. Khandelwal et al.

Table 1 Statistics representation of with-helmet and without-helmet Class name With helmet Without helmet

True positive

True negative

False positive

False negative

Accuracy (%)

639

1744

71

98

86.7

1744

639

98

71

96.08

Table 2 Comparative analysis

Class name

Precision

Recall

F1-score

With helmet

0.90

0.87

0.88

Without helmet

0.95

0.96

0.95

Micro average

0.93

0.93

0.93

Macro average

0.92

0.91

0.92

Weighted average

0.93

0.93

0.93

precision and accuracy for both the classes. Since the model is detecting the bikerider at an accuracy of 98.9% which sums up the overall accuracy of the model at 93.37%. The recall and precision value for riders with-helmet can be an increase by changing video quality.

5 Conclusion Keeping in mind the security of bike-riders we had proposed a framework to detect bike-riders with or without a helmet, which is giving precise results, as we had tested the model and achieved an accuracy of 96.8 and 98.9% for detecting defaulters and bike-riders, respectively. The accuracy accomplished for detecting the bike-riders with the helmet was 86.7% which can further be convalesced as in some frames the rider was wearing a cap like a helmet which makes it difficult to fetch features effectively. The proposed model can detect helmets in various conditions like multiple bike-riders in the same frame and even if there are multiple riders on the same bike, it can also track them, which in future will be helpful in detecting tripling on the bike. In future, this model can be used in detecting the license plate of these defaulters and send the data directly to the cloud. Also, we can implement this model along with facial recognition model which will be trained for our college students and send notices to students who are riding bike without helmet.

A Framework for Enhancing the Security of Motorbike …

281

References 1. https://timesofindia.indiatimes.com/india/india-way-off-road-safety-targets-%09for-2020road-accidents-still-kill-over-a-lakh-a-year/articleshow/65765549.cms 2. Ren S, He K, Girshick R, Sun J (2015) Faster R-CNN: towards real-time object detection with region proposal networks. In: Proceedings of conf. Neural Inf. Process. Syst., pp. 91–99, (2015) 3. Dahiya K, Singh D, Mohan CK (2016) Automatic detection of bike-riders without helmet using surveillance videos in real-time. In: Conference on neural networks (IJCNN). IEEE, pp 3046–3051 4. Chang W-J, Chen L-B (2019) Design and implementation of an intelligent motorcycle helmet for large vehicle approach intimation. IEEE Sens J 1–1. https://doi.org/10.1109/jsen.2019. 2895130 5. Talaulikar AS, Sanathanan S, Modi CN (2018) An enhanced approach for detecting helmet on motorcyclists using image processing and machine learning techniques. In: Advances in intelligent systems and computing, pp 109–119. https://doi.org/10.1007/978-981-13-06808_11 6. Wu H, Zhao J (2018) An intelligent vision-based approach for helmet identification for work safety. Comput Ind 100:267–277 7. Mistry J, Misraa AK, Agarwal M, Vyas A, Chudasama VM, Upla KP (2017) An automatic detection of helmeted and non-helmeted motorcyclist with license plate extraction using convolutional neural network. In: 2017 seventh international conference on image processing theory, tools and applications (IPTA). IEEE, pp 1–6 8. Han X, Liu H, Sun F, Yang D (2018) Active object detection using double DQN and prioritized experience replay. In: Conference on neural networks (IJCNN). IEEE, pp 1–7 9. Kulkarni Y, Kamthe A, Bodkhe S, Patil A (2018) Automatic number plate recognition for motorcyclists riding without helmet. In: 2018 international conference on current trends towards converging technologies (ICCTCT). IEEE, pp 1–6 10. Doungmala P, Klubsuwan K (2016) Helmet wearing detection in Thailand using Haar like feature and circle hough transform on image processing. In: International conference on computer and information technology (CIT). IEEE 11. Daimary A, Goswami M, Baruah RK (2018) A low power intelligent helmet system. In: Symposium on devices, circuits and systems (ISDCS). IEEE. https://doi.org/10.1109/isdcs.2018. 8379658 12. Gupta M, Solanki VK, Singh VK, García-Díaz V (2018) Data mining approach of accident occurrences identification with effective methodology and implementation. Int J Electr Comput Eng (IJECE) 8(6) 13. Gupta M, Solanki VK, Singh VK (2017) A novel framework to use association rule mining for classification of traffic accident severity. J Eng Educ 13(21):37–44 14. Sharma A, Tripathi V, Gangodkar D (2019) An effective video surveillance framework for ragging/violence recognition. In: International conference on data engineering and communication technology. Springer, Singapore, pp 239–248

Fisherman Communication at Deep Sea Using Border Alert System N. R. Rajalakshmi and K. Saravanan

Abstract Fishermen are being caught and killed by the naval forces of the neighbouring country, due to the lack of awareness about the ocean frontiers and advanced alert instruments in the boats. For this reason, maritime security has become major concerns of all coastal areas to protect the fisherman and providing the assistance about sea frontiers via alerting, tracking, and monitoring of boat vessel. Thereby, a path breaking technology of maritime border alert system using smart vessel is proposed here which would foster coast guard officials to effectively monitor fishermen and alert them if they sail inside the other country’s border. The proposed border alert system alerts the border to safeguard the fisherman and make good relationship between seaside nations. This system uses Global Positioning System which helps to find out the current latitude and longitude values of vessel. If the fishermen are very close to IMBL means then, this alert system aware the fishermen through an audio and visual alert. While receiving alerts, if the fisherman did not tag on any action and move further, then the smart vessel should be reversed automatically and information will be transmitted to the nearby coast guard officials. Next, the guards can give assistance and provide additional help to those fishermen. The proposed system takes the Gulf of Mannar maritime border which is between India and Sri Lanka as a case study to validate the proposed system. Keywords Fisherman · IMBL border · LoRa · Maritime border · IMBL alert

N. R. Rajalakshmi (B) Department of Computer Science and Engineering, Vel Tech Rangarajan Dr. Sagunthala R and D Institute of Science and Technology, Chennai, India e-mail: [email protected] K. Saravanan Department of Computer Science and Engineering, Anna University Regional Campus, Tirunelveli, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_33

283

284

N. R. Rajalakshmi and K. Saravanan

1 Introduction Historically, there is no border problem and no disagreement till the civil war in 1983. Henceforward, the territorial demarcation happens to create central notion in international relations among the seaside nations. For instance, the nations are isolated by the International Maritime Boundary Line (IMBL) borders. The maritime border system was initiated and developed by the United Nations Convention of the Law of the Seas (UNCLOS) [3]. This effective IMBL border promotes national sovereignty, economic prosperity. It also supports to prevent the terrorist attacks and criminal acts. The formulation of border for marine is the triplet of tourism–transportation network–economy [29]. The country gets an economic contribution through maritime border which supports the tourism industry. This economic prosperity from tourism gives the significant impact on enhancement of the country’s development and prosperity [4, 16]. Maritime spatial planning is implemented in the Romanian Black Sea coastal zone [8], which is another successful example to standardize the tourism coastal economic movement [31]. The state has special rights to explore and use the marine resources of exclusive economic sea zone which is prescribed by the UNCLOS. Many countries (Iran and Egypt for Saudi Arabia) share the exclusive economic zone [33] for fishing and customs in the IMBL boundary. The economic development of a country is also greatly grown through the influence of maritime industry [17]. Regarding the sea, United Nations agency is held responsible for protection and prevention of ships from marine pollution and also yields security, safety, and efficiency of international merchant shipping [15]. The International Maritime Organization (IMO) implemented a system for measuring maritime traffic, marine services, and security. It contributes its integral role to meet out the target set by United Nations for Sustainable development. A safe navigation in fishing is required in maritime boundary line border. EUROSUR [10] is smart border system, which is the concept to implement more dominant and economic system for border surveillance in a secure way of 24/7 persistence to avoid illegal intrusion [24]. The Indian coast guard addresses the national defence in terms of marine safety, security, law enforcement, lifesaving of fisherman, and fisheries. In Tamilnadu state, about 20,000 vessels make spinning usually stay into the International Maritime Boundary Line (IMBL) of India–Sri Lanka for fishing. The trans-border fishing has been going as the most outstanding problem, such as the fishermen cross the line unwillingly, even though the international laws clearly define and mark the IMBL line border. They could be easily caught or imprisoned by sea pirates or foreign navy while crossing IMBL borders inadvertently. But the identification of international borderline is fairly difficult countenance for fisherman while fishing [7]. This lack of knowledge in identifying the maritime boundaries of two countries during fishing could put the lives of the fishermen in a lot of danger such as killing and detention. The foreign trawler easily overcomes our coast guard security force by killing or imprisoning the fisherman. Regardless of the existing excellent relations and elaborate understanding between the Indo-Srilankan, the fishermen issue has

Fisherman Communication at Deep Sea Using Border Alert System

285

lifted up in many occasions [1]. Sometimes, the death of penalty and imprisonment of the Tamil Nadu fishermen, supposedly by the Sri Lankan Navy, becomes an emotive right to life and livelihood (human rights) issue in the domestic politics of maritime boundary. Next, the difference of opinion is raised between China and ASEAN countries about the Nine Dash Line Claim of China [2]. The violation in east sea happens often because China disagrees with the border with other neighbouring countries [20, 21, 23]. Hence, the maritime border identification and fishermen alert is necessary [5]. In 2015, the crossing of borderline ended hazardous to 600 fishermen. The fishermen in risk is increased every year [14]. Hence, the fishermen should be alerted before they reach the borderline to avoid hazard. But, the crossing border alerting system is still inadequate for assisting fishermen automatically. Thus, the aim of our proposed system designed here is to monitor the fishermen and encourage them to explore inside our sea within our nation’s border through smart boat vessel. The communication to coast guard officials leads to assisting the fishermen effectively. The literature survey is in Sect. 2. Section 3 describes the border alert system, GPS, coast guard communication, engine control unit. The results and scenarios are detailed in Sect. 4, followed by Sect. 5, which concludes the paper.

2 Literature Survey The Internet portfolio of technologies attains major development through (IoT) Internet of Things which connects billion of sensors across the world. The technological developments in IoT vision shapes the multiple economic sectors, including telecommunications, computing, construction and logistics. The way of human life would be greatly changed due to the fast-growing innovation of Internet of Things. The computation and storage of IOT devices are offloaded into the cloud environment to run their operation in smart manner, hence it becomes their client. The cost effective and scalable solutions are obtained through the implementation of Cloud-based IOT applications. This work discusses an IoT-based border alert system using GPS and LoRa modules. Suresh and Sharath [28, 32] designed a Low Cost Maritime Boundary Identification device using GPS System for fishermen navigation and nautical border guidance. Surekha et al. [27] proposed the system using ARM processor. The location of travelling boat is identified using GPS to alert the fisherman while crossing the boundary. This crossing information will also be passed to the control section through ZIGBEE transmitter. Naveen and Ranjith [19] proposed the work using DGPS and GSM for border alert and smart tracking with in-built alarm. The location of the boat could be tracked by using DGPS. This system raises an alarm, if the border has been crossed. Also, the crossing information is transferred to the control room and family members by the GSM at regular time intervals. Sivagnanam et al. [25, 26] proposed the coast guard alert and rescue system for intruder of international maritime line crossing, the exact location of fisherman has been identified with the help of integrated GPS

286

N. R. Rajalakshmi and K. Saravanan

(Global Positioning System) receiver. This position is communicated to the respective coast guard station via GSM (Global System for Mobile communication). Jim and Eugene [11] proposed the concept of Advanced Border Alert System Using GPS and Intelligent Engine Control Unit which is used to stop the movement of fishermen navigation towards the other country’s border. They detect the location of the boat via GPS module which is the most accurate and fastest way of locating the vessel, alert them with the help of an alarm and if smugglers and intruders neglect the alarm, the boat engine will be stopped to prevent from trespassing into the other border by an intelligent engine control unit and then alert the coast guard. Kamalakannan et al. [12] used GSM and LPC2148 Arm microcontroller and Radio Frequency IDentification (RFID) to protect the fishermen from crossing the border and from the death of penalty by the other country navy. When crossing the parental country circumstances, the fisherman has been warned by using the alerting devices such as buzzer with LCD display. Also, The motor will stop functioning automatically at the third border. The communication of border crossing information through GSM is unfeasible, because the coverage of cellular network at deep sea is unworkable and also passing information to the coast guard officials through Zigbee transmitter is impractical, due to the coverage of Zigbee transmission is within the metre. The transmission must be reliable although environment in sea area is tough. Hence, our proposed border alert system is designed with LoRa and GPS to overcome the above issues [9]. LoRa is a single-hop wireless communication technology with long range, low power and low bit rate [18]. It is primarily developed for Internet of Things (IoT) devices requiring low powered battery and low throughput. Therefore, hereby, LoRa has been used as a lightweight smart sensing device to pass the border crossing information (Fig. 1). Fig. 1 Block diagram of a border alert system

ALARM signal and Message display to Fisher man GPS in BOAT

Arduino

Engine Control Circuit to reverse the Boat

LoRa Module

Transmits crossing information to Higher official

Fisherman Communication at Deep Sea Using Border Alert System

287

3 Border Alert System The proposed border alert system is designed to protect the innocent life of fishermen, which is shown in Fig. 2. The main contribution of this system is alerting the fisherman and rescuing them from international maritime line crossing punishments. The core components of proposed system are GPS, Arduino, LoRa Module, Alarm unit, Voice module along with loud speaker, LCD and Ultra sonic sensor.

3.1 Ultrasonic Sensor The ultrasonic sensor in fisherman boat safeguards the boat from the collision. This sensor sends ultrasonic wave which gets objected back by the target. It measures the distance to the target by measuring the time between the transmission and reception, that distance is displayed in the LCD.

Fig. 2 Latitude and longitude value of GPS

288

N. R. Rajalakshmi and K. Saravanan

Fig. 3 Flow diagram for identification of trespass movement of fishermen communities

3.2 Global Positioning System Next, the proposed border alert system uses Global Positioning System (GPS) to monitor and track the movement of spinning boat vessel which is strayed in sea. The constant maritime boundary values (latitude and longitude parameters) for Gulf of Mannar are taken for this experiment (Fig. 3). Figure 2 shows the value of GPS sensor which is connected to Arduino board to give the exact location of a vessel and other vital information. The mobility of vessel aids to predict the ground speed of the spinning boat vehicle and driving behaviour of the fisherman at the current moment in the IMBL borders. The latitude and longitude positions of vessel at different time stamps envisage the trespass movement into the border. The comparison of maritime IMBL boundary reference value with current moment of fisherman aids to predict the trespass movement into the border. If the movement of fisherman is identified as beyond the first border of parental country circumstances means, the boat alarm raises the alarm signal and also the boat LCD displays the warning message (Figs. 4 and 5).

3.3 Coast Guard Communication LoRa (Long Range) allows data transferring at extremely low data rates to extremely long ranges, which uses spread spectrum modulation technique for data transferring. It is capable to transmit over very long distances with little power and support a star topology. A ‘star-of-stars’ architecture of LoRa network is designed with three major components of end devices, gateways and network server. Gateway is a powerful device, connected to a backbone infrastructure to transmit the data between the IoT end devices and server. It can also receive and decode the multiple concurrent

Fisherman Communication at Deep Sea Using Border Alert System

Fig. 4 Output screenshot for border alert system

Fig. 5 Hardware implementation of border alarm circuit

289

290

N. R. Rajalakshmi and K. Saravanan

transmissions. The Lora node has been built by using Arduino uno and LoRa communicates to the gateways over a single wireless hop. The LoRa node can have the capability of transmitting the data packets in kilometre distance range (Figs. 6 and 7). The LoRa module in fisherman boats assists to transmit the crossing information of parental country border to the gateway of coast guard officials. Gateway placed in the coast guard boat receives the data packet from LoRa. Then, the coast guard can help the fishermen accordingly. This LoRa node and Gateway is simulated in things network console with the set of 10 devices that have been taken for validation. LoRa Node

LoRa Node

LoRa Gateway

LoRa Node Fig. 6 Block diagram of LoRa communication to coast guard

Fig. 7 LoRa node data transmission to gateway

Raspberry pi Server

Fisherman Communication at Deep Sea Using Border Alert System

291

3.4 Engine Control Unit If the fisherman boat/vessel is nearer to final IMBL area means then, the engine control unit would be worked. The Pneumatic valve which is fitted with the fuel injector would be turned on. This valve reduces the flow of fuel into the engine; hence the speed of boat engine would be slowed down. Then, the direction handle is positioned astern by using actuator. Next, the engine should be turned into the astern direction by admitting compressed air into the cylinder in the correct sequence. Then the fuel is admitted if the engine arrives at its firing speed. Hence, the vessel will be reversed automatically by ceasing the air admission. From this, further trespassing of IMBL border is stopped.

4 Simulation Results 4.1 Case Study Gulf of Mannar is identified as an India and Sri Lanka maritime border. It is shown in Fig. 8. This maritime border is the arc of huge circles among locations in the sequence. Hence, the proposed system uses the latitude and longitude value of Gulf of Mannar to position the IMBL line. If the fisherman is crossing the above location, the LoRa node setup in the fishing boat transfers the crossing information to the gate way. Here, a LoRa node is being built by using Arduino uno and LoRa module. Hence, the communication takes place between coast guard and fisherman through channel. The NS3 module is used here to simulate the behaviour of LoRa node in an accurate way. Initially, the set of 10 nodes have been taken for validation, and then enhanced to 1000 end devices. Our scenario is validated using one gateway and 1000 number of end nodes. The time of simulation is constrained to 200 s and end devices are randomly plotted in the gateway area of 1000 m × 1000 m. The unconfirmed data frames have been sent by LoRa nodes. Hence, The LoRa module in proposed system send packets to the coast guard officials, the reliable communication has been made to prevent the trespass movement into the border (Fig. 9). The core measurement of proposed border alert system is prevention of trespass movement into the border. The prevention rate of trespass movement into the border is calculated by number of boats that received border support from coast guard officials divided by Total number of boats at the border, which is shown in Eq. 1. Prevention rate of Border Crossing =

Number of boats send border assist to avoid crossing Total number of boats at the IMBL border

(1)

Saravanan et al. [22] implemented AIS based border alert and prevention system with better communication coast guard station. They used the instruments with VHF

292

N. R. Rajalakshmi and K. Saravanan

Fig. 8 Gulf of Mannar—Maritime border between India and Sri Lanka Fig. 9 Packet delivery to gateway from LoRa nodes

Fisherman Communication at Deep Sea Using Border Alert System

293

to attain the reliable communication between the fishermen and coast security station. They compared the reliable communication using RF technology. They reported that the communication is missing at the mid of the sea and also the communication has been done by around 40 fishing boats. Thereby, they achieved the reliable communication only for 40 boats. The prevention percentage from border crossing is very low using RF technology. The reliable communication between the fishermen and coast security station is good in most of the boats using ECDIS [30]. Hence, the prevention percentage from border crossing is increased to 74% [13]. The android application is designed for fishermen to stick on to the system commands to accomplish the target in the Maritime boundary. But, the reliable communication using mobile application [6] at deep sea is measured as 56%. Hence the prevention percentage is 69%. Sivagnanam et al. [25] used the GPS and GSM technology to protect the fishermen from border crossing penalty. The exact position is computed using the GPS receiver and transmit the data to base station. The GSM communication at the deep sea is unfeasible. The percentage of prevention and reliable communication of GPS and GSM system is nearly 60%. Therefore, hereby the border alert system designed using LoRa network for sending packets was to make the reliable communication with the coast guard officials. Thereby, the boats at the border get support from coast guard. The percentage of trespass movement into the border is averagely around 85%. The proposed border alert system has proved through the better attainment in percentage of prevention rate of trespass movement of fishermen while fishing. Moreover, it aids the coastal guards to guarantee safeness of fishermen. The fishermen boats are also stopped from crossing borders using motor reversing mechanism of the boats.

5 Conclusion Many livelihood challenges are faced by fishermen, the major problem of fishermen is dealt in this paper. Herewith, the concept of automatically acquiring the exact location, calculating ground speed of a spinning boat vehicle at the IMBL borders and safeguard of fisherman are discussed. Thus, the border alert system using smart boat saves the innocent life of fisherman from easily caught or imprisoned by sea pirates or foreign navy.

References 1. Amarasinghe O, Bavinck M (2017) Furthering the implementation of the small-scale fisheries guidelines: strengthening fisheries cooperatives in Sri Lanka. The small-scale fisheries guidelines. Springer, Cham, pp 379–399 2. Beckman RC (2017) The China-Philippines dispute in the South China Sea: does Beijing have a legitimate claim? In: The South China sea disputes: flashpoints, turningpoints and trajectories, pp 23–26

294

N. R. Rajalakshmi and K. Saravanan

3. Binder C (2017) Border disputes through ill-defined borders: maritime territorial conflicts and their impact on security. Border politics. Springer, Cham, pp 33–50 4. Bunholi IV, da Silva Ferrette BL, De Biasi JB, de Oliveira Magalhães C, RotundoMM, Oliveira C et al (2018) The fishing and illegal trade ofthe angelshark: DNA barcoding against misleading identifications. Fish Res 206:193–197 5. Charan AS, Srinivasaverma VSM,Mahammad N (2016) Lifeline system for Fisherman. Springer India. Microelectronics, Electromagnetics and Telecommunications. https://doi.org/ 10.1007/978-81-322-2728-1_55 6. Charles J, et al (2016) Alert system for fishermen crossing border using android. In: International conference on electrical, electronics and optimization techniques (ICEEOT), pp 245–256 7. Churchill RR (2018) Maritime boundary problems in the barents Sea. Routledge revivals: maritime boundaries and ocean resources (1987). Routledge, Abingdon, pp 147–161 8. De Lessio MP, Wynn DC, Clarkson PJ (2017) Modelling the planning system in design and development. Res Eng Des:1–23 9. Haxhibeqiri J, Abeele FVD, Moerman I, Hoebeke J (2017) LoRa scalability: a simulation model based on interference measurements. Sensors 17(6):1193 10. Jeandesboz J (2017) European border policing: EUROSUR, knowledge, calculation. Glob Crime 18(3):256–285 11. Jim Isaac D, Eugene kingsley (2015) Advanced border alert system using GPS and with intelligent Engine Control Unit. Int J Electr Comput Eng 1(4) 12. Kamalakannan B, Naresh K, Sakthivel P (2016) Protecting fishermen’s by detecting and warning them while crossing sea borders using GSM and RFID technologies. In: Online international conference on green engineering and technologies 13. Kumar R, Dinesh, Shubin Aldo M, Charles Finny Joseph J (2016) Alert System for fishermen crossing border using android. In: International conference on electrical, electronics, and optimization techniques (ICEEOT) 14. Lampert T (2017) Stopping illegal fishing and seafood fraudsters: the presidential task force’s plan on tackling IUU fishing and seafood fraud. BCL Rev 58:1629 15. Lim K (2017) The role of the international maritime organization in preventing the pollution of the world’s oceans from ships and shipping. UN Chron 54(2):52–54 16. LoRaTM Alliance, “A technical overview of LoRa LoRaWAN.”[Online].Available:{https:// www.loraalliance.org/portals/0/documents/whitepapers/LoRaWAN101.pdf 17. Michalski M, Cjajewski J (2004) The accuracy of the global positioning system. Inst Electr Electron Eng, Instrum Meas Mag 7(1):56–60 18. Models in the Maritime Industry. In: Meeting the energy demands of emerging economies. 40th IAEE international conference, June 18–21, 2017. International Association for Energy Economics 19. Naveen Kumar M, Ranjith R (2014) Border alert and smart tracking system with alarm using DGPS and GSM. Int J Emerg Technol Comput Sci Electron 8(1):2014 20. Nguyen HT (2017) China’s Bhistorical evidence: Vietnam’s position on South China Sea. In: The South China sea disputes: flashpoints, turning points and trajectories, pp 243–246 21. Rosello M (2017) Cooperation and unregulated fishing: interactions between customary international law, and the European Union IUU fishing regulation. Mar Policy 84:306–312 22. Saravanan K, Aswini S, Kumar R (2019) How to prevent maritime border collision for fisheries?-A design of Real-Time automatic identification system.Earth Sci Inform:1–12 23. Senthilkumar A (2013) Portable life protection system for fishermen using global positioning system. Int J Emerg Technol Adv Eng 3(9):60–64 24. Singh DK, Kushwaha DS (2017) Automatic intruder combat system: away to smart border surveillance. Def Sci J 67(1):50 25. Sivagnanam G, Midhun AJ, Krishna N, Maria G, Reuben Samuel, Anguraj A (2015) Coast guard alert and rescue system for international maritime line crossing of fisherman. Int J Innov Res Adv Eng 2(2):2015

Fisherman Communication at Deep Sea Using Border Alert System

295

26. Sundar S, Mani, Hariharan M (2013) Deep sea fishermen patrol system for coastal intruder positioning. Int J Sci Eng Technol 2(3):129–132 27. Surekha M, Preethi R, Padmapriya K, Devika T, Divya N (2014) ARM based fishing boat security system. Int J Innov Res Electr, Electron, Instrum Control Eng 2(2):2014 28. Sureshkumar K, Sharath Kumar K (2010) Design of low cost maritime boundary identification device using GPS system. Int J Eng Sci Technol 2(9):2010 29. Tsiotas D (2017) The imprint of tourism on the topology of maritime networks: evidence from Greece. Anatolia 28(1):52–68 30. Vanparia, Pradip, Ghodasara YR (2014) Review paper on to study and enhance coastal security system using GIS/GPS tool.Int J Comput Appl and Inf Technol 1 31. Vintila DF, Filip C, Stan MI, Tenea ¸ DD (2017) Considerations regarding a sustainable economic development in the Romanian black sea coastal area in the context of maritime spatial planning (MSP). In: International conference on management and industrial engineering, vol 8. Niculescu Publishing House, Bucures, ti, pp 667–675 32. Vivek S, AravindVenkatraman TV, Aravinth L, and Raagul G (2015) Automatic border crossing detection and navigation of boat. Int J Sci Technol and Eng 1(8) 33. Yin Y, Lam JSL, Tran NK (2017) Review of emission accounting models in the maritime industry, Meeting the Energy Demands of Emerging Economies. In: 40th IAEE international conference, International association for energy economics, June 18–21

Promoting Green Products Through E-Governance Ecosystem: An Exploratory Study Arindam Chakrabarty, Mudang Tagiya and Shyamalee Sinha

Abstract Green product is the future of global sustainability. The e-governance has been emerged as a form of effective and efficient strategy of the state to optimize its resources and delivery mechanism. The green product needs serious attention, encouragement, investment, and effective promotional strategies so that it gathers the desired momentum in the market. This paper has attempted to understand the basic concept of green products and its various illustrations across diversified product segments. The paper has proposed a conceptual model which is simple but effective to encourage the consumers by appropriately exercising reward-incentive mechanism. This research paper is exploratory in nature, which has been developed using various secondary information and research outcomes. Keywords Green products · Sustainability · Green technology · E-governance · Ecosystem

1 Introduction 1.1 Green Product and Commitment Toward Environment There are products having the feature of less impact on the environment or are less detrimental to human health than traditional equivalents. Such products fall under the category of green products. These may be developed or partly developed from recycled components, manufactured in a more energy-conservative way, supplied to the market with less packaging, or manufactured from local materials to reduce the need for transportation and also reduce carbon footprints. In today’s world, the A. Chakrabarty (B) Department of Management, Rajiv Gandhi University (Central University), Itanagar, Arunachal Pradesh 791112, India e-mail: [email protected] M. Tagiya · S. Sinha North Eastern Regional Institute of Science and Technology, Nirjuli, Itanagar, Arunachal Pradesh, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_34

297

298

A. Chakrabarty et al.

planet needs to be protected. Human greed and selfish ambition has exploited the resources and put the planet in a critical predicament. By using and promoting the green products, one may contribute to the safety and preservation of the resources provided from the planet, such as metals, plastics, and even water. Today, more number of people needs to be aware about green products and its application so that it would benefit all living beings in the planet earth. The term development has been perhaps wrongly or narrowly manifested within the locus of massive infrastructure, construction, and building of engineering structures to jump from natural green to jungles of concrete. The north-eastern states are still alive with its flora and fauna. If the adoption of green products has not been incorporated by upcoming generations, the flood of indiscriminate and irresponsible consumerism would sweep the core values of sustainability for the region and for the entire nation [1].

1.2 Emerging Green Management Practices 1.2.1

Green Marketing

Green marketing incorporates a broad range of activities, including product modification, changes to the production process, packaging changes, as well as modifying advertising. Yet defining green marketing is not a simple task. Indeed, the terminology used in this area has varied; it includes: green marketing, environmental marketing, and ecological marketing. While green marketing came into prominence in the late 1980s and early 1990s, it was first discussed much earlier. The American Marketing Association (AMA) held the first workshop on “Ecological Marketing” in 1975 [2]. Green or environmental marketing consists of all activities designed to generate and facilitate any exchanges intended to satisfy human needs or wants such that the satisfaction of these needs and wants occurs, with minimal detrimental impact on the natural environment [3].

1.2.2

Green HRM

Nowadays, green HRM has become a significant thrust area for management which can have an enormous impact on people issues in an organization. It is the application of HRM policies in the way to encourage sustainable use of resources in an organization by increasing awareness and commitments among the employees toward the issues of sustainability to protect and preserve natural resources. It consists of two important elements, that is, environment-friendly HRM practices and the protection of knowledge capital. Green HRM consists of process and practices, like acquisition, induction, training, performance management, and reward system, which have a bearing on the whole carbon footprint of an organization. Green practices under green HRM that are followed by the company are power saving, internal environment and energy audit, eco-friendly or green surveys, going paperless

Promoting Green Products Through E-Governance Ecosystem

299

by using software and apps and so on, recycle waste, water saving, alternative energy sources and so on.

