Recent Advances in Internet of Things and Machine Learning: Real-World Applications (Intelligent Systems Reference Library, 215) 3030901181, 9783030901189

Citation preview

Intelligent Systems Reference Library 215

Valentina E. Balas Vijender Kumar Solanki Raghvendra Kumar   Editors

Recent Advances in Internet of Things and Machine Learning Real-World Applications

Intelligent Systems Reference Library Volume 215

Series Editors Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland Lakhmi C. Jain, KES International, Shoreham-by-Sea, UK

The aim of this series is to publish a Reference Library, including novel advances and developments in all aspects of Intelligent Systems in an easily accessible and well structured form. The series includes reference works, handbooks, compendia, textbooks, well-structured monographs, dictionaries, and encyclopedias. It contains well integrated knowledge and current information in the field of Intelligent Systems. The series covers the theory, applications, and design methods of Intelligent Systems. Virtually all disciplines such as engineering, computer science, avionics, business, e-commerce, environment, healthcare, physics and life science are included. The list of topics spans all the areas of modern intelligent systems such as: Ambient intelligence, Computational intelligence, Social intelligence, Computational neuroscience, Artificial life, Virtual society, Cognitive systems, DNA and immunity-based systems, e-Learning and teaching, Human-centred computing and Machine ethics, Intelligent control, Intelligent data analysis, Knowledge-based paradigms, Knowledge management, Intelligent agents, Intelligent decision making, Intelligent network security, Interactive entertainment, Learning paradigms, Recommender systems, Robotics and Mechatronics including human-machine teaming, Self-organizing and adaptive systems, Soft computing including Neural systems, Fuzzy systems, Evolutionary computing and the Fusion of these paradigms, Perception and Vision, Web intelligence and Multimedia. Indexed by SCOPUS, DBLP, zbMATH, SCImago. All books published in the series are submitted for consideration in Web of Science.

More information about this series at https://link.springer.com/bookseries/8578

Valentina E. Balas · Vijender Kumar Solanki · Raghvendra Kumar Editors

Recent Advances in Internet of Things and Machine Learning Real-World Applications

Editors Valentina E. Balas Department of Automatics and Applied Software Aurel Vlaicu University of Arad Arad, Romania

Vijender Kumar Solanki Department of Computer Science and Engineering CMR Institute of Technology Hyderabad, Telangana, India

Raghvendra Kumar Department of Computer Science Engineering GIET University Gunupur, India

ISSN 1868-4394 ISSN 1868-4408 (electronic) Intelligent Systems Reference Library ISBN 978-3-030-90118-9 ISBN 978-3-030-90119-6 (eBook) https://doi.org/10.1007/978-3-030-90119-6 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 This work is subject to copyright. All rights are solely and exclusively licensed 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 Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

The purpose of this edited book is to inform and educate its audience about methods, technologies, and solutions related to IoT and machine learning. The Edited Book is mainly used at Basic and Intermediary Levels for Computer Science Postgraduate Students, Researchers, and Practitioners. It presents multidisciplinary and dynamic findings in the broad fields of IoT and machine leaning. It will cover a broad range of topics like Intelligent Tourism System, Emergency Alert System, Ambulance Monitoring System, Improving Health Care, Energy Saving and Agriculture, and their interaction with computation methods and applications. The wide variety in topics it presents offers readers multiple perspectives on a variety of disciplines. The book is organized into 25 chapters. Chapter 1 proposes a crossover design framework summoning understudy facial feeling acknowledgment, eye stare checking, and head developments distinguishing pieces of proof-based breaking-down powerful understudy commitment/conduct in study hall and toward a particular course at e-learning stages. Our proposed design utilizes extraction calculations like Principal Component Analysis (PCA) for facial feeling acknowledgment, Haar Cascade for student recognition, and Local Binary Patterns for perceiving head developments. For AI approach and to give precise outcomes, we propose OpenCV. In this way dependent on the understudies, input weightage is assigned; in light of the last score, we do contrast and the edge esteem. On the off chance that the understudies’ consideration esteem is more noteworthy than the limit esteem, hypothesis-based redemption is suggested. On the off chance that the understudies’ consideration esteem is lesser than the limit esteem, video, brilliant class, persuasive video-based liberation is suggested. Experimental outcomes are actualized using the PyCharm device, an IoT-based machine learning system that can identify and monitor the learner’s emotions based on the facial expressions of the student in an e-learning environment that can determine the concentration level of learners during the session for better content delivery. Chapter 2 recognizes effectively a few initial nourishing fixings in nourishment that can profit the restoration of those ailments, and the IoT-based system records the data of the patients periodically. The Machine Learning (ML) technique assists the statistical and texture features of the patient’s diet sheet-based nutrient loss for the v

vi

Preface

particular disease are monitored and notified personally. These exploratory outcomes exhibit the adequacy of applying information mining in choosing wholesome fixings in nourishment for ailment examination. Chapter 3 discusses the system to monitor the lid of the manhole and alerts the Municipal Corporation whenever the position of the lid is changed. Similarly, the condition of the sewers are monitored periodically, and an alert information regarding the issue in the sewer system and its location details are sent to Municipal Corporation to take necessary actions whenever there is an emergency situation. The sensor details were periodically recorded and the machine learning algorithm is used to set the threshold of the sensor values with the help of Recurrent Neural Network using the Long Short-Term Memory (LSTM) algorithm. Chapter 4 discusses the process of informing the hospital to attend to the injured people, informing the police station to indicate the accident zone, and claiming the insurance for the vehicle which becomes a difficult task. The proposed Deep Forest algorithm is the Machine Learning algorithm which recognizes the accident in the most accurate way with enhanced hyper-parameters. In order to speed up this process, a smart accident recognition and rescue system is introduced which performs these tasks in an easy way. Chapter 5 presents a brief discussion on blockchain and its architecture. Specifically, we describe the classification and consensus mechanisms that are very important in the blockchain network. We also detail the transaction system in the blockchain. Later on, we discuss the fundamentals of Industry 4.0. Furthermore, we provide an explicit discussion on the usage of blockchain in the industry. Finally, we point out some issues that can create possible future research areas in blockchain for industries. Chapter 6 briefly reviews some enhanced schemes and recently proposed security mechanisms as countermeasures to various cyber-attacks. Recent references are primarily used to present smart healthcare privacy and security issues. The issues are laid out briefly based on the different architecture layers, various security attacks, and their corresponding proposed solutions along with other facets of smart health such as Wireless Body Area Network (WBAN) and healthcare data. Chapter 7 discusses settling the imbalance and forgery; a study has been made to identify the fake reviews and filter it out from the user’s sight, such that they can’t be manipulated as well as the particular company’s reputation won’t be at stake. This system majorly works on the basis of identifying fake reviews by user extracting the multiple review of single user from reviewers. Chapter 8 implements COVID-Net, an Internet of Things (IoT) hand-accessible Machine Learning (ML) network mode to identify COVID-19 cases using the chest X-ray images. This investigation utilizes the COVID cases database images from an open source that is accessible to the general public, employs Deep Neural Network (DNN) architecture for the detection, and analyzes the disease using Machine Learning (ML) e-network-based COVID-Net system. The COVID-Net data collection is referred to as COVIDx which includes 13,800 chest X-ray photos of 13,725 patients from 3 open-access data sources, one of which we launched is also addressed.

Preface

vii

Chapter 9 can get the exact location of the perpetrator so that the police can quickly locate the perpetrator, and the crash can be avoided quickly and save the family, prosecute the responsible. We will remotely track the women’s details through a map in order to use the IoT platform. This aims to minimize the slaughter and embarrassments of women. For machine learning, we used the regression analysis framework and we used this training sample to inform our algorithm about different danger and non-danger situations, as well as how to act on them. Then, based on actual evidence, a prediction is made as to whether or not danger exists. Chapter 10 presented to arrange the leaf spot, rhizome rot, powdery mildew diseases and leaf blotch diseases which are infected in the Tea-leaf plantation. The color transformed images are sharply segmented using the Watershed Transformation algorithm. Multiclass SVM classifier classifies the Tea-leaf diseases using gradient feature values of the tea-leaf images. The real-time IoT-based Machine Learning (ML) techniques help to classify the tea leaf as either a healthy or infected leaf from the pre-processed image captured using the mobile camera or video camera. In this paper, the tea-leaf infection is detected using the hybrid filter that comprises a median filter and Gaussian filter for the purpose of edge detection, color enhancement, gray pixel resizing, and noise reduction. Finally, the performance is evaluated in terms of accuracy, and it is found that the presented system is realizable and provides better classification than earlier techniques. Chapter 11 discusses preventing the theft of vehicles parked at the car parking; a two-way screening process is implemented to let the vehicles enter into the parking lots. This system also lets the users choose the parking slot based on their flexibility. This system also facilitates in monitoring the availability of the face mask on the people in order to prevent the spread of the COVID-19 disease. Chapter 12 explains the primary objective of this work which is to give detailed background knowledge of Machine Learning (ML) algorithms and explores the role of ML, Deep Reinforcement Learning (DRL), and Artificial Intelligence (AI) in the development of the smart city. The paper presents a comprehensive overview of the smart city concept and focuses on different privacy solutions in the smart city. Further, the paper highlights the role of ML in various smart city applications such as intelligent transportation system, smart grids, health care, cyber-security, and supply chain management. Finally, the work enumerates some future research directions to guide further advancements in the area. Chapter 13 aims to give a conclusive acquaintance of Artificial Intelligence, Machine Learning, and Deep Reinforcement Learning approaches that can perform an essential role in the formation of a smart city. Finally, the paper highlights the complex problems that occur in a smart city and a solution to these problems using Internet of Things, Artificial Intelligent, Machine Learning, and Deep Reinforcement Learning techniques. Chapter 14 proposes the work that shows a less time-saving, slightly more areas, and higher-speed adder efficiency between the 16-bit adders with the Brent-Kung adder. It is a robust system for machine learning-based optimal adder analysis that connects the prefix adder design synthesis to the final physical design. The work proposed tests that are carried out and Xilinx 14.7 simulations are conducted.

viii

Preface

Chapter 15 discusses that we can append the data of the new vehicle into the database with a timestamp. The details of both registered and non-registered vehicles are recorded that include vehicle number and some personal details of the vehicle owner such as license number, insurance, and mobile number. It also intimates the registered vehicle owners regarding the renewal of the pass-through papers such as insurance, driving license renewal, updation of mobile number, Aadhaar number, and vehicle number with the GPS tracker using the mobile-based text message. This machine learning network helps the vehicle owner to know each and every movement of their vehicle and acts as personal reminder. Chapter 16 proposes an efficient method for IVUS image classification using NonNegative Matrix Factorization (NNMF), and various Support Vector Machine (SVM) kernels are presented in this study. The input IVUS images are given to NNMF for feature extraction and stored in the feature database. Finally, SVM kernels like linear, polynomial, quadratic, and Radial Basis Function (RBF) are used for the prediction and classification of coronary artery lesions; IVUS-based ML algorithms show good diagnostic performance for identifying ischemia-producing lesions. An IoT-based alert is given to the patient’s database Cloud that has information of self or blood relation to alert via messages in case of emergency artery disease using wearable sensors. The system produces a classification accuracy of 94% by using NNMF and different SVM kernels. Chapter 17 proposes a system that is based on flex sensor with a controller. This circuit is installed in the waist belt, which can monitor the breathing pattern of the patient continuously. Due to the variation in the value of the flex sensor, the serial monitor displays the live status of the patients. At times of any abnormalities, an SMS is triggered to the kin and friends of the patients. The suggested Respiratory Rate Monitoring System was checked and assessed based on satisfactory findings. Chapter 18 proposes a new approach of reducing the cost of a circuit that is dealt with by representing the circuit as a non-overlapping graph called the planar graph. Boolean function representation as a planar graph is used to bring down the cost of IoT circuits. Chapter 19 proposes a marketing association comprising multi-grade subjects to the consumption of manpower (wastages) because of strategy choices with high or low wearing down rate is thought of. The time to hire is evaluated in this paper using univariate max recruiting policy for a third-level program with waste created by its policy decision. The distribution of the threshold for all three degrees is considered to be linear and the inter-decision periods are geometrical. Numerical diagrams help the theoretical findings. Chapter 20 proposes approach the real-world that constructed to perform cyberattacks. Here, the machine learning-based abnormality recognition approach can execute well in identifying these attacks. Extending these algorithms can assist to enhance the security of the system also. In order to measure the efficacy, we measured the results using the respective criteria. Chapter 21 develops an application for the needs of businesses and organizations where expenditures must be made by the workers and accepted by the members of the business with the access and authority to approve it. Using our service, customers can

Preface

ix

more efficiently handle their expenses. Not only can this program help users handle their expenditures but it can also help marketing executives organize campaigns according to user needs. With advanced features such as auto scan for receipts, it stands out as a quick but effective solution for controlling and maximizing the expenses of the company. This makes the analysis and cost flow smoother, with the recording of expenses offline without Internet will not be a burden. Chapter 22 discusses the perception of students regarding the counseling app (chatbot) which can be implemented in colleges/universities to deal with three aspects of life, namely Personal life with chatbot, Professional life with chatbot, and Genera life with chatbot. A Questionnaire was developed which measures the effect of the chatbot while counseling the students on the above-mentioned aspects of life. The results are measured and well presented in the study. So, it is recommended to colleges and universities to implement chatbots in their premises for the wellbeing of their students. Chapter 23 discusses the current study on the perception of students regarding the counseling app (chatbot) which can be implemented in colleges/universities to deal with three aspects of life, namely Personal life with chatbot, Professional life with chatbot, and Genera life with chatbot. A Questionnaire was developed which measures the effect of the chatbot while counseling the students on the abovementioned aspects of life. The results are measured and well presented in the study. So, it is recommended to colleges and universities to implement chatbots in their premises for the wellbeing of their students. Chapter 24 presents an edge-based architecture to quickly deploy a deep learningbased IDS to edge network devices regardless of the heterogeneity in hardware and deep learning model configurations. To demonstrate the effectiveness of our proposal, we also analyze various performance indicators of the architecture, deployment process, and deep learning models. Chapter 25 introduces a framework used for practically evaluating the IDS models in real time. The proposed framework also supports quickly generating different networking attacks that are similar to real scenarios. Indexes of performance (resource consumption, throughput, and detection performance) of eight attacking scenarios are recorded and analyzed to demonstrate our proposal’s effectiveness. We are sincerely thankful to Almighty for supporting and standing by at all times with us, be it good or tough times, and giving ways to conceded us. Starting from the Call for chapters till their finalization, all the editors have given their contributions amicably, which is a positive sign of significant team work. The editors are sincerely thankful to all the members of Springer (India) Private Limited, especially Prof. (Dr.) Lakhmi C. Jain, S. Tigner, and Varsha Prabakaran for providing constructive inputs and allowing the opportunity to edit this important book. We are equally thankful to a reviewer who hails from different places in and around the globe shared their support and stand firm toward quality chapter submission. We have kept the rate of acceptance as low as 16% to ensure the quality of work submitted by the author. The aim of this book is to support the computational studies at the research and postgraduation levels with open problem-solving technique. We are confident that it will bridge the gap for them by supporting novel solutions to support in their

x

Preface

problem-solving. At the end, the editors have taken utmost care while finalizing the chapters of the book, but we are open to receive your constructive feedback, which will enable us to carry out necessary points in our forthcoming books. Arad, Romania Hyderabad, India Gunupur, India

Valentina E. Balas Vijender Kumar Solanki Raghvendra Kumar

About This Book

Internet of Things (IoT)-related applications are gaining much attention; with more and more devices getting connected, they become the potential components of some smart applications. Thus, a global enthusiasm has sparked over various domains such as health, agriculture, energy, security, and retail. So, in this book, the main objective is to capture this multifaceted nature of IoT and Machine Learning in one single place. Each of the chapters of this book will cover a domain that is significantly impacted by the growth of soft computing. According to the contribution of each chapter, the book will also provide a future direction for IoT and Machine Learning research. The objectives of this book are to identify different issues, suggest feasible solutions to those identified issues, and enable researchers and practitioners from both academia and industry to interact with each other regarding emerging technologies related to IoT and Machine Learning. In this book, we look for novel chapters that recommend new methodologies, recent advancement, system architectures, and other solutions to prevail over the limitations of IoT and Machine Learning.

xi

xii

About This Book

Key Features 1. 2. 3. 4. 5. 6.

Covering the potential impactful Growth of IoT System. It will bring the interesting work in the line of IoT and its associated boundaries. It will focus on Machine Learning with recent applications. Machine learning and allied applications. Application fruitful to Industry in IoT with machine learning approaches. Machine learning models, technologies, and solutions.

Contents

Part I 1

2

3

Internet of Things

Concentration Level of Learner Using Facial Expressions on e-Learning Platform Using IoT-Based Pycharm Device . . . . . . . . . Karlapudi Rahul Sai, Bhumireddy Sohith Reddy, and S. Vijaya Kumar 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 System Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IoT-Based Machine Learning System for Nutritional Ingredient Analyzer for Food . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Thilagavathy, Tadavarthi Rishi, Veeram Deepak Reddy, and Sudesh Nimmagadda 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 Data-Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2 Statistical Algorithm (SA) . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Secured Manhole Management System Using IoT and Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. Santhana Krishnan, A. Sangeetha, D. Abitha Kumari, N. Nandhini, G. Karpagarajesh, K. Lakshmi Narayanan, and Y. Harold Robinson

3

4 5 6 6 7 7 8 9

10 11 13 13 13 15 16 17 19

xiii

xiv

Contents

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

5

Internet of Things Based Smart Accident Recognition and Rescue System Using Deep Forests ML Algorithm . . . . . . . . . . . . K. Lakshmi Narayanan, Y. Harold Robinson, Rajkumar Krishnan, C. Ramasamy Sankar Ram, R. Santhana Krishnan, R. Niranjana, and A. Essaki Muthu 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Revolutionizing the Industrial Internet of Things Using Blockchain: An Unified Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. K. M. Bahalul Haque, Bharat Bhushan, Md.Rifat Hasan, and Md.Oahiduzzaman Mondol Zihad 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Blockchain Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 The Block Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Blockchain Classification . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 Consensus Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.5 Blockchain Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.6 Transaction Stages in Blockchain . . . . . . . . . . . . . . . . . . 5.3 Blockchain for Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Why Does the Industry Need Blockchain . . . . . . . . . . . . . . . . . . . . 5.5 Blockchain Apllications in Industry . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Automation of Supply Chain . . . . . . . . . . . . . . . . . . . . . . 5.5.2 Blockchain-Based Security and Privacy . . . . . . . . . . . . . 5.5.3 Tracking and Tracing Product Manufacturing Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.4 Payment Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.5 Cloud and Edge Computing . . . . . . . . . . . . . . . . . . . . . . . 5.6 Future Issues and Research Directions . . . . . . . . . . . . . . . . . . . . . . . 5.6.1 Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.2 Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.3 Resource Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.4 Scalability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6.5 Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20 23 23 25 28 29 31

32 35 35 37 39 40 43

44 45 45 46 47 48 50 52 53 54 55 56 56 56 57 57 58 58 59 59 59 60

Contents

xv

5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Attacks and Countermeasures in IoT Based Smart Healthcare Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. K. M. Bahalul Haque, Bharat Bhushan, Afra Nawar, Khalid Raihan Talha, and Sadia Jeesan Ayesha 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Smart City Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Smart City Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 Pillars of Smart City . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Healthcare in Smart City . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Smart Healthcare Applications . . . . . . . . . . . . . . . . . . . . . 6.4 Smart Healthcare Privacy and Security . . . . . . . . . . . . . . . . . . . . . . 6.4.1 Denial of Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.2 Spyware and Worm Attacks . . . . . . . . . . . . . . . . . . . . . . . 6.4.3 Ransomware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.4 Eavesdropping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.5 Man In The Middle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.6 Side Channel Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5 Challenges and Future Research Direction of Smart City Healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.1 Wearable Technology Challenges . . . . . . . . . . . . . . . . . . 6.5.2 Smart Healthcare Data Challenges . . . . . . . . . . . . . . . . . 6.5.3 Recommendations and Opportunities . . . . . . . . . . . . . . . 6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Part II 7

60 60 67

68 69 70 71 71 72 75 76 78 78 79 79 80 80 80 81 81 83 84

Machine Learning

Online Product Review Monitoring System Using Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 J. Madhumathi, R. Aishwarya, V. Vedha Pavithra, and Sandra Johnson 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 7.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 7.3 Proposed Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 7.4 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 7.5 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 7.6 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 7.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

xvi

8

9

Contents

Deep Learning Analysis for COVID 19 Using Neural Network Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Vijaya Baskar, V. G. Sivakumar, S. P. Vimal, and M. Vadivel 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Implementation of COVID-19 Using Deep Learning Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Network Design Prototyping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Model Creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Dataset Exploration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.6 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.7 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Machine Learning Approach to Design and Develop a BEACON Device for Women’s Safety . . . . . . . . . . . . . . . . . . . . . . . . . . S. Srinivasan, P. Muthu Kannan, and R. Kumar 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Study Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Methodology and Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10 Tea Plant Leaf Disease Identification Using Hybrid Filter and Support Vector Machine Classifier Technique . . . . . . . . . . . . . . . . S. Prabu, B. R. Tapas Bapu, S. Sridhar, and V. Nagaraju 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4 Module Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.4.1 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5 Image Processing Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.1 Image Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.2 Hybrid Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.3 Median Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.4 Gaussian Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.6 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.7 Performance Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.7.1 Accuracy Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.8 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

103 104 105 105 106 107 107 107 109 109 111 112 112 113 113 114 115 117 118 118 119 119 119 120 121 121 121 122 122 125 126 126 127 127

Contents

11 Machine Learning Based Efficient and Secured Car Parking System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R. Santhana Krishnan, K. Lakshmi Narayanan, S. T. Bharathi, N. Deepa, S. Mathumitha Murali, M. Ashok Kumar, and C. R. T. Suria Prakash 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Proposed Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Machine Learning Approaches for Smart City Applications: Emergence, Challenges and Opportunities . . . . . . . . . . . . . . . . . . . . . . . Sonam Mehta, Bharat Bhushan, and Raghvendra Kumar 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Background Knowledge of Machine Learning . . . . . . . . . . . . . . . . 12.2.1 Reinforcement Learning (RL) . . . . . . . . . . . . . . . . . . . . . 12.2.2 Markov Decision Process (MDP) . . . . . . . . . . . . . . . . . . 12.2.3 Dynamic Programming (DP) . . . . . . . . . . . . . . . . . . . . . . 12.2.4 Deep Q Network (DQN) . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.5 Monte Carlo (MC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2.6 Temporal Difference (TD) Methods . . . . . . . . . . . . . . . . 12.2.7 Bayesian Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3 Smart City Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3.1 Introduction of Smart City . . . . . . . . . . . . . . . . . . . . . . . . 12.3.2 Privacy Violations in Smart City . . . . . . . . . . . . . . . . . . . 12.3.3 Driving Forces for Smart Cities . . . . . . . . . . . . . . . . . . . . 12.4 ML Based Solutions in Smart City . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4.1 Intelligent Transportation System (ITS) . . . . . . . . . . . . . 12.4.2 Smart Grids (SGs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4.3 Health Care . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4.4 Cyber Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4.5 Supply Chain Management (SCM) . . . . . . . . . . . . . . . . . 12.5 Conclusion and Future Research Directions . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Machine Learning and Deep Learning Models for Privacy Management and Data Analysis in Smart Cites . . . . . . . . . . . . . . . . . . Trisha Bhowmik, Abhishek Bhadwaj, Avinash Kumar, and Bharat Bhushan 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Smart City Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.1 Pillars of a Smart City . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.2 Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.3 Characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xvii

129

130 131 132 136 143 143 147 147 149 150 150 150 151 151 152 152 152 153 155 155 156 156 156 157 158 158 159 159 165

165 166 168 169 171

xviii

Contents

13.3 Machine Learning Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.1 Supervised Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.2 Unsupervised Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.3 Reinforcement Learning . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4 ML for Smart Cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.1 Intelligent Transportation System . . . . . . . . . . . . . . . . . . 13.4.2 Cybersecurity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.3 Smart Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.4 Healthcare Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5 Privacy-Enhancing Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.1 Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.2 Shuffling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.3 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.4 Encryption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.5 Blockchain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.6 K-Anonymity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.7 Onion Routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.5.8 Zero-Knowledge Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6 Future Research Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6.1 Connecting Technologies . . . . . . . . . . . . . . . . . . . . . . . . . 13.6.2 Management of Water and Waste . . . . . . . . . . . . . . . . . . 13.6.3 Construction and Building Technologies . . . . . . . . . . . . 13.6.4 Use of Renewable Resources . . . . . . . . . . . . . . . . . . . . . . 13.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

172 172 173 173 174 174 175 175 176 178 178 178 179 179 179 179 180 180 181 181 182 182 182 183 183

Part III Applications 14 FPGA Based Implementation of Brent Kung Parallel Prefix Adder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . K. Gunasekaran, D. Muthukumaran, K. Umapathy, and S. A. Yuvaraj 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 Study Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Vehicle Entry Management System Using Image Processing . . . . . . . R. Vallikannu, Krishna kanth, L. SaiPavan Kumar, Monisha, and Karthik 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4.1 Image Capture from Camera . . . . . . . . . . . . . . . . . . . . . .

191

191 193 195 197 197 199

200 200 202 203 203

Contents

15.4.2 RGB to Grey Conversion . . . . . . . . . . . . . . . . . . . . . . . . . 15.4.3 Blurring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4.4 Edge Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4.5 Finding and Drawing Contours . . . . . . . . . . . . . . . . . . . . 15.4.6 License Plate Detection and Text Extraction . . . . . . . . . 15.4.7 Comparing Vehicle Number with Database . . . . . . . . . . 15.4.8 Website Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4.9 Hardware Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.7 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 A Non-negative Matrix Factorization for IVUS Image Classification Using Various Kernels of SVM . . . . . . . . . . . . . . . . . . . . S. P. Vimal, M. Vadivel, V. Vijaya Baskar, and V. G. Sivakumar 16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2 Methods and Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.2.1 NNMF Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . 16.2.2 SVM Kernels Classification . . . . . . . . . . . . . . . . . . . . . . . 16.3 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Novel Approach to Monitor the Respiratory Rate for Asthma Patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. G. Sivakumar, S. P. Vimal, M. Vadivel, and V. Vijaya Baskar 17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Flow Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Representation of Boolean Function as a Planar Graph to Reduce the Cost of a Circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vinitha Navis Varuvel, A. Kanchana, and D. Samundeeswari 18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2 Material and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2.1 A. Boolean Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2.2 Planar Graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.2.3 C. Java Scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.3 A Three Variable Boolean Function . . . . . . . . . . . . . . . . . . . . . . . . . 18.4 A Four Variable Boolean Function . . . . . . . . . . . . . . . . . . . . . . . . . . 18.5 Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xix

203 203 204 204 205 205 205 207 207 209 209 210 211 212 212 213 213 213 214 214 217 217 219 219 220 221 222 225 225 226 226 226 226 226 229 233 234 234

xx

Contents

19 A Man Power Model Forthree Grade System with Univariate Policy of Recruitment Using Geometric Process for Inter Decision Times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Samundeeswari, A. Kanchana, and Vinitha Navis Varuvel 19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Denial of Service Attack in Wireless Sensor Networks . . . . . . . . . . . . D. Jeyamani Latha, P. Akshaya, M. M. Nilavarasi, N. J. Raghel, V. Utharapathi, V. M. Sanjaykumar, and P. Kishore 20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3.1 Malware Data Visualization . . . . . . . . . . . . . . . . . . . . . . . 20.3.2 Random Forest Algorithm . . . . . . . . . . . . . . . . . . . . . . . . 20.4 Integration of Web App . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Android Application for Business Expense Management . . . . . . . . . . S. Surekha, G. Swetha, Sri Ram Gowd Vuppala, Arya Vishnu Thotakura, Tejasvi Dasari, R. Imayavaramban, and T. C. Jermin Jeaunita 21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Background Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.4 Proposed Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5 Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.1 Users, Roles and Departments . . . . . . . . . . . . . . . . . . . . . 21.5.2 Recording of Expenditure . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.3 Creation of Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.4 Creation of Trips . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.5 Advance Payments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.6 Types of Approval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.7 Types of Currencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.8 Policy Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.9 Automation Using Templates . . . . . . . . . . . . . . . . . . . . . . 21.5.10 Customised Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.11 Providing Statistical Data of a Company . . . . . . . . . . . . 21.5.12 Customised Status Tracking . . . . . . . . . . . . . . . . . . . . . . . 21.5.13 Budget Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.5.14 Procurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

235 235 240 240 243

243 244 245 245 245 248 249 249 250 251

251 252 253 253 255 255 255 255 256 256 256 257 257 257 258 258 258 259 259

Contents

21.6 Technical Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.7 Future Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.7.1 Forecasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.7.2 Receipt Scanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Student Perception Regarding Chatbot for Counselling in Higher Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shivani Agarwal, Nguyen Thi Dieu Linh, and Gloria Jeanette Rincón Aponte 22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1.1 Background About Chatbot . . . . . . . . . . . . . . . . . . . . . . . 22.1.2 Application of Chatbot in Business . . . . . . . . . . . . . . . . . 22.1.3 Scope of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3.1 Participants and Procedure . . . . . . . . . . . . . . . . . . . . . . . . 22.3.2 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3.3 Objectives of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.3.4 Hypothesis of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . 22.4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.6 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Empirical Performance Evaluation of Machine Learning based DDoS Attack Detections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bao-Sam Tran, Thi-Huyen Ho, Thanh-Xuan Do, and Kim-Hung Le 23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.3 The proposed framework architecture . . . . . . . . . . . . . . . . . . . . . . . 23.4 Evaluation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.4.1 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.4.2 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Towards Remote Deployment for Intrusion Detection System to IoT Edge Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xuan-Thanh Do and Kim-Hung Le 24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.3 The Deployment Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xxi

259 259 259 260 260 261 263

263 264 265 266 266 271 271 271 271 272 272 279 280 280 283 284 285 287 289 289 290 290 297 298 301 301 303 305 305

xxii

Contents

24.3.2 Deployment Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.3.3 Feature Extractor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.4.1 Running Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 A Real-Time Evaluation Framework For Machine Learning-Based IDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anh-Hao Vu, Minh-Quan Nguyen-Khac, Xuan-Thanh Do, and Kim-Hung Le 25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.2 Available Datasets and Its Limitations . . . . . . . . . . . . . . . . . . . . . . . 25.3 The Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.3.2 Network Traffic Generator . . . . . . . . . . . . . . . . . . . . . . . . 25.3.3 Feature Extractor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.3.4 IDS Model Evaluating . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.4 Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.4.1 Running Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.4.2 Index of Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

306 307 309 309 310 314 315 317

318 319 320 320 323 323 324 324 324 325 325 328 328

About the Editors

Valentina E. Balas, Ph.D. is currently Full Professor in the Department of Automatics and Applied Software at the Faculty of Engineering, “Aurel Vlaicu” University of Arad, Romania. She holds a Ph.D. in Applied Electronics and Telecommunications from Polytechnic University of Timisoara. Dr. Balas is the author of more than 270 research papers in refereed journals and International Conferences. Her research interests are in Intelligent Systems, Fuzzy Control, Soft Computing, Smart Sensors, Information Fusion, Modeling, and Simulation. She is the Editor-in-Chief of International Journal of Advanced Intelligence Paradigms (IJAIP) and International Journal of Computational Systems Engineering (IJCSysE), Editorial Board member of several national and international journals, and is evaluator expert for national and international projects. She served as General Chair of the International Workshop Soft Computing and Applications in seven editions 2005–2016 held in Romania and Hungary. Dr. Balas participated in many international conferences as Organizer, Session Chair, and member in International Program Committee. Now, she is working in a national project with EU funding support: BioCell-NanoART = Novel Bio-inspired Cellular Nano-Architectures—For Digital Integrated Circuits, 2M Euro from National Authority for Scientific Research and Innovation. She is a member of EUSFLAT, ACM, and a Senior Member IEEE, member in TC—Fuzzy Systems (IEEE CIS), member in TC—Emergent Technologies (IEEE CIS), and member in TC—Soft Computing (IEEE SMCS). Dr. Balas was vice-president (Awards) of IFSA International Fuzzy Systems Association Council (2013–2015) and is a Joint Secretary of the Governing Council of Forum for Interdisciplinary Mathematics (FIM), A Multidisciplinary Academic Body, India. Vijender Kumar Solanki, Ph.D. is an Associate Professor in Department of Computer Science & Engineering, CMR Institute of Technology (Autonomous), Hyderabad, TS, India. He has more than 11 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, Tamil Nadu, India, and Institute of Technology and Science, Ghaziabad, UP, India. He has attended an orientation program at UGC-Academic Staff College, xxiii

xxiv

About the Editors

University of Kerala, Thiruvananthapuram, Kerala, and a Refresher course at Indian Institute of Information Technology, Allahabad, UP, India. He has authored or coauthored more than 25 research articles that are published in journals, books, and conference proceedings. He has edited or co-edited 4 books in the area of Information Technology. He teaches graduate and postgraduate-level courses in IT at ITS. 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 Editor in International Journal of Machine Learning and Networked Collaborative Engineering (IJMLNCE) ISSN 2581-3242, Associate Editor in International Journal of Information Retrieval Research (IJIRR), IGI-GLOBAL, USA, ISSN: 2155-6377|E-ISSN: 2155-6385 also serving editorial board members with many reputed journals. He has guest-edited many volumes, with IGI-Global, USA, InderScience, and many more reputed publishers. For more details, please visit https:// www.sites.google.com/site/vijenderkrsolanki. Raghvendra Kumar is working as Associate Professor in Computer Science and Engineering Department at GIET University, India. He received B.Tech, M.Tech, and Ph.D. in Computer Science and Engineering, India, and is Postdoc Fellow from Institute of Information Technology, Virtual Reality and Multimedia, Vietnam. He serves as Series Editor Internet of Everything (IOE): Security and Privacy Paradigm, Green Engineering and Technology: Concepts and Applications, publishes by CRC press, Taylor & Francis Group, USA, and Bio-Medical Engineering: Techniques and Applications, publishes by Apple Academic Press, CRC Press, Taylor & Francis Group, USA. He also serves as acquisition editor for Computer Science by Apple Academic Press, CRC Press, and Taylor & Francis Group, USA. He has published a number of research papers in international journals (SCI/SCIE/ESCI/Scopus) and conferences including IEEE and Springer as well as served as organizing chair (RICE— 2019, 2020), volume editor (RICE—2018), keynote speaker, session chair, Co-chair, publicity chair, publication chair, advisory board, Technical Program Committee members in many international and national conferences and serve as guest editors in many special issues from reputed journals (Indexed By Scopus, ESCI, and SCI). He also published 13 chapters in edited books published by IGI Global, Springer, and Elsevier. His research areas are Computer Networks, Data Mining, Cloud Computing and Secure Multiparty Computations, Theory of Computer Science, and Design of Algorithms. He authored and edited 23 computer science books in the field of Internet of Things, Data Mining, Biomedical Engineering, Big Data, Robotics, and IGI Global Publication, USA, IOS Press Netherland, Springer, Elsevier, and CRC Press, USA.