1.2.3

Green Finance

Green finance refers to financial investments flowing into sustainable development projects and initiatives, environmental products, and policies that encourage the development of a more sustainable economy. Green finance includes climate finance but is not limited to it. It also refers to a wider range of other environmental objectives, for example industrial pollution control, water sanitation, or biodiversity protection. Mitigation and adaptation finance is specifically related to climate change related activities: mitigation financial flows refer to investments in projects and programs that contribute to reducing or avoiding greenhouse gas emissions (GHGs) whereas adaptation financial flows refer to investments that contribute to reducing the vulnerability of goods and persons to the effects of climate change [4].

1.2.4

Green Technology, Green Manufacturing, and Green Services

Green technology is considered as environment-friendly based on its production process or supply chain. It also may refer to a means of energy production that is less harmful to the environment than more traditional ways of generating energy, such as burning fossil fuels. This technology is considered as young market comparatively, but investor’s interest runs very high in response to global warming fears and the increasing scarcity of many natural resources (Investopedia). It aims to conserve nature and mitigate the impact of human activities. This technology provides the benefits not only to nature but also for a clean and greener human lifestyle. This technology ensures that the earth remains well for all generations and exist. On the other hand, the “green” manufacturing is known for the renewal of production processes and the establishment of environment-friendly operations within the manufacturing field. In the process the workers use minimal natural resources, reduce pollution and waste, recycle and reuse materials, and moderate emissions in their processes.

2 Theoretical Background There was a time where many practicing managers regarded a preoccupation with green management almost exclusively as a threat. Nowadays, it is more widely accepted that green management can be profitable [5–7]. Green management can act as a vital role in the optimization of production processes and new-product development, not only in pollution-sensitive industries, such as petrochemicals and electric power and manufacturing, but also in high-tech industries [8]. The need for

300

A. Chakrabarty et al.

green management springs from a variety of sources, including societal mandates incorporated into laws, treaties, and regulations [9]. Since green management is a type of public good, whose full value a firm cannot entirely appropriate [10], government’s role in the acquisition of green capabilities is obviously important [11]. Management or managers should pre-define green goals, targets, and responsibilities for their strategic business unit, and corporates should assess number of green incidents, use of environment responsibility, and successful communication of environmental policy within their scope of their operations for improving the performance [12, 13].

3 Objectives of the Study I. To study the concept of green product and its representation across various product segment. II. To formulate comprehensive model and flowchart to increase and optimize green movement in India through efficient e-governance.

4 Research Methodology This paper is designed on the basis of various reports, articles, research papers, and information collected from varied secondary sources. The conceptual model has been proposed in order to motivate the users toward green products by establishing real-time network with the market players. The e-governance framework may retrieve adequate information about the green product and its purchase indents so that it could establish a structured reward-incentive mechanism for promoting green marketing.

5 Analysis 5.1 Analysis—I The wave of sustainable development has drawn the attention of the manufacturers, service providers, users, policy makers, and so on across the globe. It has been trickle down from the developed economies to the developing nations of the world. The affinity of the people of India has been increasing to the extent that it has found that the propensity of using green products has been significantly observed among the indigenous community of Arunachal Pradesh, the least population density state in India [1]. The study conducted by Chakrabarty and Tagiya [1] has emphasized that the attitude of the consumer toward environment and green products has combined

Promoting Green Products Through E-Governance Ecosystem

301

effect on favorable purchase intention behavior. However, price sensitivity, quality enhancement, brand familiarity, ease of access, and convenient to use are the decisive factors that influence the attitude of consumer toward green product. The availability, ease of access, and awareness of green product predominantly encourage the buyers for purchasing or availing green product or green technology. The green products are gaining popularity day-by-day and it became available in various sectors, for example, FMCG, consumer durability, health care, white goods, packaging material, and transportation. The indicative list of green products is illustrated below: FMCG Sector: Biodegradable detergents, soaps, green tea, eco-friendly disinfectants, all types of papers (writing papers, tissues, toilet). Consumer Durable Segment: Recyclable batteries, LED light bulbs and tubes, solar panels, clay-based cutlery, and crockery. Health Care Sector: Biodegradable fittings and fixtures, cotton-based consumables for dressing or bandit materials, cotton bed sheet, eco-friendly disinfectants, biodegradable gloves. White Goods Segment: Water heater tank (electric), dish washer, high-efficiency washing machine and clothes dryer, induction top cooker, energy star refrigerators, vacuum cleaner, dual-blade twin window fan. Packaging Industry: Edible package material, paper bags, tetra-pack package. Transportation sector: Bio fuel, low-carbon emission gas (CNG), recyclable tires.

5.2 Analysis—II The popularity and penetration of green product may essentially be enhanced by the collective efforts of all the stake holders, including the dominant role of the government. The strategic and interactive roles among the stake holders are the prerequisite for enabling the green products in the demand baskets of its users. The strong network needs to be established that would yield desired result for effective promotion strategy of green products. A conceptual model has been proposed where the e-governance can facilitate to promote green consumerism.

302

A. Chakrabarty et al.

5.3 Modus Operandi of Proposed Model Step 1: The Government should identify the lists of green products, green technology, and green processes. Appropriate awareness campaign may be initiated to create customer pool for this segment. Step 2: The market players may be identified and are established with the real-time network through which any transaction made at their end may send the overview of purchased details. Step 3: Based on the purchase details, the customer profile would be identified and tracked. The incentive package or any form of subsidy may be extended to the identified customer through electronic transfer in the form of “Direct Benefit Transfer” (DBT). Step 4: The real-time reward-incentive mechanism would reinforce and promote the green product among the target segments.

6 Conclusion In the dynamics of fourth industrial revolution, to apply threshold level of technology emerged, particularly in the domain of IoT ecosystem. This is high time to create appropriate interface and network between public–private interactions through new generation devices. The e-governance is quite popular and useful in augmenting the efficient delivery system across the world even in India. The success of smart card in Andhra Pradesh is the testimony of India’s success story where the system minimizes its leakage [14]. The paper has showcased how the appropriate reward-incentive mechanism can be offered to the green product users using augmented electronic governance. This model may be implemented that would essentially increase green

Promoting Green Products Through E-Governance Ecosystem

303

consumerism in the market, which in turn would fulfill the commitment of sustainable development as expressed in Brundtland Commission 1987.

References 1. Chakrabarty A, Tagiya M (2018) Awareness and affinity towards green products among young generation: a case of arunachal pradesh. In: Mathirajan M (ed) 6th international conference on business analytics and intelligence 2018 (ICBAI-2018). I.K. International Publishing House Pvt. Ltd, pp 142–154 2. Ottman J, Books NB (1998) Green marketing: opportunity for innovation. J Sustain Prod Des 60(7):136–667 3. Polonsky MJ (1995) A stakeholder theory approach to designing environmental marketing strategy. J Bus Ind Mark 10(3):29–46 4. Höhne N, Khosla S, Fekete H, Gilbert A (2011) Mapping of green finance delivered by IDFC members in 2011. Ecofys, Cologne. http://www.idfc.Org/Downloads/Publications/01_green_ finance_mappings/IDFC_Green_Finance_Mapping_Report_2012_14-06-12.Pdf 5. Porter ME, Van der Linde C (1995) Toward a new conception of the environmentcompetitiveness relationship. J Econ Perspect 9(4):97–118 6. Sharma S (2000) Managerial interpretations and organizational context as predictors of corporate choice of environmental strategy. Acad Manag J 43(4):681–697 7. Sharma S, Vredenburg H (1998) Proactive corporate environmental strategy and the development of competitively valuable organizational capabilities. Strateg Manag J 19(8):729–753 8. King A (1999) Retrieving and transferring embodied data: Implications for the management of interdependence within organizations. Manag Sci 45(7):918–935 9. Marcus AA (1980a) Promise and performance: choosing and implementing an environmental policy, Vol 39. Praeger Pub Text 10. Teece DJ (2007) Explicating dynamic capabilities: the nature and micro foundations of (sustainable) enterprise performance. Strateg Manag J 28(13):1319–1350 11. Marcus A (1980b) The Environmental protection agency. The politics of regulation 12. Renwick D, Redman T, Maguire S (2008) Green HRM: a review, process model, and research agenda. Univ Sheff Manag Sch Discuss Pap 1:1–46 13. Renwick DW, Redman T, Maguire S (2013) Green human resource management: a review and research agenda. Int J Manag Rev 15(1):1–14 14. Chakrabarty A (2019) Is India poised for M-commerce in the cashless milieu? In: Singh A, Duhan P (eds) M-commerce experiencing the phygital retail. Apple Academic Press, CRC, Taylor and Francis Group, pp 205–216. Hard ISBN: 9781771887144, E-Book ISBN: 9780429487736. https://doi.org/10.1201/9780429487736

Intervention of Smart Ecosystem in Indian Higher Education System: Inclusiveness, Quality and Accountability Arindam Chakrabarty, Mudang Tagiya and Shyamalee Sinha

Abstract In the knowledge age, the human society largely depends on both inclusive growth and superior quality of higher education system. The world is transforming very fast and it tends to celebrate the fourth industrial revolution that extends from the information processing and automation to the extent of replication of human intelligence. The emerging protocol of artificial intelligence, RFID, cloud computing, block chain and machine learning are the gamut of resources which essentially would embody the teaching–learning process more effective and result-oriented. In India, the use of e-resources like MOOC, e-learning, Swyam have been experimented and they enjoy popularity and success among the users. However, the higher education system of the country is severely compromised by regular flow of information and databases. It is affecting the quality of teaching–learning process and research. It is high time to have a centralized database reservoir which would contribute to every learning organization, irrespective of government, private or NGO. The database would be collected and preserved by a national e-resource portal which could be accessed by any individual or institution with or without any processing fees; otherwise the direction and continuum of academia and research would have to be severely affected. The present study has attempted to showcase how the various e-resources are integrated into Indian education system. The paper would also approach and present a prototype model on how a comprehensive e-resource portal can be developed and optimized to ensure collection, preservation and access of data set. Keywords Higher education · Smart ecosystem · e-resource portal · Inclusive growth · Accountability

A. Chakrabarty (B) Department of Management, Rajiv Gandhi University (Central University), Itanagar, Arunachal Pradesh 791112, India M. Tagiya · S. Sinha North Eastern Regional Institute of Science and Technology, Nirjuli, Itanagar, Arunachal Pradesh, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_35

305

306

A. Chakrabarty et al.

1 Introduction 1.1 India and Higher Education During the ancient era, the scenario of Indian education consists of ‘Gurukul System’ which mainly concentrated on education to developed knowledge. The Guru (referred to the teacher) will train their ‘Sishya’ (referred to the students) through yoga, meditations and various standards. The early education system in India eventually got unnoticed due to series of invasions and dispute in the country. In the beginning of modern age, the Islamic influences improved the outdated education centers and brought in the broad domains, like geography, administration, law, arabic, mathematics and so on, into India. Colonial rulers who ruled India brought a significant transformation in the higher education system. It was the British who set up the formal system of higher education dedicated to the disciplines, like languages, literature, history and philosophy. In India, the higher education system started to grow rapidly after independence. The study shows that during the year 1980, there were 132 universities and 4738 colleges, enrolling around 5% of the eligible age group in higher education. The total number of educational institutions in India was four times higher than the overall number of institutions present in both United States as well as Europe. Today, India is advancing toward modernization, technology, communication, education and economic growth. It is giving a tough competition to other developed nations in the field of high-tech industries, such as agriculture, medical, information technology, energy and power, and biotechnology to drive the nation to opulence. In the present day, Indian higher education system holds an important place in the global education industry. India has one of the largest networks of higher education institutions in the world and is the third largest in the world. The UGC—an apex body established in the year 1949—essentially deals with the setting up and maintenance of standards in higher education throughout the nation on a uniform basis. The Central Government has been playing a key role in providing overall policy directions and thus acts as a vital link between the policy-making bodies of the government and institutions of higher education. With the introduction of various policies on higher education and subsequent programs undertaken to operationalize the policy has significantly impacted the growth and development of higher education in India. The important landmarks in the evolution of policy in higher education are as under:

Evaluation of Higher Education Policy in India

Year

University Education Commission

1948–49

Education Commission

1964–66

National Policy on Education

1968

Policy on Education (Draft)

1978 (continued)

Intervention of Smart Ecosystem in Indian Higher Education System …

307

(continued) Evaluation of Higher Education Policy in India

Year

National Commission on Teachers-II

1984

Challenge of Education: A Policy Perspective

1985

National Policy on Education

1986

National Policy on Education: A Program of Action

1986

National Policy on Education: A Program of Action

1992

Source IGNOU study material for PGDHE (MES-101, Block-2, Unit-6, pp. 23

1.2 Emphasis on ICT in the Higher Education Policy In the year 1984–1985, the need of ‘Information and Communication Technology’ (ICT) in education sector has been recognized in India. It was realized when the program called Computer Literacy and Studies in Schools (CLASS) was introduced on experimental basis, and the project was later on adopted as a centrally sponsored scheme during the seventh Five-Year Plan (1993–1998). Eventually, the scheme was extended in eighth plan to provide financial grants to institutions covered earlier and to include new government-aided secondary and senior secondary schools. The financial assistance included annual maintenance grant and purchasing equipments for new school. During this period 2598 schools were covered. In the mid-1998, the information technology and software development (IT taskforce) came into the picture for the purpose of recommendations on introduction of IT in education sector including school. The report recommended the provision of computer system to all educational institutions up to higher secondary schools by appropriate investments (about 2–3%) of total budget during the next five years. During the year 2001–2002, a revised class scheme was introduced by making the provision of Rs. 845 million on recommendation. The applications of ICT for quality improvement were also included in Government of India flagship program on education, viz., Sarva Shiksha Abhiyaan (SSA).

308

A. Chakrabarty et al.

2 Literature Review [1] ICT in Indian University and colleges shows the revolution of higher education in the nation, in terms of access, equity and quality with the application of ICT in education. The various prospects and challenges posed by amalgamation of ICTs in various aspects of higher education in the present scenario are discussed and factors regarding future development in ICT in education sector are also highlighted. Information and communication technology (ICT) plays a major role in supporting powerful, efficient management and administration in education sector. It is stated that ICT may use right from student management to various resource management in an education institution [2]. The e-learning and pedagogical innovation framework at Leicester provided a proper stage for the number of formal and informal discussions required to develop an e-learning strategy for the university [3]. ICT evolves as an instrument toward advanced knowledge. As learning tool, that is, ICT, it enhances the human intellectuals and capabilities in solving problems, helping and benefiting the students in gaining and increasing knowledge, and promoting the faculties, teachers, trainers and administrators in improving teaching and learning. This technology has also incorporated the knowledge and skills required to effectively use ICT as a tool [4]. Even though the application of ICT is not the answer for all the challenges faced by higher education systems in the region, it does leverage and extend conventional teaching and learning activities, and has the potential to positively influence on learning [5]. The application of information and communication technology (ICT) in higher education system has resulted in shifting from teacher-centered delivery and transmissive learning to student-centered learning. ICT acts as a channel of information, and intellectual tools have been supporting and serving the students to be mature enough and become responsible toward learning [6].

3 Objectives of the Study I. To explore the application of smart eco-system in Indian higher education system. II. To formulate integrated and smart strategy framework for sharing information through the man–machine interfaces across the country.

4 Research Methodologies This study conceptual in nature is based on information collected from secondary sources like reports, journals and so on. The paper attempts to understand the present scenario of smart eco-system used in higher education system in India and attempts to

Intervention of Smart Ecosystem in Indian Higher Education System …

309

formulate a schematic model where the advance smart eco-system would be deployed the available resources to enhance quality of the higher education.

5 Analysis and Interpretation 5.1 Analysis—I

Various e-resources on education E-resources from MHRD supported programs

Other learning resources

NPTEL (https://nptel.ac.in/)

Digital Teaching and Learning Resources for PwDs (http://www.ayjnihh.nic.in/Digital_ teach_resources.asp)

Virtual Labs (http://www.vlab.co.in/)

LILA Hindi Pravah

Spoken Tutorial (https://spoken-tutorial.org/)

Physics—Mysterious Magnetism (http:// www.youtube.com/watch?v= wKdqCqTzSnI&list=PLdm-2_ AHi21QoOEbiVEMty8vy6yS3UWF3& index=4)

The Consortium for Educational Communication (http://cec.nic.in/Pages/ Home.aspx)

Astronomy—Eclipse (http://www.youtube. com/watch?v=Q1yq2LpQ-Qc&list=PLdm-2_ AHi21QoOEbiVEMty8vy6yS3UWF3& index=28)

e-Yantra (https://www.e-yantra.org/)

Astronomy—Day and Night

e-ShodhSindhu (www.inflibnet.ac.in/ess/)

Khan Academy

FOSSEE (Free and Open Software in Education) (https://fossee.in/)

CS Unplugged Coursera Udemy (http://www.youtube.com/watch?v= Q1yq2LpQ-Qc&list=PLdm-2_ AHi21QoOEbiVEMty8vy6yS3UWF3& index=28) MITOCW (https://www.edx.org/) LEARNING SPACE:THE OPEN UNIVERSITY (https://www.open.edu/ openlearn/) Vidya Online (http://www.vidyaonline.net/ index.php)

Source http://vikaspedia.in

310

A. Chakrabarty et al.

India has been using e-education resources since early twenty-first century. The application of ICT has been pioneered in a few open and distance learning program and online courses offered by both govt. and private enterprises with the pace of phenomenal growth in satellite technology, access to the internet and even highconfiguration mobile usage. The importance of e-resources has been well thought and adopted. The indicative list of popular e-education resources in India is mentioned below:

5.2 Analysis—II The growth of higher education system largely depends on creation of new knowledge, development of contents, smart dissemination process and application-based research. In India, various sectors and agencies are working in their respective domains but observations, outcome and experiences are not adequately shared among all its partner, stakeholders and users. These create massive hindrance for the learners, researchers and the implementing agencies to achieve success in their respective intellectual pursuits. The lack of data support or exchange leads the society toward policy paralysis. India is in the alarming position where all the knowledge-generating, policy-making and research organization need to interact freely with their databases, sharing of experiences and critical observations. The paper has coined this urgency and has attempted to device an integrated e-resource portal which would perform the task of continuous data collection, preservation, its uninterrupted flow of processing across the entire stakeholders.

Intervention of Smart Ecosystem in Indian Higher Education System …

311

Modus Operandi of the Proposed Model Step—1: Creation of a dedicated e-knowledge national portal/database. It can be created by appropriate enactment of law and with the consortium public–private partnership (PPP). Step—II: All the knowledge bodies, institutes, private enterprises, non-government organization, govt.–pvt. establishments and so on compulsorily need to share their database, particularly R&D, process outcomes, achievements experiences or observations to the said national portal on regular interval. All the contributing institutional entities may be connected through appropriate network topology or modern gateway. It could also use the flowchart of electronic data interchange (EDI). Step—III: All the data set/information would be collated, correlated and preserved so that the user’s community across the nation can benefit from this system. However, access to the database may be free of cost or partially chargeable as the case may be depending on the rigor and cost implication of data procurement and its preservation. However, this national portal may exclude the information pertaining to security issues of the country, as well as the product/process secrecy, and others forms. In fact, the national portal would collate and preserve all the published information or documents in an integrated, coherent and synergic orientation. Step—IV: The user may have to either subscribe with the portal or have to purchase the database if it is chargeable. In the present context of Indian higher education system, it is difficult to access panel data or cross-sectional data, due to lack of integration of data across the stakeholders on a particular field of inquiry. The standard deviation and variation in data collected from different sources on a particular set of measurable attributes and entities sometimes appear to be very high. Few organizations that generate and

312

A. Chakrabarty et al.

preserve database hardly share with the common users whether it is inaccessible, limitedly accessible or very costly for use. There are issues in generating particular set of database on regular intervals. The question of real-time data management is insignificantly exercised. In the corporate sector, the transparency and disclosure of information are limited and highly concentrated among few big players. All these catastrophic features of data generation, preservation and excess mechanism have largely affected the quality of higher education, teaching–learning and dissemination process. The research activities are severely compromised because of lack of availability or access to relevant data set. This paper has devised a comprehensive and nationwide data management ecosystem which would collate, collaborate and integrate all the relevant stakeholders for generating, preserving and sharing the platform for the users. The proposed national e-portal would ensure the authenticity, reliability of dataset and avoid data redundancy. This would ensure incremental access to such dynamic database platform which would trigger for achieving higher inclusive education. The projected model would perform the task of validating and integrating the database with higher precision and reliability which would ensure superior quality of knowledge exchange. As the system is reinforced by all the stakeholders of the country representing various segments of economy and intellectual acumen, the system is committed toward creating high-end value in the process of creating of knowledge and its dissemination, which shall reaffirm the spirit of accountability.

6 Conclusion Information is the most decisive factor for success, particularly in the era of knowledge economy. Even in the ancient time the battle was fought among kings and the winning party did not conquer not only because of its marshal but of its strength in information search. This paper has shown the growing trend of using e-resources in the process of teaching–learning dissemination and research, particularly in the Indian context. However, there is lack of integrated approach to collate, preserve and share all the pertaining data sets among its users and stakeholders. This paper portraits a model solution by integrating all contributory institution with the national portal and in return the propagation of data flow from the portals to individual and institutional users with free access, limited access or paid access mechanism so that without compromising the sovereignty, security issues of the nation and without affecting the patent, copyright and commitments the country can foster high-end academic environment and experience the frontiers of research outcome.

Intervention of Smart Ecosystem in Indian Higher Education System …

313

References 1. Snehi N (2009) ICT in Indian universities and colleges: opportunities and challenges. Manag Chang 13(2):231–244 2. Maki C (2008) Information and communication technology for administration and management for secondary schools in cyprus. J Online Learn Teach 4(3):18–20 3. Salmon G (2005) Flying not flapping: a strategic framework for e-learning and pedagogical innovation in higher education institutions. ALT-J 13(3):201–218 4. Mlitwa N (2007) Technology for teaching and learning in higher education contexts: activity theory and actor network theory analytical perspectives. Int J Educ Dev Using ICT 3(4):54–70 5. Jaffer S, Ng’ambi D, Czerniewicz L (2007) The role of ICTs in higher education in South Africa: one strategy for addressing teaching and learning challenges. Int J Educ Dev Using ICT 3(4):131–142 6. Jonassen DH (1996) Learning with technology: using computers as cognitive tools. In: Handbook of research for educational communications and technology

A Study of Epidemic Approach for Worm Propagation in Wireless Sensor Network Shashank Awasthi, Naresh Kumar and Pramod Kumar Srivastava

Abstract Wireless sensor networks (WSNs) are novel expansive-scale wireless network systems that comprise conveyed, self-sorting out, low-control, ease, smallsensor gadgets that help to gather data through framework-less wireless network system. Nowadays, propagation of worms is a major issue in wireless sensor network because a large number of attackers (virus) are present in the network by which the total network or communication will be affected. Sometimes network limitation and node boundaries are also responsible for lost information. For means of perfect propagation of worm, some data is required (about the node and network), that is, knowledge of local concept, power of node, as well as network, distribution concept, and timing. In this paper we try to represent the basic concept of different kinds of propagation model for propagation of worm using wireless sensor network. Here we study an epidemic approach for worm propagation in wireless sensor networks (WSNs) to represent the concept of worm propagation in different applications or propagation from one node to another node or from one node to the entire network. Keywords Wireless sensor networks (WSNs) · Epidemic theory · Worm propagation

1 Introduction Nowadays, communication through wireless is more popular. From this point of view, the wireless sensor network has a lot of attraction because wireless sensor network is widely used in many applications, for example, in military to detect the location of target as well as enemies, in security for checking the presence of any objectionable S. Awasthi (B) · N. Kumar Department of Computer Science Engineering, Galgotia’s University, Greater Noida, India e-mail: [email protected] Department of Computer Science Engineering, G. L. Bajaj Institute of Technology & Management, Greater Noida, India P. K. Srivastava Department of Computer Science Engineering, Rajkiya Engineering College, Azamagarh, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_36

315

316

S. Awasthi et al.

item or persons, to send the top secret information in defense, to provide care to the humans (medical) and daily needs goods in commercial applications, in interruption identification, organic discovery, and so on [1, 2]. Wireless sensor network consists of some special features and characteristics such as little measured, modest, vitality restricted, and multi-useful gadgets considered as sensors, which are sent to gather information from a domain [3]. Its architecture consists of a number of nodes which are responsible for the perfect data transmission and receptions. These nodes (known as sensor nodes) first check whether the data is correct or not; if the data is correct then the sensor nodes sense the channel to verify that channel is ready or not; for ready condition, the sensor nodes accept the information from network and forward them to the users. The architecture of wireless sensor network is a combination of various types of component, elements, devices, and instruments, for example, sensors nodes, processing devices, collected devices, distributed devices, power supply, memories, transmitters, receivers, display devices, and checking devices (as shown in Fig. 1). Nodes may or may not be a part of wireless sensor network. If the node is a part of network then the sensor node performs various functions and a node has various targets to complete whenever the sensor node is involved in a transmission and receiving process. During this process it verifies the physical structure of environment, status of nodes, condition of network, location of nodes, and reliability of communication (both for transmitting and receiving). Sometimes, it is very difficult to transmit or receive the information from network to user or user to network because of low energy level of nodes, network architecture, availability of internet, atmosphere condition, weather condition, and air pollution. Because of these reasons the whole information will be destroyed and damaged [4, 5]. So for rapid communication, a wireless device should be perfect and up-to-date with memory, software, and connecting devices because wireless devices move quickly from place to place, check the network, gather the information from network, and transmit the information to the receiver or user. In other words, we can say that the propagation of worm is directly dependent on the condition of wireless devices. If the wireless device deals in terms of ideal characteristics and feature (noise or distortion or error free), then the worm propagation will be ideal. Sometimes it is

Fig. 1 A general network architecture (used for worm propagation)

A Study of Epidemic Approach for Worm Propagation …

317

very difficult to find a right route in these conditions when a wireless device moves quickly, so the whole process of worm propagation may be corrupted. Because of this reason an epidemic model that provides a right way or route for worm propagation is proposed. This epidemic model for routing supports the worm propagation by sharing the information between all sensor nodes, which are connected within network or involve in worm propagation [6]. The epidemic model for routing also faces some problems; for example, wrong transmission, wrong receptions of information from network, useless messages, length of message, low energy level of nodes, or device. So a wireless sensor network always supports a sensor node with sufficient energy (more than threshold level) and with the sufficient size of memory in which the worm propagation or worm is stored. Each sensor node contains some properties or specifications in terms of constrained energy level and worm propagation properties (range, strength, etc.). If the wireless device has a limitation in terms of range, then the sensor node only sends the received information to nearest neighboring sensor node and this sensor node passes the received information to his next nearest sensor node; however, if a sensor node does not have a sufficient energy level, then it is required to design a hardware which is used to support a low-power rest mode [7]. A low-power rest mode means, if it is not used during worm propagation then that specific node will be moved in the rest mode for a fixed time duration (means minimum power consumption), and after sometime it will be in active condition whenever sensor node is required. As a result, large amount of energy or power will be saved and this saved energy may be used for the next operation. To take care of the issue of worm propagation, different investigations have been done in this line. An author named Cabir worm [8] has proposed a model with a Bluetooth device installed in sensor node or in sender’s device within operating range and similarly another author named Mabir worm [9] used the filtering concept with sensor node or sender device for attack (virus). To describe the basic concept of epidemic model, at first, it is required to find the applications of epidemic model. After a long research, it has been found that some authors have used the concept of epidemic model for worm propagation. Initially, the epidemic model concept was studied to find the worm propagation for disease, for example, plagues and so on. Because some diseases always propagate from persons to persons when persons communicate physically or personally, this concept represents the theory of epidemic model. Some epidemic concept also gives asymptotic scientific (numerical) results when the quantity of nodes (or populace) increases to endlessness by using a utilitarian rendition of the Law of Large Number [10]. Ball and Neal [11] also have given the concept of epidemic theory in different ways. Andersson [12] used the concept of processing on previous stage of any sensor node, and based on this concept, a new model was proposed (model based on discrete time architecture). Anderson and Djehiche [13] also described the concept of worm propagation about the infected persons with powerless assembly as far as possible procedure for a consistent time display. Sellke [14] and Ball and Barbour [15] presented the concept of vulnerable persons, where each and every node supports the two-dimensional data.

318

S. Awasthi et al.

The infection propagation through wireless sensors and the disease transmission in a large number of persons are fundamentally connected or we can say that both are similar. This concept is explained by different configurations [16]. Some researchers also investigate the basic concept and applications of epidemic model for wireless sensor network [17–21]. In [17], a susceptible-infective (SI) scourge display was created for a basic data dispersion calculation, and the effect of node thickness on data dissemination was explored logically. In [18], a topologically mindful worm engendering model (TWPM), which catches both reality proliferation elements, was created for wireless sensor network. In [19], a plague demonstration for a cell phone infection was created, which thought about the dispersion thickness, inclusion sweep, and speed of the cell phone. Pandemic hypothesis was connected in [20] to demonstrate the transmitting procedure of bargained point and distinguish main components deciding power or energy episodes in systems. In [21], a different method is used to describe the rule of regulation of a network for broadcasting necessary information in wireless sensor network. In wireless sensor network, nodes are always used in middle part to provide the routing of the data, but the transmission and reception of the data is performed by the source and the destination. Hence, PCs are used as transmitter and receiver. PC infections transmission concept is dependent on pandemic models by Kephart. He utilized the pandemic models to discover the standard in PC infections and focused on the topological properties of the system and on the propagation of infections. Numerous creators utilized the SIR and SEIR model to examine the conduct of elements of PC infection [22]. A SEIR model is used for the PC infection that incorporates an antitoxin populace compartment [23]. In addition, the effect of association model on personal computer systems to provide a platform to the spreading of PC virus is examined in [24]. Isolate may be considered as an important part of the vital healing procedures to virus malware assault in the system. Many researchers have considered the isolate as one of the critical parts in the plague models [25]. Many models have already being considered before the scientific model for the propagation of worm in a properly organized network [26]. In [27], Wan et al. considered a different model named iSIRS model, which is used for worm propagation in wireless sensor network. Mishra et al. [28] gave a unique concept of worm propagation in wireless sensor coordinate with isolate and upkeep component in the rest nodes. In Part II, survey on the basic function and operation of epidemic theory and wireless sensor network (WSN) model has been presented, while part III introduces the structure of the model and finally, part IV reports some concluding remarks.