Part I

Internet of Things

Chapter 1

Concentration Level of Learner Using Facial Expressions on e-Learning Platform Using IoT-Based Pycharm Device Karlapudi Rahul Sai, Bhumireddy Sohith Reddy, and S. Vijaya Kumar Abstract India consistently a significant job in the worldwide instruction. India is constantly considered as one of the biggest system of instructive establishments. Albeit a few imperatives are been related with our learning framework. We attempt to give a similar substance of educating to all understudies with various entomb individual aptitudes. The most significant factor is absence of understudy inspiration towards a subject, course and so forth. Versatile learning is an instructive strategy that uses PCs as an intuitive educating gadget. In existing most instructive operators don’t screen commitment expressly, yet rather accept commitment and adjust their connection dependent on the understudy’s reactions to questions and undertakings. Consequently unique understudy conduct investigation is an initial move towards a mechanized instructor input device for estimating understudy commitment. In our framework, we propose a crossover design framework summoning understudy facial feeling acknowledgment, eye stare checking, head developments distinguishing pieces of proof based breaking down powerful understudy commitment/conduct in study hall and towards a particular course at e-learning stages. Our proposed design utilizes include extraction calculations like Principal Component Analysis (PCA) for facial feeling acknowledgment, Haar Cascade for student recognition and Local Binary Patterns for perceiving head developments. For AI approach and to give precise outcomes we propose Open CV. In this way dependent on the understudies input weightage is assigned, in light of the last score, we do contrast and the edge esteem. On the off chance that the understudies consideration esteem is more noteworthy than the limit esteem, hypothesis based redemption is suggested. On the off chance that the understudies consideration esteem is lesser than the limit K. R. Sai · B. S. Reddy Computer Science & Engineering, RMK Engineering College, Chennai, India e-mail: [email protected] B. S. Reddy e-mail: [email protected] S. Vijaya Kumar (B) Department of Computer Science, RMK Engineering College, Chennai, India e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_1

3

4

K. R. Sai et al.

esteem, video, brilliant class, persuasive video based liberation is suggested. Experimental outcomes are actualized using the Pycharm device, an IoT based machine learning system that can identify and monitor the learner’s emotions based on the facial expressions of the student in an e-learning environment that can determine the concentration level of learners during the session for better content delivery. Keywords Open CV · Haar cascade · Principal Component Analysis (PCA)

1.1 Introduction In a virtual learning condition, students can lose inspiration and fixation effectively, particularly in a stage that isn’t custom-made to their necessities. Our exploration depends on considering student’s conduct on a web based learning stage to make a framework ready to grouping students dependent on their conduct, and adjusting instructive substance to their necessities. As the expense of instruction (educational costs, charges and everyday costs) has soar in the course of recent decades, delayed graduation time has become a vital contributing element to the evergrowing understudy graduation. Truth be told, late investigations show that lone 50 of the in excess of 580 open four-year organizations in the United States have on-time graduation rates at or over 50% for their full-time understudies. To make school increasingly reasonable, it is along these lines essential to guarantee that a lot more understudies graduate on time through early mediations on understudies whose exhibition will be probably not going to meet the graduation measures of the degree program on schedule. A basic advance towards powerful intercession is to assemble a framework that can constantly monitor understudies’ consideration level and precisely foresee their mind-set of tuning in and dependent on that information the educating can be conveyed. Restriction is Dynamic student conduct examination is absent in the current framework and Biometric based student conduct investigation is absent in the customary and e-learning stages likewise same substance of instructing is been conveyed to all sort of understudies. The superseding motivation behind this paper was to give an outline of dynamic understudy conduct examination is an initial move towards a computerized educator input instrument for estimating understudy commitment. Our proposed framework can be applied in both customary/e-learning frameworks. In our framework, we propose a half and half engineering framework conjuring understudy facial feeling acknowledgment, eye stare checking, head developments recognizable pieces of proof based examining dynamic understudy commitment/conduct in study hall and towards a particular course at e-learning stages.

1 Concentration Level of Learner Using Facial Expressions …

5

1.2 Literature Review Mediating the Expression of Emotion in Educational Collaborative Virtual Environments: An Experimental Study Fabri, M., Moore, D.J., Hobbs, D.J. 2014: Web based educating and e-learning philosophies have risen above higher than ever after the blast of data innovation age. Subsequently, the nature of instruction and number of online students has expanded considerably. All things considered, the modernized method of e-learning makes issue that influences an understudy’s expectation to absorb information because of inaccessibility of any immediate management. Measuring the Impact of Emotion Awareness on e-learning Situations M. Feidakis, T. Daradoumis, S. Caballé and J. Conesa 2013: A teacher can give some knowledge into understudy’s fulfillment during addresses, in this way understudy’s contribution in class has direct relationship with the expert fitness of the educator. Direct management encourages learning as well as keeps the understudy synchronized with the course destinations because of moment correspondence with the educator whenever during the talk. Absence of correspondence has demonstrated that influenced understudies may encounter elevated levels of disappointment. An Infrastructure for Real-Time Interactive Distance E-Learning Environment J. Yu 2010: Characteristic input on the substance being conveyed can be taken naturally from students by utilizing their outward appearances as an instrument to gauge intriguing quality of the substance and commitment of understudy in the online talk. Eye Tracking and e-Learning: Seeing through Your Students S. Al. Hend, G. K. Remya 2014: Hend et al. contended that the information gathered from eye GPS beacons demonstrate an individual’s advantage level and the focal point of her consideration. From eye position following and such aberrant measures as obsession numbers and length, look position, and flicker rate, data can be drawn about client levels of consideration, stress, unwinding, critical thinking, learning achievement, and exhaustion. Learner Behavior Analysis through Eye Tracking I. E. Haddioui, and M. Khaldi 2011: Ismail and Mohamed incorporated eye following innovation to gauge and examine student practices on an e-learning stage. They concentrated on the fascinating pieces of courses that reflect client feeling consideration, stress, unwinding, critical thinking, and weakness. TFacial Expression Recognition utilizing Neural Network—An Overview Pushpaja V. Saudagare, D.S. Chaudhari 2012: Pushpaja V. Saudagare and D.S Chaudhari4 approached with a strategy to recognize articulation from feelings through neural systems. It audits the different strategies of articulation identification utilizing MATLAB (neural system tool compartment). Muid Mufti and Assia Khanam10 built up a fluffy principle based feeling acknowledgment strategy utilizing outward appearance acknowledgment. In1 Local Binary Pattern has been separated from static pictures to group outward appearance utilizing PCA. In2, in view of the remaking blunder after the projection of each despite everything picture into symmetrical premise headings of various appearance subspaces, the outward appearance is perceived [12].

6

K. R. Sai et al.

1.3 System Framework This testing approach report is intended for Information and Technology Services’ moves up to PeopleSoft. The report contains an outline of the testing exercises to be performed when an overhaul or improvement is made, or a module is added to a current application. The accentuation is on trying basic business forms, while limiting the time essential for testing while likewise relieving dangers. Note that diminishing the measure of testing done in an overhaul builds the potential for issues after go-live. The board should decide how much hazard is satisfactory on an update by overhaul premise. Framework testing is basically trying the framework in general; it gets all the coordinated modules of the different segments from the combination testing stage and consolidates all the various parts into a framework which is then tried. Testing is then done on the framework as all the parts are presently coordinated into one framework the testing stage will currently must be done on the framework to check and expel any blunders or bugs. It is the first and the most essential degree of Software Testing, in which a solitary unit (for example a littlest testable piece of a product) is inspected in detachment from the rest of the source code. Unit Testing is done to check whether a unit is working appropriately. As such, it checks the littlest units of code and demonstrates that the specific unit can work impeccably in segregation. In any case, one needs to ensure that when these units are consolidated, they work in a durable way. This guides us to different degrees of programming testing. In PC programming, unit testing is a product testing technique by which singular units of source code, sets of at least one PC program modules along with related control information, utilization methodology, and working systems are tried to decide whether they are fit for use [1]. Intuitively, one can see a unit as the littlest testable piece of an application. In procedural programming, a unit could be a whole module, however it is all the more normally an individual capacity or method. In objectarranged programming, a unit is regularly a whole interface, for example, a class, however could be an individual strategy. Unit tests are short code parts made by software engineers or sporadically by white box analyzers during the advancement procedure. In a perfect world, each experiment is free from the others. Substitutes, for example, technique hits, mock articles, fakes, and test bridles can be utilized to help testing a module in seclusion. Unit tests are ordinarily composed and run by programming engineers to guarantee that code meets its structure and carries on as planned [2].

1.4 Proposed System Along these lines dynamic understudy conduct examination is an initial move towards a mechanized educator input apparatus for estimating understudy commitment. Our

1 Concentration Level of Learner Using Facial Expressions …

7

proposed framework can be applied in both conventional/e-learning frameworks. In our framework, we propose a half and half design framework summoning understudy facial feeling acknowledgment, eye stare checking, head developments recognizable pieces of proof based dissecting dynamic understudy commitment/conduct in homeroom and towards a particular course at e-learning stages [3]. Our proposed design utilizes highlight extraction calculations like Principal Component Analysis (PCA) for facial feeling acknowledgment, Haar Cascade for student discovery and Local Binary Patterns for perceiving head developments. For AI approach and to give exact outcomes we propose Open CV. Trial results are been actualized utilizing Pycharm [4]. Framework Implementation is the significant phase of venture when the hypothetical plan is fixed on useful framework. The primary stages in the usage are as per the following [5]: • • • •

Planning Training System testing and Changeover Planning.

Arranging is the main undertaking in the framework execution. Arranging implies choosing the strategy and the time scale to be received. At the hour of execution of any framework individuals from various offices and framework examination include [6]. They are affirmed to handy issue of controlling different exercises of individuals outside their own information handling divisions. The line chiefs controlled through an execution organizing advisory group. The board thinks about thoughts, issues and objections of client division, it should likewise consider [7]: • • • •

The ramifications of framework condition; Self-choice and portion for usage errands; Consultation with associations and assets accessible; Standby offices and channels of correspondence.

1.5 Results and Discussions 1.5.1 Conclusion The hybrid biometric based learner analysis does appear to be a promising new tool for evaluating learners’ behavior dynamically. This technology can provide many benefits to e-learning, such as facilitating adaptive and personalized learning thus IoT-based proposed system that examines the concentration level of the students from their facial expression, provides a real-time feedback system aids to enhance the mode of e-learning. Hence the tutor can change the method of deliverance by dynamically analyzing the learner’s attention level. This really brings a revolution in

8

K. R. Sai et al.

the education sector for the online-tutors in delivering interesting topics to improve learner’s involvement in the subject with more concentration [8]. In future, the performance of proposed method will be improved and will be extended for detecting the students fatigue by measuring the degree of openness of the eye. Further the real time implementation using video camera will be conducted. Also this project can be extended in analyzing student’s head movements and facial emotions [9–12].

References 1. Korukonda, A.R., Finn, S.: An investigation of framing and scaling as confounding variables in information outcomes: the case of technophobia. Inf. Sci. 155(1–2), 79–88 (2003) 2. Jay, M.T.: Computerphobia: what to do about it. Educational Technology (1981) 3. Scull, C.A.: Computer anxiety at a graduate computer center: computer factors, support, and situational pressures. Comput. Human Behav. 15(2), 213–226 (1999) 4. Sheeson, E.C.: Computer anxiety and perception of task complexity in learning programmingrelated skills. Comput. Human Behav. 21(5), 713–728 (2005) 5. Nwanewezi, M.C.M.: Problems in business education research in ICT-Era as perceived by business educators. Bus. Educ. J. 6(2) (2010) 6. Osual, E.C.: Business and Computer Education. Cheston Agency Limited, Enugun (2009) 7. Jude, D.A., Nosakhare, E.: Information and communication technology: challenges to effective teaching of business education. In: Book of Reading, vol. 2, no. 1 (2012) 8. Goodwin, Y.B.N., Miklich, B.A., Overall, J.U.: Perceptions and attitudes of faculty and students in two distance learning modes of delivery: online computer and telecourse. Orlando: FL. ERIC Document Reproduction Service No. ED 371 708 (1993) 9. Nithya, V., Ramesh, G.P.: Wireless EAR EEG signal analysis with stationary wavelet transform for co channel interference in schizophrenia diagnosis. In: Balas, V., Kumar, R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent Systems Reference Library, vol. 172. Springer, Cham (2020) 10. Hemanth, K., Ramesh, G. P.: Energy efficiency and Data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–995 (2020) 11. Pattnaik, M.: Infrastructure of data mining technique with big data analytics. Int. J. MC Sq. Sci. Res. 11(1), 23–30 (2019) 12. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 13. Swarnalatha, A., Manikandan, M.: Intravascular ultrasound image classification using wavelet energy features and random forest classifier. In: Intelligent Computing in Engineering, pp. 803– 810. Springer, Singapore (2020)

Chapter 2

IoT-Based Machine Learning System for Nutritional Ingredient Analyzer for Food A. Thilagavathy, Tadavarthi Rishi, Veeram Deepak Reddy, and Sudesh Nimmagadda Abstract Adequate nutritional regimes have been generally accepted as important prevention and management steps for non-communicable diseases (NCDs). In all cases, little research is now underway into safe food fixtures which support the rehabilitation of NCDs. Right now, significantly investigated the connection between healthful fixings and maladies by utilizing information mining techniques. Initially, in excess of 7,000 sicknesses were acquired, and we gathered the suggested nourishment and unthinkable nourishment for every illness. The analyses on genuine information show that our technique dependent on Data mining improves the exhibition contrast and the conventional measurable methodology, with the accuracy of 1.682. Moreover, for some basic ailments, for example, Diabetes, Hypertension and Coronary illness, our work can recognize effectively the initial a few nourishing fixings in nourishment that can profit the restoration of those ailments and the IoT-based system records the data of the patients periodically. The Machine Learning (ML) technique assists the statistical and texture features of the patient’s diet sheet based nutrient loss for the particular disease are monitored and notified personally. These exploratory outcomes exhibit the adequacy of applying information mining in choosing of wholesome fixings in nourishment for ailment examination. Keywords NCD · Noise-intensity · Data mining

A. Thilagavathy (B) · T. Rishi · V. D. Reddy · S. Nimmagadda R.M.K. Engineering College, Gummidipoondi, India e-mail: [email protected] T. Rishi e-mail: [email protected] V. D. Reddy e-mail: [email protected] S. Nimmagadda e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_2

9

10

A. Thilagavathy et al.

2.1 Introduction NCD’S are ceaseless infections, which are for the most part brought about by word related and ecological elements, ways of life and practices, including Corpulence, diabetes, Hypertension, Tumors and different maladies. As indicated by the Worldwide on Non-transferable WHO Ailments are given, the yearly loss of life from NCDs keeps including, which has made genuine financial weight the world. Around 40 million individuals passed on from NCDs every year, which is identical to 70% [1–3] of the worldwide loss of life. Measurements of Chinese Occupant’s Interminable Ailment and Sustenance show that the quantity of some other nations on the planet, and the present commonness rate has smothered. What’s more, the populace matured 60 or over in China has arrived at 230 million and around 66% of them are experiencing NCDs as per the official insights [4, 5]. Along these lines, pertinent divisions in every nation, particularly in China, for example, clinical schools, emergency clinics and infection investigation focuses all are worried about NCDs. Appropriate nourishing eating regimens assume a significant job in keeping up wellbeing and forestalling the event of NCDs [2, 6]. Continuous acknowledgment idea, China is additionally effect of nourishment on wellbeing. Be that as it may, look into dietary fixings in nourishment through information mining, which are helpful for the restoration of maladies is as yet uncommon quite recently started the IT (Data Innovation) development of keen medicinal services. Most examinations on the connection in nourishment and illnesses are still through costly accuracy preliminaries. What’s more, there are additionally numerous anticipation reports, however they concentrated on just one or a few maladies [7]. In China, considering the connection between nourishing fixings and ailments utilizing Data mining is juvenile. Most specialists just prescribe the particular nourishment to patients experiencing NCDs, without giving any important sustenance data, particularly [8]. In the period of large information, information mining has become a basic method for finding new information in different fields, particularly in malady expectation and exact medicinal services (AHC) [9]. It has become a center for preventive medication, essential medication and clinical medication examination. Regarding the infection investigation through the mining of nourishing fixings in nourishment, we mostly make the accompanying commitments: (I) We separated information identified with Chinese sicknesses, comparing suggested nourishment and forbidden nourishment for every illness however many as could be expected under the circumstances from clinical and official sites to make an important information base that are accessible on the web; (ii) Applying clamor power and data entropy to discover which healthful fixings in nourishment can apply constructive outcomes to maladies; (iii) Right now, information is ceaseless and has no choice traits. Red dependent on an unpleasant set hypothesis, which can more readily choose comparing center fixings from the positive dietary fixings in nourishment. Area II surveys the related work in the field of infection investigation and information mining. Portrays the particular information mining calculations utilized right now, why we select the calculations, just as two assessment records. Expounds the information, test results and investigation in detail.

2 IoT-Based Machine Learning System …

11

Presents conversations between techniques. A few ends and potential future research headings are additionally examined.

2.2 Literature Survey The most critical advance in the programming method is the literature review. The time factor, economics and quality of friends must be determined before constructing the unit. If these things are completed, ten next steps must decide the structure and language the device will be used to construct. Whenever the computer programmers start constructing the system, the developers require a kit of outside assistance. Senior software developers, books and websites can obtain this support. The above insights are called to construct the proposed structure before constructing the structure. • Retrospective study of diabetes testing in a mass meeting in India: consequences for the regulation of non-communicable diseases The major case of non-communicable diseases (NCD) mortality [10–12] in India is heart disease. The national strategy pernicious Development, Diabetes, Cardiovascular Disease, and Strokes Prediction and Management Program aims to extend the NCD management, screenings and references across India and include network based initiative and machine tracking engineers. The Indian government regularly finds strict public get-togethers as essential. In any event, the economic growth its administrators to provide a hyper testing provider at the Nashik and Trimbakeshwar Kumbh Mela in 2015. In this article, we look at the importance and implications of such a groundbreaking social testing. At the Kumbh, 5760 people deliberately decided on hypertension screening, and got a solitary circulatory strain estimation. All in all, the favourable findings were reviewed by 1783 (33.6%), of which 1580 had not been told. Of the 303 newly experienced hypertonic medications, 240 (79%) have earned regulatory clearance and 160 (52.8% under treatment) have been satisfying. Typical circulatory pressure assessment (BP leveled out) was in 55% (18%). The knowledge was also more common (39%) among nicotine consumers with epilepsy, compared with non-users (28%) (P < 0.001). Bad telephone chronology (0.01%) has blocked the production of a cell phone. The low levels of tuberculosis understanding, care and management highlights both Indian and Indian managers’ progressive challenge. • DIETOS: A Recommender System for Adaptive Diet Monitoring and Personalized Food Suggestion These days there is a far reaching dissemination of portable weight and diet on the board. Despite the fact that, the most famous applications are not typically tested in clinical settings, just as applications are not bolstered by clinical proof. Important to survey the adequacy of applications for weight and diet the board. Besides, there are not many instances of nourishment that give to the clients healthful realities about

12

A. Thilagavathy et al.

reasonable nourishment decisions and consider for the versatile conveyance of nourishment substance to improve the personal satisfaction of both sound individuals and people influenced by constant diet-related infections. The proposed system can assemble a user’s wellbeing profile, and gives individualized nourishing suggestions as indicated by the wellbeing profile. The profile is made using dynamic constant surveys arranged by clinical specialists and accumulated by the clients. The wellbeing profile incorporates data about wellbeing status and possible incessant infections. The main model incorporates a list of run of the mill Calabrian nourishments assembled by sustenance masters. DIETOS can propose not just the utilization of explicit nourishments good with the wellbeing status, yet in addition it might give dietary signs identified with some particular or wellbeing conditions. • Lumping versus division: the need for precise biological data mining Biological data mining is assuming an undeniably significant job all through the range of biological and biomedical research with expansive ramifications for the comprehension of life science addresses, for example, the tree of life and functional uses of such information to improve human wellbeing. Maybe no place is data mining required more than the developing order of exactness medication. The capacity to anticipate singular danger of giving an infection or reaction to treatment is at the center of the idea of exactness medication, which is picking up ever-expanding levels of footing in the period of technology-driven estimation of biological frameworks. This has gotten particularly significant with the new Presidential activity on accuracy medication in the US. It is clear to the pursuers of Bio Data Mining that this will require cautious investigations [13–16] of huge and regularly complex data sets to best make an interpretation of data into progressively individualized chance. Here we inquire as to why improved and fitting data mining isn’t just positive however an immense enhancement for most current examinations of genomic data. The appropriate response deceives some degree in clarifying the current act of - omic examinations and how we should extend it. • Aquatic Ecosyope Health Evaluation based on the Main Entropical Weight Portion assessment: A Rising Dam test case (Hainan Island, China) In order to discover if the new approach can undo the calculation coverage which existed for environmental welfare in traditional entropy-driven straight lines, a further evaluation strategy based on the main component analysis (PCA) and randomness weight for pharmacological system good was introduced in the Dwindling Dam, Hainan Area, China. The outcomes demonstrated that, the biological system wellbeing status of Waning Reservoir indicated an improvement pattern generally from 2010 to 2012; the methods for environment well being far reaching file and the biological system wellbeing status was III (medium), II (great), and II (great), separately. Furthermore, the environment wellbeing status of the reservoir showed a frail regular variety. The variety of EHCI decreased as of late, demonstrating that Waning Reservoir would in general be moderately steady. Examining the length of the modern and traditional files indicates that for the traditional, a more embedded relation of 0.382 was feasible than with the new approach, the cumulative load for

2 IoT-Based Machine Learning System …

13

the 4 files. The use of PCA with oxidation was advised to retain proper relative weight away again from cover. The interaction study between the photosynthetic status file and EHCI also revealed a big negative correlation (P < 0.05), suggesting that the modern entropic-weight PCA-dependent methodology could both further boost weighting and results precision. The new technique here is reasonable for assessing the environment strength of the reservoir.

2.3 Methodology 2.3.1 Data-Mining Data extraction is the way to locate prototypes in large data sets. Machine Learning, Measuring and Database Convergence Methods. Frameworks.-Frameworks. In order to acquire data from a dataset and to turn the information into an accessible structural for additional use, data mining is an emerging subfield of information technology and initiatives. Data mining is the analysis venture of the “knowledge discovery in databases” procedure or KDD.

2.3.2 Statistical Algorithm (SA) On the off chance that a specific illness is brought about by the absence of certain healthful fixings, at that point their qualities in suggested nourishment ought to be generally higher in principle. Along these lines, we can make sense of which wholesome fixing esteems are high. The dietary fixings with higher qualities ought to be the fixings, which advantage the restoration of that particular malady, i.e., PNIs. This strategy is alluded to as statistical algorithm (SA) in this paper. The basic documentations are characterized as following, let n be the quantity of the suggested nourishment for a specific illness, and m be the quantity of wholesome fixings. The above thought can be communicated as beneath: sort{xi1, xi2, • • • , xim |descend}

(2.1)

i = 1 i = 1 i = 1. xi1 shows the primary nutrient value for a certain infection of the suggested ith food; xi2 measures the amount of the second food nutrient of the suggested ith food and so forth. Both foods approved for this illness are added to dietary values. Equations are then sorted downward. Therefore, PNIs ought to be the highest of the classified functional foods. We also obtained different methods for research in order to show whether or not data mining tools can be used for the disease study. Since this paper is trying to

14

A. Thilagavathy et al.

Fig. 2.1 Over all work flow of proposed diagram

take care of another issue from a true application, there is no pertinent work for examination. In any case, we utilized an exploring way to tackle this issue, at the end of the day, we led an impact comparison between various strategies to choose the best one. In order to demonstrate nutrient nutrient characteristics in food, we derive nutrient estimates for four serious illnesses from recommended food and tabú foods. In Fig. 2.1, nutrient appreciation for 4 diseases (such as hypertension, coronary sickness, kidney and apoplexy) is seen in the recommended diet and tabou food. The X-axis and Y-axis are compared and correspond to the dietary ingredients. Figure 2.5 indicates the significance of any X-axis document. In comparison, blue and red stand for recommended foods and tabu foods. Figure 2.2 indicates that some nutrient estimates are very high while the others are very poor independent of diets or tabu diets. Broadly speaking, for the most infections in China, there is scarce tabou rice, so we can’t collect more. Because of the less true awareness for limited data mining interventions, we are currently testing foods to decide which nutrient additives may have beneficial effects on a specific disease. In this article, positive nutrient ingredients (PNI) are called nutritious ingredients which help the restoration of illnesses.

2 IoT-Based Machine Learning System …

15

Fig. 2.2 User section home page

2.4 Results and Discussions User module will upload the dataset below. After entering the appropriate dataset in the upload file option shown in Fig. 2.3, we have to click the upload dataset. Then user will search or insert the food name which he or she wants positive nutrients as shown in Fig. 2.4 to get cure from the disease suffering from. After entering the disease, user will be shown a list of positive nutrients and its cure for the disease as shown in Fig. 2.5.

Fig. 2.3 User viewing the dataset

16

A. Thilagavathy et al.

Fig. 2.4 User entering the food item

Fig. 2.5 Prediction graph

2.5 Conclusion The key analysis of the article can be separated into two sections: initially, the relationships involving nutrient component and disease, that primarily seek to classify which additives play a constructive role in restoring Chinese diseases, have been collected and sorted from the biomedical and formal database. To our best understanding, it’s the first research in China, that uses data mining technologies to disrupt the connection between nutrition and infection nutrient materials. Innovative findings revealed that while the positive dietary components could not be entirely identified by data mining techniques for illnesses, the first two or three have been precisely chosen. Furthermore, if we can merge our vision with tabu food, the outcomes will possibly be smoother and in line with the fact that we will be working on in the future. The

2 IoT-Based Machine Learning System …

17

IoT based website that consists of the patients’ previous record stored in the database aids to compare the current nutrient chart of the patient. This features of the nutrients in the food they nourishes utilize the ML algorithm with CNN technique for feature extraction and thus classify the deficiency based on the feature vectors providing the disease probe to infect. Hence this is efficient preventive method of analyzing system in the medical field for the nutritionist to follow and treat the patients accordingly. The two major aspects of this study include: (1) you can use our website to receive illness, prescribed food, tabuic foods and associated nutrition data; and (2) doctors and illness inspectors may be helped to identify positive nutrient additives that are as effective as possible to rehabilitate illnesses. Actually, some details are not accessible while surgical exams are still in progress. In addition, if investigators discover anything wrong in our work, we expect to notify us and enhance our study. In reality, our knowledge and experience is increasingly improving.

References 1. CNS.: Global Nutrition Report. Chinese Nutrition Society (2016) 2. WHO.: Global Status Report on Noncommunicable Diseases. World Health Organization (2014) 3. Balsari, S., Vemulapalli, P., Gofine, M., et al.: A retrospective analysis of hypertension screening at a mass gathering in India: implications for non-communicable disease control strategies. J. Hum. Hypertens. 31(11), 750–753 (2017) 4. Meng, Q., Yang, H., Chen, W., Sun, Q., Liu, X.: People’s Republic of China health system review. Health. Syst. Trans. 5(7), (2015) 5. Tellier, S., KiabyLars, A., Nissen, P., et al.: Basic Concepts and Current Challenges of Public Health in Humanitarian Action, pp. 229–317. International Hu-manitarian Action (2017) 6. Ara1, F., Saleh, F., Mumu, S. J., Afnan, F., Ali, L.: Awareness A-mong Bangladeshi Type 2 Diabetic Subjects Regarding Diabetes and Risk Factors of Non-communicable Diseases, Diabetologia, pp. S379 (2011) 7. Wu, L., Yu, Z., Zhang, P., Kan, C.: Research on the market development path of intelligent medical industry under the background of aging. Adv. Soc. Sci., Educ. Humanit. Res. 213, (2018) 8. Ling, W.H.: Progress of Nutritional prevention and control on noncommunicable chronic diseases in China. China J. Dis. Control Prev. 21(3), 215–218 (2017) 9. Margaret, M.B., Barbara, B.K., Colette, D.: Developing health promotion workforce capacity for addressing non-communicable diseases globally. In: Global Handbook on Noncommunicable Diseases and Health Promotion, pp. 417–439 (2013) 10. Williams, M., Moore, H.: Lumping versus splitting: the need for biological data mining in precision medicine. BioData Mining 8(16), 1–3 (2015) 11. Hemanth, K., Ramesh, G.P.: Energy efficiency and Data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), pp. 944–95 (2020) 12. Hemanth Kumar, G., Ramesh, G.P.: Energy efficient multi-hop routing techniques for cluster head selection in wireless sensor networks. In: Computational Intelligence and Complexity, vol. 193, pp. 297–305. Springer (2021) 13. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed. Tools Appl. 79(5), 4011–4026 (2020) 14. Manahoran, N., Srinath, M.V.: K-means clustering based marine image segmentation. Int. J. MC Sq. Sci. Res. 9(3), 26–29 (2017)

18

A. Thilagavathy et al.

15. Nithya, V., Ramesh, G.P.: Wireless EAR EEG signal analysis with stationary wavelet transform for co channel interference in schizophrenia diagnosis. In: Balas, V., Kumar, R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent Systems Reference Library, vol. 172. Springer (2020) 16. Pattnaik, M.: Infrastructure of data mining technique with Big data analytics. Int. J. MC Sq. Sci. Res. 11(1), 23–30 (2019)

Chapter 3

A Secured Manhole Management System Using IoT and Machine Learning R. Santhana Krishnan, A. Sangeetha, D. Abitha Kumari, N. Nandhini, G. Karpagarajesh, K. Lakshmi Narayanan, and Y. Harold Robinson

Abstract Improper maintenance of manholes leads to many critical issues. Periodical checking of Manholes is necessary to maintain our society with proper hygiene. Improper closing of manholes in India led to a loss of 102 lives during 2019. In addition to this, lot of people in India is dying while entering into manhole for cleaning the sewers. The death count of people involved in cleaning of sewers keeps on rapidly increasing year to year. In order to control these issues, a secured manhole management system is developed. This system monitors the lid of the manhole and alerts the Municipal Corporation whenever the position of the lid is changed. Similarly the condition of the sewers are monitored periodically and an alert information regarding the issue in the sewer system and its location details are sent to Municipal Corporation to take necessary actions whenever there is an emergency situation. The sensor details were periodically recorded and the machine learning algorithm is used to set the threshold of the sensor values with the help of Recurrent Neural Network using Long Short Term Memory (LSTM) algorithm. R. S. Krishnan (B) ECE, SCAD College of Engineering and Technology, Tirunelveli, Tamilnadu, India A. Sangeetha PSNA College of Engineering And Technology, Dindigul, Tamilnadu, India e-mail: [email protected] D. A. Kumari Sethu Institute of Technology, Virudhunaragar, Tamilnadu, India N. Nandhini KPR Institute of Engineering and Technology, Coimbatore, Tamilnadu, India G. Karpagarajesh Government College of Engineering Tirunelveli, Tamilnadu, India K. L. Narayanan (B) Francis Xavier Engineering College Tirunelveli, Tamilnadu, India Y. H. Robinson School of Information Technology and Engineering, Vellore Institute of Technology, Tamilnadu, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_3

19

20

R. S. Krishnan et al.

Keywords Manhole · IoT · GSM · GPS · Gas Sensor · Mechanical float sensor · Temperature LSTM · RNN · Cloud server

3.1 Introduction According to the report released by Social Justice and Empowerment (SJE) Ministry, the total number of lives lost while cleaning the septic tanks and sewers is found to be 620 till June 2019 [1]. Out of these 620 people, 88 people have died during the last 3 years. It is clearly depicted in the Fig. 3.1. Tamil Nadu leads the list with 144 cases constituting to around 23.2% of the total causalities during the last 3 years. SJE Ministry also released a report on February 11, 2020 [2] stating that death rate of people involved in cleaning sewers has increased to 68% in 2019 compared to the previous year [21–23]. The last 5 years the death count of people involved in cleaning of sewers is clearly explained in the Fig. 3.2. According to the article released by Times of India on September 11-2020 [3], the number of person died by entering into manhole for cleaning purpose and the accidental fall due to improper closing of manhole has not reduced over the last 5 years which is clearly depicted in Fig. 3.3. According to the survey report of NCRB (National Crime Record Bureau), the number of person died due to accident caused by open manhole is found to be 102 which are clearly explained in Fig. 3.4. Internet of Things (IoT) means a group of internet activated devices which are capable of collecting data and transferring information without any human intervention[24,

Fig. 3.1 People death count in India while cleaning of sewers between the years 2017 and 2019

3 A Secured Manhole Management System …

21

Fig. 3.2 People death count while cleaning sewers between the years 2015 and 2019

Fig. 3.3 Number of deaths due to open manhole between the years 2015 and 2019

25]. At present, IoT has gathered more attention among people since it has reduced the manual work to a great extent. IoT has its impact on various applications like agriculture [4–6], transport, logistics, health care [7], retail, industry [8–11] and many more. It has created a great impact on India. There are more than 504 million internet users in India who are above the age limit of 5. India stands next to China (805 million users) in internet usage. Indian zone wise internet user’s details during beginning and end of the year 2019 are depicted in Fig. 3.5. The main objective of our secured manhole management system is. • To monitor the Manhole lid and to inform the Municipal Corporation if there is a change in the position of the Manhole lid. • To monitor the methane gas leakage using MQ4 sensor and alerting the Municipal Corporation to take necessary actions.

22

R. S. Krishnan et al.

Fig. 3.4 Number of People died due to accidents caused by open manhole at various states of India

Fig. 3.5 System architecture

3 A Secured Manhole Management System …

23

• To monitor the level of drainage in the sewers and intimate the Municipal Corporation when 70% of the sewers is filled. • To collect and store the details of the corporation workers who attends the sewers cleaning/repair work in the database.

3.2 Related Work Haswani et al. [12] proposed an open manhole monitoring system which uses ultrasonic sensor and light sensor to identify whether the manhole lid is properly covering the manholes. Whenever the manhole lid is not covering the manhole, then the light sensor detects the presence of sunlight. Ultrasonic sensor also detects the opening of the lid whenever the observed distance of the obstacle is greater than the fixed threshold distance. By using these 2 sensors the system clearly identifies the opening of manhole. Similar works have been carried out by Nataraja et al. [13] and Nathila Anjum et al. [14]. Here the author used ultrasonic sensor and vibration sensor to identify the removal of manhole lid and alerts the surrounding people using an alarm. In addition to this a SMS is sent to the authority regarding the removal of manhole lid. LiLei et al. [15] developed a system which uses LoRa technology to continuously monitor the entire group of manholes available within a particular area. The system detects the opening of manhole lid using angle sensor. Then the system sends intimation to LoRa, which in turn uses GSM and Wi-Fi technology to send SMS to the authority and update the information in the server respectively. Similar work has been carried out by Zhang et al. [16] where the author used accelerometer sensor instead of angle sensor. Ke et al. [17] proposed a system which identifies the theft of manhole covers using a simple magnetic switch and alerts the people using an alarm. Asthana et al. [18] proposed a smart sewage monitoring system which detects the presence of poisonous gas inside the sewage using the MQ-7 and MQ-4 gas sensors and alerts the authority using a SMS. Xiucai et al. [19] proposed a smart system which gathers information like level of the wastage in the tank and availability of poisonous gas using the sensors fixed on the sewage tank. An alert message is sent to the authority using GSM module. Similar work is carried out by Salehin et al. [20].

3.3 Proposed System The Raspberry-pi processor monitors the data from the entire sensors available within the system [26]. The position of the manhole lid is monitored by accelerometer sensor and alert information is passed to raspberry-pi whenever there is any change in position of the lid. MQ4 gas sensor is used in this system which is capable of detecting the presence of methane gas in sewer system. In this system there are 2 magnetic float sensors.

24

R. S. Krishnan et al.

The magnetic float sensor 1 used to indicate that the sewer system is filled up to 70% of its holding capacity. Similarly the magnetic float sensor 2 is used to indicate that the sewer tank is filled up to 90%. Temperature sensor is used to monitor the temperature inside the sewer system and gives an alert to Raspberry-pi [27]. All these details collected from those sensors are updated into the cloud server along with the GPS location through Wi-Fi module. Then the cloud server sends a message to the concerned authority in the Municipal Corporation and to the Corporation worker allotted to that particular area. The details of the corporation worker who attends the concerned problem is also gathered by our system using RFID reader and his details and time of attending the sewer related issues are also stored in the server for further processing [28]. The system architecture is given in Fig. 3.5. The overall flow diagram of this system is explained using Fig. 3.6. The working procedure of this system is explained using the following steps. Step 1 Raspberry-Pi reads the data from all the sensors. Step 2 MQ4 Gas sensor is used to detect the presence of the methane gas. Similarly the Accelerometer sensor checks for any change in position of the sewer’s lid. In addition to this the level of waste in the sewer system is monitored by magnetic float sensor1 and magnetic float sensor 2. Magnetic Float sensor 1 and Magnetic Float sensor 2 detects the 70% and 90% of the waste filled in the sewer system respectively. Temperature sensor monitors the temperature level of the sewer system and sends an alert to Raspberry-pi when the sensor detects the temperature value greater than the threshold value. Step 3 Whenever any of these sensors detects the abnormal activities, it sends an alert information to the Raspberry-pi.

Fig. 3.6 Flow diagram

3 A Secured Manhole Management System …

25

Step 4 Then the Raspberry-pi gets the location details from GPS immediately after receiving the alert from any of these sensors and sends that information to the cloud server with the help of Wi-Fi module. Whenever the Raspberry-pi receives the data from Magnetic float sensor 2, it follows the same procedure of updating the data to the cloud and along with this it activates the buzzer for 2 min to alert the people about the worse situation of the sewer system. Step 5 Then the cloud server identifies the corporation worker allotted for the received GPS location and sends the message to the corporation worker as well as the authority responsible for monitoring these activities in municipal corporation. Step 6 Raspberry-pi also reads the RFID card available with the corporation worker using RFID card reader and sends that information to the cloud server to update the details in the database. Thus the details of the corporation worker who attends the problem and the timing by which he attends it are recorded in the database.

3.4 Results and Discussion Our system is capable of monitoring the sewer system in a smart manner and updates the abnormal activities to the cloud server using Wi-Fi module. Then the cloud server alerts the Municipal Corporation and the corporation worker based upon the location information received from the GPS. Here the system detects the waste level of the sewer system which has crossed 70% of its holding capacity using the Magnetic float sensor 1. Then the Magnetic Float sensor intimates this information to Raspberry-pi. Further the processor gets the location of this system from GPS module and updates the issue noticed in the system to the cloud server through Wi-Fi module. After analyzing the location details the cloud server identifies the person in charge for the cleaning activity and sends the alert SMS to him indicating the issue in the sewer system and its location details. This is clearly explained in Fig. 3.7a. Our system also intimates the work allotment details to Municipal Corporation through a SMS. This is helpful in monitoring the follow up activity of the corporation worker. This is explained in Fig. 3.7b. The information received by the Raspberry-pi is stored in the database with the help of Wi-Fi module. This is explained using Table 3.1. Once any issue is observed by the cloud server, it gets the location details and searches for the corporation worker assigned for that location in the database and sends the details of the issue observed in the sewer system through a message to his contact number. This is explained in Table 3.2. Once the corporation worker reaches the spot, the details of the corporation worker is collected through the RFID reader module. Then those details are updated to the cloud server through the Raspberrypi. This will confirm the work attending status of the corporation worker. This is explained using the Table 3.3 (Figs. 3.8 and 3.9).

26

R. S. Krishnan et al.

Fig. 3.7 a Intimation of the waste level in sewer system to corporation worker. b Intimation of waste level to municipal corporation

Table 3.1 Issue storage database Issue_no.

Issue_type

Location

Issue_time

01

Excess waste level

8.719505146879992, 77.73270257036525

2021/01/10 12:30:34 PM

02

Open manhole

8.730238538831168, 77.72239950022494

2021/01/19 07:00:12 PM

03

Poisonous gas

8.710356284662698, 77.74912325987525

2021/02/21 08:10:19 PM

04

Open manhole

8.707376241032382, 77.75562493450923

2021/02/26 06:23:52 PM

05

Poisonous gas

8.726677169890213, 77.69059745942927

2021/03/05 11:30:28 AM

Sensor type

Training data set

Average value

Threshold value

Temp

352

129.56

130

Float sensor

468

251.23

256

Tilt sensor

551

100

110

Gas

429

180

200

3 A Secured Manhole Management System …

27

Table 3.2 Work allotment database Report_no. Report_type Location

Worker_assigned Contact number

Report_time

01

Excess waste level

8.719505146879992, Ganeshan 77.73270257036525

9,751,395,769 2021/01/10 12:32:30 PM

02

Open manhole

8.730238538831168, Ashok 77.72239950022494

9,566,509,646 2021/01/19 07:02:11 PM

03

Poisonous Gas

8.710356284662698, Jebastin 77.74912325987525

9,894,000,020 2021/02/21 08:13:06 PM

04

Open manhole

8.707376241032382, Ulaganathan 77.75562493450923

9,003,871,815 2021/02/26 06:25:13 PM

05

Poisonous gas

8.726677169890213, Kumar 77.69059745942927

8,843,221,562 2021/03/05 11:33:24 AM

Table 3.3 Issue attending database Report_no.

Report_type

Location

Attending_time

01

Excess waste level

8.719505146879992, 77.73270257036525

2021/01/10 02:25:15 PM

02

Open manhole

8.730238538831168, 77.72239950022494

2021/01/20 10:20:16 AM

03

Poisonous gas

8.710356284662698, 77.74912325987525

2021/02/22 10:52:06 AM

04

Open manhole

8.707376241032382, 77.75562493450923

2021/02/26 06:55:19 PM

05

Poisonous gas

8.726677169890213, 77.69059745942927

2021/03/05 01:39:29 PM

Fig. 3.8 Issue attending intimation

28

R. S. Krishnan et al.

Fig. 3.9 Experimental setup

The values of the various sensors like temperature sensor, float sensor, tilt sensor and gas sensor are gathered by the processor and sent to the LSTM (Long Short Term Memory) algorithm during the training phase. Then the sensor values are obtained and processed and an average value is obtained by the LSTM. Then the threshold value is fixed based on the average value calculated by LSTM algorithm.

3.5 Conclusion Our secured manhole management system monitors the lid of the manhole continuously and sends the alert information to the Municipal Corporation whenever the position of the lid is changed. Similarly the condition of the sewers are monitored periodically and an alert information regarding the issue in the sewer system along with its location details are sent to Municipal Corporation to take necessary actions whenever there is an emergency situation. This system also monitors and intimates the higher authorities in the Municipal Corporation about the availability of the corporation workers in attending the issue reported spot. This system is very much useful in maintaining a proper hygiene in our society.

3 A Secured Manhole Management System …

29

References 1. https://www.thehindu.com/news/national/88-manual-scavenging-deaths-in-3-years/article28 336989.ece 2. https://www.thehindu.com/news/national/110-deaths-by-cleaning-sewers-septic-tanks-in2019/article30795201.ece 3. https://timesofindia.indiatimes.com/city/ahmedabad/30-of-indias-manhole-deaths-in-gujarat/ articleshow/78047520.cms 4. SanthanaKrishnan, R., Golden Julie, E., Harold Robinson, Y., Raja, S., Kumar, R., Thong, P.H., Son, L.H.: Fuzzy logic based smart irrigation system using Internet of Things. J. Clean. Prod. 252, 119902 (2020) 5. RamasamySankar Ram, C., Ravimaran, S., SanthanaKrishnan, R., Golden Julie, E., Harold Robinson, Y., Kumar, R., Son, L.H., Thong, P.H., Thanh, N.Q., Ismail, M.: Internet of Green Things with autonomous wireless wheel robots against green houses and farms. Int. J. Distrib. Sens. Netw. 16(6) (2020) 6. Hemanth, K., Ramesh, G.P.: Energy efficiency and Data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–995 (2020) 7. Harold Robinson, Y., Santhana Krishnan, R., Raja, S.: A comprehensive study for security mechanisms in healthcare information systems using Internet of Things. In: Balas V., Solanki V., Kumar R. (eds.) Internet of Things and Big Data Applications. Intelligent Systems Reference Library, vol. 180. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-39119-5_15 8. Divya, Lakshmi Narayanan, K., Ramesh, G.P.: Robust and brittle secured video for IoT. Int. J. Eng. Technol. 7(2.20), 93–96 (2018) 9. Lakshmi Narayanan, K., Ganesan, J., Kasinathaguru: Intranet based multiple equipment controller with text panel. Irish Interdiscip. J. Sci. Res. 4(3), 120–126 (2020) 10. Hemanth Kumar, G., Ramesh, G.P.: Energy efficient multi-hop routing techniques for cluster head selection in wireless sensor networks. In: Computational Intelligence and Complexity, vol. 193, pp. 297–305. Springer (2021) 11. Thirupathieswaran,R., Suria Prakash, C.R.T., Krishnan, R.S., Narayanan, K.L., Kumar, M.A., Robinson, Y.H.: Zero queue maintenance system using Smart Medi Care application for Covid19 pandemic situation. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 1068–1075 (2021). https://doi. org/10.1109/ICICV50876.2021.9388454 12. Haswani,N.G., Deore, P.J.: Web-based realtime underground drainage or sewage monitoring system using wireless sensor networks. In: 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA), Pune, India, 2018, pp. 1–5 (2018) 13. Nataraja,N., Amruthavarshini, R., Chaitra, N.L., Jyothi, K., Krupaa, N., Saqquaf, S.S.M.: Secure manhole monitoring system employing sensors and GSM techniques. In: 2018 3rd IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT), Bangalore, India, 2018, pp. 2078–2082 (2018). https://doi.org/10. 1109/RTEICT42901.2018.9012245 14. Nathila Anjum, G., Saniya Kouser, K., Pragathi, M.S., Soundarya, P.P., Prashanth Kumar, H.K.: To Design & Analysis of Underground Drainage and Manhole Monitoring System for Smart Cities, IJESC, March 2020 (2020) 15. Lei,L., Sheng, Z.H., Xuan, L.: Development of low power consumption manhole cover monitoring device using LoRa. In: 2019 IEEE International Instrumentation and Measurement Technology Conference (I2MTC), Auckland, New Zealand, 2019, pp. 1-6 (2019). https://doi.org/ 10.1109/I2MTC.2019.8826885 16. Zhang, H., Li, L., Liu, X.: Development and test of manhole cover monitoring device using LoRa and accelerometer. IEEE Trans Instrum Meas 69(5), 2570–2580 (May 2020). https://doi. org/10.1109/TIM.2020.2967854 17. Ke,B., Mao, D.: Design of manhole covers missing alarm monitoring and management system. In: 2017 IEEE 2nd Information Technology, Networking, Electronic and Automation Control

30

18.

19.

20.

21. 22. 23. 24. 25.

26.

27. 28.

R. S. Krishnan et al. Conference (ITNEC), Chengdu, 2017, pp. 682-686 (2017). https://doi.org/10.1109/ITNEC. 2017.8284819 Asthana,N., Bahl, R.: IoT device for sewage gas monitoring and alert system. In: 2019 1st International Conference on Innovations in Information and Communication Technology (ICIICT), Chennai, India, 2019, pp. 1–7 (2019). https://doi.org/10.1109/ICIICT1.2019.8741423. Guo,X., Liu, B., Wang, L.: Design and implementation of intelligent manhole cover monitoring system based on NB-IoT. In: 2019 International Conference on Robots & Intelligent System (ICRIS), Haikou, China, 2019, pp. 207–210 (2019). https://doi.org/10.1109/ICRIS.2019.00061 Salehin,S., et al.: An IoT based proposed system for monitoring manhole in context of Bangladesh. In: 2018 4th International Conference on Electrical Engineering and Information & Communication Technology (iCEEiCT), Dhaka, Bangladesh, 2018, pp. 411–415 (2018). https://doi.org/10.1109/CEEICT.2018.8628091 https://www.analyticsinsight.net/india-estimated-to-reach-1-billion-internet-users-by-2025/ https://timesofindia.indiatimes.com/business/india-business/for-the-first-time-india-hasmore-rural-net-users-than-urban/articleshow/75566025.cms https://cms.iamai.in/Content/ResearchPapers/2286f4d7-424f-4bde-be88-6415fe5021d5.pdf Vadivelu, S., Prakashraja, Saini, S.R., Ramaiyan, M.: Smart manhole coverage system using IOT. Int. J. Adv. Eng. Technol. 1(4). ISSN: 2581-3374 Krishnan,R.S., Kannan, A., Manikandan, G., Sri Sathya, K.B., Sankar, V.K., Narayanan, K.L.: Secured college bus management system using IoT for Covid-19 pandemic situation. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 376–382 (2021). https://doi.org/10.1109/ICICV50876.2021.938 8378 Krishnan,R.S., Sangeetha, A., Kumar, A., Narayanan, K.L., Robinson, Y.H.: IoT based smart rationing system. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 300–305 (2021). https://doi.org/10. 1109/ICICV50876.2021.9388451 Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multim Tools Appl 79(5), 4011–4026 (2020) Narayanan,K.L., Ram, C.R.S., Subramanian, M., Krishnan, R.S., Robinson, Y.H.: IoT based smart accident detection & insurance claiming system. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 306–311 (2021). https://doi.org/10.1109/ICICV50876.2021.9388430

Chapter 4

Internet of Things Based Smart Accident Recognition and Rescue System Using Deep Forests ML Algorithm K. Lakshmi Narayanan, Y. Harold Robinson, Rajkumar Krishnan, C. Ramasamy Sankar Ram, R. Santhana Krishnan, R. Niranjana, and A. Essaki Muthu Abstract Road accidents are very common nowadays. As more people are buying vehicles, the frequency of road accident increases. There are gargantuan amount of sensor available in the vehicles to monitor the condition of both the vehicle and the people traveling in it. To trim down the fatality rate, nowadays automobiles are manufactured with various safety measures (like air bag system) to prevent the people from accident. But it is extremely intricate to deal with injured people who are involved in the accident. The process of informing the hospital to attend the injured people, informing the police station to indicate the accident zone and claiming the insurance for the vehicle becomes a difficult task. The proposed Deep Forests algorithm is the Machine Learning Algorithm which recognizes the accident in most accurate way with enhanced hyper parameters, In order to speedup this process, a smart accident recognition and rescue system is introduced which performs these tasks in an easy way. Keywords Internet of Things · Smart card system · Biometric system · Raspberry Pi · Android application · Deep forest algorithm

K. Lakshmi Narayanan · R. Niranjana Francis Xavier Engineering College, Tirunelveli, Tamilnadu, India Y. Harold Robinson School of Information Technology and Engineering, Vellore Institute of Technology, Tamilnadu, India R. Krishnan PSNA College of Engineering and Technology, Dindigul, Tamilnadu, India C. Ramasamy Sankar Ram CSE, University College of Engineering, Anna University, Tiruchirappalli, Tamilnadu, India e-mail: [email protected] R. Santhana Krishnan (B) · A. Essaki Muthu SCAD College of Engineering and Technology, Tirunelveli, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_4

31

32

K. Lakshmi Narayanan et al.

4.1 Introduction Nowadays road accidents are increasing due to the increase in vehicle purchase. WHO report [1] states that more than 50 million people get injured and 1.25 million people expire due to road accidents. In the year 2019, nearly 77% of the people died in road accidents belong to the age group of 18–45.This is depicted in Fig. 4.1. As per the report of Ministry of Road Transport & Highways [2], the number of males died in road accident is higher while compared to that of female.The total number of males and females died due to road accident in the year 2019 is found to be 1,29,319(85.6%) and 21,794(14.4%) respectively. Figure 4.2 depicts the percentage of male and female involved under various age groups. Figure 4.3 shows the top 5 states in India with highest victim counts in road accident. From the figure it is very clear that Tamil Nadu state stands first with highest fatality count of 57,228. Figure 4.4 portrays the year wise accident count, injured people count, and fatality count between 2015 and 2019. Figure 4.5 depicts the month wise total accident count and people death count during the year 2019.

Fig. 4.1 People age profile of expired people involved in road accidents

Fig. 4.2 Age Profile and gender wise share in road accident deaths

4 Internet of Things Based Smart Accident Recognition …

Fig. 4.3 Road accident fatality count in top 5 Indian states during 2019

Fig. 4.4 Year wise accident count between 2015 and 2019

Fig. 4.5 Month wise accident count and fatality count during the year 2019

33

34

K. Lakshmi Narayanan et al.

Fig. 4.6 Accident percentage in rural and urban areas during 2019

Fig. 4.7 Number of people killed in road accident during 2019

Figure 4.6 represents accident percentage in rural and urban areas during the year 2019. From the figure it is clearly that 61% of the road accidents happen in rural area only. Hence the number of people killed during the road accident in the rural area is always greater than the number of people killed during road accident in the urban area. This is clearly depicted in Fig. 4.7. From Fig. 4.4, it is very clear that during the year 2015, the total number of road accidents is found to be 501,423, in which 500,279 were injured and 146,133 were killed. Due to drastic precautionary measures taken by the government the total accident count was reduced during 2019. The accident count is found to be 449,002, in which 451,361 people were injured and 151,113 were killed. Even though the accident count was reduced, the number of people died in the accident does not make a great difference. This is because the people who are not involved in the accidents are not treated at proper time. If the information regarding the accident location is conveyed to the ambulance at proper time, then the fatality rate can be reduced drastically. To overcome this issue only we propose smart accident recognition and rescue system using IoT. The main objectives of this system are • To inform the accident location to the nearby hospital and police station through cloud server. • To inform one family member regarding the accident location using GSM module. • To alert the insurance company regarding the accident process, so that the insurance claiming procedure may be accelerated. • The deep forests algorithm has been constructed with hyper parameters that achieves the enhanced performance in several domains.

4 Internet of Things Based Smart Accident Recognition …

35

4.2 Related Works Mulla et al. [3] developed an ARM processor supported accident detection system which detects the occurrence of accident with the help of Shock sensor and sends information to the family member regarding the accident location through GSM module. Habib et al. [4] proposed a motor vehicle monitoring scheme, in which condition of the vehicle is periodically observed using an accelerometer sensor which is capable of sensing an accident based upon the position of the vehicle. This system sends an SMS to the family regarding the accident location. In addition to this the system also stores the accident location in the memory card for future reference. Similar system was designed by Rishi et al. [5] in which it intimates to multiple members including a family member, a friend and a doctor. Sherif et al. [6] developed a smart system which identifies the accident and sends the accident details like location of the accident, speed of the vehicle and number of person inside the vehicle to nearby hospital for speeding up the rescue operation. Alexandra Fanca [7] proposed an accident detection system which identifies the accident location and sends the information to the nearby emergency help center with the help of cloud server. Similar work is carried out [8–10]. Khalil et al. [11] proposed an accident prevention system which prevents the accident using Ultrasonic sensor. This sensor senses the objects and vehicles which are approaching near our system and alerts the driver through an alarm installed inside the vehicle.

4.3 Proposed System The system architecture is depicted in Fig. 4.8. In this system MQ3 sensor is used which will alert the Raspberry-pi when the driver has consumed alcohol. In turn Raspberry-pi will turn off the vehicles engine in order to prevent the driver from driving it in a drunken state [12]. In addition to this we have a vibration sensor which sends the alert signal to the Raspberry-pi whenever the vibration value exceeds the fixed threshold limit during the occurrence of an accident [13]. As a result of this the Raspberry-pi gets the GPS location of the accident spot and sends a message immediately to the family member with the support of GSM module [14]. During the same course of time the Raspberry-pi updates the accident location information to the cloud server through Wi-Fi module [15]. Now the cloud server searches for the nearby hospital and police station details in the database and collects the contact number using which the accident information is updated. This will help the ambulance driver and cops to arrive to accident spot and perform their duties quickly [16]. Accident information is also updated to the insurance company by the server, so that a person from insurance company can visit the accident spot and speed up the insurance claiming process [17]. The flow diagram of this system is elucidated using Fig. 4.9. The working procedure is explained evidently using the following steps.

36

K. Lakshmi Narayanan et al.

Fig. 4.8 Architecture diagram

Step 1: Step 2:

Step 3: Step 4:

Step 5:

Once the vibration sensor value crosses the threshold limit, it is noticed by Raspberry-pi. Then the processor sends an alert message to family member with the help of GSM module. The alert message contains the details of vehicle’s accident location and speed. Processor also updates the cloud server with the accident related information. Now the server looks into the database and fetches the contact information of nearby police station and hospital and updates the accident details to them. During the same time, Raspberry-pi sends the accident details to insurance company for further insurance claiming activities.

The deep forests algorithm is constructed to identify the data features for analyzing the accident severity. The correlations within every feature have been used for developing the model, every level has been maintained the feature data to the network level and the training samples are used to generate the estimated values.

4 Internet of Things Based Smart Accident Recognition …

37

Fig. 4.9 Flowchart

4.4 Results and Discussion Our system sends the accident alert information to the family member through a SMS with the help of GSM module which is depicted in Fig. 4.10. The message contains the accident location details and the speed at which the vehicle met with the accident. In addition to this, our system sends the accident related information Fig. 4.10 Accident information shared with family member

38

K. Lakshmi Narayanan et al.

to the nearby hospital, police station and insurance company through a web server [18]. The cloud server searches for the contact details of insurance company, nearby hospital and nearby police station in the data base and send the accident information to them for taking necessary actions this is clearly depicted in (Figs. 4.11 and 4.12). The accident details are regularly updated in the website which is displayed in Fig. 4.12. Here the accident spot and speed at which the accident took place are mentioned clearly at the webpage [19]. The vehicles which met up with an accident are updated in the accident vehicle database. This is depicted in Table 4.1. Once accident occurs, intimation is sent to the cloud server. The cloud server in turn searches for the nearby hospital details and

Fig. 4.11 Accident information sent to authorities via cloud server

Fig. 4.12 Website updating process

4 Internet of Things Based Smart Accident Recognition …

39

Table 4.1 Accident vehicle database Vehicle ID

Vehicle name

1

Maruthi Alto

TN 72 BC 7499

Mr. Sidhu

7708084020

2

Maruthi Swift

TN 67 AB 1215

Ms. Janani

9585761467

3

Hundai i10

TN 72 AJ 5554

Mr. John

7708710731

Table 4.2 Emergency database

Table 4.3 Police station database

Vehicle number

Vehicle owner name

Vehicle owner contact number

HOS_ID HOS_NAME

HOS_NUMBER

1

Cheranmahadevi Govt. Hospital 04634–261800

2

Pathamadai Govt. Hospital

04634–260497

3

Tirunelveli medical college

0462–2572611

PS_ID

PS_NAME

PS_NUMBER

1

Cheranmahadevi Police Station

04634–260125

2

Pathamadai Police Station

04634–260165

3

KTC Nagar Police Station

0462–2541520

stores it in the emergency database (depicted in Table 4.2) based on which an SMS is sent to the hospital for arranging an ambulance [20, 21]. Similarly the details of closest police station to the accident spot are identified by the cloud server and it is hoarded in police station database. This is depicted in Table 4.3.

4.5 Conclusion Our system prevents the occurrence of accident by alcoholic drivers by turning of the vehicles engine as soon as the system detects alcoholic nature of the driver. In addition to this, the system also spots the accident using vibration sensor and the intimation is sent to the family member via SMS. The location details and the vehicle users’ primary information are shared to the closest hospital and police station to attend the accident spot quickly. Our system also facilitates the insurance company to speed up the insurance claiming process as soon as they receive the accident related information from our system.

40

K. Lakshmi Narayanan et al.

References 1. https://www.who.int/publications/i/item/9789241565684 2. https://morth.nic.in/ 3. Mulla, J.M.S., Gavade, D., Bidwai, S.S., Bidwai, S.S.: Research paper on airbag deployment and accident detection system for economic cars. In: 2017 2nd International Conference for Convergence in Technology (I2CT), Mumbai, India, 2017, pp. 846–849 (2017). https://doi.org/ 10.1109/I2CT.2017.8226248 4. Khan, M.H.U., Howlader, M.M.: Design of an intelligent autonomous accident prevention, detection and vehicle monitoring system. In: 2019 IEEE International Conference on Robotics, Automation, Artificial-intelligence and Internet-of-Things (RAAICON), Dhaka, Bangladesh, 2019, pp. 40–42 (2019). https://doi.org/10.1109/RAAICON48939.2019.6263505 5. Rishi, R., Yede, S., Kunal, K., Bansode, N.V.: Automatic messaging system for vehicle tracking and accident detection. In: 2020 International Conference on Electronics and Sustainable Communication Systems (ICESC), Coimbatore, India, 2020, pp. 831–834 (2020). https://doi. org/10.1109/ICESC48915.2020.9155836. 6. Sherif, H.M., Shedid, M.A., Senbel, S.A.: Real time traffic accident detection system using wireless sensor network. In: 2014 6th International Conference of Soft Computing and Pattern Recognition (SoCPaR), Tunis, Tunisia, 2014, pp. 59-64 (2014). https://doi.org/10.1109/SOC PAR.2014.7007982 7. Fanca, A., Pu¸sca¸siu, A., V˘alean, H.: Accident reporting and guidance system: with automatic detection of the accident. In: 2016 20th International Conference on System Theory, Control and Computing (ICSTCC), Sinaia, Romania, 2016, pp. 150–155 (2016). https://doi.org/10. 1109/ICSTCC.2016.7790657 8. Hemanth Kumar, G., Ramesh, G.P.: Energy efficient multi-hop routing techniques for cluster head selection in wireless sensor networks. In: Computational Intelligence and Complexity,vol. 193, pp. 297–305. Springer (2021) 9. Khalil, U., Nasir, A., Khan, S.M., Javid, T., Raza, S.A., Siddiqui, A.: Automatic road accident detection using ultrasonic sensor. In: 2018 IEEE 21st International Multi-Topic Conference (INMIC), Karachi, Pakistan, 2018, pp. 206–212 (2018). https://doi.org/10.1109/INMIC.2018. 8595541 10. Hemanth Kumar, G., Ramesh, G.P.: Energy efficiency and data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–995 (2020) 11. Nasr, E., Kfoury, E., Khoury, D.: An IoT approach to vehicle accident detection, reporting, and navigation. In: 2016 IEEE International Multidisciplinary Conference on Engineering Technology (IMCET), Beirut, Lebanon, 2016, pp. 231–236 (2016). https://doi.org/10.1109/ IMCET.2016.7777457 12. Thirupathieswaran, R., Suria Prakash, C.R.T., Krishnan, R.S., Narayanan, K.L., Kumar, M.A. Robinson, Y.H.: Zero queue maintenance system using Smart Medi Care Application for Covid19 pandemic situation. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 1068–1075 (2021). https://doi. org/10.1109/ICICV50876.2021.9388454 13. Narayanan, K.L., Ram, C.R.S., Subramanian, M., Krishnan, R.S., Robinson, Y.H.: IoT based smart accident detection & insurance claiming system. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 306–311 (2021). https://doi.org/10.1109/ICICV50876.2021.9388430 14. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multim. Tools Appl. 79(5), 4011–4026 (2020) 15. Krishnan, R.S., Sangeetha, A., Kumar, A., Narayanan, K.L., Robinson, Y.H.: IoT based smart rationing system. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 300–305 (2021). https://doi.org/10. 1109/ICICV50876.2021.9388451

4 Internet of Things Based Smart Accident Recognition …

41

16. Krishnan, R.S., Kannan, A., Manikandan, G., Sri Sathya, K.B., Sankar, V.K., Narayanan, K. L.: Secured college bus management system using IoT for Covid-19 pandemic situation. In: 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), 2021, pp. 376–382 (2021). https://doi.org/10.1109/ICICV50876.2021.938 8378 17. Harold Robinson, Y., Santhana Krishnan, R., Raja, S.: A comprehensive study for security mechanisms in healthcare information systems using Internet of Things. In: Balas, V., Solanki, V., Kumar, R. (eds.) Internet of Things and Big Data Applications. Intelligent Systems Reference Library, vol. 180. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-391195_15 18. Lakshmi Narayanan, K., Ganesan, J., Kasinathaguru: Intranet based multiple equipment controller with text panel. Irish Interdiscip. J. Sci. Res. 4(3), 120–126 (2020) 19. Lakshmi Narayanan, D.K., Ramesh, G.P.: Robust and brittle secured video for IoT. Int. J. Eng. Technol. 7(2.20), 93–96 (2018) 20. Ramasamy Sankar Ram, C., Ravimaran, S., SanthanaKrishnan, R., Golden Julie, E., Harold Robinson, Y., Kumar, R., Son, L.H., Thong, P.H., QuangThanh, N., Ismail, M.: Internet of Green Things with autonomous wireless wheel robots against green houses and farms. Int. J. Distrib. Sens. Netw. 16(6) (2020) 21. Santhana Krishnan, R., Golden Julie, E., Harold Robinson, Y., Raja, S., Kumar, R., Thong, P.H., Son, L.H.: Fuzzy logic based smart irrigation system using Internet of Things. J. Clean. Prod. 252, 119902 (2020)

Chapter 5

Revolutionizing the Industrial Internet of Things Using Blockchain: An Unified Approach A. K. M. Bahalul Haque, Bharat Bhushan , Md.Rifat Hasan, and Md.Oahiduzzaman Mondol Zihad Abstract Blockchain brings some significant changes in many industries implementing a shared, decentralized public ledger. Characteristics like immutability, autonomy, and transparency make this technology desirable to all sectors. Industry 4.0 becomes more effective in building upon a concrete and robust system like blockchain since its integration provides trust, transparency, and sustainability. It can deal with several aspects of the industry like supply chain, tracking products, payments, database management, security, etc. Unlike other systems, there are some issues with company-level implementation and maintenance of blockchain. The solution to these issues can pave the way for blockchain to make industrial management more successful. In this chapter, we present a brief discussion on blockchain and its architecture. Specifically, we describe the classification and consensus mechanisms that are very important in the blockchain network. We also detail the transaction system in the blockchain. Later on, we discuss the fundamentals of Industry 4.0. Furthermore, we provide an explicit discussion on the usage of blockchain in the industry. Finally, we point out some issues that can create possible future research areas in blockchain for industries. Keywords Blockchain · Industry · Distributed · Industry 4.0 · IoT

A. K. M. B. Haque LUT School of Engineering Science, LUT University, Lappeenranta, Finland e-mail: [email protected] B. Bhushan (B) Department of Computer Science and Engineering, School of Engineering and Technology, Sharda University, Greater Noida, India M. Hasan · M. M. Zihad Department of Electrical and Computer Engineering, North South University, Dhaka, Bangladesh e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_5

43

44

A. K. M. B. Haque et al.

5.1 Introduction Blockchain [1] is one of the most productive technologies that is continually growing with newer features to modernize our living and civilization. Along with the impact on our daily lifestyle, it is creating a significant effect on industrialization [2, 3]. Modern industries need some essential organizational facilities to process and perform their business [4]. Every industry needs to maintain a relationship among its partners, such as transmitting messages, processing data, and working on a mutually agreeable contract. There are also actions like manufacturing, purchase policy, marketing, payments, complying with rules and regulations, logistics, and many more. These actions are primarily based on trust and transparency [5]. Blockchain offers a complete platform for all of these actions and manages these activities as proficiently as possible. Blockchain uses cryptographic security in a P2P network [6]. This is one of the most needed features for industrial data and process security. If an industry needs to function in a large production area on different geographical locations, it must conduct business with other local and international industries with its regulation procedures. This is a challenging task to comply with varying provisions while maintaining the legislation [7]. Blockchain can record the various functions based on different legislations maintaining the actual logged transaction. These transactions and agreements are not alterable once approved and stored in blockchain by any individual [8]. An industry must excel in industrial management by developing the company status and earn a profit at the highest margin. The industrial process management efficiency ranges from workforce development to product distribution [9, 10]. Blockchain provides all facilities to operate on a large scale and maintain all the industry’s needs. However, there are some shortcomings in some areas of which solutions need to be investigated. However, the accomplishment of those solutions will increase the scope of blockchain more for future industries. Many prominent industries and businesses are taking blockchain into account more seriously as it takes some conventional system paving the way for future industries [11]. Due to the possibilities and emerging issues mentioned above, it is understandable that, blockchain has a lot of potential application in terms of modernizing IoT integration in industry. This work comprises an in-depth explanation of blockchain principles and its integration with the industry, especially industry 4.0. Since industry 4.0 is heavily dependent on IoT devices, the possibility of blockchain integration in industrial IoT is a primising sector. For this reason, a detailed analysis regarding the importance of industrial IoT and the possibility of blockchain integration in it shall provide a significant insight for potential researchers and enthusiasts. A summary of the major contributions of this work can be outlined as follows: • Blockchain characteristics have been described briefly for establishing its uniqueness. • Block structure is explained briefly, including its various components.

5 Revolutionizing the Industrial Internet of Things …

45

• Various types of blockchain, including a tabular form of comparative analysis, are laid out. • Different layers of blockchain architecture are outlined briefly. • Various types of consensus algorithms are explained with a comparative analysis. • Fundamental of industry 4.0 is explained elaborately, including its characteristics. • The feasibility of blockchain integration in the industry is described briefly. • Potential blockchain-related industrial applications are laid out. • Future research directions included finally followed by the conclusion. The rest of the work is organized as: Sect. 5.2 contains blockchain fundamentals that comprise characteristics, block structure, layers, consensus algorithms, and transaction stages. Section 5.3 contians the fundamentals of the industrial environment, including industry 4.0. Section 5.4 describes why blockchain is impactful for the industry. Section 5.5 describes the application of blockchain in related areas. Section 5.6 explains the future research direction and challenges. Finally, Sect. 5.7 concludes the paper.

5.2 Blockchain Overview Blockchain comprises principles that are crucial for its integration into any environment. These principles include the unique attributes of blockchain, transaction scenario, consensus algorithms etc. [12]. A brief and precise explanation of the blockchian fundamentals can facilitate a solid understanding of the technology. For this reason the fundamentals are written below.

5.2.1 Characteristics Blockchain has some core characteristics. These characteristics can be very crucial for industry-level management and transactions [13]. Their brief description outlined below• Decentralization: In a decentralized system, every node is connected, implementing a P2P network. There is no central authority like a centralized system and no intermediaries that eventually reduce server, operation, and development [14]. • Immutability: It is impossible to alter any of the already stored information in the public ledger. Participants can never modify or delete the ledger data [15]. This one is highly beneficial for auditing data and financial transactions [16]. • Persistency: In a blockchain, every block is verified by other nodes. So, nodes cannot contain fake transactions [17]. If any block shows any malicious behavior, other features of blockchain trace them and resolve the transaction [18].

46

A. K. M. B. Haque et al.

• Anonymity: Anonymous transaction between sender and recipient is a must to build sustainable privacy. Blockchain can complete a successful transaction using a blockchain address only. For this reason, it does not reveal the real identity of the user [19, 20]. • Traceability: Users can trace records using the information of transactions like timestamps [21]. In this way, blockchain provides increased traceability and transparency to the network data [22]. • Transparency: After completing each transaction, a public ledger makes the records visible to the users [23]. For example, blockchain self-audit the ecosystem after every ten minutes and reunite the network transactions in bitcoin. This kind of openness creates transparency, improving the trust issue [24]. • Autonomy: The nodes in a distributed network can update and transfer information securely without any interference. So, there is no central authority to audit the transactions [25].

5.2.2 The Block Structure A block structure (Fig. 5.1) is divided into two parts [26]. The block body holds the data, transactions, and transaction counter, and the block header contains metadata like a nonce, timestamp, Merkle tree root hash, parent block hash, etc. The block header size is 80 bytes, and the size of the block body is not fixed. The number of transaction capabilities depends on block size. The first block has no parent block and is named as genesis block [27]. • Hash Function: This is similar to a digital fingerprint that generates a unique value code within a fixed length. This hash number is generated based on the input data using a unique algorithm called SHA-256 containing 64 characters [28]. Fig. 5.1 Block structure

5 Revolutionizing the Industrial Internet of Things …

47

• Timestamp: Timestamp keeps the record of a block when created and keeps the record when any block data have been modified [29]. It makes the block easier to track and verify, creating a particular date and time. • Merkle Tree Root Hash: Merkle tree increases the efficiency to store the large data structure within a specific period. Every node is labeled with a block as a leaf, and every transaction is connected with a Merkle tree [30]. To process all the transactions, it needs a different Merkle root generated by the hash process. • Nonce: Nonce is a 4-byte value connected to the hash function. It is used for mining by the miners to validate the block using PoW. If the nonce is changed, the hash will also change entirely following the avalanche effect [31].

5.2.3 Blockchain Classification Blockchain can be classified into three different types [32]. A brief description is given below for each of them• Public Blockchain: Everyone gets access to this type of blockchain network without any permission. It is built upon the decentralized concept that enables everyone to see the transparent transaction history and formulates the mining process. Nevertheless, the real identity of the users always remains anonymous. The transaction in this environment follows the P2P method, enabling all network participants to control the operating system [33]. The consensus mechanism public blockchain follows protects it from attacks like node failures and Sybil attacks. Protocols like PoW or PoS restricts the freedom of every node to create the block or by malevolent participants [34]. So, public blockchains are very secured. The most valuable part of the blockchain is ensuring all the users’ anonymity and a distributed ledger [35]. But it has some drawbacks regarding cost and speed. The cost of electric power increases and the system gets slower as more nodes are added. Still, it is faster and less expensive than the systems before blockchain. Bitcoin, Ethereum, and Litecoin are some examples of public blockchain [36]. • Private Blockchain: Private blockchain is a centralized type of blockchain. It is controlled by a central organization based on some rules for the network [37]. It brings the ancient centralized system giving access permission. It follows strict management and a deterministic distributed consensus, namely “Practical Byzantine Fault Tolerance (PBFT)” [38]. A dedicated team or an organization or some groups follow this consensus in the mining process to ensure transparency and restrict unknown users. The main drawback of a private blockchain is the lack of decentralization characteristics. However, in exchange for that, it is faster and efficient in terms of power usage than other blockchain types. So, private companies or organizations and governments are perfect to use this privately-run version of blockchain [39]. • Consortium Blockchain: This type of blockchain is the conjunction of both types mentioned above of blockchain. It is partly decentralized, allowing some controlling mechanism to the nodes for authentication. Nodes have the power to

48

A. K. M. B. Haque et al.

Table 5.1 Comparison of the types of blockchain Properties

Public [34, 36]

Consortium [40, 41]

Private [38, 39]

Consensus Visualization

All miners or nodes

Selected set of nodes among multiple organization

Single organization

Controlled By

All nodes

Predetermined group

Trusted users

Execution Speed

Slower

Fast

Fast

Access

Public

public or permissioned

Permissioned Access

Security

Very Good

Average

Below average

Consensus Protocols

PoW, PoS, DPoS

PBFT

PBFT, RAFT

Network Structure

Decentralized

Partially decentralized

Centralized

Immutability

Never be altered

Could be altered or not

Could be altered or not

Privacy

Low

Medium

High

Example

Bitcoin, Litecoin, Multichain, Blockstack, Hyperledger, R3, Ethereum, Blockstream, Blockchain, Corda, Ripple, Government Dash, Factom Quorum applications

verify the transaction instead of an individual [40]. Like public blockchain, the transactions can be open source. It has some issues regarding immutability. Nodes are prone to be malicious and alter transaction. However, it serves a great deal to the organizations consist of different partners [41]. Table 5.1 presents the comparison of the aforediscussed types of blockchain.