2 Epidemic Theory Till date lot of work is done and still continuing to make smart micro grid more efficient and reliable. Akyildiz et al. [3] discuss about advantages and disadvantages of using smart grid in distributed renewal energy generation. Nishiyama et al. [4] focusses on use of Wireless sensor networks for improving efficiency and reliability

A Study of Epidemic Approach for Worm Propagation …

319

of smart grid. Sakano et al. [5] emphasis on monitoring and data collection from generation to load consumption for decision support system. Vahdat and Becker [6] discuss about providing intelligence to smart grid through multi agent system. Pantazisa et al. [7] discusses about grid security and proposes PKI (Public Key Infrastructures) to be best among various available security techniques. Ferrie et al. [8] proposes a generalized technique for based on load shifting. Chien [9] proposes distributed algorithm—Stackelberg Game which will solve Demand Response Management (key component of smart grid) problem and will take care of both companies and consumers. Billingsley [10] discusses about future aspects of smart micro grids and discuss about integrating smart grid with Electric Vehicles. Reference [11] focus on crucial issues related to integrating of smart grid with building which includes use of smart meters, interoperability and demand response capabilities. Andersson [12] proposes use of fuzzy logic for load reduction with thermal comfort considerations. Andersson and Djehiche [13] proposes integrated dynamic Response program to reduce dependency on customers in existing Dynamic Response program. Sellke [14] proposes cognitive based communication for smart grid. Ball and Barbour [15] discuss about RD & D policies for energy storage to enhance operational experience and reduce operational cost. Basically, there is three basic concept in the epidemic theory which may be consider as variables: (i) Provider or Source (beginning node or beginning sensor), (ii) entertainer, who receives or entertains the information from other people, and (iii) surroundings. All concepts are related to each other and support a lot of work function as well as components. Mostly the interaction between provider and entertainer are very quickly but sometimes the change in surrounding may affect the interaction of the provider and the entertainer. The epidemic theory also helpful to describe the concept of propagation of worm from source to destination by defining the virus as well as distortion or noise in information. Similarly we can say that the epidemic theory describe the propagation of virus or distortion or noise in the information. Usually the information always transmit or receive by common path, so if the worm is propagated through the same common path, then the virus also propagate through the same path because the direct contact defines the propagation of virus via a common channel. For example, the virus in milk is also considered the concept of worm propagation in terms of virus through any type of contact as well as medium. In case of study of science, during the irresistible infection at the populace scale, there are two essential methodologies or concepts that describe the propagation of infection, which are defined as: the stochastic methodologies and the deterministic methodologies. The stochastic methodologies can precisely depict vacillations, change variety in dangers of presentation, and different variables; however, they may turn out to be extremely mindboggling and relentless to set up. But, the deterministic methodologies portray in detail, by averaging the value, the dynamic of the malady at the populace measure with huge populaces known as deterministic models. For the most part, a populace of n people is parceled into a few compartments, and the propagation of the malady is mulled over. Roused from the study of epidemiology, propagation of worm in correspondence systems has been found to shockingly take after the transportation of epidemics [28,

320

S. Awasthi et al.

29]. The ordinary differential equations (ODEs), behave as an intelligent elements of compartmental model. Insusceptibility might be impermanent, dependable, and different models exist in epidemic theory that describes disease propagation, for example, the popular and traditional epidemics models, susceptible-infected (SI) model, susceptible-infectedsusceptible (SIS) model [30] and susceptible-infected-recovered (SIR) model [31]. In the SIS display, a susceptible node is tainted and after a hatching period, the individual node winds up and will not be influenced again. In the SIR model, the susceptible node is affected for a short duration, recoups, and after that wind up resistant to encourage contaminations, which are broadly utilized in examining the elements of worm propagation over systems.

3 Wireless Sensor Network (WSN) Model The wireless sensor networks (WSNs) are networks that are described by the sporadic nearness of the minimization. Collector nodes (sensor nodes) received the information from the environment or surrounding, and then two operations work together. First, all the supporting nodes in system try to forward the information, and second is storing the information with minimization concept [32]. The basic and common conditions of nodes are observed regularly within a network for identification of the subject (or the nonattendance). The whole process may also be repeated to observe the change or adjustment of a specific physical parameter [33]. A wireless sensor network with N-susceptible nodes (or sensor), in which some nodes (sensors) behave as an infective nodes (sensors) and some behave as an invalid nodes (sensors), is shown in Fig. 2. All nodes or sensors consist of an antenna to transmit with maximum transmission range. Information produced from a source node (sensor) can be propagated to its neighboring node within the range of transmitter as well as antenna. The neighboring nodes transfer this data to their neighboring node. In the same manner, the whole information can be transmitted from source to destination, but problem may occur when virus or distortion is also transmitted from source to destination during propagation of worm. If a node within wireless sensor network (WSN) is contaminated by an infection or virus because of assaults or neighboring effects of environment or energy level, and so on, the infection or virus can be propagated together with information through proper rule and regulation, which are similarly used to transmit the information [34], and this may damage the whole network or whole transmission. To keep an episode of infection propagation in a WSN, the planner of the network may introduce proper enemy of infection programming on the sensor nodes before sending to the network. The counter infection programming works on the web and checks the node occasionally. It may be possible that sometimes online support may not be sufficient to battle the contamination, since the sensor node is asset-restricted and a functioning node is

A Study of Epidemic Approach for Worm Propagation …

321

Fig. 2 WSN model

caught up with collaborating with its neighbors by methods for different activities, such as “tune in”, “transmit”, “get”, and so forth.

4 Comparison of Different Models Wireless sensor network works on the principle of information transmission and reception means of communication, or we can say that it provides the platform to different kind of users to share their information with each other. A wireless sensor network is a combination of different small–small units or devices. These small units or devices are known as sensor nodes. These units (sensor nodes) are connected in a group within wireless sensor network, and these units work as a provider of the information (user’s data) by providing a way to send the information from one point to another point or vice versa. These units or sensor nodes are characterized, especially because these units consist of some features itself like memories, programming units, powers supplies, decision-making properties, and so on. On the basis of these properties, a wireless sensor network consists of a lot of advantages but it also has some limitations such as overheating of power supply, memory expansion, programming database limitation and so on [35]. Wireless sensor network supports a lot of applications in the field of health care, military, defense, biodiversity, and disaster relief services [36]. A lot of research has been done by different authors or researcher to transmit or receive information using wireless sensor network in the form of epidemic model.

322

S. Awasthi et al.

The way toward propagation worm (in type of worm engendering) on the internet is fundamentally the same as that of natural infection propagation in the populace. The way toward propagation information (worm propagation) on the internet has been generally considered dependent on the epidemic theory [37–44], and in these literatures models were presented to provide the way toward propagation worm (information) in the populace. In [45, 46], the worm propagation is basically described with the irregular diagram hypothesis. The most traditional epidemic models are susceptible-infected (SI) model, susceptible-infected-susceptible (SIS) model, and susceptible-infected-recovered (SIR) model, iSIRS model, EiSIRS model, susceptible-infected-treated-recovered (SITR) model, susceptible-infected-dead-recovered (SIDR) model, susceptibleexposed-infectious (SEI) model, vulnerable-exposed-infectious-secured-vulnerable (VEISV) model and susceptible-exposed-infectious-quarantine-recovered-secured (SEIQRS) model, which are broadly utilized in breaking down the elements of worm (information) propagation over systems. The SI model, sometimes also known as SIS model, is where once an individual is never again infectious. This individual becomes susceptible once again. The basic virus is demonstrated in SI model, but the SIS model may differ on some point, meaning the SIS model represents the repeated infectious diseases, and the SIS demonstrate without imperative elements can be scientifically illuminated to comprehend the sickness elements. In any case, there are two disadvantages of SIS demonstrate, one when the disease or/and restoring is not Poisson form, and the contamination connection between two nodes can be negative and also the tainted people come back to the susceptible state after infection. There is also the SEIR model, where people are categorized as susceptible, exposed, infected, or recovered. The basis of these models is susceptible-infected-recovered (SIR) model, but this model faced some serious problem. The populace must be fixed and needs to combine homogeneously. The model does not consider any variety in the illness among individuals of various genders, races, or ages. Based on SIR model some more models are also invented, such as iSIRS model, EiSIRS model. After that, a new model named SITR model developed and it defined that if the rate of treatment expands the spreading of malicious worms diminishes and improves the life of WSN. As a result, it is discovered that performance of the proposed model is better in contrast with the SIRS, where infected nodes are immediately expelled from the network. Similarly, the SIDR studied defines the node energy consumption. Based on previous theory and assumption, some more models were developed for work to define the secure communication. For example, SEI model worked on the basic concept of predicting future epidemics with less number of assumptions. By utilizing system security phrasing, the SEIR model characterized areas for broken nodes with substitutions that empowered us to recognize the safe courses for new VEISV model. In VEISV model, all worms had the capacity to overrun if the proliferation rate is more prominent. To improve the network condition, a new model named susceptible-exposed-infectious-quarantinerecovered-secured (SEIQRE) was developed. It brings worldwide stability of the disease-free harmony and the endemic balance. A compartment demonstrates taking

A Study of Epidemic Approach for Worm Propagation …

323

no thought the multi-faceted nature of social contact organize. This model considers the complexity of social contact network. These models just improved the current models for propagation information on a given or available network by constraining the scope of information, without thinking about the above vital qualities of information data in a WSN. One of the classical models was presented by Kephart and White [47] in 1991 called SIS model. The system impact was analyzed by a fixed level of every node by the authors (Kephart and White). In Zou et al. [48], the alleged SIR model is proposed to depict the procedure of spreading information on the internet, and it is generally connected in breaking down the elements. Be that as it may, because of various examples, in a WSN, the SIR model may not be straightforwardly used to portray the procedure of worm proliferation in a WSN. In [49], the iSIRS model works with the information of dead nodes and the procedure of vitality utilization of nodes. To beat the drawback of the iSIRS model [49], a model was presented to show the working state and resting condition of nodes to grow the iSIRS model named EiSIRS model. The susceptible-infected-treated-recovered (SITR) model clarifies that some contaminated people should move from treated stage to recuperated stage, subsequent to applying the insurance system [50]. Srivastava et al. [51] proposed the susceptible-infected-removed-dead (SIDR) model by presenting the idea of dead nodes and these nodes cannot be revived in view of being situated in brutal zones, which portray the non-direct elements of a susceptible, infectious, dead, and recovered types of nodes to break down the whole powerful procedure of worm propagation. Li and Zhen [52] contemplated the worldwide asymptotical security of the susceptible-exposed-infectious (SEI) epidemic model. The worldwide asymptotical steadiness and the illness-free harmony were analyzed by utilizing the Lyapunov work, LaSalle’s invariant set hypothesis, and Poincare–Bendixson property. Toutonji et al. [53] proposed a vulnerable-exposed-infectious-secured-vulnerable (VEISV) worm network, which is fitted for estimating the impacts of security countermeasures on worm propagation. Mishra and Zha [54] presented the fixed time of transitory invulnerability in the wake of running enemy of malignant programming on PC hubs and displaying a pernicious article in a PC organized by utilizing the susceptible-exposed-infectiousquarantine-recovered-secured (SEIQRS) model.

5 Conclusion In this paper we have presented the comparative study of epidemic model for worm propagation with respect to wireless sensor network. We found that the whole information can be transmitted from source to destination but problem may occur when at the same time virus or distortion was also transmitted from source to destination during the propagation of information. This may create a problem within wireless sensor

324

S. Awasthi et al.

network. Based on the comparative study of epidemic approach, we can conclude that two conditions may be considered to overcome the problem when information (worm) may be damaged or virus is involved in information. First, a sensor node supports a threshold level of energy. It means that if a node has a sufficient threshold level of energy then information can be transmitted. Second error detection and correction capability should be installed within sensor node.

References 1. Krishnamachari B (2005) Networking wireless sensors. Cambridge University Press, Cambridge 2. Mishra BK, Keshri N (2013) Mathematical model on the transmission of worms in wireless sensor network. Appl Math Model 37:41034111 3. Akyildiz IF, Su W, Sankarasubramaniam Y, Cayirci E (2002) Wireless sensor networks: a survey. Comput Netw 38(4):393–422 4. Nishiyama H, Ito M, Kato N (2014) Relay-by-smartphone: realizing multihop device-to-device communications. IEEE Commun Mag 52(4):56–65 5. Sakano T, Kotabe S, Komukai T, Kumagai T, Shimizu Y, Takahara A, Ngo T, Fadlullah ZM, Nishiyama H, Kato N (2016) Bringing movable and deployable networks to disaster areas: development and field test of MDRU. IEEE Netw 30(1):86–91 6. Vahdat A, Becker D (2000) Epidemic routing for partially connected ad hoc networks. Technical report CS-200006, Duke University 7. Pantazisa NA, Vergadosb DJ, Vergadosa DD, Douligeris C (2009) Energy efficiency in wireless sensor networks using sleep mode TDMA scheduling. Ad Hoc Netw 7:322–343 8. Ferrie P, Szor P, Stanev R, Mouritzen R (2004) Security responses: symbos.cabir. Technical report, Symantec Corporation 9. Chien E (2005) Security response: symbos.mabir. Technical report, Symantec Corporation 10. Billingsley P (1999) Convergence of probability measures. Wiley, New York 11. Ball F, Neal P (2004) Poisson approximations for epidemics with two levels of mixing. Ann Probab 32(1):1168–1200 12. Andersson M (1999) The asymptotic final size distribution of multitype chain-binomial epidemic processes. Adv Appl Probab 31(1):220–234 13. Andersson H, Djehiche B (1994) A functional limit theorem for the total cost of a multitype standard epidemic. Adv Appl Probab 26(3):690–697 14. Sellke T (1983) On the asymptotic distribution of the size of a stochastic epidemic. J Appl Probab 20(2):390–394 15. Ball F, Barbour A (1990) Poisson approximation for some epidemic models. J Appl Probab 27(3):479–490 16. Pastor-Satorras R, Vespignani A (2001) Epidemic propagation in scale-free networks. Phys Rev Lett 86:3200–3203 17. Khelil A, Becker C, Tian J, Rothermel K (2002) Directed-graph epidemiological models of computer viruses. In: Proceedings of 5th ACM international workshop on modeling analysis and simulation of wireless and mobile systems, pp 54–60 18. Khayam SA, Radha H (2005) A topologically-aware worm propagation model for wireless sensor networks. In: Proceedings of 2nd international workshop on security in distributed computing systems, pp 210–216 19. Zheng H, Li D, Gao Z An epidemic model of mobile phone virus. In: 1st International symposium on pervasive computing and applications, August 2006, pp 1–5 20. De P, Liu Y, Das SK (2006) Modeling node compromise propagation in wireless sensor networks using epidemic theory. In: Proceedings of IEEE WoWMoM 2006, (Niagara-Falls, NY), June 2006

A Study of Epidemic Approach for Worm Propagation …

325

21. De P, Liu Y, Das SK (2007) An epidemic theoretic framework for evaluating broadcast protocols in wireless sensor networks. In: Proceedings of IEEE international conference on mobile adhoc and sensor systems (MASS), (Pisa, Italy), October 2007 22. Grassberger P (2002) Critical behavior of the general epidemic process and dynamical percolation. Math Biosci 66(3):35–101 23. Piqueira JRC, Araujo VO (2009) A modified epidemiological model for computer viruses. Appl Math Comput 213:355360 24. Yuan H, Chen G (2008) Network virus-epidemic model with the point-to-group information propagation. Appl Math Comput 206:357–367 25. Chen T, Jamil N (2006) Effectiveness of quarantine in worm epidemics. In: IEEE international conference on communications 2006. IEEE, pp 2142–2147 26. Wang FG, Zhang YK, Wang CG, Ma FF, Moon SJ (2010) Stability analysis of a SEIQV epidemic model for rapid propagation worms. Comput Secur 2(9):410–418 27. Wang XM, Li QL, Li YS (2010) EiSIRS: a formal model to analyze the dynamics of worm propagation in wireless sensor networks. J Comb Optim 20:4762 28. Mishra BK, Srivasstava SK, Mishra MK (2014) A quarantine model on the propagation behavior of worms in wireless sensor network. Trans IoT Cloud Comput 2(1):1–13 29. Cheng S-M, Karyotis V, Chen P-Y, Chen K-C, Papavassiliou S (2013) Diffusion models for information dissemination dynamics in wireless complex communication networks. J Complex Syst 2013 30. Sanders J-L (1971) Quantitative guidelines for communicable disease control programs. Biometrics 27:883–893 31. Hethcote HW (1978) An immunization model for a heterogeneous population. Theoret Popul Biol 14:338–349 32. Di Pietro R, Mancini LV, Soriente C, Spognardi A, Tsudik G (2008) Catch me (if you can): data survival in unattended sensor networks. In: 2008 sixth annual IEEE international conference on pervasive computing and com munications (PerCom), March 2008, pp 185–194 33. Di Pietro R, Oligeri G, Soriente C, Tsudik G (2010) Securing mobile unattended wsns against a mobile adversary. In: IEEE symposium on reliable distributed systems, pp 11–20 34. Boyd S, Ghosh A, Prabhakar B, Shah D (2005) Gossip algorithms: design, analysis and applications. In: Proceedings of IEEE INFOCOM’05, (Miami, FL), March 2005 35. Lai WK, Fan CS, Lin LY (2012) Arranging cluster sizes and trans-mission ranges for wireless sensor networks. Inf Sci 183:117–131 36. Koren I, Krishna CM (2007) Fault tolerant systems. Morgan Kaufman Publishers Inc., San Francisco 37. Frauenthal JC (1981) Mathematical modeling in epidemiology. Springer, New York. ISBN10:0387103287 38. Chen ZS, Gao LX, Kwiat K (2003) Modeling the spread of active worms. In: Proceedings of IEEE INFOCOM, pp 1890–1900 39. Dantu R, Cangussu JW, Patwardhan S (2007) Fast worm containment using feedback control. IEEE Trans Dependable Secure Comput 4(2):119–136 40. Eugster PT, Guerraoui R, Kermarrec AM, Massoulie L (2004) Epidemic information dissemination in distributed systems. IEEE Comput Math Model Epidemiol 37(5):60–67 41. Moore D, Paxson V, Savage S, Shannon C, Stanoford S, Weaver N (2003) Inside the slammer worm. IEEE Secur Priv 1(4):33–39 42. Okamura H, Kobayashi H, Dohi T (2005) Markovian modeling and analysis of internet worm propagation. In: Proceedings of the 16th IEEE international symposium on soft reliability engineering, pp 149– 158 43. Yang F, Duan HX, Li X (2005) Modeling and analyzing of the interaction between worms and antiworms during network worm propagation. Sci China Ser E Inf Sci 34(8):841–856 44. Zou CC, Towsley D, Gong W (2007) Modeling and simulation study of the propagation and defense of Internet e-mail worms. IEEE Trans Dependable Secure Comput 4(2):105–118 45. Syed KA, Hayder R (2006) Using signal processing techniques to model worm propagation over wireless sensor networks. IEEE Signal Process Mag 23(2):164–169

326

S. Awasthi et al.

46. Pradip D, Liu Y, Sajalk D (2007) Modeling node comprise spread in wireless sensor networks using epidemic theory. In: Proceedings of IEEE international symposium on a world of wireless, mobile and multimedia networks, pp 237–243 47. Kephart JO, White SR (1991) Direct-graph epidemiological models of computer viruses. In: Proceedings of 1991 IEEE computer society symposium on research in security and privacy, May 1991, pp 343–359 48. Zou CC, Gong G, Towsley D (2005) The monitoring and early warning for internet worms. IEEE Trans Netw 13(6):961–974 49. Wang X, Li Y (2008) A improved SIR model for worm propagation in wireless sensor networks. Chin J Electron 18(1):28–32 50. Shashank A, Pratap OR, Kumar SP, Goutam S (2016) Stability analysis of SITR model and non linear dynamics in wireless sensor network. Indian J Sci Technol 9(28). https://doi.org/10. 17485/ijst/2016/v9i28/98454 51. Srivastava AP, Awasthi S, Ojha RP et al (2016) Stability analysis of SIDR model for worm propagation in wireless sensor network. Indian J Sci Technol 9(31):1–5 52. Li G, Zhen J (2004) Global stability of an SEI epidemic model with general contact rate. Chaos Soliton Fract 23:997–1004 53. Toutonji OA, Yoo S-M, Park M (2012) Stability analysis of VEISV propagation modeling for network worm attack. Appl Math Model 36:2751–2761 54. Mishra BK, Jha N (2007) Fixed period of temporary immunity after run of anti-malicious software on computer nodes. Appl Math Comput 190:1207–1212

Adaptive Super-Twisting Sliding Mode Controller-Based PMSM Fed Four Switch Three Phase Inverter K. Balaji and R. Ashok Kumar

Abstract This paper addresses design and modeling of Adaptive Super-Twisting Sliding Mode Controller and back EMF observer-based sensorless control of PMSMfed four-switch three-phase inverter. Permanent magnet synchronous motor can affect by some component variation and external disturbances. It will help to improve the accuracy and effective operation of the motor. Single-phase short circuit fault identification and reconfiguration system for Three Phase Four Switch (TPFS) Inverter fed sensorless PMSM drive with Adaptive Super-Twisting Sliding Mode Controller (ASTSMC) is proposed. The use of reducing switches leads to reducing the switching losses. ASTSMC eliminates the chattering problem and increases the robustness motor. Sensorless technique fed Three Phase Four Switch Inverter with Adaptive Super-Twisting Sliding Mode Controller is implemented using MATLAB/Simulink. Keywords Three Phase Four Switch Inverter (TPFS) · Sensorless PMSM ASTSMC · Back EMF observer

1 Introduction The permanent magnet synchronous motor has excellent heed in many different fields like aerospace, electric ship propulsion, railway vehicle, logistics systems, tool machines and robots, battery-powered applications, and industrial automation. Because of the above reason, PMSM drives are widely employed in industrial and wind power systems [1, 2]. Closed-loop control of PMSM needs the position of rotor and rotor speed. This can be achieved by using some encoders and position sensors K. Balaji (B) · R. Ashok Kumar St. Peters Institute of Higher Education and Research, Avadi, Chennai 600054, India e-mail: [email protected] R. Ashok Kumar e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_37

327

328

K. Balaji and R. Ashok Kumar

[3]. However, the use of a position sensor increases the size, cost, wiring complexity and decreases the robustness of the system. These produce ripples in the motor speed. This leads to ruining the performance of the motor speed. This can be controlled by using the sensorless control of PMSM motor which has more surveillance [4]. Speed can be estimated based on machine parameters such as flux, voltage, current, and back EMF. The sensorless control has the following two major types, that is, flux EMF, current observer-based speeds and position control [5–7]. The observers used for sensorless speed estimation use the direct estimation method based on the back EMF of MRAS and SMO. Adaptive Super-Twisting Sliding Mode Controller replaces the Second-Order Sliding Mode Controller which provides excellent tracking of speed during the presence of variation of load torque. Fault reconfiguration system involves an additional leg with two switches and TRIAC connected in each phase of the inverter and an additional leg. Fault detected based on over current flow in any phase of the motor is due to short circuit. PMSM affected by structure variation, modeling error, and uncertainty. PMSM speed control applications require dynamic speed control with faster convergence during load disturbance. Dynamic speed controller with Adaptive Super-Twisting Sliding Mode Controller (ASTSMC) is proposed [8– 10]. It provides variable switching gain, avoids the chattering effect, and produces the output signal to be continuously chatter free. The sensorless control method of PMSM uses back EMF observer and ASTSMC to regulates Three Phase Four Switch Inverter which control the speed of the motor is proposed and implemented using MATLAB/Simulink.

2 Proposed Sensorless PMSM Model The sensorless vector controlled PMSM drive fed to the Three-Phase Four Switch Inverter with Super-Twisting Sliding Mode controller regulates the speed as shown in Fig. 1. A simplified model of PMSM is required for the implementation of advanced speed control strategies, fault diagnosis schemes, speed, and position estimation methods. Even with the presence of nonlinearity in the model, it is assumed that the developed back EMF of PMSM motor follows a sinusoidal wave shape. In general to attain faster speed response of the motor, the sliding mode controller is set to track a particular speed reference. Constant switching gain function is obtained by using a sliding mode controller while variable switching gain function is obtained by using Adaptive Super-Twisting Sliding Mode Control. This reduces the chattering problem of the motor. The power circuit includes DC source and a four-switch three-phase

Adaptive Super-Twisting Sliding Mode Controller …

329

Fault Additional Leg

TRIAC

Three-Phase Four Switch Inverter

TRIAC

PMSM

TRIAC

DC

Fault Reconfiguration

Vector control

Fault Detection Adaptive SuperTwisting Sliding mode controller

Speed estimation Sensorless Control

Fig. 1 Block diagram of fault reconfigurable sensorless PMSM drive

inverter which is used to drive the specified PMSM motor. A mathematical model of PMSM with second-order speed model in the synchronously rotating d-q frame of reference is presented in this section. id =

Rs ud i d + η p ωi q + Ld ld

i q = −η p ωi d − •

ω=

ηpψ f uq Rs iq − + Lq Lq lq

3 ∗ ηpψ f B TL iq − ω − 2∗ J J J

••

ω=U−

B K p + Ki J • Ki B ω− ω KpJ KpJ

(1) (2) (3) (4)

2.1 Back EMF Observer The main function of the observer is to estimate back EMF of PMSM motor from the predefined sensed value of terminal voltage and currents at the PMSM motor windings. The requirements to determine the position and speed of the PMSM motor for applying sensorless control technique and back EMF observer are used. It is based on a mathematical calculation of PMSM motor with a simple model, and there is no tuning process.

330

K. Balaji and R. Ashok Kumar

Fig. 2 Adaptive Super-Twisting Sliding Mode Controller

⎤⎡ ⎤⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ R 0 0 ia e i Va L−M 0 0 d ⎣ a⎦ ⎣ a⎦ ⎣ Vb ⎦ = ⎣ 0 R 0 ⎦⎣ i b ⎦⎣ 0 ⎦ L−M 0 i b + eb dt 0 0 R 0 0 L−M Vc ic ic ec ⎡

(5)

2.2 AST (Adaptive Super-Twisting) Speed Controller Design Figure 2 shows the Adaptive Super-Twisting Sliding Mode Controller (ASTSMC). The purpose of this control is to provide good performance of speed tracking during load variation, disturbances, and uncertainties. The kinetic equation of the PMSM is as written in Eq. 6. nP F d ωr = (Tem − TL ) − ωr dt J J

(6)

The rated value of torques can be expressed in Eq. 7. Tem = n p

L 2m ∗ i i sq L r sd

(7)

The controlled law is designed as follows as written in Eq. 8. i sq =

 1 −α|σ |1/2 sign (σ ) + v K

(8)

Adaptive Super-Twisting Sliding Mode Controller … Q1

Q2

331 Q3 G1

C1

G2 PMSM MOTOR

VS

C2

U1

Q4

Q5

Q6

Fig. 3 Power circuit of fault reconfigurable sensorless PMSM drive

2.3 Proposed Fault Diagnosis Scheme In the proposed fault analysis, the short-circuit fault of the Three Phase Four Switch Inverter fed PMSM drive is shown as in Fig. 3. Three Phase Four Switch Inverter (TPFS) has two parts: one is a fault-free circuit, and another one is fault redesign. Fault-free mode of operation consists of a Three Phase Four Switch Inverter fed PMSM motor. It includes a DC source, two-phase legs using four power switches, and a split capacitor leg for the third phase. The fault reconfiguration system involves an additional phase leg with three TRIACS, one for connecting additional phase leg to the motor phase during fault and the other two TRIACS for disconnecting the faulty stage from the new motor. Control implementation of the fault reconfiguration system is responsible for closing and opening of TRIAC. It involves a current sensing circuit for fault detection and fault is identified based on over current flow in a faulty phase based on a predefined threshold. TRIAC is an ac voltage control device, and the switching pulses for turning on TRIACS must synchronize with ac voltage. Phase-locked loop is employed to generate synchronized pulses to control TRIAC during pre- and post-fault mode of operations. In a pre-fault operation where the pulses to TRIAC U1 kept at active low and pulses to G1 are generated using the PLL scheme are explained. During this mode of operation, the motor is excited from the Three Phase Four Switch Inverter, and an additional leg is not used. In the post-fault operation, consider a short-circuit fault in the first phase of the machine due to power switch Q1 exhibiting over current phenomena in the faulty phase, and phase voltage reduces to zero. Hence the speed of the PMSM motor enters into an uncontrollable region and speed decreases with wide oscillations in speed response. Similarly, the torque developed by the PMSM motor also introduces oscillations, and the overall efficiency of the machine reduces. Hence after the occurrence of fault TRIAC, G1 is given a low pulse and its switching pulse is switched to U1 to include additional phase leg to excite motor and exclude faulty leg from

332

K. Balaji and R. Ashok Kumar

the new motor. Similarly switching pulses of the defective leg with power switches Q1 and Q4 are switched to additional phase legs Q3 and Q6. Now faulty phase leg with switches Q1 and Q4 becomes a redundant leg; it is not needed for running the PMSM motor.