5.2.4 Consensus Algorithm In a P2P environment like blockchain where and no central authority remains in power, reaching consensus among ledgers in different nodes is a big issue. So, there are some protocols or algorithms to reach this consensus in such a disseminated environment known as Consensus Protocol or Algorithm. Consensus algorithms also work as a defense against attackers in blockchain [42, 43]. Some of the consensus algorithms are described below.

5.2.4.1

Proof of Work (PoW)

It is one of the most well-known and used consensus protocols. The main idea is that it needs some proof for miners to mine a block for validation [44]. The algorithm verifies each block before appending to the chain. Nevertheless, the miner or node must solve the cryptographic puzzle beforehand. The PoW works by setting up a

5 Revolutionizing the Industrial Internet of Things …

49

threshold T. To verify a block; miners need to guess an appropriate hash by changing the nonce. Th previous hash and timestamp of the block are also used to verify the information. The threshold changes over time, making it difficult and authenticate. The block will only be accepted when the hash value generated by changing the nonce is below the threshold [45, 46]. Nevertheless, it is challenging to find the secret nonce due to the SHA-256 algorithm making it more secured. The node can introduce the new block to the system after solving this puzzle. The miner gets rewarded for the ‘work,’ and the block is considered successfully mined. This process efficiently prevents double-spending attack and forking problem. But there are some drawbacks regarding energy cost for mining [47, 48].

5.2.4.2

Proof of Stake (PoS)

This algorithm addresses the shortcomings of PoW, like colossal energy cost and computation power. PoW faces 51% attack scenario where a pool with higher computation power can do most of the work. So, PoS is introduced with a different approach than PoW. It uses coinage as a stake. Coinage is measured as a currency amount multiplied by holding duration. For example, if Alice poses 10 coins for 10 days, the coinage will be 100. This is considered a stake along with the currency eventually helps to reduce the difficulty of mining. The node with more stake is also an indication of valid miners who will most likely not manipulate the system. PoS can be of pure stack-based or hybrid consensus. Both provide a better result, security, and energy efficiency [49, 50].

5.2.4.3

Delegated Proof of Stake (DPoS)

Daniel Larimer first introduces DPoS in April 2014 as an updated version of PoS to faster the transaction and make up the drawbacks. DPoS can be considered as delegated or representative democracy, whereas PoS can be considered as direct democracy. DPoS algorithm works based on two roles: witnesses and stakeholders. Witnesses can generate the block and obtain revenue. The delegates adjust the process of generating these blocks and transactions of fees. The votes of the stakeholders elect the members of these two roles. The delegates also take turns in voting to maintain the authenticity of the system. The critical difference between PoS and DPoS is the person to validate the block [51]. In PoS, the stakeholder itself validates the block. However, in DPoS, the delegates chosen by the stakeholders serve the purpose. It has some extra benefits over PoS. one of them is DPoS is faster as the space for participants gets fewer. It is also a very effective, efficient, and versatile consensus protocol. A particular node needs not to consider a large number of untrusted nodes. The delegates can easily adjust block intervals and size suitably. Hence, the confirmation time reduces to seconds, updating the cryptocurrency system to the next level [52].

50

5.2.4.4

A. K. M. B. Haque et al.

Practical Byzantine Fault Tolerance (PBFT)

It is a permissioned consensus algorithm to develop the BFT protocol. PBFT allows the nodes to decide major rules reducing the complexity to polynomial rather than exponential in BFT. The entire phase of PBFT is a combination of three distinguished steps [53]. These are namely pre-prepared step, prepared step and commit step. Two types of nodes, leader and backup, operate the mining process. The leader node passes any request it receives to the backup nodes. Here, to prove the majority votes’ integrity, all the nodes can exchange contact [54]. It is sustainable to fault nodes as it can tolerate up to f fault nodes if there are at least 3f + 1 node remains in the system. It can handle 1/3 of malicious nodes. The request gets passed through the phases only after getting 2/3 of the participating nodes’ votes. PBFT meets the time required to implement it in the commercial process but has a weak synchronous protocol. So, PBFT is mostly suitable where nodes can be more reliable, like in private or consortium blockchain [55, 56]. There are other consensus algorithms for specific tasks and environments. More than thirty algorithms like PoB [57], PoC [58], Ripple [59], PoET[60, 61], etc. exists for Blockchain network. These protocols are introduced to develop the previously discussed main algorithm for the blockchain system. Some are for specific purposes in some particular environments. Nevertheless, no consensus algorithm is perfect for all the systems. So more research is still going on to solve the issues in various applications. Table 5.2 presents the comparison of various consensus protocols in terms of various metrics such as approach, processing speed, resource consumption, latency and several others.

5.2.5 Blockchain Architecture A combination of technologies like digital signature, cryptographic hash, consensus algorithm, transactions, applications, etc. are amenable to build blockchain technology. These technologies can be considered as the layers of a network. The type of layers may vary for different environments and applications [62, 63].

5.2.5.1

Data Layer

The data layer contains the procedures and techniques of the data. The layer mainly consists of the block structure such as timestamp, Merkle tree, data block, hash function, asymmetric encryption and chain structure. The data layer’s main goal is to create the data and append it with appropriate tags for verification [64]. The block has its own and parent hash function. It is validated upon addition to the blockchain from where parent hash is acquired. The chain goes on as the miners append a new block to the system. This structure is called the Merkle root tree and the root node is

5 Revolutionizing the Industrial Internet of Things …

51

Table 5.2 Comparison of some popular consensus algorithm Characteristics

Consensus algorithm PoW [44–48]

PoS [49, 50]

DPoS [51, 52]

PBFT [53–56]

Approach

Prevent attacks like double spending or forking issue

Resolve the shortcomings of Proof of Work

Faster the transaction and make up the shortcomings of PoS

Two nodes can reach consensus even if there is some tampered node

Processing Speed

Slow

Fast

Fast

High

Resource Consumption

High

Low

Low

High

Latency

High

High

Low

Low

Permission

No

No

No

Yes

Language

C++, Solidity

Native

Native

Java

Energy Efficiency

Low

High

High

High

Number of Nodes

Unlimited

Unlimited

Unlimited

Limited

Scalability

High

High

High

Low

Example

Bitcoin

Peercoin, Nextcoin

Bitshares

Hyperledger Fabric

called the genesis block. Again, checking the timestamp, the Merkle tree can verify the efficiency and integrity of the system [65].

5.2.5.2

Network Layer

The network layer has various data verification mechanisms, such as distributed networking, data dissemination, network architecture, and data forwarding [66]. Blockchain uses digital signatures in P2P network topology. For this type of public network, authenticity is ensured with asymmetric cryptography.

5.2.5.3

Consensus Layer

It is a big challenge to implement blockchain technology without a single security issue. If any data is tampered with by any malware or threats and the network becomes exposed to malicious activity, there is no means to implement blockchain technology. So, it is crucial to have a superior look to protect the architecture and prevent all kinds of threats. Many consensus algorithms are proposed in order to avoid threats and resolve security issues, such as PoW, PoS, DPoS, PBFT, PoET, etc. [67]. To set

52

A. K. M. B. Haque et al.

a verification system through which blocks will be created, which transaction will be made and security will be ensured is the consensus layer’s primary goal [68].

5.2.5.4

Incentive Layer

The incentive layer inspires the nodes to take part in the process of blockchain. Every node needs to compete with self-interest in the block creation and data validation process in return for rewards. So, this layer consists of an allocation and issuance mechanism. It also creates a secure environment for transactions [69].

5.2.5.5

Contract Layer

This layer provides a combination of rules based on which transaction of coins, assets, rights, etc. can be transmitted over the network. In this case, the contract layer uses smart contracts, algorithms, and various scripts to conduct the transaction. Moreover, smart contract work as a ledger between the nodes for a trustless environment. The contract is signed cryptographically if the nodes agreed with the rules [70].

5.2.5.6

Application Layer

The application layer indicates various scenarios like smart city, market security, business application, IoT, digital identity, intellectual property, etc. [71]. Indeed, it provides the outcome of a blockchain system that can be implemented. It has an execution layer that consists of APIs, different frameworks, user interfaces, and scripts for the user to participate in the blockchain network [72].

5.2.6 Transaction Stages in Blockchain The transaction process generally follows five steps [73]. These steps can vary from platform to platform but the basic idea is almost similar. Let us assume that Alice wants to give 5 bitcoin (BTC) to Bob. • Generate Transaction: Alice first needs to generate the transaction data following the Unspent Transaction Outputs (UTXOs). It refers to the transferrable bitcoin must be unspent. Alice then generates a hash of the source data and the public key of the receiver following SHA-256. It includes the bitcoin amount and data encrypted with the private key by Alice. Using this hash value, the input transaction can be uniquely identified. It can also cancel the transaction upon differing or altering. In this way, the hash will work as a digital signature keeping the transaction safe from tampering [74].

5 Revolutionizing the Industrial Internet of Things …

53

• Confirm Transaction: After the generation of the transaction, Bob, the receiver, needs to confirm. Now Bob has to look for several things. He has to be sure of the fact that the transaction is not a case of double-spending. It can be solved by verifying the signature of both participants. The output transaction can only be redeemed after this verification. Again, only the valid blocks can be appended into the network. So, the transaction is also considered valid if it becomes a part of the blockchain. A transaction will be legal if the total input UTXOs is equal to or greater than the total output UTXOs. This is called the law of conservation of values [75]. • Claiming Ownership: Bob now needs a private key and public key to claim the transaction’s ownership. A recipient can generate a valid signature using the private key. The hash containing the receiver’s public key from the transaction generation stage decrypts precisely as the sender sends the transaction. This digital signature must match; otherwise, the transaction will be canceled, considering it a malicious entity [76]. • Mining and Consensus: The blocks generated after encryption need to be distributed ledger where specific criteria should be made for authentication. Different algorithms set these criteria enabling the participants, Alice and Bob, to reach a consensus that the block is valid. Algorithms like PoW can work correctly as it needs to solve the cryptographic puzzle with massive work to verify the block’s authenticity. A hash value smaller than the target value is generated with the block information like previous hash, timestamp, block version, nonce, and Merkle root hash. This target value also changes from time to time to maintain the authenticity of the block [77]. • Block Validation: In the validation stage, Bob needs to ensure that the transaction’s occurrence follows the reference. Again, the current block must include the previous hash and the timestamp should be accurate. At last, Bob needs to ensure that the block is validated through PoW [78]. After completing all these stages, Bob will receive the transaction, and Alice will get the confirmation of the finished transaction.

5.3 Blockchain for Industry Blockchain is an emerging technology that is evolving rapidly in recent times. Its characteristics and benefits create a boundless possibility for different applications. Blockchain has characteristics like immutability as well as transparency that is very rare in the industrial sectors. Another critical aspect of blockchain is security, which is a must for business. Problems in the transaction, supply chain management, managing workers, and legal issues can be tackled using blockchain. Blockchain brings advanced features to the industry [79]. These features serve as the basement for

54

A. K. M. B. Haque et al.

various business models. These new models are reshaping the modern world of business, bringing enormous healthcare, finance, supply chain management, garments, IoT sectors, automobile, etc. In industry, until now there four revolutions happen. These are the first industrial revolution, the second industrial revolution, the third industrial revolution, and the latest is the fourth industrial revolution. The latest revolutions facilitate automation, optimization, efficiency, high production rate, and sustainability [80]. There are various elements (Technical enablers) that modernize the entire industrial sector [81, 82]. Industry 4.0 is production intensive; that means it had to produce more and more consumption. There was not much importance given to data generated at each step of the supply chain. Industry 4.0 is now not looking at individual computerized machines; it is looking at the whole network. Smart product is of the different components of the Industry 4.0 [83]. It consists of resource control, along with other things, to complete the manufacturing process. Another paradigm is the smart machines that replace the traditional production environment with optimized, state-of-the-art distributed systems. Some design principles in Industry 4.0 allow businesses to investigate a potential transformation from their current systems to Industry 4.0 technologies. Generally, there are six design principles [84] depending on the business scenario. It can be more or less. The six design principles are described as blow. • Interoperability: It creates a shared platform for devices, people, and other entities to communicate. It used IoT and IoS networks to communicate. • Service-Oriented: Industry 4.0 framework must be developed, taking into consideration the IoS. • Decentralization: Cyber-physical system (CPS) should work independently, which gives a more flexible environment for customized products. • Real-time Capability: CPS and other systems must work in real-time to provide live insights about the business operations and find issues related to production [85]. • Modularity: All the components should be able to plug and play with minimum configuration. This will enable businesses to upgrade or downgrade when the market needs efficiently. • Virtualization: Here is creating a digital copy of the physical products and services. CPS must be able to replicate the real world in a virtual environment. Moreover, this will help in running simulations of scenarios [86].

5.4 Why Does the Industry Need Blockchain The industrialization phase of today’s world is going through a modification phase. Different types of startups are coming with enormous potential. Some of these ideas are ahead of our time, and appropriate methods are being discussed worldwide. Blockchain can be an excellent platform to implement these ideas and very useful

5 Revolutionizing the Industrial Internet of Things …

55

for the future world. Moreover, the industries built on a strong foundation face additional security, management, and transaction problems. These issues can also be solved with developed blockchain architecture. Today’s industries are based on a shared approach. More than one partners and investors together take the decisions for proceedings. Hence, the transaction must be very secured and need to neutralize any potential threats. The shared ledger distribution system of blockchain is perfect for ensuring detailed tracking, management, and transaction tracking. It creates specific rules to reach an agreement. After the confirmation of all the shareholders, the agreement can be registered into the blockchain. There may also be activities like voting, purchasing, validating, issuing, surveying, or analyzing a particular type of product or decision. The use of blockchain keeps all of these activities secured and light-hearted [87]. Another purpose of using blockchain in the industry is cryptographic activity. The agreements that are registered in the block can only be registered when it is safe. This is ensured through the distribution network procedure as we discussed earlier. However, these registered agreements or transactions should also be maintained not to be accessed at anybody’s will or tampered with easily. The cryptographic algorithm of blockchain makes the registered agreements unknown to everyone but known only to get permission. The digital signature approach and other components of blockchain-like hash function and consensus algorithm ensure the correctness of the system and keep track of every step. All these benefits inspire the world to deploy blockchain in the industry level at a higher level that can confirm the security of data worldwide [88]. Blockchain is representing a capable technological platform that supports all industry applications. This area is included by supply chain and trade across manufacturing, health, food, and creative industries. Blockchain drives the fourth industrial revolution by storing data, securing trust, and transferring value. Blockchain is an automated process, and it will remove all manual activities by eliminating the needs of intermediaries or mediators’ activity. Characteristics of Blockchain are already experimented with and tested in the financial sector. It has performed well to verify any transactions safely and quickly without any manual involvement [89].

5.5 Blockchain Apllications in Industry Blockchain is one of the latest addition for providing automation, security, privacy, product racing, etc. In this section, the use of blockchain in industry 4.0 is explained briefly [90].

56

A. K. M. B. Haque et al.

5.5.1 Automation of Supply Chain Supply chain management can be extraordinarily complex, depending on the products. A supply chain network has various elements that are spread across various regions and phases. These various elements facilitate the product distribution network with the necessary ingredients for fulfilling user demands. Our supply chain can be broken down into several entities due to the complexity and lack of transparency. These entities can adopt blockchain. Blockchain technology’s unique characteristics can revolutionaries the sector in terms of automation, privacy, transparency, and security [91].

5.5.2 Blockchain-Based Security and Privacy Industry 4.0 cannot apply the traditional security and privacy enforcement directly due to its nature of integration with other technologies [92]. Industry 4.0, being integrated gradually with many other technologies, is a potential security attack sector for the intruders [93]. Moreover, IoT devices are crucial for the latest industrial advancement. Due to IoT’s various attributes, there are probable security risks such as a single point of failure, jamming attacks, interceptions, etc. [94]. Any technology that provides security for the industrial environment needs to be scalable, lightweight and trusted. Moreover, the technology shall have to protect confidentiality, integrity, and availability. Blockchain is a technology that is implemented in a distributed environment. It facilitates immutability, transparency, non-repudiation as its prime characteristics. Considering these attributes, blockchain can provide a very efficient, secure and trusted industrial solution for the industry. Blockchain can be used for identity management, authentication purpose and defense against cyberattacks e.g. resilience against the DDoS attacks in the IoT environment [95].

5.5.3 Tracking and Tracing Product Manufacturing Phases Every process has a lifecycle. The same goes for the manufacturing process. The lifecycle of any product manufacturing process requires various phases. It is vital to keep track of what is happening in each of these phases. From collecting raw materials, logging the items in the database, starting the manufacturing process till the waste management falls under the product life cycle management process [96]. Blockchain can be beneficial with all of these phases included in any manufacturing process. This allows the stakeholders to check the product life cycle performance in real-time without contacting or reviewing customers. However, suppliers cannot monitor the product life cycle manually on the traditional system. For this reason,

5 Revolutionizing the Industrial Internet of Things …

57

there is a lack of transparency in the whole system. Blockchain can be included in different tasks of the lifecycle. Some of them are given below [97]. • • • • • • • •

Version Controlling and Management System. Track past, present, and upcoming ideas and innovations. Indexing the raw materials. Managing customers’ expectations and product intent. Manufacturing process Tracing. Logging and arranging various physical and chemical attributes. Integrating enterprise systems. Keep updated the organization with product knowledge.

5.5.4 Payment Systems Nowadays, with the help of IoT, the payment system is made automated. This allows all devices on IoT to make decisions on its own when required. An example, devices are directly paid for the electricity that they consumed. Nowadays, we do not have enough scope to do that everywhere; the current payment method is not suitable due to the high transaction cost and limited capacity. Also, there are many complications in using the credit card system and sharing the card information with our devices for payment. In this case, blockchain can come to the rescue. Some of these reasons for that are given below• Low transaction rate: Normally, we used third-party payment services, and those are quite expensive. To reduce the transaction rate, we can avoid it and start using blockchain. Also, we can use crypto transactions, the transaction fee is low, and it will be reduced more in the future. • Immediate payment: In cryptocurrencies, the money is credit within a few minutes and does not need to wait for two or three days. The crypto transactions are processed 24/7 relentlessly. • Fair distribution: smart contract can make the total system automated. For this reason, there will be no need for human intervention [98].

5.5.5 Cloud and Edge Computing Most current modern organizations depend on applications conveyed on the neighborhood or distant distributed computing frameworks. The frameworks permit numerous industry 4.0 members to work together among them in a simple manner. However, such a sort of framework experiences the ill effects of a significant impediment [99]. If the cloud is affected by programming issues, high outstanding tasks at hand, or assaults, the entire framework may hinder each client. Although cloud frameworks empower adjusting the remaining task at hand and appropriating such a heap, most of them were not imagined without preparation as P2P frameworks.

58

A. K. M. B. Haque et al.

Interestingly, blockchain-based frameworks are planned as conveyed frameworks that can supplement distributed computing arrangements in specific perspectives for example, distributed storage processing and management [100, 101] and fog and edge computing based architectural framework [102, 103]. Apart from the researches mentioned above recently, researchers have proposed various other industry-oriented researches. Guo et al. [104] discussed blockchain for the banking industry. The authors also discussed various regulatory efficiency to provide security and transparency. The agricultural industry is vital for maintaining food supply and sustainability. Tripoli et al. [105] described the agricultural industry, supply chain and the agricultural land registry related implementations using blockchain technology. Czachorowski et al. [106] discussed a unique industrial idea related to blockchain for maritime industries. The authors proposed their ideas about facilitating control, inspection, audit, cost reduction and system security. Fraga-Lamas et al. [107] provided a use case analysis in the automotive industry. The analysis focused on increasing trust, security, integrity, authenticity and robustness. Jovovi´c et al. [108] focused on the possible use of blockchain technology for the fourth industrial concept. Among many other topics, the authors also discussed the blockchain-based storage system. Fernández-Caramés et al. [109] focused on a similar area, but the authors focused on blockchain and smart contract application use cases for industrial security purposes. Among various other notable industrial applications, Ozdemir et al. [110] discussed blockchain-based application assessment for the tourism industry, Lu et al. [111] focused on cost reduction of the oil industry. Moreover, Papathanasiou et al. [112] proposed blockchain integration in the Greek shipping industry to increase trust, confidentiality, security, and Chen et al. [113] discussed blockchain-enabled efficient and cost-effective supply chain for the retail industry.

5.6 Future Issues and Research Directions It is undeniable that blockchain has great potential for industries. However, it is also true that the extent of this potential is still unknown to us. So, it is evident that some issues will occur in the system to implement it that needs more research.

5.6.1 Security Blockchain is considered suitable for the industrial sector for having concrete security. Nevertheless, it has some issues regarding security. There are some risks like message hijack, smart contract program vulnerabilities, etc. [114]. Privacy leakage is another concern. Blockchain is an entirely internet-based application creating a vast network. Hence, it is open for attacks like hacking, stealing, spy attempts, DoS

5 Revolutionizing the Industrial Internet of Things …

59

attacks. There is also an issue of double-spending that creates an invalid transaction. Blockchain has its mechanism to prevent this, but it still needs much work to deploy in multiple platforms with many participants [115, 116].

5.6.2 Integration Blockchain is a combination of technology that works together. To implement it in industrial sectors, it has to maintain coherence with various types of components. Hence, it a very complex mechanism to integrate all of it at once. Any integration process with lots of functions should be readable and work faster. If any of the parts cannot work correctly, the other part should work as a backup. Otherwise, it will be open for vulnerability, privacy concerns, and availability issues. More research is needed to make its integration efficient and faster [117].

5.6.3 Resource Constraints To enjoy blockchain’s benefit, the system must have high computation power, scalability, and stability. Several blockchain functions, such as the mining process, block creation, validation, etc. need to be done on a continuous process. However, this is a challenging task. IoT devices that blockchain use nowadays may support it for now. Nevertheless, day by day, the need for more powerful devices and network systems will arise. It is crucial to research this area to meet blockchain resources’ future needs to implement it in big industries [118].

5.6.4 Scalability The distributed ledger of blockchain has been deployed for a specific blockchain size where transactions and security of data are ensured entirely. However, on a larger scale, like in industry, scalability is a concern of the research too on a global stage. The bigger the blockchain, the slower the process gets. The consensus mechanism needs a certain amount of time for the mining process. When it comes to the more significant scenario, it gets slower to synchronize with the system [119]. Hence, more work should be done regarding the information update, synchronization, and maintain all the entities for a larger scale.

60

A. K. M. B. Haque et al.

5.6.5 Regulations Blockchain is a decentralized network system. It has no third-party that can perform the role of authority and monitor everything. However, to implement anything on a large scale, ceratin rules and regulations should be adopted and followed along with the current industry standards. Otherwise, the sustainability of the industrial implementation of blockchain will be at stake.

5.7 Conclusion In this chapter, we focus on the basics of the blockchain and its importance in the industry. There are many possibilities to overcome the issues of the traditional and centralized industrial system. We demonstrated that blockchain applications could face many challenges in industrial domains from various studies if used efficiently. We also mention some shortcomings of blockchain in adaptation and implication. Nevertheless, blockchain can develop further solving these issues enabling vast applications in industries. Emphasis should reduce costs in consensus mechanisms, proper data management, integrate different industries, and balanced government regulation. It will bring scalability, efficiency, and reliability to the overall system. More research in the fields and applications discussed in this chapter will uplift blockchain in the industry worldwide.

References 1. Nakamoto, S., Bitcoin, A.: A peer-to-peer electronic cash system. Bitcoin. https://bitcoin.org/ bitcoin.pdf (2008). https://doi.org/10.1007/978-3-030-17740-9_3 2. Collomb, A., Sok, K.: Blockchain/distributed ledger technology (DLT): what impact on the financial sector? Digiworld Econ. J. (103) (2016) 3. Joshi, A.P., Han, M., Wang, Y.: A survey on security and privacy issues of blockchain technology. Math. Found. Comput. 1(2), 121–147 (2018). https://doi.org/10.3934/mfc.201 8007 4. Andoni, M., Robu, V., Flynn, D., Abram, S., Geach, D., Jenkins, D., McCallum, P., Peacock, A.: Blockchain technology in the energy sector: a systematic review of challenges and opportunities. Renew. Sustain. Energy Rev. 100, 143–174 (2019). https://doi.org/10.1016/j.rser. 2018.10.014 5. Saxena, S., Bhushan, B., Ahad, M.A.: Blockchain based solutions to Secure Iot: Background, integration trends and a way forward. J. Netw. Comput. Appl. 103050 (2021).https://doi.org/ 10.1016/j.jnca.2021.103050 6. Kumar, A., Abhishek, K., Bhushan, B., Chakraborty, C.: Secure access control for manufacturing sector with application of ethereum blockchain. Peer-to-Peer Netw Appl (2021). https://doi.org/10.1007/s12083-021-01108-3 7. Dyrdonova, A.N., Shinkevich, A.I., Galimulina, F.F., Malysheva, T.V., Zaraychenko, I.A., Petrov, V.I., Shinkevich, M.V.: Issues of industrial production environmental safety in modern economy. Ekoloji 27(106), 193–201 (2018)

5 Revolutionizing the Industrial Internet of Things …

61

8. Hwang, J., Choi, M.I., Lee, T., Jeon, S., Kim, S., Park, S., Park, S.: Energy prosumer business model using blockchain system to ensure transparency and safety. Energy Procedia 141, 194–198 (2017). https://doi.org/10.1016/j.egypro.2017.11.037 9. Sunaryo, D.: The effect of net profit margin, return on asset, return on equity on share prices in The Southeast Asian metal industry. Int. J. Sci., Technol. Manage. 1(3), 198–208 (2020). https://doi.org/10.46729/ijstm.v1i3.47 10. Mahdi, M., Khaddafi, M.: The influence of gross profit margin, operating profit margin and net profit margin on the stock price of consumer good industry in the Indonesia stock exchange on 2012–2014. Int. J. Business, Econ., Soc. Dev. 1(3) (2020). https://doi.org/10.46336/ijbesd. v1i3.53 11. Gupta, R., Tanwar, S., Kumar, N., Tyagi, S.: Blockchain-based security attack resilience schemes for autonomous vehicles in industry 4.0: a systematic review. Comput. Electr. Eng. 86, 106717 (2020). https://doi.org/10.1016/j.compeleceng.2020.106717 12. Goyal, S., Sharma, N., Kaushik, I., Bhushan, B.: Blockchain as a solution for security attacks in named data networking of things. Secur. Priv. Issues IoT Devices Sens. Netw. 211–243 (2021).https://doi.org/10.1016/b978-0-12-821255-4.00010-9 13. Tayal, A., Solanki, A., Kondal, R., Nayyar, A., Tanwar, S., Kumar, N.: Blockchain-based efficient communication for food supply chain industry: transparency and traceability analysis for sustainable business. Int. J. Commun. Syst. 34, e4696 (2021).https://doi.org/10.1002/dac. 4696 14. Lin, I.-C., Liao, T.-C.: A survey of blockchain security issues and challenges. I. J. Netw. Secur. 19, 653–659 (2017). https://doi.org/10.6633/IJNS.201709.19(5).01 15. Zheng, Z., Xie, S., Dai, H., Chen, X., Wang, H.: Blockchain challenges and opportunities: a survey, Int. J. Web Grid Serv. 14(4) (2018). https://doi.org/10.1504/IJWGS.2018.095647 16. Zhang, F., Ding, Y.: Research on anti-tampering simulation algorithm of block chain-based supply chain financial big data. In: 2021 IEEE 2nd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE). https://doi.org/10.1109/ ICBAIE52039.2021.9389859 17. Purohit, S., Calyam, P., Alarcon, M.L., et al.: HonestChain: consortium blockchain for protected data sharing in health information systems. Peer-to-Peer Netw. Appl. (2021). https:// doi.org/10.1007/s12083-021-01153-y 18. Li, X.H.: Blockchain-based Cross-border E-business Payment Model. https://doi.org/10.1109/ ECIT52743.2021.00022 19. Yassein, M.B., Shatnawi, F., Rawashdeh, S., Mardin, W.: Blockchain technology: characteristics, security and privacy; issues and solutions. In: 2019 IEEE/ACS 16th International Conference on Computer Systems and Applications (AICCSA), Abu Dhabi, United Arab Emirates, pp. 1–8 (2019). https://doi.org/10.1109/AICCSA47632.2019.9035216. 20. M¨oser, M.: Anonymity of bitcoin transactions: an analysis of mixing services. In: Proceedings of M¨unster Bitcoin Conference, M¨unster, Germany, pp. 17–18 (2013). 21. Yiu, N.C.K.: Toward blockchain-enabled supply chain anti-counterfeiting and traceability. Future Internet. 13(4), 86 (2021). https://doi.org/10.3390/fi13040086 22. Zhu, P., Hu, J., Li, X., Zhu, Q.: Using blockchain technology to enhance the traceability of original achievements. https://doi.org/10.1109/TEM.2021.3066090 23. AlShamsi, M., Salloum, S.A., Alshurideh, M., Abdullah, S.: “Artificial Intelligence and Blockchain for Transparency in Governance”—Part of the Studies in Computational Intelligence book series (SCI, volume 912) 24. Rathee, P.: Introduction to blockchain and IoT. In: Advanced Applications of Blockchain Technology, Studies in Big Data, vol. 60 (2020). https://doi.org/10.1007/978-981-13-877 5-3_1 25. Xu, G., Dong, J., Ma, C.: A certificateless encryption scheme based on blockchain. Peer-toPeer Netw. Appl. (2021). https://doi.org/10.1007/s12083-021-01147-w 26. Huang, S., Wang, G., Yan, Y., Fang, X.: Blockchain-based data management for digital twin of product. J. Manuf. Syst. 54, 361–371 (2020). https://doi.org/10.1016/j.jmsy.2020.01.009, ISSN 0278–6125. https://www.sciencedirect.com/science/article/pii/S0278612520300091

62

A. K. M. B. Haque et al.

27. Cui, G., Shi, K., Qin, Y., Liu, L., Qi, B., Li, B.: Application of block chain in multi-level demand response reliable mechanism. In: Proceedings of IEEE ICIM’17, Chengdu, China, pp. 337–341 (2017). https://doi.org/10.1109/INFOMAN.2017.7950404 28. Yuan, Y., Wang, F.: Towards blockchain-based intelligent transportation systems. In: 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC), Rio de Janeiro, pp. 2663–2668 (2016). https://doi.org/10.1109/ITSC.2016.7795984 29. Estevam, G., Palma, L.M., Silva, L.R., Martina, J.E., Vigil, M.: Accurate and decentralized timestamping using smart contracts on the Ethereum blockchain. Inf. Process. Manage. 58(3), 102471 (2021). https://doi.org/10.1016/j.ipm.2020.102471, ISSN 0306–4573. https://www. sciencedirect.com/science/article/pii/S0306457320309602 30. Wei, P.C., Wang, D., Zhao, Y., Tyagi, S.K.S., Kumar, N.: Blockchain data-based cloud data integrity protection mechanism. Future Gener. Comput. Syst. 102, 902–911 (2020). https:// doi.org/10.1016/j.future.2019.09.028, ISSN 0167–739X. https://www.sciencedirect.com/sci ence/article/pii/S0167739X19313494 31. Sharma, S., Jain, S., Chandavarkar, B.R.: Nonce: life cycle, issues and challenges in cryptography. In: Kumar, A., Mozar, S. (eds.) ICCCE 2020. Lecture Notes in Electrical Engineering, vol. 698. Springer, Singapore. https://doi.org/10.1007/978-981-15-7961-5_18 32. Buterin, V.: On public and private blockchains (2015). https://blog.ethereum.org/2015/08/07/ on-public-and-private-blockchains/ 33. Pincheira, M., Vecchio, M., Giaffreda, R., Kanhere, S.S.: Cost-effective IoT devices as trustworthy data sources for a blockchain-based water management system in precision agriculture. Comput. Electron. Agric. 180, 105889 (2021). https://doi.org/10.1016/j.compag.2020. 105889, ISSN 0168–1699. https://www.sciencedirect.com/science/article/pii/S01681699203 30945 34. Manimuthu, A., Sreedharan, V., R., Rejikumar, G., & Marwaha, D.: A literature review on bitcoin: transformation of crypto currency into a global phenomenon. IEEE Eng. Manage. Rev. 47(1), 28–35 (2019). https://doi.org/10.1109/emr.2019.2901431 35. Huang, K., Zhang. X., Mu, Y., Rezaeinbagha, F., Du, X.: Scalable and redactable blockchain with update and anonymity. https://doi.org/10.1016/j.ins.2020.07.016 36. Douceur, J.R.: The sybil attack. Peer-to-Peer Systems Lecture Notes in Computer Science, 251–260 (2002).https://doi.org/10.1007/3-540-45748-8_24 37. Ghosh, B.C., Bhartia, T., Addya, S.K., Chakraborty, S.: Leveraging Public-Private Blockchain Interoperability for Closed Consortium Interfacing. 10 pages, 12 figures, accepted for publication in IEEE INFOCOM 2021, Distributed, Parallel, and Cluster Computing (cs.DC). arXiv: 2104.09801 38. Puthal, D., Malik, N., Mohanty, S.P., Kougianos, E., Das, G.: Everything you wanted to know about the blockchain: its promise, components, processes, and problems. IEEE Consum. Electron. Mag. 7(4), 6–14 (2018). https://doi.org/10.1109/mce.2018.2816299 39. Castro, M., Liskov, B., et al.: Practical byzantine fault tolerance 99 (1999), OSDI173–186 40. Bhushan, B., Khamparia, A., Sagayam, K.M., Sharma, S.K., Ahad, M.A., Debnath, N.C.: Blockchain for smart cities: a review of architectures, integration trends and future research directions. Sustain. Cities Soc. 61, 102360 (2020).https://doi.org/10.1016/j.scs.2020.102360, ISSN 2210-6707 41. Huang, X., Zhang, Y., Li, D., Han, L.: An optimal scheduling algorithm for hybrid EV charging scenario using consortium blockchains. Future Gener. Comput. Syst. 91, 555–562 (2019). https://doi.org/10.1016/j.future.2018.09.046 42. Lamport, L.S.: The byzantine generals problem. ACM Trans. Program. Lang. Syst. 4(3), 382–401 (1982). https://doi.org/10.1145/357172.357176 43. Dedeoglu, V., Jurdak, R., Dorri, A., Lunardi, R.C., Michelin, R.A., Zorzo, A.F., Kanhere, S.S.: Blockchain Technologies for IoT. Springer Nature Singapore Pte Ltd. (2020). https:// doi.org/10.1007/978-981-13-8775-3_3 44. Dwork, C., Naor, M.: Pricing via processing or combatting junk mail. In: Brickell, E.F. (eds.) Advances in Cryptology—CRYPTO’ 92. CRYPTO 1992. Lecture Notes in Computer Science, vol. 740. Springer, Berlin, Heidelberg (1993). https://doi.org/10.1007/3-540-48071-4_10

5 Revolutionizing the Industrial Internet of Things …

63

45. Nakamoto, S.: Bitcoin: a peer-to-peer electronic cash system (2008). Retrieved 27 September 2019 from https://bitcoin.org/bitcoin.pdf 46. Wan, S., Li, M., Liu, G., et al.: Recent advances in consensus protocols for blockchain: a survey. Wireless Netw (2019). https://doi.org/10.1007/s11276-019-02195-0 47. Bitcoinwiki, “SHA-256,” 2016. https://en.bitcoin.it/wiki/SHA-256 48. Nguyen, G.-T., Kim, K.: A survey about consensus algorithms used in blockchain. J. Inf. Process. Syst. 14(1), 101–128 (2018).https://doi.org/10.3745/JIPS.01.0024 49. Bentov, I., Gabizon, A., Mizrahi, A.: Cryptocurrencies without proof of work. In: Financial Cryptography and Data Security. Heidelberg: Springer, pp. 142–157 (2016). https://doi.org/ 10.1007/978-3-662-53357-4_10 50. King, S., Nadal, S.: PPCoin: peer-to-peer crypto-currency with proof-of-stake. Technical Report (2012). https://bitcoin.peryaudo.org/vendor/peercoin-paper.pdf 51. Zhan, Y., Wang, B., Lu, R., Yu, Y.: DRBFT: delegated randomization Byzantine fault tolerance consensus protocol for blockchains. https://doi.org/10.1016/j.ins.2020.12.077 52. Luo, Y., Chen, Y., Chen, Q., Liang, Q.: A new election algorithm for DPos consensus mechanism in blockchain. In: 2018 7th International Conference on Digital Home (ICDH), pp. 116–120 (2018). IEEE. https://doi.org/10.1109/ICDH.2018.00029 53. Xu, X., Zhu, D., Yang, X., Wang, S., Qi, L., Dou, W.: Concurrent practical byzantine fault tolerance for integration of blockchain and supply chain. ACM Trans. Internet Technol. 7(2021). https://doi.org/10.1145/3395331 54. Wang, X., Weili, J., Chai, J.: The research on the incentive method of consortium blockchain based on practical Byzantine fault tolerant. In: 2018 11th International Symposium on Computational Intelligence and Design (ISCID) (2018). https://doi.org/10.1109/iscid.2018. 10136 55. Fan, J., Yi, L.T., Shu, J.W.: Research on the technologies of Byzantine system. J. Softw. 24(6), 1346–1360 (2013). https://doi.org/10.3724/SP.J.1001.2013.04395 56. Kotla, R., Alvisi, L., Dahlin, M., Clement, A., Wong, E.: Zyzzyva: speculative byzantine fault tolerance. ACM SIGOPS Operating Syst. Rev., ACM 41, 45–58 (2007). https://doi.org/10. 1145/1294261.1294267 57. Karantias, K., Kiayias, A., Zindros, D.: Proof-of-burn. In: International Conference on Financial Cryptography and Data Security, Springer, Cham, pp. 523–540 (2020). https://doi.org/ 10.1007/978-3-030-51280-4_28 58. Salman, T., Zolanvari, M., Erbad, A., Jain, R., Samaka, M.: Security services using blockchains: a state of the art survey. IEEE Commun. Surv. Tutorials 21(1), 858–880 (2019). https://doi.org/10.1109/comst.2018.2863956 59. Schwartz, D., Youngs, N., Britto, A.: The ripple protocol consensus algorithm. Ripple Labs Inc White Paper, vol. 5 (2014). https://ripple.com/files/ripple_consensus_whitepaper.pdf 60. Intel: Sawtooth Lake (2017). https://intelledger.github.io/ 61. Intel Software Guard Extensions (Intel SGX). https://software.intel.com/en-us/sgx 62. Xie, J., Tang, H., Huang, T., Yu, F.R., Xie, R., Liu, J., Liu, Y.: A survey of blockchain technology applied to smart cities: research issues and challenges. IEEE Commun. Surv. Tutorials 21(3), 2794–2830 (2019). https://doi.org/10.1109/comst.2019.2899617 63. Gupta, S., Sinha, S., Bhushan, B.: Emergence of blockchain technology: fundamentals, working and its various implementations. SSRN Electron. J. (2020). https://doi.org/10.2139/ ssrn.3569577 64. Wang, X., Zheng, X., Zhang, X., Zeng, K., Wang, F.: Analysis of cyber interactive behaviors using artificial community and computational experiments. IEEE Trans. Syst., Man, Cybern. Syst. 47(6), 995–1006 (2017). https://doi.org/10.1109/tsmc.2016.2615130 65. Lv, Z., Qiao, L., Hossain, M.S., Choi, B.J.: Analysis of using blockchain to protect the privacy of drone big data. IEEE Netw. 35(1) (2021).https://doi.org/10.1109/MNET.011.2000154 66. Neudecker, T., Hartenstein, H.: Network layer aspects of permissionless blockchains. IEEE Commun. Surv. Tutorials 21(1), 838–857 (2019). https://doi.org/10.1109/comst.2018.285 2480

64

A. K. M. B. Haque et al.

67. Song, H., Zhu, N., Xue, R., He, J., Zhang, K., Wang, J.: Proof-of-Contribution consensus mechanism for blockchain and its application in intellectual property protection. https://doi. org/10.1016/j.ipm.2021.102507 68. Hyla, T., Peja´s, J.: Long-term verification of signatures based on a blockchain. Comput. Electr. Eng. 81(106523) (2020). https://doi.org/10.1016/j.compeleceng.2019.106523 69. Li, J., Ni, X., Yuan, Y.: The reserve price of ad impressions in multi-channel realtime bidding markets. IEEE Trans. Comput. Soc. Syst. 5(2), 583–592 (2018). https://doi.org/10.1109/tcss. 2018.2831234 70. Kosba, A., Miller, A., Shi, E., Wen, Z., Papamanthou, C.: Hawk: the blockchain model of cryptography and privacy-preserving smart contracts. In: 2016 IEEE Symposium on Security and Privacy (SP) (2016). https://doi.org/10.1109/sp.2016.55 71. Haque, A.K., Bhushan, B., Dhiman, G.: Conceptualizing smart city applications: requirements, architecture, security issues, and emerging trends. Expert. Syst. (2021). https://doi. org/10.1111/exsy.12753 72. Bamakan, S.M.H., Faregh, N., ZareRavasan, A.: Di-ANFIS: an integrated blockchain–IoT– big data-enabled framework for evaluating service supply chain performance. J. Comput. Des. Eng. 8(2), 676–690 (2021). https://doi.org/10.1093/jcde/qwab007 73. Bhushan, B., Sahoo, C., Sinha, P., et al.: Unification of Blockchain and Internet of Things (BIoT): requirements, working model, challenges and future directions. Wireless Netw (2020). https://doi.org/10.1007/s11276-020-02445-6 74. Liu, X.F., Jiang, X., Liu, S., Tse, C.K.: Knowledge discovery in cryptocurrency transactions: a survey. IEEE Access 9. https://doi.org/10.1109/ACCESS.2021.3062652 75. Zhang, R., Xue, R., Liu, L.: Security and privacy on blockchain. ACM Comput. Surv. 1, 1, Article 1, 35 (2019). https://doi.org/10.1145/3316481 76. Tabassum, A., Jeba, H.A., Mahi, T.K., Salim Reza, S.M., Hossain, D.A.: Securely transfer information with RSA and digital signature by using the concept of fog computing and blockchain. In: 2021 International Conference on Information and Communication Technology for Sustainable Development (ICICT4SD) (2021). https://doi.org/10.1109/ICICT4 SD50815.2021.9396987 77. Yadav, A.S., Agrawal, S., Kushwaha, D.S.: Distributed Ledger Technology-based land transaction system with trusted nodes consensus mechanism. J. King Saud Univ. – Comput. Inf. Sci. (2021). https://doi.org/10.1016/j.jksuci.2021.02.002 78. Kim, H.J.: Technical aspects of blockchain. In: Baker, H.K., Nikbakht, E., Smith, S.S. (eds.) The Emerald Handbook of Blockchain for Business, Emerald Publishing Limited, Bingley, pp. 49–64 (2021). https://doi.org/10.1108/978-1-83982-198-120211006 79. Swan, M.: Blockchain: Blueprint for a New Economy. O’Reilly Media Inc., Newton, MA, USA (2015) 80. Zhou, K., Liu, T., Zhou, L.:Industry 4.0: towards future industrial opportunities and challenges. In: Proceedings of 12th International Conference on Fuzzy System and Knowledge Discovery (FSKD), pp. 2147–2152 (2015). https://doi.org/10.1109/FSKD.2015.7382284 81. Alladi, T., Chamola, V., Parizi, R.M., Raymond Choo, K.K.: Blockchain applications for industry 4.0 and industrial IoT: a review 7, 176936 (2019). https://doi.org/10.1109/ACCESS. 2019.2956748 82. Digital and Emerging Technology Strategy. intueriglobal.com/digital-and-emergingtechnology-strategy/ 83. Weyer, S., Schmitt, M., Ohmer, M., Gorecky, D.: Towards industry 4.0-standardization as the crucial challenge for highly modular,multi-vendor production systems. 48(3), no. IFACPapersonline, pp. 579–584 (2015). https://doi.org/10.1016/j.ifacol.2015.06.143 84. Mohamed, N., Al-Jaroodi, J.:Applying blockchain in industry 4.0 applications, 0853–0854 (2019). https://doi.org/10.1109/CCWC.2019.8666558 85. Tonelli, F., Demartini, M., Pacella, M., Lala, R.: Cyber-physical systems (CPS) in supply chain management: from foundations to practical implementation. https://doi.org/10.1016/j. procir.2021.03.080

5 Revolutionizing the Industrial Internet of Things …

65

86. Semenkov, K., Promyslov, V., Poletykin, A., Mengazetdinov, N.: Validation of complex control systems with heterogeneous digital models in industry 4.0 framework. Machines 9(3), 62 (2021) 87. Al-Jaroodi, J., Mohamed, N.: Blockchain in industries: a survey. IEEE Access 7, 36500–36515 (2019). https://doi.org/10.1109/ACCESS.2019.2903554 88. Sethi, R., Bhushan, B., Sharma, N., Kumar, R., Kaushik, I.: Applicability of industrial IoT in diversified sectors: evolution, applications and challenges. In: Studies in Big Data Multimedia Technologies in the Internet of Things Environment, pp. 45–67 (2020).https://doi.org/10.1007/ 978-981-15-7965-3_4 89. Pilkington, M.: 11 blockchain technology: principles and applications. In: Research Handbook on Digital Transformations, p. 225 (2016). https://doi.org/10.4337/9781784717766.00019 90. Luu, L., Chu, D.H., Olickel, H., Saxena, P., Hobor, A.: Making smart contracts smarter. In: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security, pp. 254–269 (2016). https://doi.org/10.1145/2976749.2978309 91. Hofmann, E., R¨usch, M.:Industry 4.0 and the current status as well as future prospects on logistics. Comput. Ind. 23–34 (2017). https://doi.org/10.1016/j.compind.2017.04.002 92. Saxena, S., Bhushan, B., Yadav, D.: Blockchain-powered social media analytics in supply chain management. SSRN Electron. J. (2020). https://doi.org/10.2139/ssrn.3598906 93. Sicari, S., Rizzardi, A., Grieco, L.A., Coen-Porisini, A.: Security, privacy and trust in Internet of Things: the road ahead. Comput. Netw. 146–164 (2015). https://doi.org/10.1016/j.comnet. 2014.11.008 94. Weber, R.H.:Internet of Things–new security and privacy challenges. Comput. Law Secur. Rev. 26(1), 23–30 (2010). https://doi.org/10.1016/j.clsr.2009.11.008 95. Chelladurai, U., Pandian, S.: A novel blockchain based electronic health record automation system for healthcare. J. Ambient Intell. Human Comput. (2021).https://doi.org/10.1007/s12 652-021-03163-3 96. Roman, R., Zhou, J., Lopez, J.: On the features and challenges of security and privacy in distributed internet of things. Comput. Netw. 57(10), 2266–2279 (2013). https://doi.org/10. 1016/j.comnet.2012.12.018 97. Kurkin, O., Januˇska, M.: Product life cycle in digital factory Knowledge management and innovation: a business competitive edge perspective. In: Cairo: International Business Information Management Association (IBIMA), p. 1881–1886, 2010. ISBN 978–0–9821489–4–5 98. Bradley Shaw Swathi Sambhani, “Data, data, everywhere! product lifecycle management in the world of iot.“ 99. Emanate: Emanate intro series 2: Blockchain, micro-payments and streaming. https://med ium.com/emanate-live/emanate-intro-series-2-blockchain-micro-payments-and-streaming102698859420 100. Kshetri, N.: Can blockchain strengthen the Internet of Things. IT Prof. 19, 68–72 (2017). https://doi.org/10.1109/MITP.2017.3051335 101. Li, J., Liu, Z., Chen, L., Chen, P., Wu, J.: Blockchain-based security architecture for distributed cloud storage. In: Proceedings of IEEE International Symposium on Parallel and Distributed Processing with Applications and IEEE International Conference on Ubiquitous Computing and Communications (ISPA/IUCC), Guangzhou, China, pp. 408–411 (2017). https://doi.org/ 10.1109/ISPA/IUCC.2017.00065 102. Fernández-Caramés, T.M., Fraga-Lamas, P., Suárez-Albela, M., Vilar-Montesinos, M.: A fog computing and cloudlet based augmented reality system for the industry 4.0 shipyard 18, 1798–1961 (2018). https://doi.org/10.3390/s18061798 103. Suárez-Albela, M., Fraga-Lamas, P., Fernández-Caramés, T.M.: 18, 3868 (2018).https://doi. org/10.3390/s18113868 104. Guo, Y., Liang, C.: Blockchain application and outlook in the banking industry. Financ Innov 2, 24 (2016). https://doi.org/10.1186/s40854-016-0034-9 105. Tripoli, M., Schmidhuber, J.: Emerging opportunities for the application of blockchain in the agri-food industry. FAO and ICTSD: Rome and Geneva. Licence: CC BY-NC-SA, 3 (2018)

66

A. K. M. B. Haque et al.

106. Czachorowski, K., Solesvik, M., Kondratenko, Y.: The application of blockchain technology in the maritime industry. In: Kharchenko, V., Kondratenko, Y., Kacprzyk, J. (eds.) Green IT Engineering: Social, Business and Industrial Applications. Studies in Systems, Decision and Control, vol. 171. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-00253-4_24 107. Fraga-Lamas, P., Fernández-Caramés, T.M.: A Review on Blockchain technologies for an advanced and cyber-resilient automotive industry. IEEE Access 7, 17578–17598 (2019). https://doi.org/10.1109/ACCESS.2019.2895302 108. Jovovi´c, I., Husnjak, S., Forenbacher, I., Maˇcek, S.: Innovative application of 5G and blockchain technology in industry 4.0. EAI Endorsed Trans. Indus. Netw. Intell. Syst. 6(18) (2019). https://doi.org/10.4108/eai.28-3-2019.157122 109. Fernández-Caramés, T.M., Fraga-Lamas, P.: A Review on the Application of blockchain to the next generation of cybersecure industry 4.0 smart factories. IEEE Access 7, 45201–45218 (2019). https://doi.org/10.1109/ACCESS.2019.2908780 110. Ozdemir, A.I., Ar, I.M., Erol, I.: Assessment of blockchain applications in travel and tourism industry. Qual. Quant. 1–15 (2019).https://doi.org/10.1007/s11135-019-00901-w 111. Lu, H., Huang, K., Azimi, M., Guo, L.: Blockchain technology in the oil and gas industry: a review of applications, opportunities, challenges, and risks. IEEE Access 7, 41426–41444 (2019). https://doi.org/10.1109/ACCESS.2019.2907695 112. Papathanasiou, A., Cole, R., Murray, P.: The (non-) application of blockchain technology in the Greek shipping industry. Eur. Manage. J. (2020). https://doi.org/10.1016/j.emj.2020. 04.007 113. Chen, J., Cai, T., He, W., Chen, L., Zhao, G., Zou, W., Guo, L.: A Blockchain-driven supply chain finance application for auto retail industry. Entropy 22(1), 95 (2020). https://doi.org/ 10.3390/e22010095 114. Alladi, T., Chamola, V., Parizi, R.M., Choo, K.R.: Blockchain applications for industry 4.0 and industrial IoT: a review. IEEE Access 7, 176935–176951 (2019). https://doi.org/10.1109/ ACCESS.2019.2956748 115. Madaan, G., Bhushan, B., Kumar, R.: Blockchain-based cyberthreat mitigation systems for smart vehicles and industrial automation. In: Studies in Big Data Multimedia Technologies in the Internet of Things Environment, pp. 13–32.https://doi.org/10.1007/978-981-15-7965-3_2 116. Soni, S., Bhushan, B.: A comprehensive survey on blockchain: working, security analysis, privacy threats and potential applications. In: 2019 2nd International Conference on Intelligent Computing, Instrumentation and Control Technologies (ICICICT) (2019). https://doi.org/10. 1109/icicict46008.2019.8993210 117. Yadav, A.K., Singh, K.: Comparative analysis of consensus algorithms and issues in integration of blockchain with IoT. In: Tiwari, S., Trivedi, M., Mishra, K., Misra, A., Kumar, K., Suryani, E. (eds.) Smart Innovations in Communication and Computational Sciences. Advances in Intelligent Systems and Computing, vol. 1168. Springer, Singapore (2021). https://doi.org/ 10.1007/978-981-15-5345-5_3 118. Sajid, S., Haleem, A., Bahl, S., Javaid, M., Goyal, T., Mittal, M.: Data science applications for predictive maintenance and materials science in context to Industry 4.0. In: 2021 Elsevier Ltd. All rights reserved. Second International Conference on Aspects of Materials Science and Engineering (ICAMSE 2021). https://doi.org/10.1016/j.matpr.2021.01.357 119. Karame, G.: On the security and scalability of bitcoin’s blockchain. In: Proceedings of ACM SIGSAC Conference on Computer and Communications Security (CCS), New York, NY, USA, pp. 1861–1862 (2016). https://doi.org/10.1145/2976749.2976756

Chapter 6

Attacks and Countermeasures in IoT Based Smart Healthcare Applications A. K. M. Bahalul Haque, Bharat Bhushan , Afra Nawar, Khalid Raihan Talha, and Sadia Jeesan Ayesha

Abstract The perpetual evolution of IoT continues to make cities smart beyond measure with the abundance of data transactions through expansive networks. Healthcare has been a foremost pillar of settlements and has gained particular focus in recent times owing to the pandemic and the deficiencies it has brought to light. There is an exigency to developing smart healthcare systems that make smart cities more intelligent and sustainable. Therefore, this paper aims to present a study of smart healthcare in the context of a smart city, along with recent and relevant research areas and applications. Several applications have been discussed for early disease diagnosis and emergency services with advanced health technologies. It also focuses on security and privacy issues and the challenges posed by technologies such as wearable devices and big healthcare data. This paper briefly reviews some enhanced schemes and recently proposed security mechanisms as countermeasures to various cyberattacks. Recent references are primarily used to present smart healthcare privacy and security issues. The issues are laid out briefly based on the different architecture layers, various security attacks, and their corresponding proposed solutions along with other facets of smart health such as Wireless Body Area Network (WBAN) and healthcare data.

A. K. M. Bahalul Haque LUT School of Engineering Science, LUT University, Lappeenranta, Finland e-mail: [email protected] B. Bhushan (B) Department of Computer Science and Engineering, School of Engineering and Technology, Sharda University, Greater Noida, India A. Nawar · K. R. Talha · S. J. Ayesha Department of Electrical and Computer Engineering, North South University, Dhaka, Bangladesh e-mail: [email protected] K. R. Talha e-mail: [email protected] S. J. Ayesha e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_6

67

68

A. K. M. Bahalul Haque et al.

Keywords Smart healthcare · Smart city · IoT · Security · Privacy

6.1 Introduction In this modern era, the population of urban areas is increasing rapidly. So, the needs of the citizens are also increasing. The conventional supply management is not sufficient enough to support the requirements of the urban area and citizens. Smart cities play a vital role in ensuring maximum comfort to the citizens. The concept uses information and communication technology (ICT) to fulfill all the necessary functional and environmental requirements of an urban area [1]. In other words, ICT and urban functions get integrated or combined. In a broad sense, smart cities are a combination of ICT technologies, ecological environment, energy management technologies, and supportive families within the urban and rural areas [2, 3]. The idea of a smart city was introduced due to some particular reasons. One of the reasons is that a significant number of jobs is taking place in urban areas. The majority of the citizens move to urban areas by making it denser than before [4]. Another reason is to ensure an excellent educational opportunity for their children; many families are moving to urban areas from rural areas [5]. Many problems occur in the case of facilities and urban areas’ environment to cope with this expansion. The idea of the smart city comes to play a vital role in eliminating these types of problems. It is essential to employ smart cities’ necessary infrastructure, various sensors, and supportive technologies in urban areas. The Internet of Things (IoT) is said to be one of the essential concepts that can be implemented successfully in a smart city [6]. A vast commercial objective of IoT is driven by the extraordinary growth of digital devices such as sensors, actuators, smartphones, and smart gears etc. [7]. There has been a fair amount of research that focuses on smart healthcare systems. Emergency healthcare is now creating possibilities from recent studies [8]. Remote healthcare monitoring can be used to monitor the physical condition of non-critical patients at home to reduce the workload on the hospitals. It is now possible to monitor the patient using wearable sensors and vision-based technologies (cameras) around the home to observe behavior patterns, tremors, and general activity levels [9]. In the future, machine learning can be introduced to have a more accurate system. Nowadays, a practical system has been developed to measure blood glucose level where patients have to give blood samples manually [10]. Heart attack can be detected by the utilization of ready-made components with built-in antennas [11]. Since smart healthcare deals with sensitive personal data, it needs to be secured properly. Data breach in healthcare sectors can be catastrophic. Patients personal will be out in the open and those can be used for identity theft, blackmail and many other reasons. IoT infrastructures can face various types of attacks which needs to be thoroughly studied, so that smart healthcare sector can be more resilient and provide integrity of services.

6 Attacks and Countermeasures in IoT Based …

69

This paper comprises a brief introduction to the smart city architecture and an extensive analysis of attacks and countermeasures on smart healthcare infrastructure. In addition to this, a tabular representation of the recent work done on healthcare applications has been presented, which further enhances this paper’s novelty. A summary of the contribution of this work has been listed below: • An overview of a smart city, including that of its layers and fundamental pillars • An elaborate discussion on smart healthcare in an IoT based smart city. • Categorized discussion of smart healthcare applications alongside tabularized recent applications. • Frequently occurring cyber-attacks in the healthcare domain and their existing security measures. • Challenges and future research opportunities of Smart Healthcare. The rest of this research, excluding the introduction, has been categorized into five different sections where Sect. 6.2 focuses on the smart city fundamentals, its layers, and its pillars. In Sect. 6.3, a broad description of smart healthcare and its vital characteristics have been included. Section 6.4 discusses the issues and vulnerabilities associated with the healthcare sector and also their proposed countermeasures. Section 6.5 emphasizes the challenges and future research directions. Finally, Sect. 6.6 concludes the presented work.

6.2 Smart City Fundamentals Although the descriptions fall on a broad spectrum when defining the concept of smart cities, they are generally acknowledged to be extensively interconnected settlements which modern cities are inevitably shaping into. A smart city “tracks and integrates conditions of its critical infrastructure” [12] and is characterized by IBM as one that is intelligent, interconnected, and instrumented [13]. Smart cities provide upgraded services to citizens by seeking and distinguishing intelligent solutions and creating thriving cities of the future [14]. They must make “conscious decisions to aggressively deploy technology as a catalyst to solving its social and business needs” [15]. Smart cities allow decision-makers to base their decisions on comprehensive and relevant data sources and improve every city aspect’s efficiency. The IoT network collect data while communicate with each other. As the population density in urban areas faces unprecedented growth [16], the need for this interconnection through technology has become essential. Connectivity through the internet serves as a fundamental block in the smart city architecture. Digital devices are employed in various services that rely on it to communicate with other devices [17, 18].

70

A. K. M. Bahalul Haque et al.

6.2.1 Smart City Layers The development of the internet leads to better interconnection and an enhancement in a smart city’s infrastructure. Strictly, the technical architecture of a smart city is implemented by several layers. Brief description of those layers are given below.

6.2.1.1

Sensing Layer

This layer works for data acquisition. The methods and equipment utilized in data acquisition are dependent upon the nature of the data and context. The sizable extent of smart city operations creates substantial variegated data and introduces considerable complexity in implementing the sensing layer [19]. Networks of responsive and self-regulating physical devices are employed in data collection [20]. This network of nodes captures various forms of data and environment variables like temperature, humidity, pressure, etc. which can necessitate the use of multi-featured nodes equipped with temperature sensors, cameras, GPS, and other equipment. The extent of the network in this layer has been linked to the smart city [21].

6.2.1.2

Network Layer

This layer facilitates the transmission of collected data from the sensing equipment to processing stations. Satellite, wireless, and wired communications enable transmissions within sensing networks and processing stations. Existing and developing technologies like Bluetooth, radio frequency identification (RFID), Wi-Fi, 3G, 4G, 5G, LP-WAN, and others are employed for this task.

6.2.1.3

Analysis Layer

The processing, analysis, manipulation, organization, and storage of the raw data is handled in the Analysis Layer. The processing and other tasks can be carried out on smartphones and similar terminal devices. However, a more sizable amount of data may require the usage of cloud computing platforms. Cloud computing platforms for the network layer are applicable particularly for sensor networks that generate a large quantity of data [22].

6 Attacks and Countermeasures in IoT Based …

6.2.1.4

71

Application Layer

This layer includes the services which utilize the IoT-generated data, processed for specialized applications in various domains such as healthcare, mobility, surveillance etc. Most recently collected data and historical data generated in the equivalent context can provide the basis for quick service to various phenomena through applications using decision-making algorithms.

6.2.2 Pillars of Smart City Smart cities are also characterized by fundamental pillars that constitute smart infrastructure, smart society and public, smart environment, and smart governance and management, which can be further subdivided into multiple areas of concern. • Smart infrastructure is the system of the city’s physical and organizational structures like its buildings, transportation structures, and other establishments vital to the city’s operation. • Smart society and the public utilize IoT to enhance living, education, healthcare and various societal systems. It also involves socio-economic schemes, public life, social communication, cultural activities, recreation, and tourism [23]. • Smart Mobility implies an intelligent transportation system for both individual citizens and the whole. Smart city vehicles are embedded with computational and communication devices, and vehicles exist on a network with others, which enable vehicles in the city to communicate and be monitored efficiently [24]. • Smart environment ensures the sustainability of a smart city by protecting the environment’s quality, responsible management of natural resources, maintaining ecosystem balance, monitoring of economic activity concerning its effect on the environment, and enforcing policies and regulations. Adapting to climate changes, implementing green technology, systematic reduction of pollutants, development of environmentally beneficial technologies, and efficient resource management also concern the smart environment. • Smart Governance entails the intelligent management of the smart city and its infrastructure and public. The quality and availability of public service are improved along with the public and transparency of policy decisions.

6.3 Healthcare in Smart City Smart healthcare is a system within a smart city that leverages technologies like IoT, mobile internet, and cloud to dynamically access healthcare information, create connections among people and organizations related to healthcare, and actively manage the healthcare needs intelligently. Smart healthcare enables effective

72

A. K. M. Bahalul Haque et al.

resource allocation within the healthcare sector and maintains strong interaction between all actors, ensuring that participants get proper services and make rational decisions.

6.3.1 Smart Healthcare Applications The development of computational technologies is changing the face of healthcare. Sensors of smart cities measure several that can be used to know citizens’ conditions at any point in time [25]. Frequent gathering of data and processing helps to get healthcare perpetually smarter. Many services have been proposed to support physicians and specialists’ efficient work to help prevent and diagnose diseases and provide necessary treatments and therapy [26]. The following sections can categorize significant applications of smart healthcare.

6.3.1.1

Assisting Diagnosis and Treatment

Technologies like artificial intelligence and surgical robots make diagnosis and treatment of diseases smarter. Artificial intelligence helps to make decisions about the examined result of certain diseases. The efficacy can exceed human doctors [27, 28]. Machine learning-based systems might be more accurate than experienced physicians in terms of medical imaging. The sensors play their role by working in the beginning stage of data processing. These can gather necessary information about ill persons [29]. IBM’s Watson is an intelligent cognitive system that can diagnose diabetes and cancer [30]. This supportive clinical system helps doctors give proper treatment to the patients and reduces misdiagnosis. A smart diagnosis system helps develop personalized treatment plans [31]. The Invention of surgical robots is another enticing and effective addition in assisting medical operations and treatment [32].

6.3.1.2

Health Management

Wearable smart devices and smartphones are embedded with sensors are used for patient’s medical condition monitoring. These devices can measure movement and rotation respectively in three dimensions relative to the device. Integrating Wifi, GPS and GSM sources can track an individual’s location to detect whether they are at or away from home. The compass can find the heading and the barometers can detect the altitude. Many smart devices are equipped with cameras and microphones that can sense people’s density at a particular place. Components such as LEDs and photodiodes in smartwatches use light to measure the amount of blood flowing past a patient’s wrist and determine heart rate accordingly while connecting smartphones with glucose meters to monitor blood sugar levels. Body area networks (BAN) can be created by putting sensors on or close to the physical body. It uses small sensors

6 Attacks and Countermeasures in IoT Based …

73

and control units to capture data. There are special sensors to monitor rheumatoid arthritis, heart arrhythmias, sleep apnea, cranial pressure, etc. [33]. For example, the electrochemical glucose sensor is popular in diagnosing diabetes [34].

6.3.1.3

Healthcare Based on Smart Homes

Smart homes’ role in smart healthcare is mainly divided into two parts: Firstly, health monitoring and home automation. The temperature sensor is an ambient sensor that can measure room temperature and humidity. Passive infrared (PIR) sensors can measure heat-based movement and ambient light levels. Similarly, magnetic sensors can detect the open and close action of the door. Bluetooth low energy sensors or Radio-frequency identification (RFID) can also detect object movement. Some other ambient sensors such as CO2 , power, water, and light sensors can provide outstanding services by collecting data related to patients’ health conditions without healthcare professionals’ needs. Secondly, patients can use various smart health applications and web services for self-monitoring purpose [35, 36].

6.3.1.4

Virtual Assistant

In smart healthcare, virtual assistants work using algorithms that communicate using speech recognition techniques. The inherent operations largely depend on big data. For patients, a virtual assistant works on a smart device that searches for the best healthcare service-related information by taking voice command as an input. Virtual assistants help doctors with managing and monitoring medical processes. More efficiently based on patient’s medical information, which saves more time. In medical institutions, it can be used to save workforce and reduce human labor.

6.3.1.5

Smart Hospitals

A smart hospital system can resolve the issue of the limited number of hospital staff and employees. IoT optimized environments can automate processes and introducing new features. Smart hospitals uses state of the art technologies for administrative procedure, doctors schedule management, staff management, appointment system, advance booking etc. Moreover, these types of hospitals can have a wholistic integration approach with the healthcare sensors for better efficiency [30]. Moreover, these hospitals can have state of the art research facility which can be very useful for on the spot use case research [37, 38]. Several other present-day smart health features and countermeasures have been discussed in Table 6.1.

74

A. K. M. Bahalul Haque et al.

Table 6.1 Recent advances of smart healthcare References

Years

Features

Kumar et al. [39]

2017

Patient health monitoring using non-invasive sensors; Thingspeak android app for doctors or paramedical staff

Mshali et al. [40]

2018

Adaptive context-aware e-health monitoring system for old aged and isolated persons living alone; health state prediction

Wan et al. [41]

2018

IoT-cloud-based approach; real-time and ubiquitous monitoring

Mshali et al. [42]

2018

Analysis of human behavior; smart environment for elderly and dependent people

Kang et al. [43]

2018

Cloud computing and blockchain for actively monitor patient health conditions; self-monitoring health state using wearable devices

Kharel et al. [44]

2018

Long range wireless communication and fog computing for long range connectivity between health monitoring applications

Kajornkasirat et al. [45]

2018

IoT based healthcare monitoring system using API technology. Web/mobile application created using SQL, PHP, Java, JavaScript, HTML5, Android Studio

Albahri et al. [46]

2019

Heterogeneous wearable sensors for real-time health monitoring; multi-healthcare services

Puntambekar et al. [47]

2019

Assistive band for health analysis

Li et al. [48]

2019

Factors affecting acceptance of smart wearables in elders; tested acceptance model

Islam and Shin [49]

2019

Unmanned Aerial Vehicle in outdoor health monitoring; usage of blockchain and mobile edge computing

Hartmann et al. [50]

2019

Edge computing techniques in Smart Health; edge computing data operations; challenges and future directions

Gahlot et al. [51]

2019

Smart healthcare development in villages and towns; early disease diagnosis system

Rajamohanan et al. [52]

2019

Bluetooth Low Energy (BLE) based technology in smart healthcare wearable devices; comparison of BLE with other wearable device technologies

Rayan et al. [53]

2019

Machine learning in smart health

Abdellatif et al. [54]

2020

Smart Health management using blockchain and edge computing; remote monitoring, fast response, epidemic discovery; secure medical data sharing

Allam and Jones [55]

2020

Virus outbreak from an urban standpoint; enhanced standardization protocols for increased data sharing

Zghaibeh et al. [56]

2020

Private multi-layered blockchain based Health management; smart contract and Consensus mechanism (continued)

6 Attacks and Countermeasures in IoT Based …

75

Table 6.1 (continued) References

Years

Features

Meng et al. [57]

2020

Textile based wireless biomonitoring system; noninvasive and comfortable sensor for BAN with high sensitivity

Chen et al. [58]

2020

Zero Trust architecture in 5G smart healthcare for repeated identity authentication, trustworthy dynamic access control models, monitoring access behavior, and ensuring real-time security

Ahmadi-Assalemi et al. [59]

2020

Digital Twins show real time status by matching physical objects in the health industry with digital models for enhanced patient care, risk analysis and obtaining precision healthcare

Wang et al. [60]

2020

Consortium blockchain forbids unpermitted access to stored and shared data; proxy re-encryption dynamically controls third party access while GCN recognizes malicious nodes

Tanwar et al. [61]

2020

Healthcare record management using blockchain; improved accessibility of data

Zhong et al. [62]

2021

Attribute-based strategy for access control, updating attributes, and data encryption in the healthcare domain. It’s security is confirmed by a performance check at various security levels and a DBDH assumption

Wu et al. [63]

2021

Edge-based hybrid network system to facilitate data transfer; Extended range for short-range IOT protocols like BLE

Yang et al. [64]

2021

Optimized data processing and node deployment efficiency in IoT assisted healthcare with end-edge-cloud architecture; maximized intelligence level in medical emergency

Alzubi [65]

2021

Secure healthcare IoT device authentication using block-chain and Lamport Merkle Digital Signature; identification of malicious user behaviour for improved protection of sensitive patient data

6.4 Smart Healthcare Privacy and Security As the smart healthcare industry heavily relies on smart medical devices, smart hospitals, and other smart services, it is vital to ensure these intelligent healthcare services’ securities. Some general security and privacy requirements to ensure protection are authentication, access control, availability, dependability, and flexibility. This section extensively discusses the privacy and security vulnerabilities associated with the intelligent healthcare system and highlights the importance of employing recently developed and improved security measures.

76

A. K. M. Bahalul Haque et al.

6.4.1 Denial of Service In this type of attack, intruders exploit the targeted IoMT device or system by flooding the system’s data transmission channel with undesired traffic. The attack disrupts the functionality of the IWMDs or deactivate the entire healthcare system as well as block access to emergency medical facilities and its intended users (patients or healthcare providers). The cyber threats associated are deletion and alteration of a patient’s critical health data and the insertion of false health information before being passed on to the receiver’s end (hospitals, healthcare professionals). This form of security breaches can pose a significant threat to a patient’s life. It can result in improper treatment of the patient, wrong prescription by a healthcare provider, false emergency patient alarms, and can even show an inaccurate status of the patient. In health organizations, such attacks apply to implantable devices (insulin pumps, cardiac monitors, pacemakers), wearable health monitoring technologies (Fitbit trackers), and on-site medical equipment (PET scanner, MRI machines, X-ray machines, Dialysis machines) [66, 67]. Some variants of DoS attacks have been discussed below.

6.4.1.1

Jamming Attacks

Among some of the physical layer’s common attacks are Jamming attacks in which malicious jamming nodes are used to intercept authorized wireless communications between medical sensors in WBAN systems. The attacker’s radio frequencies interfere with the RF signals of the WBAN nodes which reduces the Signal-to-noise ratio (SNR). The severity of this attack varies depending upon the attacker’s knowledge about the network. The attacker may cause functionality disturbance in a minor portion of the network or even interruption in the entire network. Due to WBAN being a small network, it is at a greater risk of getting blocked [68–70].

6.4.1.2

Node Tampering Attacks

Another subset of DoS attacks in the physical layer is tampering attacks. The attackers exploit vulnerabilities in wearable medical sensors, implantable technologies, and other hospital devices to extract patients’ confidential health data and modify them via radiofrequency electromagnetic waves.

6.4.1.3

Data Collision Attacks

These attacks mostly occur in the data link layer when two or more terminal nodes transmit data packets, causing packet collisions. The cyber attacker may intentionally escalate the number of collisions by frequently transferring messages via the network channel. This leads to a breakdown of communication and network performance deterioration [71].

6 Attacks and Countermeasures in IoT Based …

6.4.1.4

77

Exhaustion Attack

The aim of this DoS attack in the link layer is to consume the battery resources. A self-sacrificing node may exploit the vulnerabilities of the MAC layer protocols such as Request-To-Send (RTS) and Clear-To-Send (CTS) signals in the IEEE 802.11 MAC protocols. The transmitter node transfers an RTS data packet requesting access to the data transmission channel. The receiver node (victim) responds with a CTS data packet allowing the transmitter node to transmit it. This continuous packet transmission via the transmission channel causes frequent packet collisions and keeps the channel busy. Thus, exhausting the energy resources of the battery.