3 Results and Discussion Figure 4 shows the overall simulation diagram of four-switch three-phase inverter based on an Adaptive Supertwist Sliding Mode Controller. The parameter of the PMSM motor drive is given in Table 1. Single switch short circuit is applied in PMSM drive during 3–3.5 s.

Fig. 4 Fault reconfiguration systems for Three Phase Four Switch Inverter sensorless PMSM

Table 1 PMSM parameter detail

S. No.

PMSM motor parameters name

Range

1

Resistance of stator (Rs)

0.2 ohms

2

Inductance of stator (Ls)

8.5 mH

3

Rotor speed

500 rpm

4

Flux linkage

0.175 V s

5

Poles

4

6

Torque

1.05 N m/A

Adaptive Super-Twisting Sliding Mode Controller …

333

Fig. 5 a Speed response of sensorless PMSM drive with fault b fault-compensated speed response of sensorless PMSM drive

Figure 5a shows speed tracking response of sensorless PMSM drive during singleswitch short-circuit fault without fault reconfiguration system using TRIAC and the speed response shows that the system is not stable during fault with the desired speed of motor reduced from 1500 to 500 rpm. Figure 5b shows the fault compensated speed performance of PMSM using TRIAC and motor speed stays at 1500 rpm. Figure 8 shows developed electromagnetic torque response of sensorless PMSM drive during a single switch short circuit fault without a fault reconfiguration system.

334

K. Balaji and R. Ashok Kumar

Fig. 6 a Torque response of sensorless PMSM drive with fault b fault-compensated torque response of sensorless PMSM drive

Figure 6a shows the R-phase voltage of sensorless PMSM drive during a singleswitch short-circuit fault without fault reconfiguration system and phase voltage is not symmetrical and zero in a positive cycle. This leads to a flow of high current into the motor. Figure 6b shows the fault compensate R-phase voltage of the PMSM drive. Figure 7a shows the stator current response of sensorless PMSM drive during single-switch short-circuit fault without fault reconfiguration system, and the current response shows that fault current is about 200% of nominal value. The current flow during the positive half cycle is not in symmetry with the negative period. Figure 7b shows the fault-compensated stator current of the PMSM drive using a TRIAC circuit (Fig. 8).

Adaptive Super-Twisting Sliding Mode Controller …

335

Fig. 7 a R-phase voltage of the sensorless PMSM drive b fault-compensated phase voltage of sensorless PMSM drive

4 Summary Single-phase short-circuit fault detection technique and reconfiguration for Three Phase Four Switch Inverter fed sensorless PMSM drive are developed. This system is a fast response to the fault, more reliable, and mitigates the short circuit problem. Speed control problem using the Second-Order SMC is addressed, and Adaptive Super-Twisting Sliding Mode is presented for tracking the speed of the PMSM motor drive. It has a dynamic speed control response of the system with time-varying parameter and is nonlinear. ASTSMC utilizes less rise time and faster convergence. Using MATLAB/Simulink, the output of the three phase four-switch inverter is verified.

336

K. Balaji and R. Ashok Kumar

Fig. 8 a Current response of sensorless PMSM drive with fault b fault-compensated current response of sensorless PMSM drive

References 1. Barkat S, Tlemçani A, Nouri H (2011) Non-interacting adaptive control of PMSM using interval type-2 fuzzy logic systems. IEEE Trans Fuzzy Syst 19(5):925–936. https://doi.org/10.1109/ TFUZZ.2011.2152815 2. Vidhyalakshmi D, Balaji K (2018) Performance of bidirectional converter based On grid application. Indones J Electr Eng Comput Sci 12(3):1203–1210 3. Chen L, Götting G, Dietrich S, Hahn I (2016) Self-sensing control of permanent-magnet synchronous machines with multiple saliencies using pulse-voltage-injection. IEEE Trans Ind Appl 52(4):3480–3491. https://doi.org/10.1109/TIA.2016.2557299 4. Kim H, Son J, Lee J (2011) A high-speed sliding-mode observer for the sensorless speed control of a PMSM. IEEE Trans Industr Electron 58(9):4069–4077. https://doi.org/10.1109/TIE.2010. 2098357 5. Vasudevan V, Balaji, K (2018) Performance of Cuk-KY converter fed multilevel inverter for hybrid sources. Indones J Electr Eng Comput Sci 10(2):436– 445. https://pdfs.semanticscholar.org/8b33/de600503618acc4300d5fb5cc8fee7772237.pdf% 3f_ga%3d2.143331196.898165924.1558967032-54138331.1558691296 6. Kim TS, Park BG, Lee DM, Rysu JS, Hyun DS (2008) A new approach to sensorless control method for brushless DC motors. Int J Control Autom Syst 6(4):477–487. https://doi.org/10. 1049/iet-epa.2008.0168

Adaptive Super-Twisting Sliding Mode Controller …

337

7. Thiruvadi SP, Bhuvaneshwari S (2017) Hybrid PV-wind-battery based system for household applications using DC-DC converter. Int J MC Square Sci Res 9(1):54–65. http://www. jpe.or.kr/archives/view_articles.asp%3fseq%3d657 8. Galma G, Pattanaik B (2018) Current fed switched inverter using sliding mode controller (SMC) for grid application. Int J MC Square Sci Res 10(4):1–10. https://doi.org/10.1109/TAC. 2015.2450571 9. Chalanga A, Kamal S, Fridma LM, Bandyopadhyay B, Moreno JA (2016) Implementation of super-twisting control: super-twisting and higher order sliding-mode observer-based approaches. IEEE Trans Industr Electron 63(6):3677–3685. https://doi.org/10.1109/TIE.2016. 2523913 10. Manie N, Pattanaik B (2019) Zeta DC-DC converter based on MPPT technique for BLDC application. Int J MC Square Sci Res 11(2):1–12

Design of Multiplier and Accumulator Unit for Low Power Applications J. Balamurugan and M. Gnanasekaran

Abstract In recent growing of portable and multimedia devices, such as notebooks, and video phones, motivated the researchers to design low power VLSI circuits. Multiplier Accumulator Unit (MAC) is a major element of DSP. The speed of systems is based on the speed of MAC unit. This paper demonstrates the hardware-efficient MAC module with the help of using Vedic multiplier. This paper compares the proposed MAC module designs with the conventional MAC unit utilizing Vedic multiplier. The proposed scheme shows good performance in terms of low power dissipation MAC unit and analyzed using Tanner EDA tool. Keywords MAC · Vedic multiplier · Tanner EDA · Low-power dissipation

1 Introduction As the size of combination continues developing increasingly more advanced signal processing systems are being executed on a VLSI chip. These processing signal applications request incredible calculation limit as well as expend extensive measure of energy. While chip area stay to be two noteworthy structure objectives, power utilization has turned into a basic concern in the recent system design. The need of low power VLSI systems emerges from two fundamental factors, namely, the processing limit per chip and operating frequency are the first main factor. Second, battery life in compact electronic gadgets is constrained. Design of low power leads to stretched operation time in these compact gadgets. In most signals processing calculations, multiplications are the major operation. Multipliers have lengthy latency; consume considerable power and large area. In less power VLSI system structure the low power multiplier plays an essential part. Low power multiplier has broad work, namely, logical, circuit, and physical levels. These techniques of low level are not J. Balamurugan (B) Department of ECE, Vedhantha Institute of Technology, Villupuram, Tamil Nadu, India e-mail: [email protected] M. Gnanasekaran Sri Venkateshwara Polytechnic College, Vellore, Tamil Nadu, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_38

339

340

J. Balamurugan and M. Gnanasekaran

unique to modules of multiplier and they are not considered well for arithmetic calculation for the multipliers. These low-level methods are not one of a kind to multiplier modules and they are commonly pertinent to different sorts of modules. The qualities of number juggling calculation in multipliers are not viewed as well. In all the DSP the digital multipliers are the center components as well as the speed of the signal processor is generally depend on multipliers speed. In different DSP applications, the MAC function is generally used. Usually, MAC includes the multiplier and accumulator which store the products of the preceding multiplication. The performance of the system majorly depends on the time required to compute the instruction. The system performances are enhanced by the multiplication. In DSP, communication and microprocessors applications multipliers are utilized broadly. To execute the partial product addition a large number of adders are used for higher order multiplication. The demand for elevated speed and low power multiplier is raising due to the requirement of high-speed processors. Based on the 16 Vedic sutras the Vedic multiplication is dependent, which are really word formulae depicting normal methods for fathoming an entire scope of numerical issues. Vedic method is used to enhance the calculation speed of the processors which involve less hardware as well as very fast.

2 Related Works Verma [1] presented 4 × 4 bit Vedic multiplier high speed based on the crosswise and vertical techniques on the basis of one step all partial products and their sum. In terms of speed, this method offers better performances. Yuvaraj et al. [2] proposed the merits of the sampoornam sutra which is exactly named as absolute vedic multiplier which is used to design the particular logic unit that decides which multiplier is optimum results based on the types of input by enhancing the efficiency. For designing a 4-bit MAC and is extended up to 64-bit with the help of Vedic scaling scheme the Sampoornam is utilized. Katkar et al. [3] proposed MAC unit which minimize the area by lowering the number of addition and multiplication in the multiplier unit. Operation speed is increased with the help of Vedic multiplier unit. Antony et al. [4] explained the high-speed multiplication principles provided by Vedic Mathematics. From this, it is extended to a Vedic multiplier with high speed utilizing multiplexers on the basis of adder was presented. Ramesh and Rajan [5] a new radio frequency energy harvesting techniques for low power application is described. Khare et al. [6] proposed that multiplier for designing MAC unit. Various arithmetic operations are performed by the MAC unit at high speed. To design all submodules using combinational form are done in the MAC unit. In final unit, this is integrated to have better control on the circuitry clock, reset functional is provided in the final unit. Hanumantha Rao and Kumar Charlie Paul [7] optimized complex multiplier is described which provides less hardware complexity. Nivedha and

Design of Multiplier and Accumulator Unit …

341

Priyadharshini [8] vedic multiplier with carry select adder, ripple carry adder and carry save adder is described.

2.1 MAC Unit The inputs of MAC unit are obtained from the memory cell and given to the MAC multiplier block, which performs the multiplication and the results are given to the adder and the results are stored into the memory location. This entire process is to be achieved in a single clock cycle. This MAC unit design is not only reducing the standby power Consumption but also can enhance the MAC unit speed so as to attain improved system performance. MAC function is represented by Eq. (1), respectively, as follows: F=



pi Qi

(1)

The basic MAC unit includes accumulator, multiplier, and adder. The basic MAC unit block diagram is given in Fig. 1. A normal n-bit MAC unit includes 2n-bit accumulator, n-bit multiplier and 2n-bit adder. By replacing the multiplier unit various MAC unit models can be developed in different architectures. The basic MAC unit includes accumulator, multiplier, and adder. The basic MAC unit block diagram is given in Fig. 1. A normal n-bit MAC unit includes 2n-bit accumulator, n-bit multiplier and 2n-bit adder. By replacing the multiplier unit various MAC unit models can be developed in different architectures. A novel hardware-efficient scheme as well as MAC structures was designed and shown in Fig. 2. In this new MAC unit shifter, adder register and multiplier were used which improves the performance of the DSP system the high speed with less power dissipation. Compared to the conventional MAC unit the proposed method uses the barrel shifter for high performances. To perform the shifting operations the n number of data inputs, set of control inputs, and data outputs are used in the barrel shifters and it is a combinational logic and shown in Fig. 1. Barrel shifter is designed using Multiplexers. Barrel shifter design is for natural size like (2, 4, 16). Basically,

Multiplicand

Multiplier Register

Multiplier

Fig. 1 Conventional MAC unit

Add/Subractor Unit

Accumulator

O/p

342

J. Balamurugan and M. Gnanasekaran clk

Multiplicand

Multiplier register

Barrel Shifter

Adder Unit

Register

O/p

Multiplier

Fig. 2 Proposed MAC units

barrel shifter is used with logical left shift operation which is controlled by select inputs. Barrel shifter has the great advantage of high speed. Figure 3 shows the 4-bit barrel shifter using 2:1 MUX. Here Do, D1, D2, D3 are the inputs of the multiplexers and S0, S1 are the selection pins and SRL are the selection pins. If left shift the selection pin acts as 1 and if right shift the selection pin acts as 0. And Y0, Y1, Y2, Y3 are the output pins.

Fig. 3 4-bit barrel shifter using 2:1 MUX

Design of Multiplier and Accumulator Unit … Fig. 4 Structure of Vedic multiplier

x y

343

Cross products

Adders (RCA)

Fig. 5 2 × 2 bit Vedic multiplier

3 Vedic Multiplier In digital hardware, the two most common multiplications are booth multiplication as well as array multiplication algorithm. The array multiplication has less computation time due to the execution of partial products in parallel manner. With the help of array multiplier, the delay is associated with time taken by the signal to propagate through the gate which models the array multiplication. In booth multiplication, for highspeed multiplication large booth arrays are essential, also partial carry register and large partial sum are require for the exponential operations. To generate the least significant final half product multiplication of two n-bit operands utilizing booth recording multiplier involve about n/(2m) clock cycles. Where m is the number of stages in the booth recorder adder. Therefore, it associated with large propagation delay. To overcome this problem the Vedic multiplier is used. Using the sutra the partial products are attained by crosswise and vertical functions. Here the adder delay is similar to the normal delay. In the cross products, the maximum number of bits is added with the critical path adders. Vedic multiplier with the width N × N will produce the 2N−1 cross products of different widths which is the combined forms of (log2N + 1) partial products. Basic structure of Vedic multiplier is shown in Fig. 4. Block diagram of 2 × 2 bit Vedic multiplier architecture was shown in Fig. 5. Here two half adders were used and Fig. 6 shows the block diagram of the 4 × 4 bit Vedic multiplier. Figure 7 shows the 4 bit Ripple carry adder.

344

J. Balamurugan and M. Gnanasekaran

Fig. 6 Block diagram of 4 × 4 bit Vedic multiplier

4 Simulation Results In this paper, the projected MAC unit using 4-bit VM is designed using tanner EDA tool. The design of conventional MAC unit was designed using VM which is shown in Fig. 8 and the proposed design is illustrated in Fig. 9 and simulation results are shown in Fig. 10 for proposed scheme.

Design of Multiplier and Accumulator Unit …

345

Fig. 7 4-bit Ripple carry adder

Figure 10 shows the R-phase voltage of sensorless PMSM drive during single switch short circuit fault without fault reconfiguration system is shown in figure and phase voltage is not symmetrical and zero in a positive cycle. This leads to a flow of high current into the motor. Figure 11 shows the fault compensate R-phase voltage of PMSM drive.

Method

Time (s)

Average power

Conventional method [2]

5.15

8.512e-005

Proposed method

3.04

6.7851e-006

346

J. Balamurugan and M. Gnanasekaran

Fig. 8 Conventional MAC unit using VM

From Fig. 11 the performance analysis of the proposed and conventional method is shown. From this, it clearly shows that the proposed scheme is found to be efficient.

5 Summary A new method of MAC-based Vedic Mathematics was presented in the paper. The proposed 4 × 4 bit Vedic multiplier is realized in tanner EDA tool. Vedic multiplier time path delay is found to be 3.04 s. When compared to the traditional method, the proposed MAC unit using 4 × 4 bit Vedic multiplier is found to be tremendous speed with good performances. For many applications minimizing the time delay is very essential concern for this necessity the proposed scheme is much suitable.

Design of Multiplier and Accumulator Unit …

Fig. 9 Proposed MAC unit using VM

Fig. 10 Simulation analysis of proposed MAC unit using VM

347

348

J. Balamurugan and M. Gnanasekaran

Fig. 11 Performance analysis of conventional and proposed method

References 1. Verma P (2012) Design of 4 × 4 bit vedic multiplier using EDA tool. Int J Comput Appl 48(20):32–35 2. Yuvaraj M, Kailath BJ, Bhaskhar N (2017) Design of optimized MAC unit using integrated vedic multiplier. In: International conference on microelectronic devices, circuits and systems (ICMDCS) (pp 1–6). IEEE, pp 3677–3685. https://doi.org/10.1109/icmdcs.2017.8211704 3. Katkar SC, Kene P, Ugale S (2015) Design of efficient 64 bit mac unit using vedic multiplier for DSP application-a review. ISSN 2348-9928 IJAICT, 1(9) 4. Antony SM, Prasanthi SSR, Indu S, Pandey R (2015) Design of high speed Vedic multiplier using multiplexer based adder. In: 2015 International conference on control communication & computing India (ICCC). IEEE, pp 448–453. https://doi.org/10.1109/iccc.2015.7432938 5. Ramesh GP, Rajan A (2013) RF energy harvesting systems for low power applications. Int J Technol Eng Sci 1085–1091 6. Khare N, Rao D, Mohan R (2016) VLSI implementation of high speed MAC unit using Karatsuba multiplication technique. J Netw Commun Emerg Technol (JNCET) 6(1):2249–6149. https:// doi.org/10.9790/4200-05131724 7. Hanumantha Rao K, Dr. Kumar Charlie Paul C (2017) Design of area efficient R2MDC FFT using optimized complex multiplier. Int J MC Square Sci Res 9(2) 8. Nivedha M, Priyadharshini V (2017) Design and implementation of high speed and area efficient MAC unit. Int J MC Square Sci Res 9(1):117–129

Design and Implementation of IoT-Based Wireless Sensors for Ecological Monitoring System G. Santhosh, Basava Dhanne and G. Upender

Abstract The climate change topic has attributed a lot of significance in the recent past in ecological observation, making the analysis dynamically incredible. This field is proclaimed on remote sensing and wireless sensing element networks for collecting knowledge regarding the setting. The research trends in the area of the cloud computing system model, the Internet of Things (IoT), and the cyber-physical model help in aiding for the broadcast and managing of huge amount of knowledge concerning the patterns discovered from the environmental parameters. In this research, environment completely monitoring by IoT-based wireless sensors. The framework is based on two methodologies, the first being User Datagram Protocol (UDP)-based Wi-Fi communication, and the other functions through Wi-Fi as well as machine-readable text transfer protocol (HTTP). Both systems offer the likelihood of transcription of information at distant location and envisioning them from each device involving an online association, sanctioning the observation of geographically massive areas. Keywords Internet of Things (IoT) · HTTP · UDP · Wi-Fi module

1 Introduction The Internet of Things (IoT) is an interconnection of entities, wherein the components of lifestyle region unit are embedded with microcontrollers, sensors. These entities are packaged in a manner, permitting those to accumulate and share information with every different customer, converting into the crucial part of the Internet. The IoT version targets at developing the Internet even extra continual. Moreover, by sanctioning smooth accessibility as well as conversation with a variety of gadgets, G. Santhosh (B) · B. Dhanne · G. Upender Department of ECE, CMR Engineering College, Hyderabad 501401, Telangana, India e-mail: [email protected] B. Dhanne e-mail: [email protected] G. Upender e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_39

349

350

G. Santhosh et al.

including home appliances, police work cameras, watching sensors, etc., the IoT is implemented for the applications that make use of the large quantity of data generated by such objects providing services. This technique finds application in numerous areas, like mobile health care, home automation, traffic administration, and lot others [1].

2 Advantages of IoT ThingSpeak 1. 2. 3. 4.

Access information Communication Cost Effective Privacy and Security.

ThingSpeak is an IoT analytics system solution that enables one to accumulate, envisage, and evaluate online information streams in the cloud. The cloud makes use of the procedures of graphical visualization, that are readily available through digital web server for the customers and the things that relate to the cloud through feasible “cordless net links”. These cordless net links are offered to the individuals as well as the bulk items utilize the sensing units to inform the ecological analogue information. In the proposed system, the ecological criteria could straight be accessed by the customer, thereby getting rid of the requirement of third parties [2]. It is of paramount importance that the long-time occupants of the buildings are provided with hygienic environmental environments for avoiding health-related issues [3]. It is also essential to control the greenhouse gas emissions [4] for environmental sustainability and clean ambience. The device supports up to 4 GB of memory card, [7] to store the audio and image files required. It is capable of using VS1011e, aDSPbased audio codec chip to generate high quality MP3 audio. The ultimate goal of the project is to [8] quickly spread the information about the disaster warning through the internet and make it available to those who need it as soon as possible. The gas booking/order is done with the help IOT and a load cell [9] that is interfaced with a microcontroller is done using the continuous weight calculation.

3 Over All Implementation of the Project Explanation The sensors communicate by Wi-Fi technology employs the similar hardware for usage of both the UDP and HTTP protocols. However, different protocols, UDP, or HTTP may be used as per the suitability of implementation. The general architecture of the devices rooted in Wi-Fi technology is presented in Fig. 1. This work is primarily aimed at implementing a sensor system for detecting multiple phenomena. In this Temperature Sensor, Raspberry-Pi, Humidity Sensors, and

Design and Implementation of IoT-Based Wireless Sensors …

351

Fig. 1 Block diagram

Gas Sensor are used. The power supply is given to the board through the rectifier circuit, which supplies regulated voltage of 5 V to the Raspberry-Pi Model. 40 generalpurpose input–output (GPIO) pins are available in Raspberry-Pi. Devices interfaced with Raspberry-Pi are connected through these pins. Gas Sensor, Temperature Sensor, and Humidity Sensors are connected to the different pins of Raspberry-Pi. When the temperature increases above the adjusted limit, the temperature sensor senses the access temperature and reports to the system through Raspberry-Pi. When the poisonous gases are discharged in the environment, the sensor detects the poisonous gas presence and alarms the buzzer. Similarly, on incidence of fire, the temperature sensor detects it and provides the information to the Raspberry-Pi controller to alarm the Buzzer. Temperature sensor is used to detect the variations in the environment temperature and inform the Raspberry-Pi. This information is used to determine the temperature of solids, liquids, and gases. Humidity refers to the amount of moisture present in the surrounding air. If the humidity is high for a certain environment, then the water content of air for that environment is also high. The block diagram shows the present work for dissimilar IoT-Based wireless sensors used [6] for ecological as well as ambient monitoring, through User Datagram Protocol and Hypertext Transfer Protocol. UDP facilitates the low-power process of the Wi-Fi sensors, as it is connectionless in the environment [3]. The UDP protocol features with increased speeds, lower packet sizes as well as lower latency. The sensors are connected to a WLAN, that

352

G. Santhosh et al.

sends measured data all the way through HTTP requests. HTTP application protocol increases the reliability of data transmission.

4 Flow Chart See Figs. 2, 3 and 4. There are four sensors used for implementation of the system as shown in Fig. 2. The gas sensor detects the gas leakage if the gas content in the atmosphere exceeds certain permissible level. When gas leaks in the room, Raspberry-Pi sends the message to the user using Wi-Fi module and activates the buzzer. In case of fire accidents, the temperature sensor detects and provides the information to the Raspberry-Pi Controller, which alarms the buzzer, and switches the exhaust fan (Fig. 5). Fig. 2 Flow chart of gas sensor

Design and Implementation of IoT-Based Wireless Sensors … Fig. 3 Flow chart of fire sensor

Fig. 4 Flow chart of humidity sensor

353

354

G. Santhosh et al.

Fig. 5 Flow chart of temp sensor

5 Design Methodology 5.1 Raspberry-PI Raspberry-Pi3 Model B is a tiny single-board minicomputer. It is a 1.2 GHz QuadCore Cortex-A53 Broadcom BCM2837 64-bit ARMv8 processor. 40 pins for general purpose input and output as well as 1 GB RAM are available in Raspberry-pi. Raspberry-Pi is called as Visa movement solitary board COMPUTER, because of the level of the board. The Raspberry-Pi could uncover the framework besides the code vernacular.

5.2 Liquid-Crystal Display Liquid-crystal display makes utilize of beam modulation properties of liquid crystals for displaying alphanumeric characters and images. It is a level panel electronic visual display unit. The liquid crystals do not emit light straight, rather utilize the light modulation techniques. Liquid-crystal displays are popularly known as LCDs and are available for display of arbitrary images or fixed images.

Design and Implementation of IoT-Based Wireless Sensors …

355

5.3 RS232 Standards To permit similarity among substances, gadgets are made through endless producers. An interfacing trendy alluded to as RS232 wound up being set up through the Electronics Industries Association (EIA) in 1960. RS232 is popularly used as the serial I/O interface.

6 Results and Discussion An intelligent universal remote control for environmental monitoring appliances circuit, connected with respected block diagram is shown in Fig. 1. Figures 6, 7, 8, and 9 are the outputs of the project. Figure 6 is the overall hardware kit, where all the components are connected. Here LCD is used to display the sensor levels and motor ON-OFF conditions. At first the mobile application (Telnet) is opened and the IP address is set. The IP address is taken from the Wi-Fi module, which is unique, after taking the sensors inputs and outputs on the mobile application. Temperature sensor is used to detect the variations in temperature the environment monitoring. It is used to determine the temperature of solids, liquids, and gases. If the temperature increases, then the humidity sensor is ON, by that temperature is minimized which is shown in Fig. 7. When Fire accidents take place fire sensor detects the incidence of fire. Fire sensor can sound an alarm to alert where the fire accident has taken place, which is shown in Fig. 8. When the poisonous gases spread around environment the gas sensor detects it, and alarms the buzzer, which is shown in Fig. 9. Figure 10 TELNET mobile application is used for giving display conditions of the sensors data. Fig. 6 Working model

356

G. Santhosh et al.

Fig. 7 High temperature and humidity get on position

7 Conclusion In this, it is possible to check weather condition criteria like daytime, rains, fire, and gas leak and so on. The information could be kept online, which could be utilized to anticipate weather conditions and environment patterns. For the very first time, information of various kinds as well as locations could be combined with each other as well as accessed from wherever. Several considerable progression has been transformed in the last couple of years in order to connect the void among the academic advancements as well as actual releases.

Design and Implementation of IoT-Based Wireless Sensors … Fig. 8 Fire sensor detected time

357

358

G. Santhosh et al.

Fig. 9 Smoke detected time

Fig. 10 Output results by using Telnet app

References 1. Bellavista P, Cardone G, Corradi A, Foschini L (2013) Convergence of MANET and WSN in IoT urban scenarios. IEEE Sens J 13(10):3558–3567. https://doi.org/10.1109/JSEN.2013.2272099 2. Mois G, Folea S, Sanislav T (2017) Analysis of three IoT-based wireless sensors for environmental monitoring. IEEE Trans Instrum Well Meas 66(8):2056–2064. https://doi.org/10.1109/ TIM.2017.2677619 3. Zhang L, Tian F (2014) Performance study of multilayer perceptrons in a low-cost electronic nose. IEEE Trans Instrum Meas 63(7):1670–1679. https://doi.org/10.1109/TIM.2014.2298691

Design and Implementation of IoT-Based Wireless Sensors …

359

4. Kumar A, Kim H, Hancke GP (2013) Environmental monitoring systems: a review. IEEE Sens J 13(4):1329–1339. https://doi.org/10.1109/JSEN.2012.2233469 5. de Donno D, Catarinucci L, Tarricone L (2014) RAMSES: RFID augmented module for smart environmental sensing. IEEE Trans Instrum Meas 63(7):1701–1708. https://doi.org/10.1109/ tim.2014.2298692 6. Ramesh GP, Kumar NM (2019) Design of RZF antenna for ECG monitoring using IoT. Multimed Tools Appl 1–6 7. Khan A, Prakash G (2017) Design and implementation of smart glass with voice detection capability to help visually impaired people. Int J MC Square Sci Res 9(3):53–59 8. Amjath Ali J, Thangalakshmi B, Vincy Beaulah A (2017) IoT based disaster detection and early warning device. Int J MC Square Sci Res 9(3):20–25 9. Ravichandran S (2017) Cloud connected smart gas cylinder platform senses LPG gas leakage using IOT application. Int J MC Square Sci Res 9(1):324–330

Enhancing Security in Smart Homes-A Review Bhuvana Janita, R. Jagadeesh Kannan and N. Kumaratharan

Abstract Internet of Things (IoT) is growing rapidly as the number of users using smart “things” for communication has increased. The Internet of Things is the interconnection of individually recognizable smart embedded computing gadgets inside the previous web or Internet foundation. Applications find their way in Health Care, Building Smart cities, Smart Vehicles, Agriculture, and Smart Home. There are huge profits of smart home system like being able to lock/unlock homes from a long distance, get notified in case of smoke/fire in the house and automating gadget maintenance. It is especially beneficial to assist elderly people who are sick and prefer to stay home rather than being in a Hospital. Despite the benefits, these devices bring in lot of security and privacy issues. IoT devices have firmware with less computing power and battery. They mostly do not have antivirus software installed. These devices communicate seemingly large amount of data and they are most vulnerable for security attacks like Denial of Service (DOS), Malware attacks and Ransomware attacks. A lot of research work is taking place to address the security issues in IoT Platform. This paper aims to provide a detailed review of the Methods or Techniques that are proposed to secure smart home systems by providing better design, access control techniques, encryption algorithms, and secure software applications. Keywords Security in IoT · Smart home security · Home automation

B. Janita (B) · R. J. Kannan Vellore Institute of Technology, Chennai, India e-mail: [email protected] R. J. Kannan e-mail: [email protected] N. Kumaratharan Sri Venkateswara College of Engineering, Sriperubathur, India © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_40

361

362

B. Janita et al.

1 IoT Home Security Applications and Architecture 1.1 Home Security Applications A smart home framework usually includes three elements, [1] home server, [2] tools in the smart home, and [3] home gateway. The space for the storage is provided by home server. The integration and sharing task of the data gathered from a variety of media in the house. Next, a wired/wireless home network is interconnected with the Internet using home gateway. Smart home tools collect data through different sensors and exchange data between the devices and the gateway devices. Michael Schiefer [4] has categorized smart home devices in Table 1. Internet of Things Home automation solution automates the home environment and saves energy. The user can control the lighting, hue according to the environment. The electric and water meter reading will be extremely useful for utilization of resources. The wear and tear of the devices can also be monitored which makes its maintenance easy. A smart home solution often is very easy to install, and the retailer provides two elements: one is software development kit (SDK) and the other one is a framework for management. SDK contains general functions and protocols that are executed to help various IoT structural designs (e.g., Advanced RISC Machine (ARM), Peripheral Interface Controller (PIC), etc.). Framework of Management contains an online-based platform or gateway. And these two elements are employed Table 1 Smart home devices Safety systems (SafS)

Detecting gas/water leakages, and smoke detectors

Sensors and measurements and (SaM)

Reading of water meter and electric meter

Heating, ventilation, and airconditioning (HVAC)

To manage ventilation of air and the temperature of room related to ventilator or thermostats

Light and shadow (LaS)

A device that generates or stops the light, for instance, lamps

Devices in kitchen (DK)

This type contains the products destined to be employed in a kitchen like coffee maker, electric cooker, and refrigerator

Water scheme (WS)

This class contains the products like bath tub, Tap, toilet with grass sprinkler, etc.…

Cleaning scheme (CleS)

These classes contain the systems to cleaning, like dishwasher, washer, and vacuum cleaner

e-Pet systems (ePSy)

This type contains all tools about pets or animals. It can be a locating collar, a system for automated feeding, etc.

Agility devices (AgiD)

This type contains the devices for shifting individuals, like bicycle, bikes, cars, etc.…

Enhancing Security in Smart Homes-A Review

363

to connect various tools in a smart home system. The SDK is not designed particularly for one yet for various tools [2] and there is no security framework for all devices. Smart home without proper security mechanisms can be the target of Hackers whose motivation would be intimidating housemates by tracking their movements, to get overall control of the house, to steal sensitive user data or use smart home devices as mediating devices to plan attack on some other area. There is a reported case in 2018 [5] where a woman’s baby monitor was hacked. The Hue light bulb from Philips was hacked, and the Hacker used the device’s bridge to issue blackout commands to light bulb. When the users realize that their movements are being watched over, it sends panic signals to them. The August Smart Lock also is a victim for security attacks [6]. Mengmei Ye et al. identified four types of attacks in Smart Lock (1) Leakage assault of Handshake key, (2) Leakage attack of Owner account, (3) Leakage assault of the Personal Information, and (4) Service assault denial [7–9]. In home environment there is no trained proprietor present and the majority of the customers are not familiar adequate to ensure emergency conditions. Therefore, the common necessity is that smart tools must try hard for auto-configuration anywhere possible [10]. Vulnerabilities of smart home devices: In order to understand the vulnerabilities exposed by the smart home devices, Sukhvir et al. conducted an experiment highlighting the communication pattern of the devices and have tried to analyze the packets using Wireshark [11]. The experiment that was conducted on few smart home tools such as Hue light bulb from Phillips, WEMO power switch from the Belkin (Direct communication as in Fig. 1), and the Nest smoke-alarm (Fig. 1) has showcased the simplicity which protection and confidentiality can be compromised in these devices. (i)

The Nest smoke-alarm is available with a smoke sensor with photoelectric, a speaker, a carbon monoxide sensor, and four other sensors that senses the heat, light, and movement. The Authors have concluded that Nest is quite a protected item. In the smoke-alarm, all the communication is encoded. Protection is additionally enhanced. It can be done by the service of Single Sign-On by means of OAuth2. It is very tough for someone to spy and gather data about a client from the communication between the servers and nest sensor. The user’s activity is monitored and is logged on to server which raises questions related to privacy concerns. (ii) The Hue connected bulb of Philips allows the customers to control the wireless lighting support in the home. The client can be capable of regulating the intensity, color customization, combination of colors, and plan via Android and iOS App of Philips Hue. It is also very customizable and it can be configured to alter the bulb state based on other platform actions like facebook by using the service called If This Then That (IFTTT). When the app is launched Username is generated according to the android app Mobile Phone’s MAC address. But still the communications are in plain text which contains the light status (brightness/color/hue/alarm status, etc.) which gives the attacker a clue of the inside activities of the home.

364

B. Janita et al.

Fig. 1 Communication pattern between IoT devices

(iii) The WEMO switch is a power socket and notifies the user about the devices that is connected to it. It has motion sensor. All the communication between the app and device is in plain text and both the app and the device have to be in the same LAN. It does not have mechanisms to authenticate a legitimate user. According to paper [11] the operational models of communication prototype between IoT Devices, User, and Cloud-depended Server maintained by the producer is as follows: (a) Direct: IoT device directly communicates with users via the mobile app and updates the server about the current status. (b) Transit: The devices communicate with the server using the user’s mobile app via Bluetooth or NFC. (c) In-direct: The communication between the user and the IoT device is only through the external server. In all the models, we can see that the smart home solution suppliers initiate a cloud computing platform for gathering and analyzing user data. (i) This extra remote server raises privacy concerns of user data. (ii) The data in the cloud server is often maintained by third-party vendors and the mechanisms used to protect the data is not transparent. (iii) In Direct Model every event is logged in server as well as the user gets notification which is duplicating data and wastes bandwidth.

Enhancing Security in Smart Homes-A Review

365

The authors of the paper have proposed a solution at the network level. They have introduced an external SaaS provider with access control rules. In order to authenticate a legitimate user the request is checked against the SaaS provider database. In October 2014, meters in Spain were attacked to provide fraudulent bills for financial gains. Motivated by such incidents Chia et al. in their paper [12] have comprehensively provided data provenance related security attacks. Their proposed solution aims in providing Source Identity Authenticity, Source Data Authenticity, Data Integrity, Data consistency, and Location Authenticity. A magnetic sensor was integrated along with the smart meter and by examining the data sent by both the devices (v1 and v2, respectively) simultaneously, the authors have tried to achieve Source Data Authenticity using cross-check mechanism. They concluded that if v1 = v2, data is not hampered or if v1 = v2 either the sensor or the smart plug has been tampered. To identify the device that is sending the incorrect data, another component called location generator was introduced in the network. All the three components—Magnetic Sensor, Smart plug, and location generator—shared a single private key for their transaction. University of Birmingham taught its students to do Penetration testing on commercial IoT devices and asked them to expose the vulnerabilities of the devices [13]. The students not only found out known vulnerabilities but also exposed critical vulnerabilities of the devices (Table 2).

2 Security Techniques Identified for Smart Home Devices 2.1 Security by Design Smart Home devices are embedded systems with limited battery and memory power. The hardware does not support running sophisticated operating systems or antivirus and these devices have firmware with a limited set of operations. When there is a software bug or security leak, the firmware cannot be updated like mobile phones or computers. Therefore, it is crucial to apply security features to the devices or network at the design level itself. Schneier [14] in his article states that “security” does not have meaning except we know things similar to “Se-cure from whom?” or “Secure for how long?”. Threat Modelling is necessary for designing Home automation systems as it helps to identify potential risks when implemented and helps in designing the Home automation environment avoiding them. A threat model was proposed by Dominik et al. [15] using attack trees. Attack trees are formal, logical methods of representing security assaults in a tree like organization with the aim as the root hub and different approaches of achieving that aim as leaf hubs [14]. The structure is a house or building with various actuators and sensors, management units, smart tools, and interface of management. This representation mainly focuses on the smart home structural components and not on the details of implementation. Graphs and Attack trees-based

366

B. Janita et al.

Table 2 Vulnerabilities of commercial Iot Devices Setup Aria scale (fit bit) Wireless cloud camera iKettle 1.0 (smarter) Indoor video camera Home security system Jumping Sumo (Parrot) Coffee machine (smarter) Wi-Fi doorbell

App cloud

√ √

Device cloud √

Device app

√

√

√

Control

Leak

Password

× √

√

√

√ √

√

√

√

√

√

√

√

√

√

√

models face scalability issue and Mengmeng Ge in his manuscript has addressed this issue through his Extended HARM (Hierarchical Attack Representation Model) model for Security Analysis of IoT systems. The protection evaluator utilizes a variety of protection metrics to evaluate the protection and interrelates with a tool of assessment and logical modeling, Symbolic Hierarchical Automated Reliability and Performance Evaluator (SHARPE) [16] which accepts an arithmetical model of the framework and investigates it utilizing a variety of methods (Fig. 2).

2.2 Authentication and Authorization Schemes in the Internet of Things Authentication and Authorization of potential users in a distributed environment like the Internet of Things is a challenge in itself. Intelligent Virtual Agents (IVA) can be used for Home automation and Amazon have listed lot of skill APIs in its developer portal to connect devices to Alexa Echo Dot. Chung et al. have demonstrated how IVA-enabled devices can be hacked even though the communication is encrypted [17]. Alexa employed with several devices of smart home from brands including Philips Hue, Leviton, Eco bee, and Ring. The majority of communication among

Enhancing Security in Smart Homes-A Review

367

Fig. 2 SHARPE tool-security evaluator model

IVA is permitted in devices and the IVA is fixed by using HTTPS. On the other hand, a selection of methods in machine learning to classify the traffic of network and can reveal the dimensions of payload, other patterns, and data rates in encoded traffic that might be employed to distinguish the status of device’s or the behavior of customer’s, for instance, listening to music, turning on or off of the machine, talking to the ally, and items ordering. Role-based access control (RBAC) is proposed by Jing Liu et al. to authenticate legitimate users. The users of the “Things” or end nodes should register with its nearby Gateway device (known as Registry Agent—RA). RA verifies User ID and with the Role-based access control policy rules checks if the user is the legitimate user. Users get access to resources based on roles. Roles can be set of actions or responsibilities assigned against a member. Analysis of IoT systems using the above approach has shown that it can prevent assaults like eavesdropping, the man-in-the-middle assault, and replay assaults. The drawbacks in the permission models of IoT Home automation platforms like Weave/Brillo from Google, SmartThings from Samsung, and HomeKit from Apple [3], is that there are no meaningful interactions between the user and the apps that are designed for User Control. Other than initial installation and setup, only notifications or alerts are sent to the user when a permission is requested. Context-based permission control is essential in the IoT environment which has minimum user interaction [18]. Most of the sensitive authorizations requests occur when the client is not connected with the application. Current designs request permission to the user for the first time and the further correspondence does not request permission aiming to minimize user interaction [19]. A perspective-based permission structure for the platforms of IoT that helps effective recognition of perspective for responsive action and attempts in the runtime with loaded perspective data is proposed by Yunhan et al. to help provide contextual integrity [19]. In their work they applied patch on existing SmartApp of Samsung SmartThings to context-based SmartApp. The context information is collected first based on which the permission is allowed or denied. The action is triggered based

368

B. Janita et al.

on this permission. Decisions made are stored in Context IoT cloud background. The effectiveness of this approach is checked against IoT dataset with 25 smart applications each representing unique attack class and execution path. A security protocol and database design for IoT systems are proposed by Trio Adiono et al. The smart home system of the author had two sub-systems outdoor and indoor. The outdoor system consists of the Users, Server and the Home representative which is the host [20]. The Indoor framework has smart home devices. The users communicated/controlled the smart home devices via smart home app. Considering the Asynchronous communication pattern of IoT systems the author has proposed Advanced Message Queuing Protocol (AMQP) over HTTP. The publisher and the subscriber communicate via message queues. Whenever the publisher has to send data it sends it to the subscriber using the name of the queue. AMQP has one-way basic communication queue, RPC communication queue and also Topic exchange queue where the publisher broadcasts its message to all its subscribers. With respect to the communication scheme, the author proposes a combined encryption scheme bringing together the robustness of RSA algorithm and the flexibility of the AES encryption scheme. Further to improve the performance the device ID is stored in the cloud server whereas the real-time user data is stored in the local Host i.e., within smart home system. The user data will be updated in the server once in every 24 h to avoid the risk of losing the data in case of physical damage to the smart home devices. Separate Database design for Server data and Host Data was used and the experimental result worked as the expected without hampering the privacy of the users.

3 Summary This paper has a brief introduction about smart home systems and their nature, The different threats to which IoT applications are exposed and the necessity for security mechanisms to handle them.

References 1. Han JH, Jeon Y, Kim J (2015) Security considerations for secure and trustworthy smart home system in the IoT environment. In: 2015 international conference on information and communication technology convergence (ICTC). IEEE, pp 1116–1118 2. Liu H, Li C, Jin X, Li J, Zhang Y, Gu D (2017) Smart solution, poor protection: an empirical study of security and privacy issues in developing and deploying smart home devices. In: Proceedings of the 2017 workshop on internet of things security and privacy. ACM, pp 13–18 3. Erfani S, Ahmadi M, Chen L (2017) The Internet of Things for smart homes: an example. In: 2017 8th annual industrial automation and electromechanical engineering conference (IEMECON). IEEE, pp 153–157 4. Schiefer M (2015) Smart home definition and security threats. In: 2015 ninth international conference on IT security incident management & IT forensics, pp 114–118

Enhancing Security in Smart Homes-A Review

369

5. abcNEWS (2013) Baby monitor hacking alarms houston parents. http://goo.gl/LpuJzg 6. Finamore A, Saha S, Modelo-Howard G, Lee SJ, Bocchi E, Grimaudo L, Mellia M, Baralis E (2015) Macroscopic view of malware in home networks. In: 2015 12th annual IEEE consumer communications and networking conference (CCNC), pp 262–266 7. Ye M, Jiang N, Yang H, Yan Q (2017) Security analysis of internet-of-things: a case study of August smart lock. In: 2017 IEEE conference on computer communications workshops (INFOCOM WKSHPS), pp 499–504 8. Ramesh GP, Kumar NM (2019) Design of RZF antenna for ECG monitoring using IoT. Multimed Tools Appl 1–6 9. Adiono T, Marthensa R, Muttaqin R, Fuada S, Harimurti S, Adijarto W (2017) Design of database and secure communication protocols for Internet-of-things-based smart home system’. In: TENCON 2017—2017 IEEE region 10 conference, pp 1273–1278 10. Hoque ME, Rahman F, Ahamed SI, Liu L (2009) Trust based security auto-configuration for smart assisted living environments. In: Proceedings of the 2nd ACM workshop on assurable and usable security configuration. ACM, pp 7–12 11. Salman T (2015) Internet of Things protocols and standards 12. Chia MH, Keoh SL, Tang Z (2017) Secure data provenance in home energy monitoring networks. In: Proceedings of the 3rd annual industrial control system security workshop, pp 7–14 13. Chothia T, de Ruiter J (2016) Learning from others’ mistakes: penetration testing iot devices in the classroom. In: 2016 {USENIX} workshop on advances in security education ({ASE} 16) 14. https://www.schneier.com/academic/archives/1999/12/attack_trees.html 15. Meyer D, Haase J, Eckert M, Klauer B (2016) A threat-model for building and home automation. In: 2016 IEEE 14th international conference on industrial informatics (INDIN), pp 860–866 16. Ge M, Hong JB, Guttmann W, Kim DS (2017) A framework for automating security analysis of the internet of things. J Netw Comput Appl 83:12–27 17. Chung H, Iorga M, Voas J, Lee S (2017) Alexa, can I trust you? Computer 50(9):100–104 18. Notra S, Siddiqi M, Gharakheili HH, Sivaraman V, Boreli R (2014) An experimental study of security and privacy risks with emerging household appliances. In: 2014 IEEE conference on communications and network security. IEEE, pp 79–84 19. Jia YJ, Chen QA, Wang S, Rahmati A, Fernandes E, Mao ZM, Prakash A, Unviersity SJ (2017) ContexloT: Towards providing contextual integrity to appified IoT platforms. In: NDSS 20. Liu J, Xiao Y, Chen CP (2012) Authentication and access control in the internet of things. In: 2012 32nd international conference on distributed computing systems workshops, pp 588–592

Accident Detection Using GPS Sensing with Cloud-Offloading D. Srilatha, B. Papachary and N. Sai Akhila

Abstract In this paper Accident detection using GPS sensing with cloud-offloading has designed. This paper gives an approach to continuous monitoring of the specified objects. This gives an improved security as at whatever point a signal from sensor happens, an SMS that utilizes GSM module, GPS module, and Raspberry Pi is sent to the desired mobile number to take important action. The system is deployed to monitor any deviations in the surrounding IR rays’ Regular GPS receivers, although extensively readily available for navigating objectives, could take in excessive power to be valuable for lots of applications. Seeing that in many noticing conditions, the data could be effectively enhanced by the post-process of GPS and the data is circulated to a web server generates a service called cloud-offloaded GPS (CO-GPS), which allows a gadget to strongly calibrate its GPS receiver and in addition to that it logs the enough raw signal of GPS for the post-processing. By using easily accessible information, like an information source about Earth’s altitude, and also ephemeris of GNSS satellite, an arrangement of cloud might obtain the high-quality location by GPS from a raw data in few nanoseconds. Keywords Location · GSM · GPS · Assisted GPS · Raspberry Pi · IR sensor · Wi-Fi

1 Introduction Over the previous decade, remote gadget systems have progressed as far as equipment style, correspondence conventions, asset power, and various perspectives. As of late, there has been developing enthusiasm for versatile remote gadget systems, D. Srilatha (B) · B. Papachary · N. S. Akhila Department of ECE, CMR Engineering College, Hyderabad, India e-mail: [email protected] B. Papachary e-mail: [email protected] N. S. Akhila e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_41

371

372

D. Srilatha et al.

and a few other little profiles detecting gadgets that territory unit ready to the board their own development has just been created. Localization is realized utilizing radio interferometric point of landing estimation, in which course to a mobile node from few framework mobile nodes is evaluated dependent on the watched phase difference of an RF interference signal. The situation of the mobile node is then decided to utilizing triangulation [1]. Remote gadget hubs should utilize either keep energy (e.g., batteries) or gathered energy (e.g., star cells). The speed at whichenergy will be expended is influenced by either the hubs’ required time span for keep energy or by the typical rate of energy collected through gathering, detecting, registering, putting away, and human action. Every one of these procedures devours a one of a kind amount of energy. The transmitting and receiving functions in communication system have different energy consumption [2]. For effective indoor fingerprinting, FM broadcast radio signals are employed [3]. Therefore with the low frequency, the signals of FM are less vulnerable to presence of human. They show the notable indoor penetration. When compared with WiFi signals they change with less over time. The radio signal of FM with the RSSI qualities can be employed to achieve limitation in room-level with comparable or better accuracy to the one achieved by Wi-Fi signals. A new method based on embedded GPS detecting known as CO-GPS [4]. By using a common time route procedure and utilizing data that is as of now it is webaccessible, for instance, ephemeris of satellite, we display that 2 ms of common GPS signals is enough to acquire fixing of an area. By averaging a brief time frame with different short pieces, CO-GPS can all things considered accomplish =THRESHOLD

ACTIVATE GPS

GET LOATION CO-ORINATES AND DISPLAY ON LCD

INITIALIZE GSM/GPS

SENSOR VALUES AND LOCATION DETAILS SENT TO i) REGISTERED MOBILE NUMBER ii) WEB SERVER PAGE & DISPLAY ON LCD

STOP

using an IP address given to the module or alerted if any changes in the sensing data rise above the threshold level. Figure 4 shows the positioning is done in the form of latitude and longitude along with the location of the place.

5 Conclusion A new method based on embedded GPS detecting known as CO-GPS is proposed. By using a technique of coarse time navigation and leveraging data that can be accessed on the web, for instance, ephemeris of satellite, we display that 2 ms of common GPS signals is enough to acquire fixing of an area. By averaging a brief time frame with different short pieces, CO-GPS can all things considered accomplish threshold value, motor is turned ON. If measure value > threshold value, motor is turned OFF.

394

Md. Ankushavali et al.

Fig. 4 Hardware kit image

Fig. 5 Nitrate sensor actuated time

Figure 7 TELNET mobile applications are used for giving display conditions of sensors data.

6 Conclusion More to our earlier job, a temp made up interdigital capacitive sensing unit has actually been cultivated in the present research to gauge nitrate at reduced focus.

Automatic Nitrate Level Recognition in Agriculture Industry Fig. 6 Soil moisture sensor

Fig. 7 Output results by using Telnet app

395

396

Md. Ankushavali et al.

A transportable, unique noticing device has actually been established that can be utilized on-site as a gadget, and also IoT-based distant tracking intelligent sensing unit nodule, to determine nitrate focus in area as well as ground water. Electrochemical Impedance Spectroscopy was actually utilized to sense as well as present nitrate attentions, through assessing the resistance modification checked out due to the interdigital transducer submersed in the area water examples. The exam examples were actually analyzed through industrial devices (LCR gauge) and also the developed device. These end results were actually additionally verified making use of typical lab methods to examine nitrate attentions in water examples. The created body presented a really good linear partnership in between the evaluated nitrate focus (varied coming from 0.01 to 0.5 mg/L) to those gauged due to the business tools in the picked up water examples. Nonetheless, the existing unit possesses the prospective to become made use of to predict nitrate attentions in water examples, in real-time. The body can easily publish the assessed nitrate records on a site based upon IoT. This device might be utilized to include water top quality tracking internet sites within ranches, or even in between flows, waterways, as well as pond. For the in situ instalment, a sturdy container having the entire device will should be actually mounted at the checking web site.

References 1. Da Xu L, He W, Li S (2014) Internet of things in industries: a survey. IEEE Trans Industr Inf 10(4):2233–2243. https://doi.org/10.1109/TII.2014.2300753 2. Ramesh GP, Kumar NM (2019) Design of RZF antenna for ECG monitoring using IoT. Multimed Tools Appl 1–6. https://doi.org/10.1007/s11042-019-7581-9 3. Li L, Xiaoguang H, Ke C, Ketai H (2011) The applications of WiFi-based wireless sensor network in internet of things and smart grid. In: 6th IEEE conference on industrial electronics and applications (ICIEA), pp 789–793. https://doi.org/10.1109/iciea.2011.5975693 4. Thingspeak: the open data platform for the Internet of Things (2016). https://thingspeak.com/ 5. Dymond J, Ausseil AG, Parfitt R, Herzig A, McDowell R (2013) Nitrate and phosphorus leaching in New Zealand: a national perspective. N Z J Agric Res 56(1):49–59. https://doi.org/10.1080/ 00288233.2012.747185 6. Kellman L, Hillaire-Marcel C (2003) Evaluation of nitrogen isotopes as indicators of nitrate contamination sources in an agricultural watershed. Agr Ecosyst Environ 95(1):87–102. https:// doi.org/10.1016/S0167-8809(02)00168-8 7. Thorburn PJ, Biggs JS, Weier KL, Keating BA (2003) Nitrate in groundwaters of intensive agricultural areas in coastal Northeastern Australia. Agr Ecosyst Environ 94(1):49–58. https:// doi.org/10.1016/S0167-8809(02)00018-X 8. Horita K, Satake M (1997) Column pre concentration analysis spectrophotometric determination of nitrate and nitrite by a diazotization–coupling reaction. Analyst 122(12):1569–1574. https:// doi.org/10.1039/A703838K 9. Dinesh Kumar R, Ramesh GP (2018) Reconfigurable antenna design for soil testing to improve soil quality. Int J Eng Technol 7(2.20):70–73. https://doi.org/10.14419/ijet.v7i2.20.11756

Edge Detection-Based Depth Analysis Using TD-WHOG Scheme P. Epsiba, G. Suresh and N. Kumaratharan

Abstract In current trends, detection and tracking of moving object is significant for video surveillance services. The surveillance concentrates on highly prohibited regions and secured environments. These machine vision systems were widely incorporated in automotive surveillance and traffic monitoring. There are two most key challenges need to be addressed to solve the real-time issues in video surveillance systems are memory requirements to store sequence of information and moving object detection. Hence, the edge detection-based moving object tracking provides better feature vector and higher classification rate for further process. This work discusses the research contribution on edge detection-based depth analysis using TD-WHOG scheme for compression. Also, the conventional standards were failed to exploit better spatio-temporal redundancies, which may not produce considerable coding efficiency for dynamic texture video sequences. So, the prime objective is to suppress the redundancy; dimensionality reduction is the mode of process for increasing the effectiveness. Keywords Tucker decomposition · DCT · WHOG · Compression

P. Epsiba (B) Pallavi Engineering College, Hyderabad, India e-mail: [email protected] G. Suresh Sri Indu College of Engineering and Technology, Hyderabad, India e-mail: [email protected] N. Kumaratharan Sri Venkateswara College of Engineering, Sriperubathur, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_44

397

398

P. Epsiba et al.

1 Proposed Edge Detection-Based Depth Analysis Using TD-WHOG Scheme This chapter discusses the new video coding scheme to exploit better spatio-temporal redundancies, which produces considerable coding efficiency for dynamic texture video sequences. The prime motivation is to introduce TD principle for extracting sub-image features to acquire the spatial and spectral relationship at the same time [1, 2]. Numerous classes of tensor decomposition techniques have been evaluated; among the most popularly preferred tensor decomposition is TD method [3–5]. TD permits the choice of whichever values for every dimensionality of the core tensor is to achieve compression ratio at greater scale. Since, Tucker decomposition (TD) technique gets a benefit of more regular sequence and dynamic structure with lower ranked approximation [5–9]. By applying TD approach to dynamic sequence, first it calculates low-ranked approximation and subspace distance is evaluated among successive column vectors using basis components [10–13]. Then, HOG descriptor scheme detects the change in events and from the calculated gradient vector with threshold; edge classification is achieved for detection and tracking and also to improve the coding efficiency (Fig. 1). Depth analysis of the image or video sequence has various statistical characteristics which distinguish the dynamic texture features in an image. In addition to that, minimal quantities of diverse depth level of the image object take place owing to the more powerful quantization. Hence, the objective of depth map approach is for compressing and preserving with better quality. However, the DCT is applied in various natures of environments to image signal recognition, biomedical, machine vision, scientific computational analysis, etc. The basic purpose of DCT is to calculate the inner product among the source signal and the curvelet basis function to realize the sparse coefficient representation of the input sequence. A DCT coefficient be able to be stated as,

Fig. 1 Block diagram of the proposed TD-WHOG scheme

Edge Detection-Based Depth Analysis Using TD-WHOG Scheme

399

  C( j, l, k) = f, ϕ j,l,k . where j, l = 0, 1, 2, 3… are the scalable and orientation factor and k is the translation factor. Then the Fourier Transform of ϕ j (x) is ϕ j (w) = U j (w). U j (r, θ) = 2

−2 j/4



2 j/2θ W (2 r )V 2π −j

 (1)

Here, W, V is radial and angular window, and the DCT is carried out in three stages of operation. Firstly, the frame is split into three-level sub-bands. Then, tiling is done with level 1 and level 2 sub-bands, followed by discrete Ridgelet Transform (RT) applied on every piecewise tile. The RT was intended to code linear singularities to handle smooth edges by tiling sub-bands 1 and 2 by magnification. Therefore the continuous Ridgelet Transform is given as: ¨ R f (a, b, θ) =

ψa,b,θ (x, y) f (x, y)d xd y

(2)

where 

x cos θ + y sin θ − b ψa,b,θ (x, y) = a ψ a −1 2

 (3)

The digital curvelet transform is derived on a 2-D Cartesian grid. D (x, y) = Ca,b,θ



D f (m, n)ψa,b,θ [m, n]

(4)

0≤m U1 , U3 ), Pr (User) = Pr (U3 > U1 , U2 ) where U1 =car, U2 =taxi, U3 =bike. With the above-mentioned table the calculations are made and the minimum amount among the alternatives can be suggested to the users. With these two algorithms in consideration we use MLN model as it produces more generalized output with the formula calculated, and from the results produced we can obtain the mode of transport which are to be used with cost normalization. For instance, here we calculated the cost by multiplying the total distance with cost involved per km and

Suggesting Alternate Traffic Mode and Cost Optimization …

679

obtained the result. Here we calculated the cost of each transport using the distance provided during the route detection along the travel time. With the distance obtained, cost is calculated by a general formula so that it provides an approximate cost of travel of a particular mode on every time the user clicks the provided mode of transport.

3 Conclusion The current study discusses the application of random forest classifier for identifying travel or transportation means from the obtained dataset and suggested the alternate mode of transport to be used in order to obtain the efficiency in travelling. Compared with other classifiers such as KNN and Naïve Bayes, Random Forest has a great capability to perform better and to lessen noise due to the random selection of variables and data to generate more amount of classification trees. Our work brings in cost optimizing feature such as suggesting the minimal amount of cost to be used for travelling from one place to another. Further studies in future may focus on inclusion of GIS sources and take real-time data from the users and detect the travel mode and hence increase overall detection precision. For example, bus mode detection may be made much better by matching the total bus networks and choosing the best out of it, and cost optimization can be improved based on the vehicle with shared or unshared basis. Another novel strategy can be used to enhance the prediction accuracy by examining the rationality and logic of the identified trip chain. Acknowledgements The authors express their gratitude towards the TATA Realty-SASTRA Srinivasa Ramanujan Research chair grant for promoting this work.