6.4.1.5

Vampire Attacks

It is also called resource depletion attacks, target the medical sensor nodes’ batteries by dissipating their power resources, decreasing their expected lifetime. This form of energy-draining DoS attacks in the network layer uses malicious nodes to generate and transfer protocol compliant messages. It consumes more power by routing and processing (by producing longer routes and constructing loops) than legitimate nodes that transfer messages of the same size to the same destination. This increase in energy consumption arises from the Vampire nodes modifying the packets’ header information before bombarding them on to the victim nodes to add extra load. Moreover, these attacks are not easily detectable and preventable as they are protocol-independent [72].

6.4.1.6

Black Hole

This type of DoS attack in the network layer involves adversary nodes taking advantage of the routing protocol. The adversary nodes can then misroute all the data traffic towards themselves before consuming them. For instance, when an attacker node receives a Route REQuest, RREQ data packet from a source node, it sends a Route REPly, RREP data packet making a false claim of having the shortest route to the destination node. Once the source node transfers its packet to the attacker node, the packets are dropped or destroyed instead of being forwarded to their expected destination without notifying the source node [73].

6.4.1.7

TCP SYN Flooding Attack

This one is a transport layer attack that utilizes the three-way handshake mechanism to send a spoofed package to its targeted server. This is implemented when the victim responds with a SYN/ACK packet placing the connection request in a queue. The SYN/ACK packet is transmitted to another host instead of the original client, so it does not respond as it did not send SYN packets. Hence the adversary does not

78

A. K. M. Bahalul Haque et al.

complete the third step of the mechanism. In this way, the adversary can iteratively send numerous SYN packets causing its target to open several TCP connections and respond to them. So, the victim cannot handle any new incoming requests as its queue is already filled with large volumes of partially-open TCP connections. This causes the depletion of memory resources by reaching a threshold limit [74]. CoAP based enhanced DTLS scheme, Software Defined Networking based defense, learning automata-based approach, and some machine learning techniques such as deep learning are some countermeasures against DoS and DDoS. A scheme called Secure-DAD can be used against DoS attacks in IPv6 Duplicate Address Detection, DAD processes [75–78]. Some blockchain approaches can also be used against these attacks. Combining IoMT devices with Ethereum helps minimize the risk of DDoS attacks [79, 80].

6.4.2 Spyware and Worm Attacks Spyware, a malware class, is used to gather confidential ePHI of patients without their knowledge and relay them to third party entities or sell them for the attacker’s gain. Its malicious activities involve spying on their targets (healthcare sectors) via covet surveillance and monitoring their online activities to steal sensitive information. Among some of the most disastrous malware attacks are Worm attacks which can easily exploit the vulnerabilities in IWMDs as the manufacturers of these devices do not spend sufficient resources and time on strengthening their security. The worms can then replicate vertically to extract confidential information or even destroy the targeted IoMT devices leading to the loss of crucial data that can be life-threatening to devices’ users. Moreover, integrating these malicious attacks with other forms of malware attacks (Ransomware, Botnet, Trojan) can infect the entire medical network. Possible solutions are anti-virus, anti-spyware, anti-malware, intrusion detection, and machine learning algorithms [81, 82].

6.4.3 Ransomware These malware attacks are launched towards the health industry via phishing emails, malicious links, attachments, and mail advertisements. A series of steps are carried out to implement the encryption key successfully. Firstly, the malicious program fixes the external IP address, deletes copies, creates a single ID, etc. Next, important patient files and hospital documents are searched within the system with extensions like .docx, .pptx, .pdf, .jpg, .png, .xlsx. They are then shifted to different locations, renamed, and their original extensions are changed to ransomware extensions such as .crypto, _crypt, locked, RDM, RRK etc. the original files are deleted. Finally, these files are made inaccessible by the process of encryption in crypto-ransomware or locking in locker-ransomware and the victimized hospitals are threatened to pay

6 Attacks and Countermeasures in IoT Based …

79

a ransom fee to unlock their files. The attacker may increment their financial claim, misuse or delete vital health records, appointment and surgery schedules, and crucial hospital documents if the ransom fee is not paid in time. As several patients’ health records are exposed to the attacker and hospitals have sufficient financial resources, they are more likely to pay the ransom. Hence, making healthcare industries more vulnerable to ransomware attacks. Honeypot, Intrusion Detection Honeypot (IDH), and machine learning approaches are some effective defense mechanisms to deal with the growing concerns of ransomware [83–85].

6.4.4 Eavesdropping Eavesdropping is when an illegitimate entity secretly intercepts the user’s sensitive information (patient-related information or their EMR) without their knowledge or permission. The stolen information may be used to perform malevolent activities leading to privacy breaches. This attack acts as an access point prerequisite to implement many other attacks such as spyware, MITM, side-channel attacks, etc. Eavesdropping attacks can be subdivided into Passive Eavesdropping. The attacker can scan to detect which medical devices (IWMDs and on-site medical equipment) are connected to the wireless access points, and Active Eavesdropping. The attacker can spy on the data sent and received while in transit. These attacks are difficult to diagnose as no deletion or modification of data is involved and also no issues are detected in the network transmission channel. Encryption techniques such as SecureVibe, and lightweight key agreement and mutual authentication mechanisms along with a software tool, ProVerif can help to mitigate these attacks and also several other attacks like spoofing, jamming, replay attacks, etc. [86, 87].

6.4.5 Man In The Middle In healthcare, MITM involves Protected Health Information. PHI is transferred from one point to another between two legitimate parties (between patient and cloud, between cloud and smart healthcare industry, or between IWMDs and on-site medical devices). They think they are directly communicating with each other. An unauthorized entity who hacks the data secretly during the transmission process decrypts it, reviews it and then re-encrypts it before passing it on to the receiver (passive attack). The hacked data can be selectively altered, duplicated, manipulated, or even have malicious codes injected into them (active attack) to be used against the attacker’s victims for the attacker’s benefit, such as obtaining sensitive medical information of patients. Some effective countermeasures are advanced ECG based authentication techniques that eliminate the acoustic interferences unlike traditional ECGs while preserving privacy [88–90].

80

A. K. M. Bahalul Haque et al.

6.4.6 Side Channel Attacks The adversary utilizes side-channel information such as execution time of operations, fault frequencies, and power-related data to obtain IoT encryption key. Similarly, forged signals are introduced in an EM Injection attack, which are strong enough to dominate the original signals generated by implantable electronic devices. Another subset of side-channel attacks is Differential Power Analysis, DPA attack that utilizes the statistical data to overcome the cryptographic barriers and gain access to the desired implantable or wearable devices. When a smartwatch is infected with some malicious software, and a smartphone comes nearby, side-channel keystroke inference attacks can occur. Motion sensors like accelerometers and gyroscopes in the infected smartwatch can detect the wrist motions while tapping each keystroke in the smartphone keypad. Solutions include advanced cryptographic techniques such as Elliptic Curve Cryptography, ECC, software-based solutions such as TinySec. To further prevent these close-range attacks, patients may wear wrist bands or carry cards with secret keys of their IWMDs imprinted on them, or even have the keys tattooed on to their skin using ultraviolet-reactive pigmentation [91–94].

6.5 Challenges and Future Research Direction of Smart City Healthcare Implementing smart healthcare systems in the smart city poses challenges and has limitations outside of the privacy and security dimension. Technological, financial, psychological and administrative demands are required to be met when adopting smart healthcare [95], alongside the challenges they present. These challenges are exhibited in the development of proper smart healthcare architecture, raising the public awareness and engagement and interoperability with other pillars on a macrolevel and improving accessibility, data handling, efficiency etc. of healthcare IoT devices and sensors micro-level. This section aims to present the challenges and opportunities in the technological domain.

6.5.1 Wearable Technology Challenges Sensors and other connected devices are fundamental blocks of a smart healthcare system and pose their limitations. Wearable devices employed in healthcare require environments that support ambient intelligence and in which heterogeneous systems can operate simultaneously [96]. Body area network sensors have complexities brought on by their need to be operated and maintained by humans. The challenges in normalizing this technology lie in making the devices more comfortable for humans, easy to operate, and secure [97, 98]. The wearable sensors also require more accuracy

6 Attacks and Countermeasures in IoT Based …

81

and fault tolerance to be on par with hospital-grade equipment. There is a tradeoff between the accuracy and energy efficiency and wearability of sensors currently available, and finds reducing the effect of motion on sensors such as the respiratory rate and pulse sensors is a workable area to improve accuracy [99]. Applying encryption on healthcare wearable devices is also an important issue.

6.5.2 Smart Healthcare Data Challenges Another essential side of smart healthcare, which poses several challenges and also numerous research scopes and future opportunities, is big data. The lack of a standardized data format and transfer protocol increases data handling complexity [100]. Researchers have also pointed towards the disparity in progress between data storage and data processing, which is less developed. Cloud-based algorithms are emphasized when it comes to big data processing. The sensitive nature of healthcare data also warrants addressing issues in transmission. The long route of data transmission, through multiple stations, affects the transmitted data due to noise and poses the challenge of developing effective noise removal techniques.

6.5.3 Recommendations and Opportunities The smart healthcare system should move towards accurate, cost effective, personalized and efficient service provider [101, 102]. Recently, blockchain technology integration into smart cities is being researched to ensure more robust security [103, 104]. As these superior technologies integrate with healthcare to form a smart healthcare system, many research and development opportunities arise. Recent research directions and recommendations are outlined in this section.

6.5.3.1

Specialized Algorithms and Machine Learning in Smart Healthcare

Machine learning and development of algorithms for specific purposes are heavily recommended in recent papers, and there are numerous scopes of application of ML in smart healthcare. For diagnostics, the researchers identified clustering and logistic regression algorithms. Application of machine learning algorithms on signal processing on ECG monitoring to reduce supervision costs and the development of optimization algorithms to minimize wearable devices’ energy consumption is recommended in the paper. Reference [105] uses multilayered extreme machine learning to identify human activities, which can have implications on health monitoring applications’ scope. Machine learning techniques should also be developed to uphold the quality of cloud services in healthcare [106].

82

6.5.3.2

A. K. M. Bahalul Haque et al.

Energy Harvesting for Wearable Devices

One rapidly growing area in the development of wearable sensors is energy harvesting. Adapting the energy harvesting mechanisms of autonomous wireless sensors for very low power and lightweight sensors employed in smart health applications like body area networks would significantly improve them. Kinetic energybased energy harvesting [107] is more popular, but its scope does not encompass all wearable devices, for which thermoelectric energy harvesting may be explored [108, 109]. Reference [110] presents experimental validation for piezoelectric energy harvesting, which can power wearables through the human knee movements. Research is also being done to harvest low-frequency biomechanical movements using nanogenerators [111, 112]. Reference [113] discusses a possible radiofrequency energy harvester and storing system for wearable sensors.

6.5.3.3

Wearable Device Development

Wearable devices have a broad range of applications and demand usage-specific technologies besides common ones. Implementation of identification and localization mechanisms into wearable devices without affecting their lightweight and ensuring proper privacy measures is one necessary area to research. Body area networks with inbuilt drug pumps and other closed-loop systems, for example, necessitate the development of proper user identification [97]. Near field communication and RFIDs can be utilized in smart health applications to localize users in small areas such as elderly care homes [114]. Encryption of data from wearable devices is another complex area with the scope of research. For cloud-based smart healthcare security, the paper [99] recommends an Attribute-Based Encryption (ABE) and fully homomorphic encryption (FHE) hybrid scheme which would be lightweight implementing on wearable devices. To enhance the response time and critical situating management, Tuli et al. [106] propose integrating various frameworks such as sensor networks, serverless computing, data analytics etc. and shifting computation to the wearable devices. The researchers also put forward quantum computing, targeted operating systems, 6G, processing-in-memory etc. as future research directions in smart healthcare.

6.5.3.4

Big Data Analytics

The big data of smart healthcare also opens a myriad of opportunities for research, application development, and healthcare improvement. As the data provides a clearer picture of critical situations, more adept intervention methods can be used. Analysis of aggregate data will be crucial in solving medical questions and a quicker and more accurate diagnosis of diseases and identifying the most effective treatments. Health threat identification, epidemics detection and management can be streamlined through historical and globally aggregated health data [96].

6 Attacks and Countermeasures in IoT Based …

6.5.3.5

83

Smartphone Applications for Smart Healthcare

Smartphone-based healthcare tools are a rising wave of applications that utilize and generate big data of smart healthcare. With billions of individuals operating smartphones, the P4 goal of healthcare is getting closer to actualization. Smartphones present a versatile platform, and many research opportunities exist in the development of smartphone-based health monitoring. The area is being explored by recent researches such as smartphone ECG monitoring [115], respiratory monitoring, testing for respiratory disease symptoms [116], oral health monitoring using noninvasive periodontal diagnosis [117], and so on. There is immense potential for providing mental healthcare through smartphones. User engagement issues need to be resolved through innovative means to improve the usage of these applications [118]; research into improving these applications’ trustworthiness and transparency could transform user acceptance [119, 120]. Smartphones are becoming the apex of IoT networks in healthcare. Their already massive users and research into smoother integration with cloud and improved processing of the enormous could revolutionize smart healthcare.

6.6 Conclusion This chapter discuss the smart healthcare based on smart city perspectives including the application of IoT in smart healthcare applications. Healthcare is a basic human need and one of the most crucial sector to focus. The recent pandemic has hit countries worldwide in unprecedented ways and has asserted the importance of an intelligent healthcare system. Smart cities should have vital healthcare and preventive ones that utilize the constant stream of data generated by its citizens to identify patterns of diseases and the effectiveness of treatments and therapies. Attacks on smart healthcare can cripple the total system. Due to the attacks, significant data will be lost and personal data will be at risk. Since health data is one of the most sensitive data falling it into the wrong had can bring catastrophic effects. For this reason, possible countermeasures are needed. These are discussed in this paper. The scope of future research in smart healthcare is extensive. Smart health applications require better-specialized algorithms and integration of machine learning for more intelligent systems. Wearable devices and BANs need improved energy efficiency and added features without compromising lightweights or security. The big data of healthcare also opens opportunities for research into data operations targeted for healthcare and efficient storing mechanisms. The importance and exigency of research into smart healthcare and the plethora of opportunities present make it a prime research area. A foundation for understanding smart healthcare in smart cities has been laid out in this paper in recent years.

84

A. K. M. Bahalul Haque et al.

References 1. Cirillo, F., Gómez, D., Diez, L., Elicegui Maestro, I., Gilbert, T.B.J., Akhavan, R.: Smart city IoT services creation through large-scale collaboration. IEEE Internet Things J. 7(6), 5267–5275 (2020). https://doi.org/10.1109/JIOT.2020.2978770 2. Haque, A.K., Bhushan, B., Dhiman, G.: Conceptualizing smart city applications: requirements, architecture, security issues, and emerging trends. Expert Syst. (2021). https://doi.org/ 10.1111/exsy.12753 3. Laurini, R.: A primer of knowledge management for smart city governance. Land Use Policy 104832 (2020).https://doi.org/10.1016/j.landusepol.2020.104832 4. Vaquero, M.G., Saiz-Alvarez, J.M.: Smart cities in Spain—policy, sustainability, and the national plan. In: Economic Modeling, Analysis, and Policy for Sustainability, pp. 266–283 (2016). https://doi.org/10.4018/978-1-5225-0094-0.ch014 5. Park, B.-J., et al.: Long-term warming trends in Korea and contribution of urbanization: an updated assessment. J. Geophys. Res.: Atmos. 122(20), 10637–10654 (2017). https://doi.org/ 10.1002/2017jd027167 6. Roopa, M.S., Pattar, S., Buyya, R., Venugopal, K.R., Iyengar, S.S., Patnaik, L.M.: Social Internet of Things (SIoT): foundations, thrust areas, systematic review and future directions, Comput. Commun. 139, 32–57 (2019). ISSN 0140-3664. https://doi.org/10.1016/j.comcom. 2019.03.009 7. Saxena, S., Bhushan, B., Ahad, M.A.: Blockchain based solutions to secure IoT: background, integration trends and a way forward. J. Netw. Comput. Appl. 103050 (2021).https://doi.org/ 10.1016/j.jnca.2021.103050 8. Fan, Y.J., Yin, Y.H., Da Xu, L., Zeng, Y., Wu, F.: IoT-based smart rehabilitation system. IEEE Trans. Ind. Inform. 10(2), 1568–1577 (2014). https://doi.org/10.1109/tii.2014.2302583 9. Câmara Gradim, L.C., Archanjo José, M., Marinho Cezar da Cruz, D., de Deus Lopes, R.: IoT services and applications in rehabilitation: an interdisciplinary and meta-analysis review. IEEE Trans. Neural Syst. Rehabil. Eng. 28(9), 2043–2052 (2020). https://doi.org/10.1109/ TNSRE.2020.3005616 10. Chang, S.-H., Chiang, R.-D., Wu, S.-J., Chang, W.-T.: A context-aware, interactive M-health system for diabetics. IT Prof. 18(3), 14–22 (2016). https://doi.org/10.1109/mitp.2016.48 11. Wolgast, G., Ehrenborg, C., Israelsson, A., Helander, J., Johansson, E., Manefjord, H.: Wireless body area network for heart attack detection [education corner]. IEEE Antennas Propag. Mag. 58(5), 84–92 (2016). https://doi.org/10.1109/map.2016.2594004 12. Hall, R.E., Bowerman, B., Braverman, J., Taylor, J., Todosow, H., Von Wimmersperg, U.: The vision of a smart city, Brookhaven National Laboratory, Upton, NY (US), BNL-67902; 04042, Sept 2000 13. Harrison, C.: Roads to smarter cities. In: Concept-Oriented Research and Development in Information Technology, pp. 55–69 (2014). https://doi.org/10.1002/9781118753972.ch4 14. Giffinger, R., Fertner, C., Kramar, H., Kalasek, R., Pichler-Milanovic, N., Meijers, E.: Smart cities: ranking of European medium-sized cities, Vienna UT, Jan 2007 15. Eger, J.M.: Smart growth, smart cities, and the crisis at the pump a worldwide phenomenon. I-WAYS Dig. Electron. Commer. Policy Regul. 32(1), 47–53 (2009). https://doi.org/10.3233/ iwa-2009-0164 16. U. N. D. of E. A. S. Affairs and United Nations Department of Economic and Social Affairs: World Urbanization Prospects: The 2018 Revision (2019). https://doi.org/10.18356/b9e995 fe-en 17. Chen, M.: Towards smart city: M2M communications with software agent intelligence. Multimed. Tools Appl. 67(1), 167–178 (2013). https://doi.org/10.1007/s11042-012-1013-4 18. Aujla, G.S., Singh, M., Bose, A., Kumar, N., Han, G., Buyya, R.: BlockSDN: blockchainas-a-service for software defined networking in smart city applications. IEEE Netw. 34(2), 83–91 (2020). https://doi.org/10.1109/MNET.001.1900151

6 Attacks and Countermeasures in IoT Based …

85

19. Silva, B.N., Khan, M., Han, K.: Towards sustainable smart cities: a review of trends, architectures, components, and open challenges in smart cities. Sustain. Cities Soc. 38, 697–713 (2018). https://doi.org/10.1016/j.scs.2018.01.053 20. Haque, A.B., Bhushan, B.: Security attacks and countermeasures in wireless sensor networks. In: Integration of WSNs into Internet of Things, pp. 17–43 (2021).https://doi.org/10.1201/ 9781003107521-2 21. Kandris, D., Nakas, C., Vomvas, D., Koulouras, G.: Applications of wireless sensor networks: an up-to-date survey. Appl. Syst. Innov. 3(1), 14 (2020). https://doi.org/10.3390/asi3010014 22. Wazid, M., Das, A.K., Hussain, R., Succi, G., Joel, J.P.: Authentication in cloud-driven IoTbased big data environment: survey and outlook. J. Syst. Archit. 97, 185–196 (2019). https:// doi.org/10.1016/j.sysarc.2018.12.005 23. Gretzel, U., Werthner, H., Koo, C., Lamsfus, C.: Conceptual foundations for understanding smart tourism ecosystems. Comput. Hum. Behav. 50, 558–563 (2015). https://doi.org/10. 1016/j.chb.2015.03.043 24. Kirimtat, A., Krejcar, O., Kertesz, A., Fatih Tasgetiren, M.: Future trends and current state of smart city concepts: a survey. IEEE Access 8, 86448–86467 (2020). https://doi.org/10.1109/ access.2020.2992441 25. Bedogni, L., Bononi, L., Di Felice, M., D’Elia, A., Cinotti, T.S.: A route planner service with recharging reservation: electric itinerary with a click. IEEE Intell. Transp. Syst. Mag. 8(3), 75–84 (2016). https://doi.org/10.1109/mits.2016.2573418 26. Page, A., et al.: Support systems for health monitoring using internet-of-things driven data acquisition. Serv. Trans. Serv. Comput. 4(4), 18–34 (2016). https://doi.org/10.29268/stsc. 2016.4.4.2 27. Dhar, J., Ranganathan, A.: Machine learning capabilities in medical diagnosis applications: computational results for hepatitis disease. Int. J. Biomed. Eng. Technol. 17(4), 330 (2015). https://doi.org/10.1504/ijbet.2015.069398 28. Esteva, A., et al.: Dermatologist-level classification of skin cancer with deep neural networks. Nature 542(7639), 115–118 (2017). https://doi.org/10.1038/nature21056 29. Jindal, M., Gupta, J., Bhushan, B.: Machine learning methods for IoT and their future applications. In: 2019 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS) (2019). https://doi.org/10.1109/icccis48478.2019.8974551 30. Tian, S., Yang, W., Le Grange, J.M., Wang, P., Huang, W., Ye, Z.: Smart healthcare: making medical care more intelligent. Glob. Health J. 3(3), 62–65 (2019). https://doi.org/10.1016/j. glohj.2019.07.001 31. To˘gaçar, M., Özkurt, K.B., Ergen, B., Cömert, Z.: BreastNet: a novel convolutional neural network model through histopathological images for the diagnosis of breast cancer. Phys. A 545, 123592 (2020). https://doi.org/10.1016/j.physa.2019.123592 32. Peters, B.S., Armijo, P.R., Krause, C., Choudhury, S.A., Oleynikov, D.: Review of emerging surgical robotic technology. Surg. Endosc. 32(4), 1636–1655 (2018). https://doi.org/10.1007/ s00464-018-6079-2 33. Cook, D.J., Duncan, G., Sprint, G., Fritz, R.: Using smart city technology to make healthcare smarter. Proc. IEEE Inst. Electr. Electron. Eng. 106(4), 708–722 (2018). https://doi.org/10. 1109/JPROC.2017.2787688 34. Lucisano, J.Y., Routh, T.L., Lin, J.T., Gough, D.A.: Glucose monitoring in individuals with diabetes using a long-term implanted sensor/telemetry system and model. IEEE Trans. Biomed. Eng. 64(9), 1982–1993 (2017). https://doi.org/10.1109/TBME.2016.2619333 35. Akmandor, A.O., Jha, N.K.: Keep the stress away with SoDA: stress detection and alleviation system. IEEE Trans. Multi-Scale Comput. Syst. 3(4), 269–282 (2017). https://doi.org/10. 1109/tmscs.2017.2703613 36. Redfern, J.: Smart health and innovation: facilitating health-related behaviour change. Proc. Nutr. Soc. 76(3), 328–332 (2017). https://doi.org/10.1017/s0029665117001094 37. Oncologists partner with Watson on genomics. Cancer Discov. 5(8), 788–788 (2015). https:// doi.org/10.1158/2159-8290.cd-nb2015-090. Epub 2015 Jun 16. PMID: 26080837

86

A. K. M. Bahalul Haque et al.

38. Zhong, F., et al.: Artificial intelligence in drug design. Sci. China Life Sci. 61(10), 1191–1204 (2018). https://doi.org/10.1007/s11427-018-9342-2 39. Kumar, S.P., Samson, V.R.R., Sai, U.B., Rao, P.L.S.D.M., Eswar, K.K.: Smart health monitoring system of patient through IoT. In: 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), Palladam, Tamilnadu, India, pp. 551–556, Feb 2017. https://doi.org/10.1109/I-SMAC.2017.8058240 40. Mshali, H., Lemlouma, T., Magoni, D.: Adaptive monitoring system for e-health smart homes. Pervasive Mob. Comput. 43, 1–19 (2018). https://doi.org/10.1016/j.pmcj.2017.11.001 41. Wan, J., et al.: Wearable IoT enabled real-time health monitoring system. EURASIP J. Wirel. Commun. Netw. 2018(1), 59 (2018). https://doi.org/10.1186/s13638-018-1308-x 42. Mshali, H., Lemlouma, T., Moloney, M., Magoni, D.: A survey on health monitoring systems for health smart homes. Int. J. Ind. Ergon. 66, 26–56 (2018). https://doi.org/10.1016/j.ergon. 2018.02.002 43. Kang, M., Park, E., Cho, B.H., Lee, K.-S.: Recent patient health monitoring platforms incorporating internet of things-enabled smart devices. Int. Neurourol. J. 22(Suppl 2), S76-82 (2018). https://doi.org/10.5213/inj.1820corr.001 44. Kharel, J., Reda, H.T., Shin, S.Y.: Fog computing-based smart health monitoring system deploying LoRa wireless communication. IETE Tech. Rev. 36(1), 69–82 (2018). https://doi. org/10.1080/02564602.2017.1406828 45. Kajornkasirat, S., Chanapai, N., Hnusuwan, B.: Smart health monitoring system with IoT. In: 2018 IEEE Symposium on Computer Applications & Industrial Electronics (ISCAIE), Penang, pp. 206–211, Apr 2018. https://doi.org/10.1109/ISCAIE.2018.8405471 46. Albahri, A.S., et al.: Based multiple heterogeneous wearable sensors: a smart real-time health monitoring structured for hospitals distributor. IEEE Access 7, 37269–37323 (2019). https:// doi.org/10.1109/ACCESS.2019.2898214 47. Puntambekar, V., Agarwal, S., Mahalakshmi, P.: Dynamic monitoring of health using smart health band: SocProS 2018, Volume 2. In: Soft Computing for Problem Solving, vol. 1057, pp. 453–462. Springer Singapore (2020).https://doi.org/10.1007/978-981-15-0184-5_39 48. Li, J., Ma, Q., Chan, A.H., Man, S.S.: Health monitoring through wearable technologies for older adults: smart wearables acceptance model. Appl. Ergon. 75, 162–169 (2019). https:// doi.org/10.1016/j.apergo.2018.10.006 49. Islam, A., Shin, S.Y.: BHMUS: blockchain based secure outdoor health monitoring scheme using UAV in smart city. In: 2019 7th International Conference on Information and Communication Technology (ICoICT), Kuala Lumpur, Malaysia, pp. 1–6, July 2019. https://doi.org/ 10.1109/ICoICT.2019.8835373 50. Hartmann, M., Hashmi, U.S., Imran, A.: Edge computing in smart health care systems: review, challenges, and research directions. Trans. Emerg. Telecommun. Technol. 71, 503 (2019). https://doi.org/10.1002/ett.3710 51. Gahlot, S., Reddy, S.R.N., Kumar, D.: Review of smart health monitoring approaches with survey analysis and proposed framework. IEEE Internet Things J. 6(2), 2116–2127 (2019). https://doi.org/10.1109/JIOT.2018.2872389 52. Rajamohanan, D., Hariharan, B., Unnikrishna Menon, K.A.: Survey on smart health management using BLE and BLE beacons. In: 2019 9th International Symposium on Embedded Computing and System Design (ISED), Kollam, India, pp. 1–5, Dec 2019. https://doi.org/10. 1109/ISED48680.2019.9096227 53. Rayan, Z., Alfonse, M., Salem, A.-B.M.: Machine learning approaches in smart health. Procedia Comput. Sci. 154, 361–368 (2019). https://doi.org/10.1016/j.procs.2019.06.052 54. Abdellatif, A.A., Al-Marridi, A.Z., Mohamed, A., Erbad, A., Chiasserini, C.F., Refaey, A.: ssHealth: toward secure, blockchain-enabled healthcare systems. IEEE Netw. 34(4), 312–319 (2020). https://doi.org/10.1109/MNET.011.1900553 55. Allam, Z., Jones, D.S.: On the coronavirus (COVID-19) outbreak and the smart city network: universal data sharing standards coupled with Artificial Intelligence (AI) to benefit urban health monitoring and management. Healthcare (Basel) 8(1) (2020). https://doi.org/10.3390/ healthcare8010046

6 Attacks and Countermeasures in IoT Based …

87

56. Zghaibeh, M., Farooq, U., Hasan, N.U., Baig, I.: SHealth: a blockchain-based health system with smart contracts capabilities. IEEE Access 8, 70030–70043 (2020). https://doi.org/10. 1109/access.2020.2986789 57. Meng, K., et al.: A wireless textile-based sensor system for self-powered personalized health care. Matter 2(4), 896–907 (2020). https://doi.org/10.1016/j.matt.2019.12.025 58. Chen, B., et al.: A security awareness and protection system for 5G smart healthcare based on zero-trust architecture. IEEE Internet Things J. 1–1 (2020). https://doi.org/10.1109/jiot. 2020.3041042 59. Ahmadi-Assalemi, G., et al.: Digital twins for precision healthcare. In: Advanced Sciences and Technologies for Security Applications, pp. 133–158 (2020). https://doi.org/10.1007/9783-030-35746-7_8 60. Wang, Z., Luo, N., Zhou, P.: GuardHealth: blockchain empowered secure data management and Graph Convolutional Network enabled anomaly detection in smart healthcare. J. Parallel Distrib. Comput. 142, 1–12 (2020). https://doi.org/10.1016/j.jpdc.2020.03.004 61. Tanwar, S., Parekh, K., Evans, R.: Blockchain-based electronic healthcare record system for healthcare 4.0 applications. J. Inf. Secur. Appl. 50, 102407 (2020). https://doi.org/10.1016/j. jisa.2019.102407 62. Zhong, H., Zhou, Y., Zhang, Q., Xu, Y., Cui, J.: An efficient and outsourcing-supported attribute-based access control scheme for edge-enabled smart healthcare. Future Gener. Comput. Syst. 115, 486–496 (2021). https://doi.org/10.1016/j.future.2020.09.021 63. Wu, F., Qiu, C., Wu, T., Yuce, M.R.: Edge-based hybrid system implementation for long-range safety and healthcare IoT applications. IEEE Internet Things J. 1–1 (2021). https://doi.org/ 10.1109/jiot.2021.3050445 64. Yang, Z., Liang, B., Ji, W.: An intelligent end-edge-cloud architecture for visual IoT assisted healthcare systems. IEEE Internet Things J. https://doi.org/10.1109/JIOT.2021.3052778 65. Alzubi, J.A.: Blockchain-based Lamport Merkle Digital Signature: authentication tool in IoT healthcare. Comput. Commun. (2021). https://doi.org/10.1016/j.comcom.2021.02.002 66. Butt, S.A., Diaz-Martinez, J.L., Jamal, T., Ali, A., De-La-Hoz-Franco, E., Shoaib, M.: IoT smart health security threats. In: 2019 19th International Conference on Computational Science and Its Applications (ICCSA) (2019). https://doi.org/10.1109/iccsa.2019.000-8 67. Hassija, V., Chamola, V., Bajpai, B.C., Zeadally, S.: Security issues in implantable medical devices: fact or fiction? Sustain. Cities Soc. 102552 (2020).https://doi.org/10.1016/j.scs.2020. 102552 68. Alam, S., De, D.: Analysis of security threats in wireless sensor network. Int. J. Wirel. Mob. Netw. 6(2), 35–46 (2014). https://doi.org/10.5121/ijwmn.2014.6204 69. Habibzadeh, H., Soyata, T.: Toward uniform smart healthcare ecosystems: a survey on prospects, security, and privacy considerations. In: Connected Health in Smart Cities, pp. 75–112 (2020). https://doi.org/10.1007/978-3-030-27844-1_5 70. Kumar, P., Lee, H.-J.: Security issues in healthcare applications using wireless medical sensor networks: a survey. Sensors 12(1), 55–91 (2012). https://doi.org/10.3390/s120100055 71. Ahmed, I., Mousa, A.: Security and privacy issues in Ehealthcare systems: towards trusted services. Int. J. Adv. Comput. Sci. Appl. 7(9) (2016). https://doi.org/10.14569/ijacsa.2016. 070933 72. Sharma, M.K., Joshi, B.K.: Detection & prevention of vampire attack in wireless sensor networks. In: 2017 International Conference on Information, Communication, Instrumentation and Control (ICICIC), Indore, pp. 1–5, Aug 2017. https://doi.org/10.1109/ICOMICON. 2017.8279174 73. Tseng, F.-H., Chou, L.-D., Chao, H.-C.: A survey of black hole attacks in wireless mobile ad hoc networks. Hum. Centric Comput. Inf. Sci. 1(1), 4 (2011). https://doi.org/10.1186/21921962-1-4 74. Latif, R., Abbas, H., Assar, S.: Distributed denial of service (DDoS) attack in cloud-assisted wireless body area networks: a systematic literature review. J. Med. Syst. 38(11), 128 (2014). https://doi.org/10.1007/s10916-014-0128-8

88

A. K. M. Bahalul Haque et al.

75. Kumar, P.M., Gandhi, U.D.: Enhanced DTLS with CoAP-based authentication scheme for the internet of things in healthcare application. J. Supercomput. 76(6), 3963–3983 (2020). https://doi.org/10.1007/s11227-017-2169-5 76. Vishwakarma, R., Jain, A.K.: A survey of DDoS attacking techniques and defence mechanisms in the IoT network. Telecommun. Syst. 73(1), 3–25 (2020). https://doi.org/10.1007/s11235019-00599-z 77. Javaid, U., Siang, A.K., Aman, M.N., Sikdar, B.: Mitigating loT device based DDoS attacks using blockchain. In: Proceedings of the 1st Workshop on Cryptocurrencies and Blockchains for Distributed Systems—CryBlock’18 (2018). https://doi.org/10.1145/3211933.3211946 78. Ul, S., Manickam, S.: Improved mechanism to prevent denial of service attack in IPv6 duplicate address detection process. Int. J. Adv. Comput. Sci. Appl. 8(2) (2017). https://doi.org/ 10.14569/ijacsa.2017.080209 79. Biswal, A., Bhushan, B.: Blockchain for internet of things: architecture, consensus advancements, challenges and application areas. In: 2019 5th International Conference On Computing, Communication, Control And Automation (ICCUBEA), Pune, India, pp. 1–6, Sept 2019. https://doi.org/10.1109/ICCUBEA47591.2019.9129181 80. Arora, D., Gautham, S., Gupta, H., Bhushan, B.: Blockchain-based security solutions to preserve data privacy and integrity. In: 2019 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), Greater Noida, India, pp. 468–472, Oct 2019. https://doi.org/10.1109/ICCCIS48478.2019.8974503. 81. Yaacoub, J.P.A., Noura, M., Noura, H.N., Salman, O., Yaacoub, E.,Couturier, R., Chehab, A.: Securing internet of medical things systems: limitations, issues and recommendations. Future Gener. Comput. Syst. 105, 581–606 (2020). ISSN 0167-739X. https://doi.org/10. 1016/j.future.2019.12.028 82. Shakeel, P.M., Mohamed Shakeel, P., Baskar, S., Sarma Dhulipala, V.R., Mishra, S., Jaber, M.M.: Maintaining security and privacy in health care system using learning based deep-Qnetworks. J. Med. Syst. 42(10) (2018). https://doi.org/10.1007/s10916-018-1045-z 83. Podder, P., Mondal, M.R.H., Bharati, S., Paul, P.K.: Review on the security threats of internet of things. IJCAI 176(41), 37–45 (2020). https://doi.org/10.5120/ijca2020920548 84. Mohammad, A.H.: Ransomware evolution, growth and recommendation for detection. Mod. Appl. Sci. 14(3), 68 (2020). https://doi.org/10.5539/mas.v14n3p68 85. Chakkaravarthy, S.S., Sibi Chakkaravarthy, S., Sangeetha, D., Cruz, M.V., Vaidehi, V., Raman, B.: Design of intrusion detection honeypot using social leopard algorithm to detect IoT ransomware attacks. IEEE Access 8, 169944–169956 (2020). https://doi.org/10.1109/access. 2020.3023764 86. Xu, Z., Xu, C., Liang, W., Xu, J., Chen, H.: A lightweight mutual authentication and key agreement scheme for medical internet of things. IEEE Access 7, 53922–53931 (2019). https:// doi.org/10.1109/access.2019.2912870 87. Papaioannou, M., et al.: A survey on security threats and countermeasures in Internet of Medical Things (IoMT). Trans. Emerg. Telecommun. Technol. (2020). https://doi.org/10. 1002/ett.4049 88. Huang, P., Guo, L., Li, M., Fang, Y.: Practical privacy-preserving ECG-based authentication for IoT-based healthcare. IEEE Internet Things J. 6(5), 9200–9210 (2019). https://doi.org/10. 1109/jiot.2019.2929087 89. Shuwandy, M.L., et al.: mHealth authentication approach based 3D touchscreen and microphone sensors for real-time remote healthcare monitoring system: comprehensive review, open issues and methodological aspects. Comput. Sci. Rev. 38, 100300 (2020). https://doi. org/10.1016/j.cosrev.2020.100300 90. Shakil, K.A., Zareen, F.J., Alam, M., Jabin, S.: BAMHealthCloud: a biometric authentication and data management system for healthcare data in cloud. J. King Saud Univ. Comput. Inf. Sci. 32(1), 57–64 (2020). https://doi.org/10.1016/j.jksuci.2017.07.001 91. Deogirikar, J., Vidhate, A.: Security attacks in IoT: a survey. In: 2017 International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), Palladam, pp. 32–37 (2017). https://doi.org/10.1109/I-SMAC.2017.8058363