References 1. Wang Q, Feng X, Liu Y, Wang X, Zhang H (2014) Urban travel mode split optimization based on travel costs. Procedia-Soc Behav Sci 14(138):706–714 2. Bai T, Li X, Sun Z (2017) Effects of cost adjustment on travel mode choice: analysis and comparison of different logit models. Transp Res Proc 1(25):2649–2659 3. Rogalska M, Bo˙zejko W, Hejducki Z (2008) Time/cost optimization using hybrid evolutionary algorithm in construction project scheduling. Autom Constr 18(1):24–31 4. Jamshidnejad A, Lin S, Xi Y, De Schutter B (2018) Corrections to “integrated urban traffic control for the reduction of travel delays and emissions” [IEEE Trans Intell Transp Syst 14:1609–1619 (2013)]. IEEE Trans Intell Transp Syst 6(99):1–6 5. Wang B, Gao L, Juan Z (2017) A trip detection model for individual smartphone-based GPS records with a novel evaluation method. Adv Mech Eng 9(6):1687814017705066 6. Jing P, Zhao M, He M, Chen L (2018) Travel mode and travel route choice behavior based on random regret minimization: a systematic review. Sustainability 10(4):1185

Rare Lazy Learning Associative Classification Using Cogency Measure for Heart Disease Prediction S. P. Siddique Ibrahim and M. Sivabalakrishnan

Abstract Discovery of class association rules from the enormous amount of database is considered as a vital task in data mining. Nowadays, the development of computer technologies, for example, database management system and data storage has given the stage for gathering and dealing with a lot of information. Finding frequent class association rules combines two significant data mining techniques such as association rule mining and classification for maximum accuracy. Since this hybrid mining yields a large number of class rules during database scan, the processing time will increases significantly. Moreover, many of these rules may be irrelevant and sometimes it misses the most important rules during classifier construction. Rare lazy learning associative classification is the process of recognizing rare class rules without generating classifier construction by focusing on the useful features from patient data of the test instance which is important for rule generation. This method is profoundly appropriate for biomedical field like heart disease prediction system which requires higher accuracy. The proposed algorithm uses simple local caching mechanism on heart disease data and constructing classifier using cogency measure that will make the lazy associative classification fast and achieve higher accuracy than traditional algorithms. Experimental results show that the proposed cogencybased algorithm is more efficient than traditional algorithms for heart disease with high prediction accuracy and the proposed algorithm may also be useful in anomaly detection, fraud detection, detection of network failures and many more bio medical areas. Keywords Association rule mining · Rare rules · Classification · Associative classification · Cogency measure · Heart diseases S. P. S. Ibrahim (B) Research Scholar, School of Computing Science and Engineering, Vellore Institute of Technology, Chennai Campus, India e-mail: [email protected] M. Sivabalakrishnan Associate Professor, School of Computing Science and Engineering, Vellore Institute of Technology, Chennai Campus, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_74

681

682

S. P. S. Ibrahim and M. Sivabalakrishnan

1 Introduction Data has become the most famous word nowadays; in this world life depends on these data. Every day people take photos, make video recording, share millions Gigabytes of data through social media and also keep these data in their own data storage devices or in a cloud platform. The functional information from these data can be useful to the business for developing many applications. Data mining is a process used everywhere for decision-making. Frequent, in-frequent, association rule mining (ARM), closed itemset mining, rare itemset mining, etc., are some important domains in data mining. [1] ARM is defined in the form of A -> B, which implies a dataset comprises itemset A then it will prone to hold itemset B similarly. Classification is another imperative task in data mining. Numerous models have been projected by different researchers over the years to exactly guess the intention class. This task includes statistical [2], neural networks [3, 4], divide and conquer [5], decision tree [6], PART [7], prism [8], and Naïve Bayes [2]; among all these algorithm, decision tree is well suited for classification. Since the algorithm process is simple and easy to understand by the user. Associative classification [AC] is an alternate to decision tree algorithm, AC is a recent rewarding technique and a similar approach except that the classification involves prediction with class attributes on the consequent side of the database, conversely, association rule discovery correlation on any of the attributes present in the datasets [9]. Plastino and Merschmann [10] have proposed AC working in the following two ways namely eager learning and lazy learning algorithm. At first the method builds classifier and predicts the class whereas lazy learning methods delay the prediction phase until fresh sample wants to be classified and does not generate class rules for all instances and it will project the data in the training datasets only on those features in the test instance. In eager associative classifier the support measure plays a major role if it is too low, and then large number of class rules including rare rules can be generated. Thus, generating high-quality rules for constructing classifier in the outsized database without compromising the accuracy and efficiency of classification is the key challenge. This challenging task was achieved by lazy learning method [11] for protein classification proposed by Elena et al., where it does not insist on training process, instead it plays out a recursive projection of large scale datasets for its macro items for nearest neighbors prediction that speed up the classification method. The Syed and Chandran [12] proposed lazy associative classification method using information gain measure which yields better classification accuracy with less computation cost. The motivation of the proposed work is based on rare lazy learning associative classification, which reduces the hard part in the class rule generation by incorporate conditional probability based measure is the information processing model that strength the rules generation based on co-occurrence of symbols being true. Many applications are demands for such rare rules for their business improvement like intrusion detection, fraud analysis, market basket analysis, medical diagnosis, automobile sales prediction, and material engineering. Our proposed work will not

Rare Lazy Learning Associative Classification…

683

generate too many rules as it only projects the attributes in the training sets based on the given testing instances. The remaining parts of the paper are organized as follows: In the next section, we describe the previous work in the proposed work. In Sect. 3, proposed rare lazy learning method is explained. Section 4 provides sample computation that gives details about how the proposed algorithm deals both frequent and rare rules. The final section presents with different analysis of the proposed work and outline direction for future work.

2 Background Association rule mining likewise be utilized with the classification for efficient classifier construction called CBA, which used famous Apriori algorithm [1] whose consequent is a class label that needs numerous sweeps over the database. On the next stage selecting the best rules and discording the uninteresting rules from the generated rules takes place; final prediction stage, the CBA algorithm tries to predict the unknown instances. This algorithm still creates countless CARs due to Apriori algorithm. This process creates over fitting in most of the applications and memory overload [11]. CMAR [13] was proposed by Han et al. to improve the searching for frequent rule items by compact data structure to achieve the efficiency of the algorithm. The single support and confident measure are standard mechanisms used in ARM to find out frequent item sets. On the other hand, many real-time applications requires rare items, such as network instructor detection, fraud detection in credit card, rare disease prediction, competitor item prediction, etc., hence selecting high support value might be missing some rare rules. To address the problem of missing infrequent items that have minimum support values, at the same time those rules that score higher confidence is well handled by multiple minimum item support was proposed by Liu et al. In this method each individual items are assigned with different minimum support according to its frequency of appearance in the database. Ordonez [14] has introduced a technique that uses decision tree for diagnosis of heart disease. Lee [15] proposed application of data mining technique for investigation of heart disease prediction system. Association rule-based heart disease model is proposed by carlos Ordonez [16]. Wasan and Kaur [17] presented the artificial neural network-based intelligent system for diagnosis of diabetic patients. Navie Bayes-based automated system for heart disease system was proposed by Awang and Palaniappan [18], here the main role depends on the algorithm to improve classification accuracy of a given heart disease datasets. It has been distinguished from literature review, that sure there are issues which should be still improved for restorative datasets. In addition, association classification suffers from ineffectiveness, since it usually produce tremendous number of class rules. Regularly that is ahead to create more number of inconsequential rules and in the meantime rare rules having moderately low support is not delivered rather it is just disposed of. If there should be a heart disease prediction the accuracy of correctly

684

S. P. S. Ibrahim and M. Sivabalakrishnan

classified instances are not sufficient on the basis of training dataset even if it takes larger time for processing The proposed strategy was exceedingly suitable for dataset which required higher accuracy and less preparing time.

3 RLLAC Algorithm 3.1 Problem Definition Let T be the set of all transactions with n occurrences in the form , Where AT1 , AT2 …, ATm are attributes and given C a list of class labels. The learning stage algorithm is based on uncovering association in the form {A1 , A2 , Am } -> {C1 , C2 Cm }. A rule X -> Ci , here X is different rule combination and the consequent is class label. Consider the instance given in Table 1 where minimum support is 0.3 and confidence is 0.4 which means the only rule with -> C1 is greater than the minimum threshold will consider for further conclusion.

3.2 Heart Disease Heart diseases and stroke will become leading cause of death in all over the world including India. It is also known as cardiovascular disease (CVD) [18]. As per the World Health Organization (WHO) [19] report, at present 4 out of 10 people die in India, pretentious by heart disease. In rural India, 7 out of every 10 death is based on cardio vascular disease [20]. In upcoming years the death ratio will be increased about 24 million by 2030. Even though the heart disease is a serious problem for human life, it could be resolved by two ways namely bypass surgery and medication during the initial stage identification of the disease. Otherwise it leads to death and takes longer time to recover from the disease. Nowadays different types of classification algorithm have been developed for identification of heart disease [21]. But all the methods are suffering when a patient come up with exceptional or rare symptoms. So a new methodology is required to solve such problem. The proposed lazy method handles well the rare symptoms by applying the lazy method with cogency measure that helps the physician to predict the heart disease in earlier stage. Table 1 Sample training data

ID

AT1

AT2

Class

1

A1

A3

C1

2

A1

A4

C1

Rare Lazy Learning Associative Classification…

685

3.3 Rare Lazy Learning Associative Classification (RLLAC) A novel cogency measure-based RLLAC is proposed in this section. The existing algorithm uses confidence measure for controlling rules. It may not be an ideal measure in some cases, which results a few precious rules that may be missed due to its high threshold. Cogency-based class association rule will generate more instinctive rules without generating the classifier [22]. Thus, the accuracy and computation of the classification algorithm will be increased. The following procedures of the proposed method is described as follows: • Identifying the items confidence based on the sample data • Identifying the 2 items class rules based on the information • Mining rare and frequent rules by means of measures like support, confidence, and cogency threshold.

3.4 The Procedure of Proposed RLLAC The means of general RLLAC calculations are appeared as pursues: Step 1: Initially the algorithm projects only the key features on training samples from D and finds S1 and S2 frequent class rules(According to their support and confidence measure) Step 2: Find S3 class rules from S1 and S2 (According to their cog1 and cog2 measure) Step 3: Repeat the step 2 for generating S4, S5 Step 4: Calculate the algorithm accuracy.

3.5 RLLAC Algorithm Input: Let us consider that T be the set of n distinct training instances Let D be a variable length in test samples 1. 2. 3. 4. 5. 6. 7. 8. 9.

Begin for each ti ε D do Let Tti be the features prediction of Ti from ti Cti is the set of {T -> C} all CARs extracted from Tti Function cogency x -> y for each Cti Cogency = 1 Cogency* = Lay/Lyy; End for

686

S. P. S. Ibrahim and M. Sivabalakrishnan

End function cogency.

4 Sample Computation Let us consider an example heart disease dataset as given in Table 2, that stores information about 15 instances and it has 2 class values. Table 3 shows patient test instance. Here the assignment is to predict the correct diagnosis for the given test instance. Table 4 shows the projected training data given in Table 2, projected based on the sample features in the test instance showed in the Table 3. On the next steps the algorithm acquires the all frequent 1-itemsets (S1) from Table 4 training data using interestingness measure support and confidence. In the next phase the algorithm finds new 2-itemsets based on cogency measure. If min_supp = 2, confidence = 50%, minCog1, and min Cog2 are considered as 0.4 and 0.2, respectively. The following are the results for S1 found by projection of test instance with sample datasets. The following are the possible rules projected from sample datasets: Table 2 Sample training data Age

Chest pain

Blood sugar

Rest_ecg

Diagnosis

>45

typ

F

Normal

Yes

>45

typ

F

PLV

Yes

45

typ

T

PLV

Yes

30…45

asympt

T

ST_T

Yes

>45

typ

T

Normal

Yes

yes=2 PLV−>no=2 typ,T−Yes=3 typ,PLV−>yes=3

4.1 Calculation of Cogency (X -> Y) It is necessary to use the S2 itemsets for this phase to generate S3. Suppose if we can calculate X = {typ, T} -> ?(Yes or No Class) Cogency ((typ, T)− > Yes) (typ)− > Yes and (T - > Yes) Cog1 = 1 ∗ count of ((typ)/count of Yes) Cog1 = 1 ∗ 7/8 = 0.9 Cog2 = cog1 ∗ count of ((T/count of Yes) Cog2 = 0.9 ∗ 7/8 = 0.81 The rule {typ, T} -> Yes would satisfy the minCog1 and Cog2, so it inserts in the frequent 2-itemsets. Likewise, all Sm+1 iteratively are generated from Sm; this process will be continued till Sm = Ø. There are some difference in the rule results for Sθ = 25% and Cθ = 0.6. The sample computation clearly shows that the proposed cogency-based rare associative classification algorithm effectively handles the rare class problem and generates the minimal number of rules than traditional algorithm.

5 Evaluation of the RLLAC Algorithm The proposed algorithm was tested using heart disease dataset and several UCI [23] benchmark data sets, illustrated in Table 5. To check the effectiveness of our RLLAC for heart disease prediction, we judge numerous medical data sets from UCI. These datasets have different properties like various number of attributes, instances, classes, different numeric attributes, total number of rules, missing values, etc. All the datasets are implemented using C++ programming. In the dataset, Holdout approach [24] was

688

S. P. S. Ibrahim and M. Sivabalakrishnan

Table 5 Dataset descriptions Dataset

Number of transactions

Number of classes

Heart disease

303

2

Breast cancer

286

2

Breast-w

699

2

Diabetes

768

2

used where 90% of the datasets are utilized as training data and remaining 10% of the datasets are used as testing data. The system performance was examined by different metrics such as accuracy, precision, and recall for heart disease and other datasets. We used the following rule matching procedure for classifier construction: If confidence (Rk) > confidence (Rm) Else if confidence (Rk) = confidence (Rm) and support (Rk) > support (Rm) Else if confidence (Rk) = confidence (Rm) and support (Rk) = support (Rm) and length (Rk) < length (Rm). At that point the algorithm picks the correct rule dependent on the above situation and send it to figure the class label of the novel item. All the Experiments were carried out on a system with Windows 7 Operating System, AMD A4 Processor with 2.4 GHz Clock Rate with 8 GB DDR3 RAM.

5.1 Accuracy Computation Precision and recall measure is the limit of a classifier to appropriately characterize unlabeled information. It will take the count proportion of the quantity of accurately classified information data over the total number of argument in the test instances. The accuracy computation is given in Table 5 which indicates the proposed RLLAC for heart disease prediction that has greater accuracy than traditional classification algorithms such as J48, Navie Bayes, and neural networks.

Table 6 Accuracy computations Dataset

J48

Navie Bayes

KNN

RLLAC

Heart disease

80

76.1

75.4

94

Breast cancer

85

75

73

87.4

Breast-w

84

78

76

85

Diabetes

80

77

70

89

Average

82.25

77

74

89

Rare Lazy Learning Associative Classification… Table 7 Execution time comparison

689

Dataset

Time taken by LACI

Time taken by RLLAC

Heart disease

2.045

1.045

Breast cancer

1.356

0.789

Breast-w

5.343

2.450

Diabetes

8.045

2.897

Average

4.197

1.795

According to Table 6, RLLAC improves the accuracy of classification by +7% when compared to J48, 12% with Navie Bayes and 15% more accuracy than neural networks (Table 7). In Figs. 1 and 2, a diagram is drawn looking at the accuracy through RLLAC and other conventional methods. The X-axis denotes the diverse kinds of medical datasets and the Y pivot indicates the corresponding precision of the algorithms. What’s more, clearly the proposed RLLAC outperforms the traditional algorithms in both accuracy and execution time.

Average

Diabetes

Breast-w

J48 Navie bayes NN RLLAC

10 8 6 4

Time taken by LACI

2 Diabetes

Breast-w

Breast Cancer

0 Heart Disease

Fig. 2 Comparison of time efficiency

Breast Cancer

100 80 60 40 20 0 Heart Disease

Fig. 1 Different rule mining algorithms

Time taken by RLLAC

690

S. P. S. Ibrahim and M. Sivabalakrishnan

6 Conclusion A new rare lazy learning associative classification is proposed in this paper. The lazy algorithm is used as local search method to project the heart disease patient symptoms as test instance in the dataset for predicting the original condition. This work adopted the cogency measure which is a clearness measure and the main key is to generate more intuitive class rule for improving the heart disease prediction. The two essential assignments associated with this technique are initial rule which is generation based on testing instance and then the further class rules which will be generated based on cogency measure. The investigations were done on four therapeutic datasets from UCI repository and the evaluated accuracy ensured that the proposed algorithm improved significantly compared to the existing associative classification. As a future work we will work on reducing the number of attributes present in the test instance that significantly reduce the processing time.

References 1. Rakesh A, Ramakrishnan S (1994) Fast algorithms for mining association rule. In: Proceedings of the twentieth international conference on VLDB, pp 487–499 2. John GH, Langley P (1994) Estimating continuous distributions in Bayesian classifiers. In: Proceedings of international conference on uncertainty in artificial intelligence, pp 338–345 3. Ivy Y, Lien C-H (2010) Cosmetics purchasing behavior-an analysis using association reasoning neural networks. Expert Syst Appl 37(10):7219–7226 4. EI-Ghazawi T (2000) Parallel mining of association rules with a hopfield type neural network. In: Proceedings of 12th IEEE international conference on tools with artificial intelligence (ICTAI) 5. Fürnkranz J (1999) Separate and conquer rule learning. Artif Intell Rev 13:3–54 6. Ross Quinlan J (2010) Data mining tools: see 5.0. https://www.rulequest.com/see5-info.html. Accessed May 2010 7. Frank E, Witten IH (1998) Generating accurate rule sets without global optimization. In: Proceeding of international conference on ML, pp 144–151 8. Cendrowska J (1987) PRISM: an algorithm for inducing modular rules. Int J Man-Mach Stud 27:349–370 9. Claudio F, Lobo, David (1990) Genetic algorithms in search, optimization and machine learning. Addition-Wesley Publishing, Boston 10. Merschmann LHC, Plastino A (2006) Bayesian approach for protein classification. In: International proceeding of the 2006 ACM symposium on applied computing protein classification 11. Baralis E, Garza P, Chiusano S (2008) A lazy approach to associative classification. IEEE Trans Knowl Data Eng 20(2):156–171 12. Syed SP, Chandran R (2012) LACI: lazy associative classification using information gain. IACSEIT-Int J Eng Technol 4(1):1–6 13. Li W, Han J, Pei J (2001) Accurate and efficient classification based on multipleclass association rule. In: Proceedings of 2001 IEEE international conference on datamining, pp 370–377 14. Ordonez C (2006) Comparing association rules and decision trees for heart disease prediction. ACM, HICOM 15. Lee HG (2007) Mining bio signal data: CAD diagnosis using linear and non linear features of ARV, pp 56–66

Rare Lazy Learning Associative Classification…

691

16. Ordonez C (2006) Comparing association rules and decision trees for disease prediction. In: Proceedings of the international workshop on healthcare information and knowledge management, pp 38–46 17. Wasan SK and Kaur H (2006) Empirical study on applications of data mining techniques in healthcare. J Comput Sci (JCS) 194 18. Awang R, Palaniappan S (2009) Intelligent heart disease prediction system using data mining techniques. In: International conference of computer systems and applications, vol 8, pp 109– 116 19. World Health Organization, South East Countries. https://www.searo.who.int/india/en/ 20. The Times of India, 14 August 2011 21. Govrdhan A, Srinivas K, Kavitha Rani B (2010) Applications of data mining techniques in health care and prediction of heart attacks. Int J Comput Sci Eng 2(2):250–255 22. Smith A, Minnett R, Nielsen (2007) Confabulation theory. Confabulation Neuroscience Laboratory. Springer, pp 106–120 23. UCI machine learning repository. https://archive.ics.uci.edu/ml/datasets/Heart+Disease 24. Reitermanova Z (2010) Data splitting proceeding of contributed papers wds-10. Part 1:31–36

Intensify of Metrics with the Integration of Software Testing Compatibility S. Vaithyasubramanian, P. M. S. S. Chandu and D. Saravanan

Abstract The raising thickness of the present programming items joined with consistently expanding expenses of programming breakdown has pushed the requirement for testing to new pinnacles. The effective execution of the control over programming quality requires programming measurements. Utilizing compelling programming measurements we can screen necessities, foresee advancement assets, following improvement advance, and limit the support cost. The primary target is to execute the testing process with different traits associated with measurements to optimize effort and software performance. The proposed look into work is to recognize the conceivable measuring properties of programming test execution and test audit forms. This work presents a novel structure called vector space show, to perceive programming measurements identified with test execution and test audit stages additionally to distinguish the help of such measurements for the quantifiable characteristics. In addition, it is essential to break down the suspicions in the computation of the measurements. The measurements examined against each ascribe should be evaluated for their common sense as far as venture’s unique circumstance and advantages to the testing group. Keywords Software metrics · Test case · Errors · Complexity · Life cycle · Ranking · Fault recognition

S. Vaithyasubramanian Department of Mathematics, Sathyabama Institute of Science and Technology, Chennai, India e-mail: [email protected] P. M. S. S. Chandu (B) Department of CSE, S.V Engineering College, Tirupathi, India e-mail: [email protected] D. Saravanan Operations & IT, IBS University, Hyderabad, Telangana, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_75

693

694

S. Vaithyasubramanian et al.

1 Introduction Increment in rivalry and jumps in innovation have constrained organizations to receive inventive ways to deal with survey themselves concerning procedures, items, and administrations. This appraisal encourages them to enhance their business with the goal that they succeed and make more benefits and obtain a higher level of market. Numerous associations around the world are creating and executing distinctive standards to enhance the quality needs of their software. With a specific end goal to confront the creative and quickly changing difficulties postured by programming industry, the testing procedure ought to have the capacity to discover the timetables and costs that enhance the productivity, proficiency, and adequacy of the business [1]. Along these lines, the primary concern of powerful testing exertion is measuring and recognizing [2]. Software measurements are the foundation in appraisal and furthermore establishment for any business change. The destinations of this work are (i) Resolving the principle stages in the Software Testing Life Cycle (ii) Understanding the estimations part in programming testing process change (iii) Investigation of test audit and test execution process characteristics (iv) Analysis of resolutions bolstered by the measurements and when to accumulate them (v) Observing the present metric help for the distinguished test execution and test survey forms.

2 Related Works At high development associations, measurements are required to assume a key part in general process administration and in addition to dealing with the procedure of a task. It is the cornerstone of assessment and establishment for any business change. It helps in an association for gaining the data required and to enhance the profitability, items, and benefits, and to accomplish the coveted objective in the product life cycle show. The nature of programming is relating to our desires brings about advancement process [3]. Programming measurements are basic to keep up the high caliber of venture and furthermore savvy. It tells the advance of the task, so it keeps up the principles. To keep up the measurements, it is essential to have a correspondence between the groups to get insights about undertaking. Established researchers have consolidated a huge writing study on programming measurements [4]. The investigation of programming measurements total has been introduced in the past work. These total capacities of measurements likewise can be utilized to audit the product practicality list [3]. Chidamber and Kemerer’s [5] discussed essential functionality of software and its development in information technology, anticipating the product measurements the issues relies on the direct character between the needful and autonomous factors of the software. It is expanding the reliance among those factors. The quantity of imperfections is unlimited over the first incentive in the product measurements. One approach to lessen cost through deformities forecast is in utilizing programming

Intensify of Metrics with the Integration of Software …

695

measurements by and large and in light of the call chart specifically to anticipate and enhance conceivable issues in the product outline [6]. An instinctive desire is given into class expanding from unfriendly number of imperfections than expanding from measurements. Programming quality models are every now and again used to ascertain the edge estimation of programming quality [7]. Another prevalent approach in choosing disseminations and fitting its parameters is to rough the metric esteems watched [8]. Solidifying the diverse perspectives drives us to classified programming testing life cycle into test arranging, test configuration, test execution, and test audit stages. In past work, there are quantifiable qualities and related measurements for test arranging and test configuration was examined [9]. The investigation of this paper is to distinguish quantifiable characteristics for programming test execution and test audit stages. Furthermore, to explore the distinguished measurements with related conditions which can be utilized helpful and significant to the association increasing minimal effort and excellent programming. Such upgrades brings about expanded profitability and quality, and lessened process duration all of which make an organization focused on the product business.

3 Software Testing Life Cycle In past work, there are quantifiable traits and related measurements for test arranging and test configuration was examined [9]. The diverse assessments lead us to consolidate the sorted programming testing life cycle into test arranging, configuration, implementation, and audit stages. Elements of the articles are described by the specified symbols or numbers so as to portray them by clearly characterized unambiguous principle in the process of measurement. The three Ps of classification of Software metrics are (i) Product (ii) Process, and (iii) Project metrics. The product scope, strategy, structures, class, and difficulty are characterized in product metrics level. Process metrics is very well utilized to improve programming advancement and preservation. Project metrics is to optimize and to evade complications in development plans.

3.1 Test Case Prioritization Comparing with the non-prioritized order of test cases, prioritization improves the execution process of software testing and it depends on cost and coverage. Programming is tested by executing different test cases under various conditions [10]. However, in the event that where all experiments are executed for same sort of modules changed or unaltered, it results in no longer useful. By scheduling test case prioritization the interpretations of the testing modules result in better performance. It is unproductive to monotonously execute each experiment for each program work. Along these lines, experiments are organized in a sequence for execution depending

696

S. Vaithyasubramanian et al.

on various standards and parameters [2, 11, 12]. The implementation process of prioritization techniques is based on code coverage information [13, 14]. Assessing the test case cost we have evaluated cost for each test case and by estimating the time required to execute each test case, for example, machine or human interface expenses on test case execution. In each test case cost estimation is carried out initially and followed by fault severity. The relation between components Risk (R), Cost (C), and Impact (I) is given by 100I = RC. Dependency and Value based are the factors required to compute which depends on low—high priority, size, and portability.

3.2 Estimating Fault Severity Evaluating module and test quality are the two feasible ways to estimate fault severity. And to determine fault severity it is required to generate errors and to execute mutation testing. In fault generation process errors are produced by finding normally happening deficiencies and by seeding error cases. Errors happening offer outer legitimacy, however, they are costly to find and can’t be found in numbers adequate to help controlled experimentation [15]. The effectiveness of different prioritization methods Fig. 1 Overview of test case prioritization

Test case prioritization

• • • •

Value based test cases prioritization

Dependency based test case prioritization

Prioritization techniques Size loss Portability loss Risk exposure Risk reduction leverage

Prioritization techniques Tight dependency Loose dependency

Value based priorities test suite

Dependency based priorities test suite

Measuring effectiveness

Measuring effectiveness

Intensify of Metrics with the Integration of Software …

697

is measured using Average Percentage of Fault Detection (APFD) metrics [15, 16]. The measurement is as follows if there are n test cases, T test suite, and a set of m faults F revealed by T. APFD for test suite is estimated by A P F D(T ) = 1 − (T F1+T F2+···+T Fm) 1 + 2n , where TFi represents the test case T that reveals fault i. The nm overview of test case prioritization is represented as in Fig. 1.

4 Compatibility Testing Process How phases and conditionality works on applications are characterized in the compatibility testing process. To understand the customary application of various systems, test analysts should have enough information about the software, programming, equipment, hardware, etc. On course of testing process well defined, ordered conditions should be framed to check whether the application runs well under various levels. As an outcome of the testing process bugs are reported, faults are fixed, and imperfections are sent for retest.

5 Results and Discussion In this research four scenarios namely cost, coverage, value, and dependency-based metrics to evaluate the test cases to find out the priority are analyzed. Coverage is measured based on the metrics average percentage of statement, loop, condition, and branch coverage. Cost is measured based on following metrics cost of tool, regression and maintenance testing, Portability of test case failure, and Cost related to risk. Risk exposure, size, and portability of loss are the metrics measured for the factor value. The number of test cases planned, executed, test case passed, failure, and total dependency of loose are the metrics considered for dependency measure.

Fig. 2 Comparison between manual testing with two different operating system

698

S. Vaithyasubramanian et al.

The test case is written in a way that text field, combo box field should not be empty and date of birth should be in specified format. And finally of those metrics gave the result as cost is going to get the priority. Comparison between manual testing with two different operating system is represented in Fig. 2.

6 Conclusion It concludes the execution of cost-based prioritization methods and inspected viability of prioritization procedures regarding rate of fault identification for some standard java programs under various operating systems using compatibility testing. In this examination finally it demonstrates that cost-based prioritization strategies enhanced the rate of fault identification. The financial knowledge-based systems were strategies for test cases prioritization on the premise of cost factors, for instance cost of investigation and prioritization cost. The review of exact outcomes has demonstrated that these procedures are essentially more successful than irregular requesting systems. In identifying and improving the failures of the software testing compatibility, this method can be effectively used as traditional coverage-based prioritization method.