6 Attacks and Countermeasures in IoT Based …

89

92. Zhang, M., Raghunathan, A., Jha, N.K.: Trustworthiness of medical devices and body area networks. Proc. IEEE 102(8), 1174–1188 (2014). https://doi.org/10.1109/JPROC.2014.232 2103 93. Maiti, A., Jadliwala, M., He, J., Bilogrevic, I.: (Smart)watch your taps. In: Proceedings of the 2015 ACM International Symposium on Wearable Computers—ISWC ’15 (2015). https:// doi.org/10.1145/2802083.2808397 94. Al-Janabi, S., Al-Shourbaji, I., Shojafar, M., Shamshirband, S.: Survey of main challenges (security and privacy) in wireless body area networks for healthcare applications. Egypt. Inform. J. 18(2), 113–122 (2017). https://doi.org/10.1016/j.eij.2016.11.001 95. Goyal, S., Sharma, N., Bhushan, B., Shankar, A., Sagayam, M.: IoT enabled technology in secured healthcare: applications, challenges and future directions. In: Cognitive Internet of Medical Things for Smart Healthcare, pp. 25–48 (2020). https://doi.org/10.1007/978-3-03055833-8_2 96. Solanas, A., et al.: Smart health: a context-aware health paradigm within smart cities. IEEE Commun. Mag. 52(8), 74–81 (2014). https://doi.org/10.1109/mcom.2014.6871673 97. Jovanov, E., Milenkovic, A.: Body Area Networks for ubiquitous healthcare applications: opportunities and challenges. J. Med. Syst. 35(5), 1245–1254 (2011). https://doi.org/10.1007/ s10916-011-9661-x 98. Selvaraj, S., Sundaravaradhan, S.: Challenges and opportunities in IoT healthcare systems: a systematic review. SN Appl. Sci. 2(1) (2020). https://doi.org/10.1007/s42452-019-1925-y 99. Baker, S.B., Xiang, W., Atkinson, I.: Internet of things for smart healthcare: technologies, challenges, and opportunities. IEEE Access 5, 26521–26544 (2017). https://doi.org/10.1109/ access.2017.2775180 100. Rizwan, A., et al.: A review on the role of nano-communication in future healthcare systems: a big data analytics perspective. IEEE Access 6, 41903–41920 (2018). https://doi.org/10.1109/ access.2018.2859340 101. Sagner, M., et al.: The P4 health spectrum—a predictive, preventive, personalized and participatory continuum for promoting healthspan. Prog. Prev. Med. 2(1), e0002 (2017). https:// doi.org/10.1097/pp9.0000000000000002 102. Varshney, U., Chang, C.K.: Smart health and well-being. Computer 49(11), 11–13 (2016). https://doi.org/10.1145/2555810.2555811 103. Bhushan, B., Sahoo, C., Sinha, P., Khamparia, A.: Unification of Blockchain and Internet of Things (BIoT): requirements, working model, challenges and future directions. Wirel. Netw. (2020). https://doi.org/10.1007/s11276-020-02445-6 104. Bhushan, B., Khamparia, A., Martin Sagayam, K., Sharma, S.K., Ahad, M.A., Debnath, N.C.: Blockchain for smart cities: a review of architectures, integration trends and future research directions. Sustain. Cities Soc. 61, 102360 (2020). https://doi.org/10.1016/j.scs.2020.102360 105. Chen, M., Li, Y., Luo, X., Wang, W., Wang, L., Zhao, W.: A novel human activity recognition scheme for smart health using multilayer extreme learning machine. In: Cyber-Enabled Intelligence, pp. 239–258 (2019). https://doi.org/10.1201/9780429196621-12 106. Goyal, S., Sharma, N., Bhushan, B., Shankar, A., Sagayam, M.: IoT enabled technology in secured healthcare: applications, challenges and future directions. In: Cognitive Internet of Medical Things for Smart Healthcare Studies in Systems, Decision and Control, pp. 25–48 (2020).https://doi.org/10.1007/978-3-030-55833-8_2 107. Gljuš´ci´c, P., Zelenika, S., Blaževi´c, D., Kamenar, E.: Kinetic energy harvesting for wearable medical sensors. Sensors 19(22) (2019). https://doi.org/10.3390/s19224922 108. Nozariasbmarz, A., et al.: Review of wearable thermoelectric energy harvesting: from body temperature to electronic systems. Appl. Energy 258, 114069 (2020). https://doi.org/10.1016/ j.apenergy.2019.114069 109. Bahk, J.-H., Fang, H., Yazawa, K., Shakouri, A.: Flexible thermoelectric materials and device optimization for wearable energy harvesting. J. Mater. Chem. C 3(40), 10362–10374 (2015). https://doi.org/10.1039/c5tc01644d 110. Tuncel, Y., Bandyopadhyay, S., Kulshrestha, S.V., Mendez, A., Ogras, U.Y.: Towards wearable piezoelectric energy harvesting. In: Proceedings of the ACM/IEEE International Symposium on Low Power Electronics and Design (2020).https://doi.org/10.1145/3370748.3406578

90

A. K. M. Bahalul Haque et al.

111. Yan, C., et al.: A linear-to-rotary hybrid nanogenerator for high-performance wearable biomechanical energy harvesting. Nano Energy 67, 104235 (2020). https://doi.org/10.1016/j.nan oen.2019.104235 112. Zou, Y., Raveendran, V., Chen, J.: Wearable triboelectric nanogenerators for biomechanical energy harvesting. Nano Energy 77, 105303 (2020). https://doi.org/10.1016/j.nanoen.2020. 105303 113. Borges, L.M., Chávez-Santiago, R., Barroca, N., Velez, F.J., Balasingham, I.: Radio-frequency energy harvesting for wearable sensors. Healthc. Technol. Lett. 2(1), 22–27 (2015). https:// doi.org/10.1049/htl.2014.0096 114. Dohr, A., Modre-Opsrian, R., Drobics, M., Hayn, D., Schreier, G.: The internet of things for ambient assisted living. In: 2010 Seventh International Conference on Information Technology: New Generations (2010). https://doi.org/10.1109/itng.2010.104 115. Faezipour, M., Faezipour, M.: System dynamics modeling for smartphone-based healthcare tools: case study on ECG monitoring. IEEE Syst. J. 1–10 (2020). https://doi.org/10.1109/ jsyst.2020.3009187 116. Faezipour, M., Faezipour, M.: Sustainable smartphone-based healthcare systems: a systems engineering approach to assess the efficacy of respiratory monitoring apps. Sustainability 12(12), 5061 (2020). https://doi.org/10.3390/su12125061 117. Veeralingam, S., Sahatiya, P., Kadu, A., Mattela, V., Badhulika, S.: Direct, one-step growth of NiSe2 on cellulose paper: a low-cost, flexible, and wearable with smartphone enabled multifunctional sensing platform for customized noninvasive personal healthcare monitoring. ACS Appl. Electron. Mater. 1(4), 558–568 (2019). https://doi.org/10.1021/acsaelm.9b00022 118. Torous, J., Nicholas, J., Larsen, M.E., Firth, J., Christensen, H.: Clinical review of user engagement with mental health smartphone apps: evidence, theory and improvements. Evid. Based Ment. Health 21(3), 116–119 (2018). https://doi.org/10.1136/eb-2018-102891 119. Sharma, N., Kaushik, I., Bhushan, B., Gautam, S., Khamparia, A.: Applicability of WSN and biometric models in the field of healthcare. In: Deep Learning Strategies for Security Enhancement in Wireless Sensor Networks Advances in Information Security, Privacy, and Ethics, pp. 304–329 (2020).https://doi.org/10.4018/978-1-7998-5068-7.ch016 120. Haque, A.B., Muniat, A., Ullah, P.R., Mushsharat, S.: An automated approach towards smart healthcare with blockchain and smart contracts. In: 2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), pp. 250–255 (2021). https:// doi.org/10.1109/ICCCIS51004.2021.9397158

Part II

Machine Learning

Chapter 7

Online Product Review Monitoring System Using Machine Learning J. Madhumathi, R. Aishwarya, V. Vedha Pavithra, and Sandra Johnson

Abstract In this fast moving world, having major proportion of people depending on online websites for purchase of their day to day necessities, their only scope of trust is the reviews available in the about the particular product. Counterfeit audit discovery and its disposal from the given dataset utilizing distinctive Natural Language Processing strategies is significant in a few perspectives. These reviews can either be authentic made by loyal buyers or even it may be forged for marketing purposes. According to survey, almost sixty percent of the reviews present in amazon are fake and nearly fifteen percent of companies as sellers, wage individuals for creating fake reviews. In order to settle the imbalance and forgery a study has been made to identify the fake reviews and filter it out from the user’s sight, such that they can’t be manipulated as well as the particular company’s reputation won’t be at stake. This system majorly works on the basis of identifying fake reviews by user extracting the multiple review of single user from reviewers. Keywords Reviews · Machine learning · Fake · Signed inference algorithm · Genetic algorithm

J. Madhumathi (B) · R. Aishwarya · V. Vedha Pavithra · S. Johnson Departments of Computer Science and Engineering, R.M.K. Engineering College, Kavaraipettai, Chennai, India e-mail: [email protected] R. Aishwarya e-mail: [email protected] V. Vedha Pavithra e-mail: [email protected] S. Johnson e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_7

93

94

J. Madhumathi et al.

7.1 Introduction Reliability is a common and major concept which that everyone has come across as an evaluation constraint in most scenarios, in case on online shopping the only source that buyers rely on are the reviews made by users and the star rating of the product. More that star rating, the reviews are that make up the buying or denying decisions of users. The user rating and reviews serves as a major factor in driving sales, though these reviews sound and are projected as progressive improvement for purchaser benefit, the research and statistical output says the otherwise. The scholars found into being that average consumer ratings associated off-color with the tallies from Consumer Reports when compared, having alteration in average user rating between twosomes of yields larger than one star, the item with the upper user rating was appraised more favorably by Consumer Reports only around two-thirds of values provided. Additionally, for instance when comparing a laptop provided with average rating of four out of five stars along with one more laptop which has average rating of three out of five stars, the first laptop would only be objectively better 65% and not the expected 100% at most of the situation. It makes a far difference in quality of product; this applies even more apt for reviews than star rating. There are many reasons as to why the reviews are manipulated; the reasons would either be in favour of the marketer or competitor. For the marketer, in order to have positive reputation in market, as it serves as a most needed asset for driving sales, they would generate positive reviews. On the other hand, for competitors to increase their sales and to plummet the success of their rival companies, they would write negative reviews about the product by their rivals. In either of these cases, customers are those who suffer the most. To overcome this problem, a new and novel online product review monitoring system to detect the net spam is proposed. The system serves the purpose by evaluating the reviews and then labeling them to be fake or genuine, in a way that fake reviews are hidden from the sight of users, protecting the customers from being manipulate from spending their money on unworthy products. The system uses signed inference algorithm and genetic algorithm to achieve this evaluation and labeling of reviews.

7.2 Literature Review Liang et al. [1] explained a mechanism to detect spam reviews by taking both feature and relationship of the reviewer in consideration, as at the present time, most of the customers can gain copious information and help for making decisions in purchasing products and service from online review assets, through reviews in social media. It on other hand stimulates some manufactures in appointing spammers for writing fake criticisms as well on some target products. The concept of how to perceive fake review

7 Online Product Review Monitoring System …

95

as well as review creator is seeking the consideration of marketers and ecommerce. A fresh and unique graph model with multi edge, having for each node demonstrating a referee and apiece edge embodies a reviewer’s inter-relationship on each one of special product. Reviewers’ unreliability score as the feature based on which combing is done, an unendorsed continuous working out framework is proposed. Moreover, it is by far the first set of rules where both the reviewer’s possibilities and interrelationships between them are considered together. Experimental results show that the method is effective in detecting spam reviewers with a satisfied precision, but requires additional implementation requirements such as, dataset powered up with detailed attributes. Lin et al. [2] discussed on spotting of the bogus reviews from the provided sequential reviews in online social sites, as distinguishing and identifying spam in review is way more significant for present e-commerce bids. However, the displayed order of review has been abandoned by the previous research works. It considers the problem on fake review uncovering in review categorization it is crucial for instigating online anti-opinion spam implementation. The characteristics of fake reviews are first needed to be analyzed. Then based on valuation contents and critics behavior, a six period profound details are suggested for fake reviews highlighting. Following that the process devises an administered elucidation followed by threshold-based solution for spot the fake reviews at earliest possible situation. The trial results indicate that undertaken methodology is capable of identifying the fake reviews methodical with higher accuracy and recollection, but the limitations are that it requires training data sets, as well as the model works only on sequential data. Istiaq Ahsan et al. [3] discussed about the issues in online marketing such that, entire of the e-commerce has begun to get mammoth as the days passes by, even if it doesn’t by every passing minute. Online Evaluations has a crucial role in the online marketing arena, as well as, it has proved itself for being promising in terms of judgment constructing from the eyes of shoppers. The shoppers only scope of trust were the reviews. Moreover, these are precisely profound and substantial facts according to the customer, that would make certain the genuineness of user-generated gratified discussion groups, Reviews, blogs, media, blogs and so on is unpredictably perceptible. The limitations where that, it allowed spam recognition in only those contents with word count above 150, shorter reviews where not taken into consideration. Rajamohana et al. [4] has performed a survey on techniques that can be used in spam detection of reviews having our commonplace activities getting majorly impacted by internets influence, e-commerce is facing the rapid development zones in the Internet era. The survey has a detailed survey is done using various mechanism learning practices for sensing spam and sincere reviews, but the limitations is that it provides multiple approaches but not the specify the efficient and suitable one. Zhang et al. [5] proposed a collective hyping mechanism for identify the forged reviews in online activities as ever since advent of online shopping bogus reviews always misinform consumers shopping online. He proposes a new Concerted Promotion Hyping Recognition solution. But the limitation is that it cannot identify the individual spammers in social media as it focuses on masses.

96

J. Madhumathi et al.

Jia et al. [6] articulated a LDA based system for detecting spam in online reviews as it is obligatory for latent consume to contribute a conclusion based on reviews been made by users online. The disadvantage is that it requires more evaluation time, as the system checks the reviews. Artificial Intelligence is the ability to process information properly in a complex environment. The criteria of properness are not predefined and hence not available beforehand [7, 8]. Shehnepoor et al. [9] portrays the influential nature of online social media in information transmission among people. The limitation is that it just focuses on classification of fake reviews and it doesn’t work on users [10–13].

7.3 Proposed Work In simple terms, this project “online product review monitoring system” considers the delinquent of perceiving fake commentators, thereby as a result identifying fake reviews in online review datasets obtained. The datasets of online review predominantly comprise of customers or reviewers as users, a set of products made available such as mobile phones, laptops and finally the reviews. Individually each criticism is transcribed from a particular user to a particular product, and contains a star-rating, often an integer from 1 to 5. As such, a Bayesian network is used to represent the review data set. The network is displayed in such a way that, user nodules are associated with the merchandise nodes, having the links signify the “reviewed” connections and a review rating attained for each review. The manipulators, goods, and assessment object in the analysis grid, is grouped into certain classes, such that, two classes for each object type: yields are what’s more good or bad quality, users are either truthful or deceit, and finally the reviews as are either genuine or forged. The different phases and internal flows within the system are portrayed in the figure, the system functioning begins with registration and login if in case of user level access and only login if the user has admin level access. Initially the admin logins into the online product review monitoring system and adds products along with its description and images. Later when user logins and searches for products, list of products registered by admin are displayed to the user, from which user can select their desired products after going through the description, images and reviews made available for the product by admin. All these are functioning that take place in the surficial level in open eyes. Internally, after reviews are made by user for products, the reviews are evaluated and labeled to be of deceit or authentic by the system using esteemed algorithms. These algorithms label a review to be fake on three bases, if a review is made by user before buying or product, if the reviews originate from same user numerous times and if the contents of review doesn’t match with the metadata of particular product for which review was formed. After the reviews are labeled, only the reviews categorized as genuine are listed at the user side, the rest are made available only under admin level access login. The admin has additional functionalities such that, the admin can remove a product from

7 Online Product Review Monitoring System …

97

list and also delete the reviews marked as fake. The admin is also provided with user management rights, such that the admin can monitor the functioning of users based on the extracted user id.

7.4 System Architecture From Fig. 7.1, this system takes products and reviews as input for the user and admin, the provided data are sequentially processed by signed inference algorithm and genetic algorithm, these algorithms works on processing the reviews and products by labeling them and on how to display the processed data. The signed inference algorithm is majorly about labeling and the genetic algorithm is inclined towards display mechanism of the labeled contents. The output of the system would hide the fake reviews from customers, as well as the system provides an interactive display of feedback to user, by including an friend option in system, where in case for any product searched by customer, contains reviews made by their friends from the application, their reviews will be displayed on top, provided with higher priority, whereas to admin all fake and honest reviews will be displayed but in a classified manner, where fake reviews are categorized to find out if the fake reviews are created by particular users repeatedly, subsequently tracing out the spammers as well.

Fig. 7.1 System architecture

98

J. Madhumathi et al.

7.5 Methodology The system depicts the real world online shopping problem, the product review which on one hand serves as scope of trust for customers but on other hand also as a weak spot, which the manipulators target to change the rate of profit in their favor. This system, online product review monitoring system to detect the net spam, uses two algorithms signed inference algorithm and the genetic algorithm. Both these algorithms together as a single entity is implemented in multiple functionalities and in mapping of modules. As the system as whole works in evaluation, identification and labeling of reviews, through multiple sub modules of process, the execution of same algorithm exists in multiple sub modules. To ideate and display this real world working a basic online shopping website is developed and incorporated. From Fig. 7.2, the signed inference algorithm, has its algorithms pseudo code involved all the way along in program in numerous functionalities, the algorithm serves many purposes such that, it helps in concealing the fake and bogus reviews from users, it evaluates if the product has been brought by the user for the particular review, if the particular review is the first review made by the esteemed user for that particular product and finally the algorithm check for the relevance of metadata of product for with review is made along with the review itself. The genetic algorithm is implied for this system in two aspects in low key rating and for user review sorting purposes. The algorithm is applied to the low key rating functionality of the system in a way that, the reviews with lower star rating provided that less than 2 would lead to product being removed from system automatically without intervention from admin. This functionality included acts as an added advantage to the system, as it automatically removed product of lower star rating level, than being removed manually by the admin. Then, the algorithm is applied for user review sorting functionality in order to sort the way reviews are displayed to the user, the algorithm works in a manner that all the reviews with similar values are displayed for a given product and also the friend functionality executes on the basis of this algorithm, if any reviews for the selected product are from friends of the user, then these reviews are displayed with higher priority above the reviews from other users.

7.6 Results and Discussion Thus the reviews are classified into fake and genuine by considering the user id, product id, number of products bought by the user in a particular time, review content and product metadata. In Table 7.1, the user details are displayed to the admin. In Table 7.2, the reviews given by the user for particular products they bought are displayed. In Table 7.3, the reviews which are evaluated as fake are displayed to the admin.

7 Online Product Review Monitoring System …

99

Fig. 7.2 Methodology

7.7 Conclusion As the pattern to shop online is expanding step by step and more individuals are keen on purchasing the results of their need from the online stores. This sort of shopping doesn’t take a great deal of season of a client. Client goes to online store, search the thing of his/her need and submit the request. In any case, the thing by which individuals face trouble in purchasing the items from online store is the terrible

100

J. Madhumathi et al.

Table 7.1 User details Profile

Username

Mail id

Number

Location

Gender

Block

anu

[email protected]

123456778

Chennai

Female

Block

kiran

[email protected]

123456778

Chennai

Male

Block

pavithra

[email protected]

232323231

Kallakkurichi

Female

Block

harini

[email protected]

434343432

Villupuram

Female

Block

kubra

[email protected]

545454543

Hyderabad

Female

Block

indira

[email protected]

525252522

Villupuram

Female

Block

indu

[email protected]

343434342

Chennai

Female

Block

jen

[email protected]

131313132

Chennai

Male

Block

Table 7.2 Product reviews Product id

User id

Username

Ratings

Feed back

6

3

pavithra

5

Good

4

3

pavithra

3

Nice

5

1

anu

5

Very nice

6

1

anu

5

Very nice

2

1

and

5

Good

1

1

anu

5

Supper

1

2

kiran

4

Not bad

1

2

kiran

5

Good

1

2

Kiran

5

Very good

Table 7.3 Fake reviews Fake product feed back Product id

User id

Username

Ratings

Feed back

5

1

Anu

5

Very nice

6

1

anu

5

Very nice

2

1

anu

5

Good

1

1

anu

5

Supper

3

1

anu

1

Laptop is good

8

5

Kubra

5

Amazing movie

7

1

anu

4

Good

Null

Null

anu

3

Good

Null

Null

anu

2

khdgflad

7 Online Product Review Monitoring System …

101

nature of the item. Thus this is used to identify the genuine reviews from all other reviews given by the customers for a particular product by concealing the fake reviews from the users using machine learning. This helps the customer to identify quality product and helps the online shopping website to achieve customer satisfaction. Specifically, apparently Machine Learning strategies have gotten more famous and prompted preferable outcomes over more conventional perception orientated models, for example, relapse and time-arrangement examinations.

References 1. Liang, D., Liu, X., Shen, H.: Detecting spam reviewers by combing reviewer feature and relationship. In: International Conference on Informative and Cybernetics for Computational Social Systems (ICCSS) (2014) 2. Lin, Y., Zhu, T., Wu, H., Zhang, J., Wang, X., Zhou, A.: Towards online anti-opinion spam: spotting fake reviews from the review sequence. In: 2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) (2014) 3. Istiaq Ahsan, M.N., Nahian, T., Kafi, A.A., Hossain, M.I., Shah, F.M.: Review spam detection using active learning. In: IEEE 7th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON) (2016) 4. Rajamohana, S.P., Umamaheswari, K., Dharani, M., Vedackshya, R.: A survey on online review spam detection techniques. In: IEEE International Conference on Innovations in Green Energy and Healthcare Technologies (ICIGEHT’17) (2017) 5. Zhang, Q., Wu, J., Zhang, P., Long, G., Zhang, C.: Collective hyping detection system for identifying online spam activities. IEEE Intell. Syst. 32(5), September/October (2017) 6. Jia, S., Zhang, X., Wang, X., Liu, Y.: Fake reviews detection based on LDA. In: 4th IEEE International Conference on Information Management (2018) 7. Ramesh, G.P., Mohammed Irshad, S.: Hybrid renewable energy based CFSI for and motor application using ANFIS based MPPT and IFOC controller. In: IOP Conference Series: Materials Science and Engineering, vol. 925, pp. 1–10 (2020) 8. Mohammad Irshad, S., Ramesh, G.P.: Performance of CFSI with PI PID and fuzzy control strategies for hybrid power supply. J. Adv. Res. Dyn. Control Syst. 12 (2020) 9. Shehnepoor, S., Salehi, M., Farahbakhsh, R., Crespi, N.: NetSpam: a network-based spam detection framework for reviews in online social media. IEEE Trans. Inf. Forensics Secur. 12(7) (2017) 10. Sundaram, B.V.: Review of software architectural styles for artificial intelligence systems. Int. J. MC Sq. Sci. Res. 1(1), 96–112 (2009) 11. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 12. Swarnalatha, A., Manikandan, M.: Intravascular ultrasound image classification using wavelet energy features and random forest classifier. In: Intelligent Computing in Engineering, pp. 803– 810. Springer, Singapore (2020) 13. Magesh, P., Ramesh, G.P.: Fuzzy logic control implementation of ultra sparse matrix converter for renewable energy applications. In: 2017 International Conference on Information Communication and Embedded Systems (ICICES), pp. 1–3 (2017)

Chapter 8

Deep Learning Analysis for COVID 19 Using Neural Network Algorithms V. Vijaya Baskar, V. G. Sivakumar, S. P. Vimal, and M. Vadivel

Abstract The COVID-19 pandemic threatens to devastatingly impact the global population’s safety. A successful surveillance of contaminated patients is a crucial move in the battle against COVID-19 and radiological photographs via chest X-ray are one of the main screening strategies. Recent research showed that patients have abnormalities in photographs of chest X-ray that are characteristic of COVID-19 infects. This has inspired a set of deep learning artificial intelligence (AI) programs, and it has been seen that the precision of the identification of COVID-19 contaminated patients utilizing chest X-rays has been quite positive. However, these built AI schemes, to the extent of their author’s awareness, have become closed sources and not accessible for further learning and expansion by the scientific community, so they are not open to the general public. This thesis therefore implements COVIDNet, an Internet of Things (IoT) hand-accessible Machine Learning (ML) network mode to identify COVID-19 cases using the chest X-ray images. This investigation utilize the COVID cases database images from an open source that are accessible to the general public, employs Deep Neural Network (DNN) architecture for the detection and analyzing the disease using Machine Learning (ML) e-network based COVID-Net system. The COVID-Net data collection, which is referred to as COVIDx which includes 13,800 chest X-ray photos of 13,725 patients from 3 open-access data sources, one of which we launched, are also addressed. Keywords AI · Chest X-ray · Covid19 · Neural network · Screening V. Vijaya Baskar (B) School of EEE, Sathyabama Institute of Science and Technology, Chennai, Tamil Nadu, India V. G. Sivakumar Department of Electronics and Communication Engineering, Vidya Jyothi Institute of Technology, Hyderabad, Telangana, India S. P. Vimal Department of Electronics and Communication Engineering, Sri Ramakrishna Engineering College, Coimbatore 641022, India e-mail: [email protected] M. Vadivel Vidya Jyothi Institute of Technology, Hyderabad, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_8

103

104

V. Vijaya Baskar et al.

8.1 Introduction Data Science incorporates domain experience, programming capabilities and quantitative and computational understanding in order to achieve meaningful insights into the information process. In order to produce artificial intelligence systems, data scientists apply machine-learning algorithms to numbers, texts, images, videos and audio for usual human intelligence tasks. In essence, these programs produce concepts that can be converted into real market value by analysts and business consumers. Particularly for organizations with nearly limitless funding; it is difficult to intensify data science activities. In addition to making data scientists more efficient, the DataRobot autonomous machines’ learning network democratizes data analysis and IP, which allows researchers, industrial customers and other technological experts to become people and data science engineers. It automates repeated modeling activities which once took up much of the time and brain power of data scientists. DataRobot closes the divide between data scientists and the rest of the business and allows the learning of industry more available than ever. The outbreak continues to get a devastating effect on the welfare of the global population and well-being as a consequence of people’s contamination with the frequency of corona virus ARSCoV2 (SARSCoV2). Active surveillance of infectious patients is essential for mitigating COVID-19, allowing all affected to seek prompt diagnosis and care and to be separated to reduce the disease transmission. The primary screening technique to detect COVID-19 is to detect SARS-CoV-2 RNA in the reverse transcriptase-polymerase chain reaction (RTPCR) [1] from respiratory specimens. While the simplest gold standard is RT-PCR checking, the manual process is time-consuming, laborious and complex. The alternate mode of monitoring used by COVID-19 scans was the radiogram study, which conducts and analyzes visual markers of SARS-CoV-2 virus exposure via X-ray (e.g. chest X-ray) or computed tomography (CT) imaging by radiologists [2]. Early studies suggest that patients have chest X-ray defects typical of those afflicted with COVID-19 [3, 4] and that radiography evaluation can be used as a crucial tool for COVID-19 screening in epidemic areas [5]. The use of CXR images for COVID-19 screening in accordance with the global COVID-19 pandemic is particularly advantageous. CXR allows immediate triage of suspected COVID-19 patients and can be conducted in combination with viral testing (which takes time) in order to relieve vast numbers of individual patients that have been more affected by their potential (e.g. New York, Spain and Italy) or by viral testing (low supply) as an alternative means of relief [3, 6]. In regional areas where patients have to remain home before advanced signs arise CXR may also be very useful for research, as anomalies are frequently found When COVID-19 prone implementation forms at hospital sites at the time of diagnosis [7, 8]. The CXR is readily affordable and used in various hospital facilities and imagery centers, which several healthcare organizations deem basic devices. The presence of compact CXR-systems means that images can be carried in a single isolation space, thus significantly minimizing the possibility that the COVID-19 can be transmitted during transport to particular structures like a CT scanner and rooms

8 Deep Learning Analysis for COVID 19 Using Neural Network …

105

comprising specific frames [7]. Thus, radiology can be tested more easily and are more applicable in current health-care facilities despite the prevalence of chest Xray imaging devices, which renders them an ideal substitute for PCR research (in some instances much more sensitive [9]). However, one of the main bottlenecks confronting radiologists, as visual markers may be ambiguous, is the need to classify the radio graphical pictures. Of this purpose it is strongly desirable to have computerized diagnoses that will enable radiologists view radiographic images more quickly and reliably to diagnose COVID-19 events. Motivated by the need to view radiographic images more easily, a variety of deeplearning (DI) technologies [10–12] were introduced and the findings were very strong with respect to the precision in identifying COVID-19 patients through radiography [13–15]. However, in order to gain greater understanding and extension of these systems, the best knowledge of the author was that these built Ai technologies were secret sources and unavailable to the scientific community. In fact, such programs cannot be viewed and utilized by the public. As a consequence, recent attempts were made to press for open source and free software AI approaches for radio graphicallydriven COVID-19 case identification [6, 16]. The paper [15, 17–19] has made an excellent attempt to build a COVID-19 case dataset with annotated CXR to enable the researcher to locate a dataset comprising COVID-19 cases and SARS and MERS cases.

8.2 Implementation of COVID-19 Using Deep Learning Algorithms There we address the implement for programmer design planned COVID-Net, the subsequent configuration of the system, the COVIDx database construction process, and the specifics of deployment of COVID-Net. Architectural design: In this research, COVID-Net is generated using a human–machine collaborative design approach, where human instructions merge prototyping of the configuration of a network with machine-driven system experimentation in order to produce a network architecture designed to detect CXR images in the case of COVID-19 as shown in Fig. 8.1. The following are listed in each of the two design phases.

8.3 Network Design Prototyping A primary network design prototyping stage is used in the first step of a joint design approach for the human–machine project, which constructs an initial network design system focused upon concepts and best practices in human design. In this analysis,

106

V. Vijaya Baskar et al.

Conv 1

Pooling layer

Conv 2

Pooling layer

Feature Extraction Pooling layer

Normal Image

Fully-Connected Layer

Pooling layer

Conv 3

Conv 4

Output layer

Predicted Image

Fig. 8.1 Structure diagram for Covid-19 prediction using deep CNN algorithm

we have used the concepts of designing residual architecture, as demonstrated time and again in order to allow stable, high-performance trainable neural network architectures to be successfully built. The initial network design concept is built for (a) no infection (normal), (b) non-COVID19 and (c) COVID-19. Original designs are developed to generate one of the 3 predictions. The rationale behind these three possible predictions is to help clinicians decide better, as the treatment approach depends on the cause of infection, who has to be identified for PCR testing for COVID-19 Validation as COVID 19 and non-COVID19 requires different plans to treat.

8.4 Model Creation A machine-led research process is the second stage in the collective application approach for the human–machine interface used by the new COVID-Net. At this stage, more specifically, the initial network architecture prototype and details, along with the design criteria, serve as a guide to a design discovery strategy to learn and define the optimum micro-architecture designs with which a deep neural networkbuilt architecture will be developed. Such an engineering mechanism powered by the computer allows much greater granularity and greater versatility than achievable through manual software design with a human controlled handling while ensuring that the resultant deep neural network software satisfies operating criteria unique to domains. This is particularly important in the design of the COVID-Net, where the sensitivity of COVID-19 is needed for the number of cases lost to COVID-19 as much as possible.

8 Deep Learning Analysis for COVID 19 Using Neural Network …

107

8.5 Dataset Exploration A minimum of 13,800 CXR photos for 13,725 patients form the datasets for training and assessment of COVID-Net we recommend. In order to generate the COVIDx dataset, three separate publicly accessible datasets have been combined and revised. The most notable pattern being the small number of cases of COVID-19 infection and related CXR photos, which illustrates the lack of publicly available COVID19 case data, but also demonstrates the desire to collect further COVID-19 data as more case data becomes available. More precisely, from 121 COVID-19 medical cases, the COVIDx archive contains images. Much more hospital records and CXR images of no pneumonia and non-COVID19 pneumonia are available. There are 8,066 patients with no common pneumonia at all and 5,538 non-COVID19 patients with pneumonia.

8.6 Pre-processing The suggested COVID-Net were pre-trained on ImageNet [20] and instead practiced using a learning rate technique on the COVIDx dataset, which reduces the learning rate when for a certain amount of time the learning stagnates. The learning rate = 2e-5, epoch number = 22), batch volume = 8, factor = 0.7, patience = 5 for testing. The following hyper parameters were used for training. The data increased was also leveraged by encoding, flipping, horizontal flip and pressure changing the following forms of increments. Finally, we have implemented a technique to increase the distribution of increasing form of infection by lots. The first COVID-Net implementation was built and tested with a Tensorflow backend using the Keras deep learning library.

8.7 Results and Discussions They conduct quantitative and qualitative research in order to determine the efficacy of COVID-Net recommended, in order to achieve an increase in their identification efficiency and decision-making. We assessed the precision and vulnerability for each type of infection in the above COVIDx data set in the quantitative analysis of the proposed COVID-Net [18]. The measurement reliable and statistical sophistication (in comparison to the number of parameters) are worked properly. Those can be noted that by obtaining 92.6% test accuracy, COVID-Net achieves reasonable precision, thereby demonstrating the efficacy of using a joint human– machine modeling approach to speed up, tailor-made mission, data and organizational necessity development of deep-neural network architectures. Some of the sample images are shown in Fig. 8.2. It is especially relevant for situations like the diagnosis of illness, which constantly accumulate new cases and new data and

108

V. Vijaya Baskar et al.

a).Normal Image

b).Predicted Image

Fig. 8.2 Sample images in dataset directory

greatly value the opportunity to rapidly produce profound neural network architectures, which are adapted to the ever-changing knowledge base over time. Firstly, For COVID-19 cases (87.1%) that are published; COVID-Net will achieve strong sensitivity so we want to reduce the number of missing COVID-19 cases as much as possible. While promising, the number of COVID-19 cases available in COVIDx is limited in comparison with other types of infections, which improves efficacy with the additional COVID-19 patient cases. Furthermore, COVIDNet achieves high PPV in COVID-19 instances, suggesting relatively little incorrect COVID-19 positive identification. The high PPV is significant because the pressure on the health sector is compounded by many too many false positives, increasing the requirement for more PCR research and further care. Third, the frequency of common infections is significantly higher than that of COVID-19 events [19]. For regular and non-COVID19 instances, this finding can primarily be due to the significantly greater number of pictures. Thus, based on these findings, while COVID-Net is well supported as a whole in the identification of CXR images in COVID-19 instances, some areas in improvements will gain from the additional knowledge gathering and enhancement of the underlying teaching methodology to generalize them more broadly. The prediction results are showed in Fig. 8.3. In no way a production-ready approach is hoped that COVID-Net’s promising findings on COVIDx test datasets, together with its open source model and definition on the construction of an open-source dataset, would enable both researchers and people to speed up the development of higher-quality data systems. Future recommendations include constantly increasing susceptibility and PPV for COVID-19 infections, gathering new data, and applying the COVID-Net proposal for clinical diagnosis for safety study, estimation of threat status of a patient and estimation of hospitalization period to help in triage study, patient experience management and personalized care planning.