References 1. Farooq SU, Quadri SMK, Ahmad N (2011) Software measurements and metrics: role in effective software testing. Int J Eng Sci Technol 3(1) 2. Bryce RC, Colbourn C (2006) Prioritized interaction testing for pair-wise coverage with seeding and constraints. J Inf Softw Technol 48:960–970 3. Oman P, Hagemeister J (2008) Construction and testing of polynomials predicting software maintainability. J Syst Softw 24:251–266. IEEE Computer Society 4. Kitchenham BA (2010) What’s up with software metrics?—a preliminary mapping study. J Syst Softw 83(1):37–51. Elsevier 5. Chidamber SR, Kemerer F (1994) A metrics suite for object oriented design. IEEE Trans Softw Eng 20(6):476–493 6. Abandah H, Alsmadi I (2013) Call graph based metrics to evaluate software design quality. Int J Softw Eng Appl 7(1):525–548 7. Heitlager I, Kuipers T, Visser J (2007) A practical model for measuring of information and communications technology. IEEE Computer Society, pp 30–39 8. Concas G, Marchesi M (2012) An empirical study of software metrics for assessing the phases of an agile project. Int J Softw Eng Knowl Eng 22:525–548 9. Afzal W, Torkar R (2008) Incorporating metrics in an organizational test strategy. In: International conference on software verification and validation workshop, pp 236–245 10. Rothermel G, Untch RH, Chu C, Harrold MJ (1999) Test case prioritization: an empirical study. In: Proceeding of the 15th international conference on software maintenance, Oxford, England, pp 179–188 11. Leon D, Podgurski A (2003) A comparison of coverage-based and distribution-based techniques for filtering and prioritizing test cases. In: Proceeding of the 14th international symposium on software reliability engineering. IEEE Computer Society, Washington, DC, USA, pp 442–453

Intensify of Metrics with the Integration of Software …

699

12. Malishevsky A, Rothernel G, Elbaum S (2002) Modelling the cost-benefits tradeoffs for regression testing techniques. In: Proceeding of the international conference on software maintenance. IEEE Computer Society, Washington, DC, USA, p 204 13. Elbaum S, Rothermel G, Kanduri S, Malishevsky AG (2004) Selecting a cost- effective test case prioritization Techniques. Softw Qual J 12:185–210 14. Malishevsky AG, Ruthruff JR, Rothermel G, Elbaum S (2006) Cost-cognizant test case prioritization technical report, TR-UNL-CSE-2006-0004, Department of Computer Science and Engineering, University of Nebraska-Lincoln, Lincoln, Nebraska, USA 15. Li Z, Harman M, Hierons RM (2007) Search algorithms for regression test case prioritization. IEEE Trans Softw Eng 33(4):225–237 16. Hoffman D (1999) Cost benefits analysis of test automation. STARW, Software Quality Methods LLC

Petri Nets for Pasting Tiles M. I. Mary Metilda and D. Lalitha

Abstract Image synthesis is the process of creating new images from some form of image description. This paper presents an effective approach to generate a set of patterns from an input of small tiles that can be glued together. Small tiles can be useful in tiling a large area realistically and efficiently. Tile pasting system has been introduced in many ways. In this paper, tiling is done using Petri nets. Tiles are used as tokens in place of black dots. Keywords Petri nets · Tile pasting system · Tile tokens · Tiling

1 Introduction The craft of tiling is an important aspect which was developed very early in human civilization. Complex tiling designs have been utilized in finishing the floors, dividers in the houses just as in the meeting places like hotels, exhibition halls. In the improvement of complex tiling, the perfect tiling rises sharply due to the distinctive parameters portrayed in the system [1]. Many tile pasting strategies are introduced to generate massive patterns. Place transition nets have been presented for creating a vast example of kolams from little tiles taking them as tokens in the information places. Place transition nets were first introduced by Karl Adam Petri in the year 1962 as a mathematical tool for concurrent activities of system generation [2]. Petri nets were even used to generate array languages [3–5]. Hexagonal picture languages were also generated by Petri nets [6]. Petri nets and CFG have also been used in generation of the password, CAPTCHA and ATM PIN [7–10]. Color Petri nets and timed color Petri nets were introduced for password generation [11, 12]. Motivated by these concepts a new model for pasting tiles using Petri nets is introduced in this paper. M. I. Mary Metilda (B) · D. Lalitha Department of Mathematics, Sathyabama Institute of Science and Technology, Chennai, India e-mail: [email protected] D. Lalitha e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_76

701

702

M. I. Mary Metilda and D. Lalitha

Fig. 1 Before firing

Fig. 2 After firing

2 Preliminaries Definition 1 A place transition net structure (PNS) is defined as C = (A, B, D, E) where A = {A1 , A2 … An } is a finite non empty set of places where n > 0, B = {B1 , B2 …… Bm } is a set of transitions such that m > 0, D is defined as an input function from input place to transitions and E is defined as an output faction from transition to place. Definition 2 A place transition net [Petri net] is called as a marked Petri net if the tokens are assigned to some places of the net. These tokens will determine the implementation or performance of the net. Number of tokens and their positions will change during the performance of the net. We define the marking as M = {M1 , M2 … Mn }, where Mi denotes the number of tokens in the place Ai . We define it as M (Ai ) = Mi . Example 1 Figures 1 and 2. An examples for firing of transitions and the position of tokens before and after firing.

3 Right-Angled Triangular Tile Pasting System Triangular tile pasting system is studied earlier in [13]. We introduce the right-angled triangular tile pasting system. I, R, T, H be the four types of tiles used in the paper. Notations used: (i) We use the right-angled isosceles triangle tiles and trapezoidal tile which are given in Fig. 3. The lengths of the edges of the basic tiles are explained as follows: Fig. 3 Basic tiles

7 ʋ =

1

3 I

2

,

4

T

5

6

Petri Nets for Pasting Tiles

703

Fig. 4 Extended tiles—rhombus and hexagon

R 8

12

H

17

11

*

ʋ =

, 9

10

16

13 14

15

The lengths √ of the sides 1, 2 of the triangle and 4, 6 of the trapezium are defined as 1/ 2 unit each and side 3 and 5 are defined 1 unit, respectively. (ii) The extended tiles are given in Fig. 4. When we join the right-angled triangles four times we get a rhombus. On joining the sides (1, 1) gives the upper half of the rhombus which is an isosceles triangle. On joining the sides (2, 2) produces the rhombus as in Fig. 4. In the same way on joining the edges (7, 7) of trapezoid results in hexagon. The length of all edges of rhombus is equal to 1 unit. These sides are also equal to the sides of the hexogen 13, 16. The dimensions of the edges 12, 14, 15, 17 are equal to the dimensions 1 and 2 of the triangle. Catenation Rules for pasting Tiles: Context free rules: Here the edges are joined freely without any limitations. If edge 1 is to be joined with edge 2 then all the edges with the label 1 in the net is joined with the edge 2. It is explained in the following. The catenation rules used to join the tiles are defined here. The tiles which are used like right-angled triangular tile, trapezoid, rhombus, and hexagon are defined earlier. The edges are given names as 1, 2, 3, … 17. Joining the edges 1 with 1 or 2 with 2… will form an array of tiles as explained below (Fig. 5). Firing rules: The transitions of the net can be defined with any of the following catenation rules. 1. The input places of the transitions must contain same array. 2. If there is no catenation rule associated to the transition and the initial (input) places have the distinct arrays then the transition is disabled. 3. If there are deferent arrays in the input places and a condition is assigned to the transition then the catenation will take place in a parallel manner and the out-coming array will be moved into all the output places.

Joining 1 with 1 is (1,1)

Joining 3 with 9 is (3,9)

Joining 11 with 13 is (11,13)

Joining 16 with 13(16,13)

Fig. 5 Example for tile pasting using context free rules

704 Fig. 6 Transition assigned with a condition

M. I. Mary Metilda and D. Lalitha

A1

B (A2)

A1

B (A2)

A2

A2

After firing

Before firing

Example 2 Figure 6 an example how the firing rules are followed during the execution of firing. Definition 3 A right-angled triangular tile array token Petri Net (RTATPN) is defined as 6 tuple K = (π, C, M, R, P, F) where π is the set of alphabet of tiles, C is a Petri Net Structure (PNS), M is an initial marking of the net which is formed by the tiles and is assigned to any one of the place in the net, R is the set of catenation rules that are assigned to the transitions, P is defined as a partial mapping from T to the set of catenation rules, F is the final set which is the subset of A. Definition 4 If K is a right-angled triangular tile array generating Petri net structure then we define K as (K ) = set of all arrays generated by the Petri net. It is the collection of tiles which reach the final place F. Example 3 Consider RTATPN which is defined as K 1 = (π, C, M, R, P, F) where π = (I, T), C = (A, B, D, E) is a PNS, where A = (A1 , A2 , A3 , A4 , A5 , A6 ) and B = (B1 , B2 , B3 , B4 , B5 ). The initial marking M is a right-angled triangle [ I ] in the initial place A1 . The set of catenation rules are defined as R = (1, 1), (2, 2), (3, 4), (6, 6), (4, 3). P is defined as a partial mapping from T to the set of catenation rules as shown in Fig. 7. F = (A6 ) is the final set. Initially an array I is in the initial place A1 . On firing the transition B1 , edge 1 of the array I is catenated with the edge 1 of I itself and moves it to the place A2 . On firing B2 , edge 2 is glued with edge 2, this resulting array deposits it in A3 . When B3 fires edge 3 is pated with edge 4, and moves it to A4 enabling B4 for firing. On firing B4 , edge 6 is joined with 6 and the array is moved to A5 . On firing B5 the side 4 of R is pasted with an edge 3 of I and the resulting array is deposited in the final place A6 as well as in place A1 enabling B1 again for firing. The array which is put in the place A6 is the set of arrays generated by the net K 1 . On firing the transitions Fig. 7 Net generating the language of squares

A1

B1 (1,1)

B2:(2,2) A2

A3

B3:(3,5) A6

B5:(5,3)

A5

B4:(7,7)

A4

Petri Nets for Pasting Tiles

B1 A1

705

B4

B3

B2 A2

A3

B5

A5

A4

A6

Fig. 8 Language 1 of squares

continuously, the tiles are joined in an order and the language produced by the net is shown in Fig. 8. Example 4 Consider RTATPN which is defined as K 2 = (π, C, M, R, P, F) where π = (R), C = (A, B, D, E) is a PNS, where A = (A1 , A2 , A3 ,) and B = (B1 , B2 ). The initial marking M is a Rhombus [ R ] in the initial place A1 . The catenation rules are defined as R = [(11, 9), (10, 8)]. P is defined as a partial mapping from T to the catenation rules as shown in Fig. 9. F = (A3 ) which is the final set. The language produced by the net is known as languages of rhombus that are defined as 2 (K 2 ). A tile R is assigned to the initial place. On firing B1 , side 11 is joined with 9 and the resulting array is moved to the place A2 enabling B2 for firing. On firing B2 edge 10 is joined with 8 and the resulting array is moved to the place A3 and to A1 . As soon as an array reaches A1 B1 is enabled for firing. The arrays that are collected in the final place A3 is the (set of rhombus) language 2 produced by K 2 (Fig. 10). Context sensitive rules: Here the catenation is defined with some restrictions. We define the context sensitive catenation rule as ([a/b], c), where a, b and c are the edges of the tile. If we have two or more edges with the same identification, we want to join specifically with B1: (11,9)

A

A

B2: (10,8)

A

Fig. 9 Petri net generating language of rhombus

A1

B1B2

B2

B1

A2

Fig. 10 Petri net generating Language of rhombus

A3

A3

706

M. I. Mary Metilda and D. Lalitha 1

Fig. 11 Example for context free rules Consider the tile

6 H

2

5

3 1

Joining 5 with 2 gives the resulting array as

4 6

6

1

5

2 3

4

3

4

Here we have two edges with name 6. We consider the edge 6 facing (right side) 1. We denote it as. [6/1].The sides of the Hexogen 3 is glued in between this sides 6 and 1. The resulting array is

one edge alone then we concentrate on the edge adjacent to the right side. We define this edge and its right side as [a/b]. Now the edge a facing b is joined with c. Other edges having label a will not be catenated with c when this rule is applied. Example 5 See Fig. 11. Example 6 Consider an RTATPN which is defined as K 3 = (π*, C, M, R*, P, F) where π* = (H), C = (A, B, D, E) is a Petri net structure, where A = (A1 , A2 , A3 , A4 , A5 ,) and B = (B1 , B2 , B3 , B4, ). The initial marking M is a hexagon [ H ] in the initial place A1 . The set of context sensitive rules are defined as R*=(16, 13), (13, 16), ([15/14], 12), ([17/12], 14). P is defined as a partial mapping from T to the set of catenation rules. F = (A5 ) which is the final set. The array collected in the place A6 is the language 3 produced by K 3 (Fig. 12). A tile H is assigned to the initial place A1 . On firing B1 , side 16 is joined with 13 and the resulting array is moved to the place A2 enabling B2 for firing. On firing B2 edge 13 is joined with 16 and the resulting array is moved to the place A3 . Transition B3 is assigned with the special catenation rule ([15/14, 12). The arrays in A3 have many edges with the label 15. In such a case a restriction is assigned to the transition such that an edge which faces edge 14 is alone chosen and an edge 12 is joined with 14. Similarly on firing B4 edge 17 which is facing edge 12 alone is chosen and is

A1

B1 :( 16, 13)

A2

B2 :( 13, 16)

A3

B3 :

([17 / 12],14)

A4

B 4:

Fig. 12 Petri net generating language of honey comb which is shown in Fig. 13

([15 / 14],12)

A5

Petri Nets for Pasting Tiles B1 , B2

A1

707 B3 , B4

A3

B1 , B2 , B3 , B4

A5 A 5 , A1

Fig. 13 Language 3 of honey comb network

joined with 14. The array is moved to A1 enabling B1 for firing. Similar array is also collected in the final place A5 . The arrays that are collected in the final place A5 is known as the language produced by K 2 . This language is also known as language of Honey comb of size n + 1 where n = 0, 1, 2, 3… . Theorem 1 The language generated by the Nets using context sensitive rules cannot be generated by the nets using context free rules. Proof Consider the RTATPN system K= (π, C, M, R, P, F) which generates the picture language 1 where π = {I, T}. The initial array in the initial place is T. On firing the transitions, the context free rules explained in example 3 are applied one by one. The resulting language is 1 . This language is shown in Fig. 8. On the other hand consider the RTATPN which is defined as K 3 = (π*, C, M, R*, P, F) admits the language produced using hexagon and the context sensitive catenation rules. In RTATPN the catenation rules can produces the family of languages. The languages produced by the context sensitive catenation rules as shown in Fig. 13 which cannot be produced by the context free rules. The catenation rules in context sensitive can do the combination of patterns of languages produced in Fig. 13. But by the catenation rules used in the context free rules, these language of Honey Comb network cannot be produced. Similarly the language of rectangles generated by the context free rules as seen in Fig. 8 cannot be produced by the context sensitive rules. Hence concludes the theorem.

4 Conclusion In this paper a new type of generating system called tile pasting using Petri nets has been introduced. A large pattern of tiling is derived in a simple manner by assigning conditions and catenation rules to the transitions of the net.

References 1. Bhuvaneswari K, Kalyani T, Gnanaraj Thomas D, Nagar AK, Robinson T (2014) Iso-array rewriting P systems with context-free rules. Int J Appl Math 3(1):1–16

708

M. I. Mary Metilda and D. Lalitha

2. Peterson JL (1981) Petri net theory and modeling of systems. Prentice-Hall, Englewood Cliffs 3. Lalitha D, Rangarajan K (2010) Column and row catenation Petri net systems. In: Proceedings of fifth IEEE international conference on bio-inspired computing: theories and applications, pp 1382–1387 4. Lalitha D (2015) Rectangular array languages generated by a Colored Petri net. In: Proceedings of IEEE international conference on electrical, computer and communication technologies, ICECCT 2015 5. Lalitha D (2015) Rectangular array languages generated by a Petri net. In: Computational vision and robotics. Advances in intelligent systems and computing, vol 332, pp 17–27 6. Lalitha D, Rangarajan K (2011) Petri net generating hexagonal arrays. In: International workshop on combinatorial image analysis. Lecture notes in computer science, vol 6636, pp 235–247 7. Vaithyasubramanian S, Christy A, Lalitha D Generation of array passwords using Petri net for effective network and information security. In: Advances in intelligent systems and computing, July 2014, vol 1. Springer, India, pp 189 – 200. ISSN: 2194-5357, https://doi.org/10.1007/97881-322-2012-1_20 8. Vaithyasubramanian S, Christy A, Lalitha D (2015) Two factor authentication for secured login using array password engender by Petri net. Proc Comput Sci 48:313–318. https://doi.org/10. 1016/j.procs.2015.04.187 9. Vaithyasubramanian S (2017) Array CAPTCHAs (completely automated public turing test to tell computer and human apart). Asia Life Sci-Asian Int J Life Sci 1:247–254 10. Vaithyasubramanian S, Christy A (2019) ATM PIN generation – a formal mathematical model to generate PIN using regular grammar, context free grammar and recognition through finite state machine, pushdown automata. Inderscience Int J Internet Protoc Technol 12(1):11–15 11. Lalitha D, Vaithyasubramanian S, Vengatakrishnan K, Christy A, Mary Metilda MI (2018). A novel authentication procedure for secured web login using colored Petri net. Int J Simul Syst Sci Technol 19(6). https://doi.org/10.5013/ijssst.a.19.06.33 12. Lalitha D, Vaithyasubramanian S (2018). Timed colored Petri net generating arrays. In: IEEE conference -on power, energy, signal and automation (ICPESA ‘18), conference ID: 43992XP, May 2018 13. Lalitha D, Rangarajan K (2012) Petri nets generating kolam patterns. Indian J Comput Sci Eng 3(1). ISSN:0976-5166

Ad Hoc Wireless Networks as Technology of Support for Ubiquitous Computation Amjed Abbas Ahmed

Abstract The main objective of ubiquitous computation is the establishment of environments where devices with processing and communications capabilities (mobile telephones, PDAs, sensing devices, electrical appliances, electronic books, etc.) can cooperate and communicate intelligently and be aware of the environment that surrounds the user transparently. Communication systems play a fundamental role in the area of the ubiquitous computation. In particular, ad hoc wireless networks also known as MANETs (mobile ad hoc networks) are presented as an ideal communication technology in this type of environments and applications. In this work, an experimental application is presented as an example of technology utilization, Wireless Bluetooth and IEEE 802.11 in the area of ubiquitous computation.

1 Introduction The term ubiquitous computation refers to that which would be able to take advantage of the information offered by computing devices distributed in the environment, in a transparent way to the user [1]. The continuous technological advancements have encouraged the development of devices with capacities of wireless communication increasingly smaller, more powerful and with a more efficient battery consumption that makes each day more realistic on the concept of ubiquitous computing. Strongly linked to the concept of ubiquitous computation, we find the environmentally dependent applications also known as context-aware applications. These applications are characterized by being able to adapt their functions in a transparent way depending on the context, the type of user, and the device used [2]. In the area of ubiquitous computing, communications play a fundamental role. In particular, the characteristics of ad hoc wireless networks can offer great flexibility to the communication system. Ad hoc networks, also known as MANETs, are wireless networks that do not require any type of fixed infrastructure or centralized administration, where the stations, A. A. Ahmed (B) Imam Al-Kadhum College, Baghdad, Iraq e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_77

709

710

A. A. Ahmed

in addition to offering functionalities of the final station, must also provide routing services, retransmitting packages between those stations that do not have direct wireless connection. Ad hoc networks can be deployed completely autonomously or combined with networks’ wireless sites to connect to the Internet using wireless access points. These networks must be able to adapt dynamically to the continuous changes in the characteristics of the network, such as station position, signal strength, network traffic, and distribution of the load. Among these characteristics, the main challenge of ad hoc networks lies in the continuous unpredictable changes in the topology of the network. New algorithms, protocols, and middleware are needed, which overcome the previously presented limitations and allow to establish independent and decentralized networks. These protocols should be completely adaptive, anticipating the future behavior of the network from parameters such as the level of congestion, the error rate, changes in routes used, etc. In addition, the resources of the network must be able to be located and used automatically without the need for a previously established manual configuration. Aspects related to security and the privacy of the information should be considered in order to provide access that allows ensuring the privacy of devices and users. Finally, oriented techniques must be incorporated to offer quality of service (Quality of Service, QoS) that allow, for example, to offer guarantees of service on certain network traffic. Numerous groups, research centers, and communications companies are working actively in projects related to the field of ubiquitous computation and context-aware applications [3]. Research related to intelligent environments and domestic spaces is increasingly more common in universities and corporate research centers [4–6]. Context-aware applications necessarily require some type of communication technology wireless network that provides wireless connection between the different devices in the network. Bliss wireless technology, together with sensor devices such as motion sensors and labels electronic systems are the basis that should allow the establishment of new intelligent environments where ubiquitous computing applications are able to interact with the environment in a transparent way without need for preset configurations or user intervention. There are several wireless communication technologies available from wireless technologies in the area broad third generation (3G), wireless networks of local area (Wireless LANs) or networks of personal area (‘Personal Area Networks, PANs’). We base our proposal on Bluetooth technology [7]. Bluetooth is a flexible and versatile wireless network technology of short range and low power [8]. Bluetooth has been specially designed to have a size and reduced cost, with the purpose of being able to incorporate into practically any everyday object. This paper describes an experimental context-aware application called UbiqMuseum, which provides information dependent on the environment to museum visitors. The application offers in each moment personalized information regarding the work of art that the visitor is observing. Bliss information is automatically adapted according to the selected language, the level of knowledge, and the type of device that the user uses. As devices, the application allows mobile telephones, portable computers, and PDAs. The application can also be used by the administrators of the museum in order to reduce costs derived from user inquiries, as well as for other purposes such as identifying the most visited museum pieces, obtaining patterns of behavior of the users, etc. The rest of

Ad Hoc Wireless Networks as Technology of Support …

711

the work is structured as follows. Section 2 summarizes the state of the art of ad hoc networks. Section 3 briefly introduces Bluetooth technology. Section 4 describes the application developed by presenting the architecture of it. Section 5 presents details related to the implemented implementation. Future extensions of the application are detailed in Sect. 6 followed by the conclusions in Sect. 7.

2 Ad Hoc Wireless Networks The term ad hoc, although it could be interpreted with negative connotations such as “improvised” or “disorganized”, in context of the wireless networks refers to flexible networks, in which all stations offer routing services to allow the communication of stations that do not have direct wireless connection. In relation to wired networks, ad hoc networks present frequent topology changes unpredictable due to the mobility of their stations. These characteristics prevent the use of Routing protocols developed for wired networks and create new research challenges that allow to offer efficient routing solutions that overcome problems such as topology dynamics, limited bandwidth and battery resources, and reduced security. Routing protocols developed for wired networks do not adapt to the highly environment dynamic of ad hoc networks. These protocols make use of periodic route update messages3 that offer a high overload even in networks with reduced traffic. In this design methodology in dynamic environments with frequent topology changes, these approaches offer an excessive overload. Recently, given the interest aroused by ad hoc networks, it has been established within the Internet Engineering Task Force (IETF), a new working group called Mobile Ad Networking group (MANET) [10], whose main objective is to stimulate research in the field of the ad hoc networks. A couple of years ago, about 60 proposals were being evaluated among the research community of different routing. However, today only four of these proposals have resisted the strong competition.

3 Bluetooth Technology in Ad Hoc Networks Recently, Bluetooth technology has been shown as a promising support platform in ad hoc networks. Ad hoc networks using Bluetooth as a base technology offer advantages considerable in the field of ubiquitous computing due to the ability of the Bluetooth to localize in a transparent way both nearby devices and the services they offer. Bluetooth is a low-cost, short-range wireless standard, especially aimed at connecting devices such as PDAs, laptops, and mobile phones without the need to additional wiring. Operating in the ISM frequency band and it is the technology used by the IEEE 802.15.1 working group dedicated to the wireless networks of personal area (WPAN).

712

A. A. Ahmed

Bluetooth uses a connection-oriented operation scheme based on a configuration master–slave in which, a single master coordinates the access to the medium of up to seven slave devices through periodic question shifts. This basic configuration is known in the standard as apiconet. Bluetooth defines two types of different links called Synchronous Connection Oriented (SCO) and Asynchronous ConnectionLess (ACL), which support applications with needs of real time such as sending video and audio and data applications, respectively. Bluetooth adapts to the requirements of context-aware applications not only because of its ability to group network stations into piconets, but also because of their ability to discover services in a transparent way. Thus, stations close to each other can locate neighboring stations using a procedure known as inquiry.

4 Architecture of the UbiqMuseum Application The UbiqMuseum application uses a network architecture that combines the use of a network dorsal with a network application. The application network uses only Bluetooth technology, while the backbone can be based on local Ethernet network technology or local network technology wireless 802.11 in infrastructure mode, as a global routing protocol. The system uses three different types of stations: museum clients (MICs), information points of the museum (MIPs), and a central server. A visitor to the museum with a PDA device with Bluetooth interface is an example of a client. In addition, there must be an associated information point with one or more pieces of art from the museum. Finally, the different MIPs of the museum will connect with the central server using technology, Ethernet, 802.11 or Bluetooth, depending on the facilities of the museum where the application is deployed. Figure 1 shows a possible configuration of the architecture of the application. As a client visits the different works of the museum, the application tries to continuously locate new information points using the primitive Bluetooth inquiry. Every time you locate a new information point, the application will check the services that it can offer using the Service Discovery Protocol (SDP) protocol. If the client wishes to receive the information that the new information point can offer, this should send you your profile, which was introduced at the beginning of the application on the client device. From user profile, the information point processes the request by combining the said profile with the identifier of the object that the client is visiting and finally sends the request to the central server. The central server stores the request and processes it, sending the requested information to the information point, which will finally send it to the customer. The search for new information points can be done automatically or by default upon request of the user. In addition, at any time the user can modify their profile, for example, in the case of considering that the information received is too advanced or very basic for its knowledge. Thus, the information received in subsequent accesses will be adapted more to your needs and requirements. UbiqMuseum has the following characteristics: Java-based implementation—The Java programming API has been used for the technology wireless Bluetooth proposed

Ad Hoc Wireless Networks as Technology of Support …

713

Fig. 1 Architecture of UbiqMuseum

by the Java Expert Group JSR-82 [9]. Around 20 companies which are leaders in the communication sector have adopted such a standard in their devices. JSR-82 offers an open and nonproprietary Bluetooth application development environment. Database with SQL support—all the information related to the objects of art in the museum is stored in a relational database. This solution offers flexibility, ease of use, efficient storage, maintenance procedures, and a high level of security.

5 Implementation of the Scatternet Protocol As previously mentioned, the global topology of the system uses a mechanism routing a modified version of the OLSR protocol. Client devices with technology Bluetooth will be connected via a scatternet topology around each information point. He proposes a new scatternet creation protocol, which is based on a creation algorithm of clusters proposed. In our implementation, each of the MIP devices acts as the master device of its own piconet, allocating channel utilization slots for all its slaves. In crowded environments, multiple piconets have to be interconnected to create a scatternet. Although the Bluetooth standard contemplates the concept of scatternet, this does not specify a certain protocol of creation of said structure. Getting a optimal scatternet structure is being the objective of numerous investigations [10–12]. Most of these studies have not focused on implementation problems. So, to be able to connect piconets 1 and 3 of Fig. 2, the master/slave bridge (M/S) device should switch to hold mode in piconet 3 and move to active mode with respect to piconet 1. This implies that communications in piconet 3 will be suspended until

714

A. A. Ahmed

Fig. 2 Example topology. Three piconets connected in a scatternet structure

the station time expires in hold mode. On the other hand, to connect piconets 1 and 2, the slave/slave bridge (S/S) device will switch to mode hold in piconet 2 and active mode in piconet 1. During hold time the device piconet master 2 will not send to POLL packets destined to assign access slots to the channel, bridge device. An active bridge device in a piconet, stores targeted data packets devices of the adjacent piconet, subsequently delivering them to the destination stations when the time in hold mode ends. When a client device cannot join the MIP piconet, it will try to locate some station that acts as a bridge station with the piconet of the information point. If it cannot find no bridge device, the client station will create a new piconet of which will be the master station and at the same time station bridge with the piconet of the MIP. To allow new stations to locate the new bridge station, it will register a new service called Bridge with the MIP. The new master station will assign the channel periodically to each of its slave stations. When a determined client requires information relating to a piece of the museum, its master the bridge device using the hold mode in piconet enables INQUIRY SCAN mode to the piconet of the MIP. The MIP locates the bridge station periodically using INQUIRY messages. When the time in hold mode ends, the bridge station leaves the piconet of the information point and send the information stored to the client station that made the request. The bridge that should stay away from your piconet should be calculated in the function using the following parameters: the time required to activate hold mode, the time required to join the piconet of the MIP, and the time necessary to obtain the information requested by the client. Throughout case, this period of time cannot exceed the maximum specified in the standard and that is 40.9 s or 65,440 slots.

Ad Hoc Wireless Networks as Technology of Support …

715

6 Future Extensions The developed prototype does not yet consider some implementation characteristics that would make our application a more viable solution. As the processing capacity of mobile devices increases, the possibility of sending multimedia data starts to be much more viable. To provide multimedia information with sufficient quality, mechanisms that guarantee quality of service must be incorporated. Using the UbiqMuseum application as an experimental environment, the problems that arise at the incorporation of service quality techniques is part of the level of access to the environment. It has been observed that even in less congested scenarios where you only have a single multimedia data stream, the benefits are not the minimum desirable due to the mobility of the stations. It has also been warned that as the length of the routes increases, the interruptions of the video sequence(video gaps) increase, reflecting these in the quality observed by the user. In [13–15] the direct relationship between video interrupts and route discovery procedures of the routing protocols was demonstrated.

7 Conclusions The main objective of this work was to demonstrate that the Bluetooth can be a candidate technology to provide access to the network to ubiquitous computing applications. Even though the interfaces of development such as Blue and JSR-82 are still in an early stage of development, these are mature enough for use in ubiquitous computing applications. UbiqMuseum, an experimental context-aware application based on Bluetooth, which has been presented is developed in Java. UbiqMuseum combines the productivity and flexibility of the development platform Java with the Bluetooth wireless connectivity features. The application has been specially designed to offer visitors to a museum information, precise of the pieces of arts that you are visiting. In addition, such information is adapted in a transparent way depending on the selected language, the level of knowledge, and the type of device that the user uses, thus significantly improving the user experience. The UbiqMuseum application uses a network architecture that combines the use of a backbone network with an application network. The application network uses only Bluetooth technology, while the network dorsal can be based on local Ethernet network technology or wireless local network technology 802.11 in infrastructure mode. The application network integrates one or several devices of nearby users to each of the pieces of art. From the user’s point of view the associated information points with the objects of the museum are detected automatically without the need of any manual intervention, obtaining accurate information of the objects visited. The concept of a device has been extended Bluetooth to create a scatternet structure, and an algorithm for creating scatternets has been proposed which allows to make more flexible the mechanisms for establishing the topology of the system. The evaluation of the application carried

716

A. A. Ahmed

out in the laboratory has focused on studying the productivity of the application and the delay of the inquiry mechanism. It has been observed that Bluetooth technology offers a stable productivity for distances of up to 14 m. The tests carried out have also shown that the delay of the inquiry mechanism shows stable behavior as the distance increases between the stations, being possible to set the average time of the inquiry procedure in 5 s. Finally, the authors of this work are aware that UbiqMuseum still requires improvements to be able to be used in a real environment. However, the developed application can be used as a test bank that allows the evaluation of new contributions in real applications related to the ubiquitous computation. Section 6 has presented some of the work lines in which theis working.

References 1. Marvel LM, Boncelet CG, Retter CT (1999) Spread spectrum image steganography. IEEE Trans Image Process 8:1075–1083 2. Wang Z, Bovik AC, Sheikh HR, Simoncelli EP (2004) Image quality assessment: from error visibility to structural similarity. IEEE Trans Image Process 13:600–612 3. Moon S, Kawitkar R (2007) Data security using data hiding. In: International conference on conference on computational intelligence and multimedia applications, pp 247–251 4. Parthasarathy C, Srivatsa S (2005) Increased robustness of LSB audio steganography by reduced distortion LSB coding. J Theoret Appl Inf Technol 7:080–086 5. Ouni T, Ktata I, Abid M (2014) Adapted method of slideshow processing 6. Nosrati M, Karimi R, Hariri M (2012) Audio steganography: a survey on recent approaches. World Appl Program 2:202–205 7. Hooda P, Ranga K (2013) Steganography in multiple data: review. Int J Latest Trends Eng Technol (IJLTET) 8. Jain AK (1989) Fundamentals of digital image processing. Prentice Hall, Englewood Cliffs 9. Chandra S, Paira S, Alam SS, Sanyal G (2014) A comparative survey of symmetric and asymmetric key cryptography. International conference on electronics, communication and computational engineering (ICECCE) 2014:83–93 10. Cole E (2002) Hiding in plain sight. Wiley, New York 11. Katzenbeisser S, Petitcolas F (2000) Information hiding techniques for steganography and digital watermarking. Artech house, Boston 12. Lee J-H, Chang B-H, Kim S-D (1994) Comparison of colour transformations for image segmentation. Electron Lett 30:1660–1661 13. Evans JJ, Brown CE (2002) Web graphics & the fireworks MX interface. Fireworks MX: zero to hero. Springer, pp 6–27 14. Cvejic N, Seppanen T (2004) Increasing robustness of LSB audio steganography using a novel embedding method. In: International conference on information technology: coding and computing. Proceedings. ITCC 2004, pp 533–537 15. Cvejic N (2004) Algorithms for audio watermarking and steganography: Oulun yliopisto

An Experimental Study on Optimization of a Photovoltaic Solar Pumping System Used for Solar Domestic Hot Water System Under Iraqi Climate Mahmoud Maustafa Mahdi and A. Gaddoa

Abstract The performance of a photovoltaic solar water pumping system (PVSWPS) promising in a solar water heating system has been studied experimentally. The design of a photovoltaic array configuration can affect the performance of the (PVSWPS), the water pump characteristic, the flow rate of the water and the overall system efficiency. The aim of the present work is to determine an optimum photovoltaic array configuration that can supply a water pump with an optimal amount of energy. Three different photovoltaic array configurations have been tested (4S, 2S × 2P and 4P). The experiments have been carried out in a sunny daylight hours under the conditions of Iraq climate and for a constant head level of 6 m. The results showed that the second photovoltaic array configuration (2P × 2S) is suitable to provide an optimal energy. Also, the second photovoltaic array configuration (2P × 2S) that powered the water pump delivered a maximum average water volume (2.298 m3 ) during the day among the three different configurations. Keywords Photovoltaic pumping system · Photovoltaic configuration optimization · Outdoor testing · Domestic hot water system

Nomenclature Apr E Ee FF Im P

Area of photovoltaic array, m2 Solar energy intensity, W/m2 Pump electric power, Wh/day Fill factor Maximum current of the pump, A Pump power required, W

M. M. Mahdi (B) · A. Gaddoa Electromechanical Engineering, University of Technology, Baghdad, Iraq e-mail: [email protected] A. Gaddoa e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_78

717

718

Q Voc ηp Ch Ei Eh H Isc Pm Vm ηpv ηoverall

M. M. Mahdi and A. Gaddoa

Water flow rate, m3 /s Photovoltaic array open-circuit voltage, V Pump efficiency, % The constant = 9800 kg/m2 s2 Energy of incident solar radiation, Kwh/day Hydraulic energy of the pump, Wh/day Water level head, m Photovoltaic array short-circuit current, A Photovoltaic array maximum power Pump nominal voltage, V Photovoltaic array efficiency, % Overall system efficiency, %

1 Introduction A hot water production for house uses is produced through the years by using a device with the benefit of using solar energy, this device is called a domestic solar water heating system (DSWHS). The hot water is circulated within the system by two methods either a thermo syphon method or a force-circulation method. The latter need an electrical energy and it cannot save it. A photovoltaic solar water pumping system was introduced in order to save the electrical power that used to power the pump and to circulate the water. The photovoltaic solar pumping systems are useful for intermediate and low applications like domestic hot water systems. Many researchers over the world have tested and studied the photovoltaic solar pumping systems [1, 2]. The performance of the photovoltaic pumping systems has been tested under solar insolation and different climate conditions using various types of pump [3–5]. A hybrid system (powered by photovoltaic arrays and by wind generators) has been used in the world today. These systems are cold solar powered systems [6–8]. Experimental and theoretical studies about the photovoltaic solar pumping systems were fabricated in order to provide water for irrigation and drinking [9–12]. Clark, studied the performance of a photovoltaic solar water pumping system. A diaphragm water pump was used. Vick and Clark compared the performance of a photovoltaic solar pumping system with a wind powered water pumping system. Also, Vick and Clark studied the difference in performance of fixed photovoltaic panel’s one-axis tracking solar pumping system. The results showed that the power increased for tracking system in comparison with the fixed system. Additionally, Vick and Clark studied the impact of using a helical pump on the performance of a photovoltaic solar water pumping system. It was concluded that the helical pumps standard (24 V) crystalline silicon photovoltaic cells will be more efficient than crystalline with a high voltage, and they manifested that the helical pump improved the performance of the pumping system significantly. Kihiwoot et al. developed a new version of a SWHS coupled with a solar water pumping system and found that

An Experimental Study on Optimization of a Photovoltaic Solar …

719

the circulated water varied from 15 to 65 L/day according to the discharge head and the isolation intensity. Roonprasung et al. presented a solar heating system coupled with a solar water pumping system and concluded that the water pump can work when the solar intensity is greater than 580 W/m2 , 600 W/m2, and 630 W/m2 with a discharge head of 1 m, 1.5 m and 2 m, respectively. The photovoltaic system performance depends on the operating condition and the system configuration. The condition at which the system operates depends on the system location, which governs the radiation intensity received, the outdoor temperature, and other aspects that affect the performance of the system. The system performance determines its economic and technical feasibility and whether the system represents the best solution in terms of electricity sources for any given application.

2 Experimental Setup The experimental test facility of photovoltaic solar water pumping system consists of a photovoltaic array of four photovoltaic modules. The photovoltaic array is configured in three different configurations. Each photovoltaic array configuration with disconnected switches contains four photovoltaic modules. The photovoltaic array is directly coupled with submersible water pump (type March 890-BR-HS 42 V). The water pump is put in the lower storage tank and circulates the fluid median between the lower storage tank and the upper one, water is the fluid medium which is circulating through a piping circuit, as shown in Fig. 1. The measured data include the solar radiation intensity in the plane of the photovoltaic array, the water flow rate, the array output voltage, and the array output current from the photovoltaic array configuration. All the measured data were half-hourly

Fig. 1 Photovoltaic water pumping system 1-photovoltaic (300 W), 2-measurement system (V, J, W), 3-adjustable stand, 4-water pump (type March 890 BR-HS 24 V), 5-water, 6-lower tank, 7-pipe line, 8-upper tank

720

M. M. Mahdi and A. Gaddoa

Table 1 Parameters of the photovoltaic module Photovoltaic module parameter

Value

Open-circuit voltage (Voc )

25.2 V

Maximum power voltage (Vmax )

20 V

Short-circuit current (Isc )

4.15 A

Maximum power current (Ipm )

3.75 A

Maximum power (Pm )

75 W

Table 2 Characteristic of Baghdad city Latitude

Longitude

Daily light

Ambient temperature

Yearly insolation

33.34o

44.1o

10–12 h

5 °C January 48 °C August

4.2–8.1 KWh/m2 /day

averaged. Water volumetric flow rate was measured as cumulative value during each half an hour. The photovoltaic module parameters are shown in Table 1. A typical commercial photovoltaic cell has an efficiency of 15%, the best efficiency of the photovoltaic module researches for different technologies was published by NREL [National Renewable Energy Laboratory] center. The photovoltaic solar water pumping system was evaluated under actual climate conditions. The experimental measured data were obtained from a series of tests at University Technology in Baghdad city. The characteristics of Baghdad city are shown in Table 2.

3 Characteristic of a Photovoltaic Configuration In order to provide the required current (I) and voltage (V), a photovoltaic array was connected in parallel, in series-parallel combination, and in series. Based on the photovoltaic modules, a proposed design of the photovoltaic array consisting of the three different configurations was selected. A 45° was selected as the photovoltaic tilt angle and facing to south. The configuration of three different arrays are, namely, C1 (4S), C2 (2S × 2P), and C3 (4P) which mean: • C1 (4S): 4 PV modules connected in series. • C2 (2S × 2P): 2 PV modules connected in two parallel rows with two serial PV modules in each. • C3 (4P): 4 PV modules connected in parallel. The aim of the above three different configurations is to reach the optimal performance of the water pump under the three different photovoltaic array configurations tested individually. The photovoltaic arrays were put into test under the condition of a sunny day at Baghdad city. The photovoltaic arrays tested under the outdoor

An Experimental Study on Optimization of a Photovoltaic Solar …

721

Table 3 Characteristic of three different photovoltaic array configurations Pv configuration

Shunt Isc (A)

Open voltage Voc (V)

Power Pm (W)

Apv (m2 )

FF

ηpv (%)

C1 4 in series

4.15

100.8

300

4.34

70.1

14.4

C2 2 in parallel 2 in series

8.2

50.4

300

4.34

70.3

14

C3 4 in parallel

16.4

25.2

300

4.34

70.1

15.17

condition then the fill factors (FF) and the maximum photovoltaic array power point of, (MPP), pump efficiency (ηp ), and the overall efficiency of the system will be determined. Maximum power point (MPP) is calculated by multiplying the maximum current and the maximum voltage that the photovoltaic array can provide at irradiance level, each photovoltaic array performance is characterized by the curve quality of its (voltage and current) sharpness. The photovoltaic energy rating has many models developed. Then, the photovoltaic energy is determined by fill factor: FF = Vm × Im /Voc × Isc

(1)

The photovoltaic performance module can be calculated by measuring the efficiency of conversion from sunlight to electric power. The photovoltaic efficiency is determined from the output power that was produced from the sun solar radiation. The photovoltaic standard measurement conditions are revered to the temperature, the area of the photovoltaic and the radiation intensity. By making a comparison between the output energy generated by the photovoltaic array and the incident solar energy on it, the efficiency of the individual photovoltaic configuration can be determined with following relation: ηpv = Vm × Im /E × APV

(2)

The characteristic of three different photovoltaic array configurations are shown in Table 3.

4 Methodology In the present work, the pump is powered by the photovoltaic array configuration. A fixed head of (6 m) was chosen for testing the water pump during the sunny days (April 2018). Each configuration was tested along three days. The measured power output, the volumetric flow rate of the water, the current and the voltage of the circuit and solar radiation intensity were measured, a typical solar intensity variation

722

M. M. Mahdi and A. Gaddoa

during the day. For each array configuration, the electrical energy, pumping hydraulic energy, the energy received by the photovoltaic surface area, pump efficiency, and the overall efficiency of the system were determined. The pumping consumes daily electrical energy, which can be determined by the power integration (during the operating period 11.5 h) demanded by the water pump. This integration is given by the following equation: hr=18 

Ee =

p · dt

(3)

hr=6.5

The pump provides a hydraulic energy during the day, and it can be determined by the following equation: hr=18 

Eh = Ch · H

Q · dt

(4)

hr=6.5

The amount of the energy received by the photovoltaic array that incidents on the surface area of it and during the daylight hours is calculated by the following equation: hr=18 

Ei = Apr

E · dt

(5)

hr=6.5

The efficiency of the water pump is obtained by dividing the pump hydraulic energy Eq. (4) by the photovoltaic electrical energy Eq. (3). This is given by the following equation: η=

Eh Ee

(6)

While the overall efficiency of the system is determined by dividing the hydraulic energy of the pump Eq. (4) by the incident radiation energy on the surface area of the photovoltaic during the operating time and it is given by the following equation: ηoverall =

Eh Ei

(7)

This optimal configuration can reach the maximum volumetric flow rate of water.

An Experimental Study on Optimization of a Photovoltaic Solar …

723

5 Results and Discussion Experiments were conducted with the aim of evaluating the behavior of a photovoltaic solar water pumping system (PVSWPS). The photovoltaic array system with three different configurations was tested. From the results of experimental data of the three different (PVSWPS) configurations, a comparison analysis is first carried out to obtain the overall performance of each configuration. The next step is to select the appropriate optimal design among photovoltaic array configurations. Table 4 displays the results of the three different (PVSWPS) configurations, (PVSWPS1), (PVSWPS2), and (PVSWPS). The average volume of the water during the day was between (1.7 m3 and 2.3 m3 ), for different configurations (C1, C2, and C3). From the obtained results of the performance of the system and through the three different configurations comparison, there is an attempt trying to select the most optimal system configuration. This optimal configuration can be able to reach the maximum volumetric rate of water. First, the comparison of the volumetric water flow rates and the electric power supplied by the three (PVSWPS) configuration (C1, C2 and C3) is made. A comparison between the three different configurations at the same level head is depicted. This work reveals the three curves which represent the variation of the daily volumetric water flow rates supplied by the three different photovoltaic solar water pumping system configurations. Also, this work elucidates that the provided volumetric flow rate by the (C2) configuration is the most significant around the mid of the day and a maximum value is reached, while the volumetric water flow rate supplied by the other two configurations (C1 and C2), significantly less than the volumetric water flow rate provided by the photovoltaic solar water systems. The volumetric water flow rate supplied by (C1 and C3) configurations has maximum values at the early morning operating time and then remains constant through the whole operating time. Table 4 Performance of the photovoltaic solar water pumping system Photovoltaic configuration

Maximum flow rate (m3 /day)

Maximum power (W)

Maximum pump efficiency (ηp %)

Average pump efficiency (ηavp %)

Maximum system efficiency (ηsys %)

Average system efficiency (ηa sys %)

C1 (4s)

1.79

117.1

72.2

42.47

7.12

3.61

C2 (2s × 2p)

2.298

152.4

67.6

46.51

7.44

4.22

C3 (4p)

1.811

167.7

58.3

36.14

8.92

4.69

724

M. M. Mahdi and A. Gaddoa

This work evinces the variation in the corresponding electrical powers with time of the day provided for the three photovoltaic solar water pumping system configurations (C1, C2, C3). This work represents the typical electrical power distribution of the photovoltaic module. There is a sharply steep electrical power gradient between (6.5 h) and (8 h), and the electrical power reaches its maximum values in the early morning. After this period, the electrical power is nearly constant in the model operation time. A second drop in the electrical power is noticed at the late daylight hours, approximately, at 16 h. The second photovoltaic configuration (C2) provides appropriate and enough electrical current, due to its parallelism arrangement, this can lead the photovoltaic array to meet the operating point of the pump. When the pump meets the operating point, this allows it to reach the maximum power of the pump quickly at the beginning hours of the day, i.e., rapid increase, then the pump still operates constant rate till the hours of the late daylight time, at which the power dropped rapidly (a steep drop in the power level). The power provided by the third configuration (C3) is over the maximum power of the pump. The second step is to make a comparison study of the pumping system efficiency for the three different configurations (C1, C2 and C3). This comparison is made in terms of the efficiency of the pump and the overall efficiency of the system trend. The work demonstrates the efficiency of the pump curves of the three different photovoltaic configurations. This work is plotted at the water level head of 6 m. From this work, it can be noticed that in the early morning and in the late daylight, the pumping system achieves the same values of efficiency. The second photovoltaic pumping system configuration (C2) provides a pumping efficiency more significant than the pumping efficiency that was provided by the other two configurations (C1 and C3). Table 4 exhibits briefly the outline results including the average pumping efficiency. It is observed from this table that the second photovoltaic pumping system configuration provides the best average pumping efficiency and the maximum water flow rate during the day (2.298 m3 /day). In order to assess the effect of the photovoltaic configuration on the overall system efficiency is plotted. This work illustrates the variation of the overall system efficiency during the operation period for ahead level of 6 m. This work clarifies a comparison of the overall system efficiency, the three different photovoltaic configurations. It can be noted from this work that the second photovoltaic solar water pumping system configuration (C2) provides an overall efficiency with a significant and constant values along the operating period hours of the days except at the early morning hours and at the late hours during afternoon. The overall efficiency of the system is provided by the first photovoltaic pumping system configuration (C1) which is less significant through the day time except in the early morning hours and in the late afternoon hours, at which the overall efficiency was marked as higher efficiency level.

An Experimental Study on Optimization of a Photovoltaic Solar …

725

6 Conclusion An experiment was carried out on a photovoltaic solar water pumping system to study its performance under three different photovoltaic module configurations. In general, the photovoltaic solar pumping system is sensitive to the variation of the solar irradiance failing on the photovoltaic array and the voltage and current that needed to run the pumping load. Besides, the volume of water provided by the photovoltaic module configuration of the second configuration (C2) is more significant around noon when compared with the other two photovoltaic system configurations. Also, the second system configuration (C2) is suitable for supplying a maximum flow of water through the daily operating period. Furthermore, the overall system efficiency with the second photovoltaic configuration (C2) is more appropriate in comparison with the low photovoltaic array systems during the optimum sun light. That means, in order to run the pumping system, the photovoltaic configuration (C2) is an appropriate choice.

References 1. Ghoneim AA (2006) Design optimization of Photovoltaic water pumping system. Energy Convers Manag 47:1449–1463 2. Amer EH, Younes MA (2006) Estimating the monthly discharge of a Photovoltaic water pumping system, Energy Convers Manag 47:2092–2102 3. Hadi AA, Benghanem M, Chenlo F (2006) Motor-pump system modelization. Renew Energy 31:905–913 4. Moechatra M, Juwono M, Kanstosa E (1991) Performance evaluation of A.C and D.C direct coupled photovoltaic water pumping system. Energy Convers Manag 31(6):512–527 5. Hamid M, Metwally B (1996) Dynamic performance of directly coupled photovoltaic water pumping system using D.C shunt motor. Energy Convers Manag 35:1405–1416 6. Grassi T, Macgregor K, Kubi I (2002) Design of photovoltaic driven low flow solar domestic hot water system and modeling of the system controller outlet temperature. Energy Convers Manag 43(8):1063–1078 7. Lava D, Merion GG, Pavez BI, Tapia LA (2011) Efficiency assessments of a wind pumping system. Energy Convers Manag 52:793–803 8. Anagunostopulos JS, Papantouis DE (2007) Pumping station design for a pumped storage wind-hydro power plant. Energy Convers Manag 48(11):3009–3017 9. Abdelmalek M, Abdelhamid M, Kadri D, Hiadsi S, Raja JA (2011) Performance of a directly coupled photovoltaic water pumping system. Energy Convers Manag 52(10):3089–3095 10. Skreta SB, Papadopoulos DP (2008) Systematic procedure for efficient design of electric water pumping system fed by photovoltaic or/and WECS application using measured meteorological data for the city of Xanthi/Thrace. Energy Convers Manag 49(4):596–607 11. Alajlan SA, Smiai MS (1996) Performance and development for remote area Saudi Arabia. In: International proceeding of WREC-IV, June 1996, vol 8, no (1–4), pp 441–446 12. Benghanem M, Hadi AA (2007) Photovoltaic water pumping systems for Algeria desalination. Renew Energy 290(1):50–57

A Novel Technique for Web Pages Clustering Using LSA and K-Medoids Algorithm Nora Omran Alkaam, Noor A. Neamah and Faris Sahib Al-Rammahi

Abstract The extensibility of various web documents available on the web made a critical challenge for many serious tasks such as information retrieval (IR), content monitoring, and indexing. Web documents could be any type of data that can be requested by user and delivered from web server through several web browsers. Most of web documents contain textual contents and are typically called web pages. However, in order to perceive and discover knowledge from these pages, novel techniques are required that have been never applied in other domains. In this paper, a new approach has been proposed by performed latent semantic analysis (LSA) on the result of VSM, which involves the correlation among web pages to their extracted features. The result of LSA involves the matrices that reflect the correlation between the web pages to their related concepts, which were used frequently for retrieving process. PAM (K-Medoids) algorithm was used with respect to semantic space, to portion the web pages into coherent groups. One of the most challenges in any clustering algorithm is to identify the correct number of clusters for the given data. Hence, two approaches are used for this manner: Elbow graph analysis to estimate the number of cluster range based on (SSE) values and clustering evaluation metrics. Calinski–Harabasz criterion (CH) and Silhouette Coefficient (SC) are the best wellknown evaluation metrics commonly used in partitioning-based algorithms. UOT has been considered to evaluate the proposed system, and the results are shown in the proposed system to achieve high accuracy results to separate the similar pages into coherent groups. Keywords Clustering · Web mining · Data mining · Web content mining · PAM · K-Medoids · Silhouette coefficient · Calinski–Harabasz criterion

N. O. Alkaam (B) Department of Higher Studies, Ministry of Higher Education and Scientific Research, Baghdad, Iraq e-mail: [email protected] N. A. Neamah · F. S. Al-Rammahi Department of Computer Techniques Engineering, Imam Al-Kadhem College, Baghdad, Iraq © Springer Nature Singapore Pte Ltd. 2020 V. K. Solanki et al. (eds.), Intelligent Computing in Engineering, Advances in Intelligent Systems and Computing 1125, https://doi.org/10.1007/978-981-15-2780-7_79

727

728

N. O. Alkaam et al.

1 Introduction With the massive growth of web documents on the World Wide Web, and enormous increasing of these documents on the web introduces a big challenge to understand its contents. However, with million if not billion interconnected web documents created by millions of Authors around the world, the task of understanding it requires hundreds of years. These documents either typically includes descriptions property in their contents (e.g., titles, tags, keywords, meta-descriptions, etc.,) or not, the processing of web documents should be performed. In addition, the web documents do not exist in unique form, and frequently exist in three common forms which are structured, semi-structured, and unstructured forms. Web documents could be image, text files, videos, xml files, etc., and these contents are varied based on the type of its content. The process of discovering knowledge and expressing useful content from large raw data is called data mining. In web scenario, the process of extracting knowledge from large web documents is called web mining. Hence, the main goal of web mining is to providing techniques that make the web data more convenient and efficient for applying advanced techniques (e.g., classification, clustering, association rule, indexing, etc.) [1]. In recent years, there has been an urgent need to apply web technologies to make search engines work best, and help variant users around the world to get their interested contents [2]. Web mining is categorized into three classes based on type of data that should processed which are Web Content Mining, Web Structured Mining, and Web Usage Mining as shown in the following block diagram in the figure (Fig. 1) [3].

Fig. 1 Web mining categorization [2]

A Novel Technique for Web Pages Clustering Using LSA …

729

2 Proposed System The main primary part in proposed system is latent semantic analysis of the text contents in web pages. VSM typically involves thousands of independent words, where every word is treated as a single feature. However, the dimensionality becomes very sparse with the present of all these words; VSM contain many (0’s) values, for most of the existing words that may belong to a small subset of web pages. However, the mean values of these columns become much smaller than nonzero columns, and this produces other challenges when clustering algorithm has been used with traditional similarity/dissimilarity measures. LSA helps to avoid some critical challenges in the text, like lexical semantic issues in some words such as synonymy, polysemy, and homonymy.

2.1 Latent Semantic Analysis of the Text LSA is a technique in Natural Language Processing (NLP), which uses a mathematical method called (SVD). SVD, actually, is a matrix factorization method for a given VSM, which aim to project the original dimensional space into lower dimension space commonly called (k-space). LSA assumes that words (features) that are close to each other in their meaning will appear in similar pieces of text. However, by reducing the dimension into semantic space with preserving the similarity structure among web pages, the words will appear into their related semantic concepts. By assumption WM × N is a word-web page matrix where rows (M) are web pages, and (N) columns are the result Words in these web pages, the entries of the contents are the words weights measured by using (TF-IDF), then the decomposition of the matrix W is given as a three matrices as in the following equation: L S A(W ) ≡ SV D(W ) = U V T

(1)

where • U is (M × r) matrix which represents the correlation between (M) web pages to the new projected space, which is called (semantic/concepts) space. columns of U matrix that represent left singular vectors and indicate to the degree of pertinence of  each word to its extracted concepts. • is (r × r) is a diagonal matrix and is called singular values matrix. U matrix actually contains non-negative square roots of r, and the values arranged are in decreasing order. However, by selecting top (σ ) values, the U and V T matrix transformed into (K) space where the value of (K) reflects the importance of concepts to their related words, and web pages to their concepts. • V T is (N × r) matrix and typically shows the semantic correlation of words to their extracted concepts. This matrix includes (negative and positive) values; the negative values refer to the words includes either poor pertinence degree or not

730

N. O. Alkaam et al.

related to that concept, V T columns named right singular vectors. The work shows the decomposition process of VSM matrix.

2.2 Elbow Graph Cutting (EGC) Method Elbow graph analysis is a method typically used in semantic analysis like LSA, PLSA, LDA, etc. The main goal of EGC is to finding the optimal truncated value that is used in semantic space after reduction  value (K) in LSA represents  step.  The T , to produce new transformed the new truncated value for the matrices of U V  k V kT . Elbow graph cutting value is detected when the matrices K  k = U k singular values curve changes with insignificant value as shown in the work, where the expected number of clusters can be detected significantly as shown below.

2.3 Partitioning-Based Algorithm Partitioning-based algorithms involve the process of assigning the points to its nearest centroid, which is also called (centroid-based clustering). However, these algorithms initially choose K points as representative points to initialize cluster centroids, and then every data point is assigned to the closest one. Hence, for a given dataset (D) with m , the main goal (M) data points that is presented in (N) dimensional space D = { pi }i=1 of partitioning algorithms is to find k clusters C = {C1 , C1 , . . . , Ck } by partitioning the dataset D into K clusters. In initialization step, the cluster centroids are chosen randomly, and then the process is repeated for every not assigned data points and updated cluster centroids. The process will continue until the function converges or requirements are met. However, the result clusters are present with spherical shape because the data points are assigned to its closest centroid. The centroid (mean) is defined as in the following equation: μi =

1  pj n i p ∈C j

(2)

i

where n i = |Ci | and refers to whole number of data points in cluster Ci . The main goal of portioning based algorithm is to find clusters with minimal sum of squared error (SSE). K-means, CLARA, PAM, and variation of K-means algorithm are some example of this type.

A Novel Technique for Web Pages Clustering Using LSA …

731

2.4 Partitioning Around Medoids (PAM) Algorithm PAM is commonly known as K-Medoids algorithm, and it is one of the most popular partitioning-based algorithm, which is related to K-means algorithm. PAM attempts to minimize the distance among the points within one cluster, and designates a point to be the center (centroid) of cluster. PAM typically chooses K-medoids as cluster centers, and mostly works with Manhattan Norm (I) to calculate the distance among data points. Hence, to calculate the distance between two web pages (p, q) by using Manhattan Distance is as follow: Dis([ p1 , p2 , . . . , pm ], [q1 , q2 , . . . , qm ]) =

m  

pj − qj



(3)

j=1

One of the most difficult questions in partitioning-based clustering algorithm is how many clusters can be found (correct number of clusters). However, there are some useful tools that could be used to know the correct number of clusters by considering the optimal evaluation values that correspond to the cluster number. One of most common tool used with K-means and its variation is Silhouette Coefficient measure. The following algorithm describes the works of K-Medoid clustering algorithm which is partitioned by the data points around the Medoids, and use greedy search technique to faster the searching process.

732

N. O. Alkaam et al.

2.5 Clustering Results Evaluation To evaluate the clustering results, either there are some significant measures used with whole algorithm or certain type of algorithms. However, there are two measures used in this paper (HC) and (SC): • Calinski–Harabasz Criterion (CH) This measure is best suited for partitioning clustering algorithm, where user defines the number of clusters. CH measure is some time called variance ratio criterion (VRC) and is used to measure the cluster validity depending on the average between and within cluster sum of square, and can be calculated as in the following equation: C Hk =

SS B (N − K ) . SSw (K − 1)

(4)

where SS B is sum of squares between clusters. SSW is sum of squares within clusters. K is the number of clusters, N is the number of instances.

SSw =

k      P − µ2  i

(5)

i=1 P∈ci

SS B =

k    |Ci |µi − µ2 

(6)

i=1

• Silhouette Coefficient (SC) SC can be defined as a measure for both cohesion and separation of clusters. SC measures the variation between the average distances of data points in their closest cluster to others. SC values ranged from (−1, to 1), where large positive values refer to high separating clustering and point is in optimal cluster, while lower values refer bad clustering results or separation. SC can be calculated based on the following equation: Si =

m MOut (Pi ) − Min (Pi )  m  Max MOut (Pi ), Min (Pi )

(7)

A Novel Technique for Web Pages Clustering Using LSA …

733

3 Experimental Results The implementation of the proposed system, starting from preprocessing of web pages in real SGML documents, actually these documents contain massive instruction lines as shown in the following figure (Fig. 1), where every web page involves around (300–1000) lines, and the useful text content arise only in 5–10 lines. Removing a lot of SGML content could face several challenges. However, the best way is to identify the significant SGML tags. In this paper, only three tags are considered (Title, Paragraphs in Body