8 Deep Learning Analysis for COVID 19 Using Neural Network …

109

6 5 4 Training

3

validation 2 1 0 Normal

Pneumonia

Covid-19

Fig. 8.3 Analysis report for covid-19

8.8 Conclusion During this research, we presented COVID-Net, a profoundly convolutional, opensource. As such, we were able to check that COVID-Net did not use inappropriate decision-making information for the detection of COVID-19 cases (e.g. wrong visual markers, embedded body signals, objects of imagery, etc.) which may give rise, for the wrong reasons, to situations when the right decisions are made. We identified internet based COVIDx ML network, an open access data archive with CXR data collection for COVID-Net that composed of 13,800 CXR photos in 13,725 patient instances. In addition, we explored how COVID-Net using the DL based neural net that diagnose the COVID infection in the lungs using the Chest X-rays and makes forecasting using an explainable approach to obtain greater understanding of essential variables for COVID situations, which will help clinicians, enhance their screening and raise their trust and clarity by utilizing COVID-Net for rapid device assistance screening.

References 1. Wong, A., Shafiee, M.J., Chwyl, B., Li, F.: Ferminets: learning generative machines to generate efficient neural networks via generative synthesis (2018). arXiv preprint arXiv:1809.05989 2. Lin, Z.Q., Shafiee, M.J., Bochkarev, S., Jules, M.S., Wang, X.Y., Wong, A.: Explaining with impact: a machine-centric strategy to quantify the performance of explainability algorithms (2019). arXiv preprint arXiv:1910.07387 3. Huang, C., Wang, Y., Li, X., Ren, L., Zhao, J., Hu, Y., Zhang, L., Fan, G., Xu, J., Gu, X., Cheng, Z.: Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. The Lancet 395(10223), 497–506 (2020) 4. Ng, M.Y., Lee, E.Y., Yang, J., Yang, F., Li, X., Wang, H., Lui, M.M.S., Lo, C.S.Y., Leung, B., Khong, P.L., Hui, C.K.M.: Imaging profile of the COVID-19 infection: radiologic findings and literature review. Radiol. Cardiothorac. Imaging 2(1), e200034 (2020)

110

V. Vijaya Baskar et al.

5. Ai, T., Yang, Z., Hou, H., Zhan, C., Chen, C., Lv, W., Tao, Q., Sun, Z., Xia, L.: Correlation of chest CT and RT-PCR testing in coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases. Radiology 200642 (2020) 6. Lescure, F.X., Bouadma, L., Nguyen, D., Parisey, M., Wicky, P.H., Behillil, S., Gaymard, A., Bouscambert-Duchamp, M., Donati, F., Le Hingrat, Q., Enouf, V.: Clinical and virological data of the first cases of COVID-19 in Europe: a case series. Lancet Infect. Dis. 20(6), 697–706 (2020) 7. Rubin, G.D., Ryerson, C.J., Haramati, L.B., Sverzellati, N., Kanne, J.P., Raoof, S., Schluger, N.W., Volpi, A., Yim, J.J., Martin, I.B., Anderson, D.J.: The role of chest imaging in patient management during the Covid-19 pandemic: a multinational consensus statement from the Fleischer society. Chest (2020) 8. Sendhilkumar, N.C., Ramesh, G.P.: Analysis of digital FIR filter using RLS and FT-RLS. In: Advances in Intelligent Systems and Computing (2020) 9. Fang, Y., Zhang, H., Xie, J., Lin, M., Ying, L., Pang, P., Ji, W.: Sensitivity of chest CT for COVID-19: comparison to RT-PCR. Radiology 200432 (2020) 10. LeCun, Y., Bengio Y., Hinton, G.: Deep Learning. Nature 521(7553), 436–444 (2015) 11. Narayanan, K.L., Ramesh, G.P.: VLSI architecture for multi-band wavelet transform based image compression and image reconstruction. J. Eng. Appl. Sci. 12, 6281–6285 (2017) 12. Sujatha, S., Sindhu, M., Priyanka, M.: Cryptography based secured LIFI for patient privacy and emergency healthcare service. Int. J. MC Sq. Sci. Res. 9(1),86–97 (2017) 13. Gozes, O., Frid-Adar, M., Greenspan, H., Browning, P.D., Zhang, H., Ji, W., Bernheim, A., Siegel, E.: Rapid AI development cycle for the coronavirus (Covid-19) pandemic: initial results for automated detection and patient monitoring using deep learning CT image analysis (2020). arXiv preprint arXiv:2003.05037 14. Li, L., Qin, L., Xu, Z., Yin, Y., Wang, X., Kong, B., Bai, J., Lu, Y., Fang, Z., Song, Q., Cao, K.: Artificial intelligence distinguishes Covid-19 from community acquired pneumonia on chest CT. Radiology 200905 (2020) 15. Wang, W., Xu, Y., Gao, R., Lu, R., Han, K., Wu, G., Tan, W.: Detection of SARS-CoV-2 in different types of clinical specimens. JAMA (2020) 16. Cohen, J.P., Morrison, P., Dao, L.: COVID-19 image data collection (2020). arXiv preprint arXiv:2003.11597 17. Narayanan, K., Ramesh, G.: Discrete wavelet transform based image compression using frequency band suppression and throughput enhancement. 9(2), 176–182 (2017) 18. Xu, X., Jiang, X., Ma, C., Du, P., Li, X., Lv, S., Yu, L., Chen, Y., Su, J., Lang, G., Li, Y.: Deep learning system to screen coronavirus disease 2019 pneumonia (2020). arXiv preprint arXiv: 2002.09334 19. Pandey, A, Prakash, G.: Deduplication with attribute based encryption in e-health care systems. Int. J. MC Sq. Sci. Res. 11(4), 16–24 (2019) 20. Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: a large-scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 248–255. IEEE (2009)

Chapter 9

A Machine Learning Approach to Design and Develop a BEACON Device for Women’s Safety S. Srinivasan, P. Muthu Kannan, and R. Kumar

Abstract Online safety seems to have big concern for everyone today, particularly women. Latest WHO survey shows that 38% of women worldwide experience some form of harassment and physical violence. Victim numbers are growing slowly. We’re implementing a program here that ensures security for women. The tool is easy to carry, and can be taken anytime they feel the risk. The aim of the program is to provide women with a rapid response and safety monitoring tool. The framework helps women cope with anxiety, and therefore can call on their guardians to help. It automatically identifies a circumstance by detecting a woman’s body posture using the BEACON device. While she wears this band or a watch, when she experiences any kind of abuse or when she thinks something is in danger, she may click the button on the prototype, when she falls down or her movement postures changes from usual to unusual due to threats, and different information such as position, body posture, pulse rate is updated to a BEACON server transmitted by Bluetooth. Using the Bluetooth tool it tracks the nearby users GPS location which might also from the kidnapper and it sends the gps coordinates, We can get the exact location of the perpetrator so that the police can quickly locate the perpetrator, and the crash can be avoided quickly and save the family, prosecute the responsible. We will remotely track the women’s details through a map in order to use the IoT platform. This aims to minimise the slaughter and embarrassments of women.For machine learning, we used the regression analysis framework and used this training samples to inform our algorithm about different danger and non-danger situations, as well as how to act in them. Then, based on actual evidence, a prediction is made as to whether or not danger exists. S. Srinivasan (B) Institute of Biomedical Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Thandalam, Chennai, India P. Muthu Kannan Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Thandalam, Chennai, India R. Kumar P.B. College of Engineering, Chennai, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_9

111

112

S. Srinivasan et al.

Keywords Controller · Bluetooth low energy · Relay · Gyroscope · Accelerometer · Machine learning

9.1 Introduction IOT is the system of hardware devices, vehicles, and other items equipped with electronics, technology, sensors, actuators, and network connectivity that allow data collection and sharing of these objects. Via its integrated computing device, every item is uniquely recognizable, but can interoperate within the current Internet infrastructure. Experts forecast that by 2020, the IoT would contain around 30 billion objects. Owing to the integration of various innovations, including pervasive wireless networking, actual-time computing, machine learning, commodity sensors, and embedded devices, the concept of the Internet of Things has developed since 2016. That implies the conventional fields of embedded devices, WSN, control systems, automation (including home and building automation), and others all lead to allowing the Iot. IoT is a part of our work. In our study we have used this modern technology in such a way that it supports the women or a girl when in difficulty. It deals with both the machinery and the application. The prototype is a blend of both that makes it unique.

9.2 Study Work Hasmah Mansor suggested a body temp measurement system using Online Safety Monitoring [1]. In order to calculate temperature and heart rate, a wearable thermal sensor and a wireless sensor were used. It has created the outpatient devices [2] based on the directional sensor to track and diagnose the behaviour of the person with PD (Parkinson’s disease) in daily life and facilitate early treatment. The disease will be established in brief time. Observe the illness from the free setting and take care. Vijayalakshmi [3] suggested a women’s health enhancement scheme that use the GPS and GSM models. A compact device with a buzzer and microcontroller is fitted and can be mounted on a band or watch. A scheme using image metadata to identify the locations of individuals was defined in [4, 5]. Using background metadata, a device GPS mapper is used to identify the location of an individual using photographs and movies. Charranzhou offered a way to find the end of the trip [6] whether commuting or not commuting. The developer built a com-computer using technology and data-driven machine language to find the speed, range, direction of going. Dawei fan suggested a system for tracking, documenting and evaluating the psychological person, the behavioural features of a person and environmental change In Dawei fan suggested a system for

9 A Machine Learning Approach to Design and Develop …

113

monitoring, recording and evaluating the psychological person, the behavioural qualities of a person and environmental influences in indoor and outdoor acts. Locationbased A Groups [7–9] P. The solution is dispersed, lightweight and resistant to attacks by N. Mahalle, B. Anggorojati, N. R. Prasad, A [10–13].

9.3 Methodology and Constraints The block diagram of the suggested structure is shown in Fig. 9.1. And the device suggested consists of a bluetooth BLE, a relay, LCD display, controller MEMES accelerometer and an emergency switch for instantly sharing their co-ordinates location to the server.

9.4 Results and Discussions Figure 9.2 shows the android application for outdoor navigation tracking by using low-energy Bluetooth to locate the individuals. And in the main server the location and range are changed. The main server serves as a conduit between the centralized server and the client. Figure 9.3 shows the server which is used to view the location of the various beacon devices. It will be updated whenever the prototype receives network connection.The article have used Regression Analysis machine learning model for estimation. The machine learning algorithm will use data from an excel sheet that is connected directly to a PIC 32 bit MCU.

Power Supply

Emergency manual Track Switch ON/OFF

PIC CONTROLLER

Liquid Crystal Display

Fig. 9.1 Block diagram of the proposed system

ON/off Control Driver Circuit

IBEACON DEVICE

114

S. Srinivasan et al.

Fig. 9.2 Tracking using android application

Fig. 9.3 Server developed for monitoring through maps

9.5 Conclusion The support of the virtual setup to connect and alert the user through a server and even to their smartphones with coordinates, this project proposed a framework for women’s protection. Once the user presses the emergency switch, the sensors gathers information and then sent to the server via low energy BLE Bluetooth. In using the latest novel approach relative to GPS and GSM, this device would help speed up the monitoring for woman safety. It is too quick and more accurate in location wise

9 A Machine Learning Approach to Design and Develop …

115

compared with the existing method. Thus we conclude that ML based regression analysis is used for the safety system and analyzed the data and send back to the server which is predicted from the data.

References 1. Vijaylashmi, B., Renuka, S., Chennur, P., Patil, S.: Self defense system for women safety with location tracking and SMS alerting through GSM network. Int. J. Res. Eng. Technol. (IJRET) 4(5) (2015) 2. Nagamma, H.: IoT based smart security gadget for women’s safety. In: 2019 1st International Conference on Advances in Information Technology (ICAIT), pp. 348–352. IEEE (2019) 3. Gadhave, S.N., Kale, S.D., Shinde, S.N., Bhosale, A.C.: Electronic jacket for women safety. IRJET (2017) 4. Nguyen, H., Lebel, K., Bogard, S., Goubault, E., Boissy, P., Duval, C.: Using inertial sensors to automatically detect and segment activities of daily living in people with Parkinson’s disease. IEEE Trans. Neural Syst. Rehabil. Eng. 26(1), 197–204 (2017) 5. Kumar, H., Ramesh, G.P.: Energy efficiency and data packet security for wireless sensor networks using African Buffalo optimization. Int. J. Control Autom. 13(02), 944–954 (2020) 6. Sharma, S., Ayaz, F., Sharma, R., Jain, D.: IoT based women safety device using ARM7. IJESC 7(5), 11465–11466 (2017) 7. Hemanth Kumar, G., et al.: Reducing power feasting and extend network life time of IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 8. Jain, R.A., Patil, A., Nikam, P., More, S., Totewar, S.: Women’s safety using IOT. Int. Res. J. Eng. Technol. (IRJET) 4(5), 2336–2338 (2017) 9. Arias, O., Wurm, J., Hoang, K., Jin, Y.: Privacy and security in internet of things and wearable devices. IEEE Trans. Multi-Scale Comput. Syst. 1(2), 99–109 (2015) 10. Mahalle, P.N., Anggorojati, B., Prasad, N.R., Prasad, R.: Identity authentication and capability based access control (IACAC) for the internet of things. J. Cyber Secur. Mobil. 1(4), 309–348 (2013) 11. Shamsudheen, S., Mubarakali, A.: Smart agriculture using IoT. Int. J. MC Sq. Sci. Res. 11(4) (2019) 12. Zeleke, B., Demissie, M.: IOT based lawn cutter. Int. J. MC Sq. Sci. Res. 11(2), 13–21 (2019) 13. Badawi, W.A.: Underground pipeline water leakage monitoring based on IoT. Int. J. MC Sq. Sci. Res. 11(3), 01–08 (2019)

Chapter 10

Tea Plant Leaf Disease Identification Using Hybrid Filter and Support Vector Machine Classifier Technique S. Prabu, B. R. Tapas Bapu, S. Sridhar, and V. Nagaraju

Abstract Agriculture is the main occupation of our country. If a plant affected by any disease in long time, then there is a shortage of agriculture crops. It is therefore essential to diagnose and analyze the disease. Tea leaf cultivation is highly labor intensive and provides employment to about 2.0 million families engaged in tea cultivation, trade and trade across India. During cultivation, tea is most affected by the disease. The aim of this paper is to study and recognize various diseases in the tea plants and also procedure to recognize diseases at early infected stage using digital image processing and pattern recognition techniques. The system presented here is to arrange the leaf spot, rhizome rot, powdery mildew diseases and leaf blotch diseases which are infected in the Tea leaf plantation. The color transformed images are sharply segmented using Watershed transformation algorithm. Multiclass SVM classifier classifies the Tea leaf diseases using gradient feature values of the tea leaf images. The real-time IoT based Machine Learning (ML) techniques helps to classify the tea leaf as healthy or infected leaf from the pre-processed image captured using the mobile camera or video camera. In this paper the tea-leaf infection is detected using the hybrid filter that comprise of median filter and Gaussian filter for the purpose of edge detection, color enhancement, gray pixel resizing and noise reduction. Finally the performance evaluated in terms of accuracy and it is found that the presented system is realizable and provides better classification than earlier techniques. Keywords Image segmentation · Crop disease · Machine vision · GLCM · SVM

S. Prabu (B) Department of ECE, Mahendra Institute of Technology, Namakkal, India B. R. T. Bapu Department of ECE, S.A. Engineering College, Chennai, Tamil Nadu, India S. Sridhar Easwari Engineering College, Chennai, Tamil Nadu, India V. Nagaraju Department of CSE, Saveetha School of Engineering, SIMATS, Chennai, Tamil Nadu, India e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_10

117

118

S. Prabu et al.

10.1 Introduction Tea Plant illness mainly on leaves is the main crucial regeneration of depletion in the food crops. The quantity and quality of food creation is minimized only due to pest’s appearance in the crops and leaves. Therefore it comes to rise in struggling, food fragility and death rate. Nowadays, in order to recognize the plant disease, more number of image processing mechanisms is proposed. One of the presiding methods is to lessen and destroy the pests infecting the crop harvest. The usual pests like fungus, alphids, caterpillars, flies, snails, slugs etc. are generally considered in the plant disease. All the agriculturists are recognize pest’s systemically over assessment by physical verification but this approach consumes more time. The application of digital image processing in the agriculture field to perform analysis on various agricultural applications. The crop decease detected by SVM classifier and hybrid filter with GLCM approach. The texture feature id extracted using SVM classifier with GLCM. From the literature analysis, many researchers proposed methods which are concentrates only on crop disease classification and not providing preventive measures. So the system presented in this paper is provide preventive measures along with disease name. With the help of GLCM method along with first order statistical moments, the textures are extracted. In GLCM, pixel’s spatial relationship is extracted to obtain texture classification. From the original image, GLCM is obtained and the differences obtained from the first non-singleton dimension of the input texture image. GLCM matrix indicates the frequency of one gray level occur in a stated structural linear connection with second gray level within the monitoring portion. The contents of this GLCM matrix can be used for evaluation of the texture characteristics with monitoring the measure of inequality in intensity at the pixel of interest.

10.2 Literature Survey Recently much research work has done to detect the disease in leaf and the impact of diseases in leaf is major issue in agriculture domain. Some already developed systems in the problem area are explained below: 1. 2.

3.

Rastogi et al. [1] about disease detection in cucumber plant by using machine vision technology and artificial neural network (ANN). Pawar et al. [2] proposed an method to detect crop disease at earlier stage and this method is derived with help of digital image processing and ANN. This method is applied in cucumber crop disease to diagnose. Proposed algorithm gives 80.45% classification accuracy. Zhang et al. [3] This paper explained about feature selection of cotton disease leaves images by using fuzzy features (fuzzy curve and fuzzy surface) selection techniques without using non-linear techniques.

10 Tea Plant Leaf Disease Identification Using Hybrid Filter …

4.

5.

6.

7. 8. 9.

10.

119

Wu et al. [4] Presents leaf recognition algorithm for plant classification using probabilistic neural network (PNN). That classifies the 32 types of plants with help of leaf images. Meunkaewjinda et al. [5] study of this paper is to detect the grape leaf disease using back-propagation neural network (BPNN). BPNN effectively used for leaf color extraction with complex background. Wang et al. [6] paper explained about the study of applications of neural networks which is used in the image recognition and based on the texture and shape features. Weizheng et al. [7] paper presents the evaluation of leaf spot disease grading using image segmentation techniques. Fujita et al. [8–10] objective of this paper is to robust diagnostic of cucumber viral disease using convolution neural network. Phadikar et al. [11–13] explained about the rice disease identification by using pattern recognition techniques. Images are classified using SOM neural network. Zhang et al. [14–16] explained about the image feature extraction of tobacco leaf using machine vision technique. Image extraction technique used for grading.

10.3 Proposed Method Tea leaf plants are infected different types of diseases in the whole plantation without any forewarning of the diseases. The aim of this work is to detection and classification of leaf spot, leaf rot and powdery mildew diseases in the variety of Tea leaf plants (leaves) in an earlier stage using digital image processing techniques as shown in Fig. 10.1. This chapter describes about processes in the current work. Architecture of the proposed method is shown in Fig. 10.2.

10.4 Module Description 10.4.1 Proposed Method The primary aim of the presented system is observation of leaf disease using SVM and hybrid filter. For experimental analysis Tea leaves are used as many types of diseases.

120

S. Prabu et al.

Image Acquisition

1. Pick up a leaf from plants 2. Capture leaf image

Image Segmentation 1. LAB color space model 2. Watershed transformation

Image Preprocessing Hybrid Filter

Image classification Multiclass SVM

Feature Extraction Leaf feature extraction using cross sectional area

Fig. 10.1 Block diagram of the proposed system

Fig. 10.2 Architecture of the proposed system

10.5 Image Processing Block Diagram The modules in the current work are given below • Image collection

10 Tea Plant Leaf Disease Identification Using Hybrid Filter …

• • • • •

121

Image preprocessing Color transformation Image segmentation Feature extraction Classification.

10.5.1 Image Preprocessing Image pre-processing is the term for functioning on images at the subordinate level of abstraction. These functioning do not rise image data content but they reduces it if entropy is a data measure. The aim of pre-processing is an development of the image data that subdue unacceptable warp or raises some relevant image features for closer processing and analysis task. The enrichment includes purifying which detach the noise and process the image accurately. The purification is done by using Hybrid filter.

10.5.2 Hybrid Filter • Median filter. • Gaussian filter.

10.5.3 Median Filter The median filter is controlled by arranging all the pixel values from the window numerical order, and then restored by middle pixel value. Figures 10.3 and 10.4 Fig. 10.3 Original image

122

S. Prabu et al.

Fig. 10.4 Median filtered image

shows the original image, median filtered image. In order to execute median filtering, initial window is shifted and all the pixels surrounded by the windows are classified. After then median is calculated and this value is set to center pixel. If the integer of constituent in K*K window is odd, middle value is set as median value, else mean of two middle values is set as median value.

10.5.4 Gaussian Filter A Gaussian filter is a type of linear filter. Purpose of this filter is to blur the image or to reduce noise. The Gaussian filter alone will fade edges and decrease contrast. The simple way to lessen noise in an image is Non-linear filter. It’s declared to fame (over Gaussian for noise reduction) is that it extracts noise while keeping edges comparatively sharp.

10.6 Classification (a) (b) (c) (d) (e)

It incorporates K binary SVM classifiers, a single classifier for each class. Each Support vector machine is instructed to split one class from the remaining classes. A hyper plane is calculated for each class, considering this class as positive (+1) class and the other classes as negative (−1). Redo the process up to all classes are divided from the rest of classes as shown in Fig. 10.5. A specimen is checked for each classifier and is set to the class that corresponds to the Support vector machine have greatest output. The multiclass classification in one versus all method is shown in Fig. 10.6.

SVM is a supervised machine learning algorithm which can be used for categorization or regression drawbacks. It uses a approach is known as kernel trick to

10 Tea Plant Leaf Disease Identification Using Hybrid Filter …

123

Fig. 10.5 Extracted Tea leaf after watershed segmentation

Fig. 10.6 One versus all method

convert your information and then based on these conversions it notice an optimal margins between the feasible outputs. The merits of SVM and support vector regression cover that they can be used to remove the complexity of using linear duty in the high-dimensional characteristic expansion, and the optimization issue is converted into twin convex quadratic programs. If the given values are close to the powdery mildew class, then the classifier recognized that untrained Tea leaf belongs to powdery mildew diseases as shown in Fig. 10.7. Multiclass SVM classifier is used to identify whether the Tea leaf is affected or not and also to identify the type of diseases in Fig. 10.8 Table 10.1 shows Comparison

124

S. Prabu et al.

Fig. 10.7 Classifier output for powdery mildew diseases

Fig. 10.8 Classifier output for healthy leaves

Table 10.1 Comparison of results between watershed transformation algorithm and Color space model Watershed transformation algorithm No

Predicted

Healthy

1

8

1

Diseased

2

0

11

No

Predicted

Healthy

1

3

6

Diseased

2

0

11

Healthy True

Diseased

Color space model Healthy True

Diseased

10 Tea Plant Leaf Disease Identification Using Hybrid Filter …

125

Fig. 10.9 Experimental result for leaf rot disease

Fig. 10.10 Classifier result for leaf rot disease

of results between watershed transformation algorithm and Color space model and Fig. 10.9 shows Experimental result for leaf rot disease. If the given values are near to the leaf rot class, then the classifier recognized that inexperienced Tea leaf belongs to leaf rot diseases as shown in Fig. 10.10. The outcomes are calculated in terms of Sensitivity and Specificity and demonstrated that the existing technique with 82.35% accuracy and the proposed method can improve the Tea leaf disease detection with 95.85% accuracy.

10.7 Performance Evaluation In the evaluation metrics, Confusion matrix is calculated to produce outcome. Table 10.2 shows confusion matrix, where. • TP (True Positive) indicates the no. of healthy leaves are perfectly classified, • FN (False Negative) indicates the no. of infected leaves are misclassified as healthy leaves, • FP (False Positive) indicates the no. of healthy leaves misclassified as infected leaves,

126

S. Prabu et al.

Table 10.2 Confusion matrix Predicted True

Healthy

Diseased

1

Healthy

TP

FP

2

Diseased

FN

TN

• TN (True Negative) indicates the no. of infected leaves are perfectly classified.

10.7.1 Accuracy Calculation • • • • • • •

TP = sum (prediction == True AND ground truth == True) FP = sum (prediction == True AND ground truth == False) TN = sum (prediction == False AND ground truth == False) FN = sum (prediction == False AND ground truth == True) Sensitivity = (Tp./(Tp + Fn)). * 100; Specificity = (Tn./(Tn + Fp)). * 100; Accuracy = ((Tp + Tn)./(Tp + Tn + Fp + Fn)). * 100.

The confusion matrix for the existing algorithm (Color space model) and proposed algorithm (Watershed transformation) are given below. Total number of Tea leaf is 30. In that 20 for testing data and 10 for training data. For testing images, based on the True records and Predicted records the TP, TN, FP, FN values are calculated for all. According to the confusion matrix, a set of metrics commonly evaluated by using the evaluation metrics are Sensitivity and Specificity.

10.8 Result Analysis Based on the confusion matrix of existing and proposed work, the sensitivity and specificity and accuracy are calculated. Figure 10.11 shows comparison table among color space model and watershed transformation algorithm in terms of sensitivity, specificity and accuracy. Compared with color space model, watershed transformation algorithm provide better sensitivity, specificity and accuracy. The pooling layer output features are then finally applied into fully connected neural networks for reducing the error rate in the intermediate layers in Convolutional layers, as illustrated in Fig. 10.5. The output from this fully connected neural network is either stroke image or non-stroke image.

10 Tea Plant Leaf Disease Identification Using Hybrid Filter …

127

Fig. 10.11 Graphical results of sensitivity, specificity and accuracy

10.9 Conclusion The system presented here is to classify the leaf spot, rhizome rot, powdery mildew diseases and leaf blotch diseases which are infected in the Tea leaf plantation. With the help of Watershed transformation algorithm, the color transformed images are segmented sharply. After that, A channel is released from l*a*b color transformed images (RGB to l*a*b). The gradient characteristics value of Tea leaf images are obtained using HOG technique based on the shapes of the Tea leaf. The watershed segmentation and Multiclass SVM classifier embedded with the real-time IoT based ML techniques that classify the tea leaf as healthy or infected leaf from the image captured using the mobile or video camera. In this paper detecting the tea-leaf infection with mobile based application which compares the captured image from the stored database in online may be the future work using the image processing method. It is concluded that accuracy of watershed transformation technique is detect the tea leaf diseases at 95.85% accuracy rate.

References 1. Rastogi, A., Arora, R., Sharma, S.: Leaf disease detection and grading using computer vision technology and fuzzy logic. In: 2015 2nd International Conference on Signal Processing and Integrated Networks (SPIN), pp. 500–505. IEEE (2015) 2. Pawar, P., Turkar, V., Patil, P.: Algorithm for detecting crop disease early and exactly, this system is developed using image processing techniques and artificial neural network 3. Zhang, Y.-C., Mao, H.-P., Hu, B., Li, M.-X.: Features selection of cotton disease leaves image based on fuzzy feature selection techniques. In: 2007 International Conference on Wavelet Analysis and Pattern Recognition, vol. 1, pp. 124–129. IEEE (2007) 4. Wu, S.G., Bao, F.S., Xu, E.Y., Wang, Y.X., Chang, Y.F.: A leaf recognition algorithm for plant classification using probabilistic neural network. In: IEEE 7th International Symposium on

128

S. Prabu et al.

Signal Processing and Information Technology (2007) 5. Meunkaewjinda, A., Kumsawat, P., Attakitmongcol, K., Srikaew, A.: Grape leaf disease detection from color imagery system using hybrid intelligent system. In: Proceedings of ECTICON, pp. 513–516. IEEE (2008) 6. Wang, H.G., Li, G.L., Ma, Z.H., Li, X.L.: Application of neural networks to image recognition of plant diseases. In: International Conference on Systems and Informatics (2012) 7. Weizheng, S., Yachun, W., Zhanliang, C., Hongda, W.: Grading method of leaf spot disease based on image processing. In: Proceedings of 2008 International Conference on Computer Science and Software Engineering, vol. 06 (2008) 8. Fujita, E., Kawasaki, Y., Uga, H., Kagiwada, S., Lyatomi, H.: Basic investigation on a robust and practical plant diagnostic system 9. Pydipati, R., Burks, T.F., Lee, W.S.: Identification of citrus disease using color texture features and discriminant analysis. Comput. Electron. Agric. (Elsevier) 52(2), 49–59 (2009) 10. Magesh, P., Ramesh, G.P.: Fuzzy logic control implementation of ultra sparse matrix converter for renewable energy applications. In: 2017 International Conference on Information Communication and Embedded Systems (ICICES), pp. 1–3 (2017) 11. Ramesh, G.P., Irshad, S.M.: Hybrid renewable energy based CFSI for and motor application using ANFIS based MPPT and IFOC controller. In: IOP Conference Series: Materials Science and Engineering, vol. 925, pp. 1–10 (2020) 12. Phadikar, S., Sil, J.: Rice disease identification using pattern recognition techniques. In: Proceedings of 11th International Conference on Computer and Information Technology, pp. 25–27 (2008) 13. Satpathy R.B., Ramesh G.P.: Advance approach for effective EEG artefacts removal. In: Balas, V., Kumar, R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent Systems Reference Library, vol. 172. Springer, Cham (2020) 14. Zhang, X., Zhang, F.: Images features extraction of tobacco leaves. In: Congress on Image and Signal Processing. IEEE Computer Society (2008) 15. Ramesh, G.P., Prabhu, S.: FPGA implementation of 3D NOC using anti-hebbian for multicast routing algorithm. In: Sharma, D.K., Son, L.H., Sharma, R., Cengiz, K. (eds.) MicroElectronics and Telecommunication Engineering. Lecture Notes in Networks and Systems, vol. 179. Springer, Singapore (2021) 16. Pandey, A., Prakash, G.: Deduplication with attribute based encryption in e-health care systems. Int. J. MC Sq. Sci. Res. 11(4), 16–24 (2019)

Chapter 11

Machine Learning Based Efficient and Secured Car Parking System R. Santhana Krishnan, K. Lakshmi Narayanan, S. T. Bharathi, N. Deepa, S. Mathumitha Murali, M. Ashok Kumar, and C. R. T. Suria Prakash

Abstract As we people move towards sophisticated living, the usage of automobiles become the part and parcel of our life. That too the usage of cars during this decade has taken a steep stride. As a result, the traffic increases during the peak time and there is an insufficiency of parking lots near the jam-packed areas. This insufficient parking lots create a hectic problem in day-to-day life. The car drivers look for a parking place by roaming round the city which leads to unwanted fuel consumption and massive pollution. To solve this problem a mobile application is proposed which facilitates the people to book the parking lot in advance for their pre-planned journey. In order to prevent the theft vehicle being parked at the car parking, a two-way screening process is implemented to let the vehicles enter into the parking lots. This system also let the users to choose the parking slot based on their flexibility. This system also facilitates in monitoring the availability of the face mask with the people in order to prevent the spread of COVID-19 disease. Keywords Machine Learning · Mobile Application · Car Parking · Secured Parking · Arduino · Cloud Server · Wi-Fi · COVID-19 R. S. Krishnan (B) · M. A. Kumar Department of ECE, SCAD College of Engineering and Technology, Tirunelveli, Tamilnadu, India K. L. Narayanan Department of ECE, Francis Xavier Engineering College, Tirunelveli, Tamilnadu, India S. T. Bharathi PSNA College of Engineering and Technology, Dindigul, India e-mail: [email protected] N. Deepa RVS College of Engineering, Dindigul, Tamilnadu, India S. M. Murali Department of CSE, Francis Xavier Engineering College, Tirunelveli, Tamilnadu, India C. R. T. S. Prakash Department of Civil Engineering, SCAD College of Engineering and Technology, Cheranmahadevi, Tirunelveli, TN, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_11

129

130

R. S. Krishnan et al.

11.1 Introduction Since we all live in a sophisticated world, the car usage becomes an integral part of our life. The purchase of cars has increased to a great extent which is clearly explained in Fig. 11.1 [1]. India stands fifth in car manufacturing by the year 2019 [2]. Nearly 34 lakhs cars are sold in India during the financial year 2019 [3]. Figure 11.2a, b shows the top 10 states and its percentage of contribution in total car sales in India respectively during the 2019 financial year. Thus, increase in vehicle usage leads to insufficient parking lots during many occasions. This may lead to various issues like increase in road traffic, raise in air and noise pollution level, increase in fuel consumption level etc., Hence it is very clear that there should be an efficient mechanism in identifying the free parking lots before reaching a particular location. This will drastically reduce the abovementioned issues linked with insufficient parking lots. Another important issue to address is the rate of vehicle being theft during the last few decades. A statistic was released in November 2019 [4], which reveals the top 5 vehicle theft prone states in India. This is depicted in Fig. 11.3. In order to prevent the theft happening in the parking lot or to prevent the theft vehicle entry into the parking lot has to be restricted. Hence a novel system is required to support people with both parking assistance and preventing the vehicle theft. To facilitate these requirements, we propose “A mobile application based efficient and secured car parking system” which provides the following facilities. • Booking of free parking slot in advance using a mobile app based on our flexibility using cashless money transfer. • Screening the car details and its owner details before the car enters the carparkingarea. • Facilitates the additional payment option if user consumes more time at parking lot than the reserved time.

Fig. 11.1 Year wise total number of registered vehicles between 2009 and 2019

11 Machine Learning Based Efficient and Secured Car Parking System

131

a

% of Contribution in total car sales

b

5.1%

5.6%

4.0%

10.5% 8.9%

5.8% 6.5% 6.7%

8.4% 7.5%

Maharashtra

Uttar Pradesh

Gujarat

Kerala

Karnataka

Tamil Nadu

Delhi

Haryana

Rajasthan

Madhya Pradesh

Number of stolen vehicles

Fig. 11.2 a Top 10 car purchasing Indian states and b Top 10 states’ contribution in car purchase

Total Number of Stolen Vehicles 50000 40000 30000 20000 10000 0 Delhi

Uttar Pradesh

Maharashtra Rajasthan

Karnataka

Top 5 Indian States/Unition teritories Fig. 11.3 Top 5 Indian States/Union territories involving in vehicle theft activities

11.2 Related Works Chan Wei Hsu et al. [5], proposed an intelligent parking system which uses Bluetooth supported short range communication device which guides the drivers to park the vehicle in the allotted parking slot. This will prevent the drivers to park in the pathway

132

R. S. Krishnan et al.

or at another person’s slot. Using this system, advance booking of parking slots can also be done. Singh et al. [6], proposed a pre-paid parking system where a unique parking card is given to individual vehicle owners who can use the card to make payment in the parking area. This reduces the payment time and once the balance in the card it gets reduced the user can recharge that card and use it for future processing. Faris Alshehri et al. [7], proposed a system which monitors the parking status using various sensors like ultrasonic sensors and PIR sensors. It alerts the parking place in charge if the car is not parked at the correct position in the allotted slot. This system will be useful for maintaining fair distance between the nearby parked vehicles so that the damage occurring inside the parking place can be prevented. Robin Grodi et al. [8], proposed a smart parking system which identifies the free parking slots in every parking area with the help of ultrasonic sensors and those details are updated to database via local zig bee gateway. The user can view the updated parking lot status and use it for parking purpose without enquiring to anyone at the parking place. In order to reduce the traffic congestion a smart parking system was introduced [9] to book the parking slot for the user by getting their pre-planned visit details in advance and it also automates the payment process through mobile application. Mejri et al. [10], introduced a parking slot reservation scheme which assists the drivers to locate the free parking lot within 400 m of their visiting area. Aswathy James et al. [11], introduced a smart parking system which generates the QR code after booking the parking slot in advance with the help of mobile application. The user uses this QR code to enter into the parking place. Ramasamy et al. [12], introduced a time saving car parking system which uses ultrasonic sensors to detect the free slots available in the parking place and updates it to the cloud server via Wi-fi module. It also helps to locate the parking place via mobile application. Nushra et al. [13], developed an automatic car parking system where the robot parks the car in the parking lot without any human intervention [14]. Here a Line following robot is used which differentiate the commands passed to it using various light colors placed at its pathway. An intelligent car parking system was introduced by Olowolayemo et al. [15], where gathering the free parking slots at various malls and cashless transaction processes are implemented using mobile application. Hence it is very clear that there should be an efficient mechanism in identifying the free parking lots before reaching a particular location [16–18] as shown in (Fig. 11.4).

11.3 Proposed Work First the user checks for the parking slot availability and books the free slot by making the payment through Smart Drive Mobile Application. After locking their allotment, the QR code will be sent to the user through the mobile application. The parking slot booking procedure is explained in Fig. 11.5. After this process, the user reaches the parking place. Here the two-stage screening process takes place. First the RFID tag reader scans the unique RFID tag attached to the car [19, 20, 21]. Then it matches the obtained details with user information gathered while installing the application. If it

11 Machine Learning Based Efficient and Secured Car Parking System

133

Fig. 11.4 Proposed system architecture

gets matched, then it leads to second stage of screening. During this stage, the QR code received through the mobile application present in user mobile is cross checked to validate the parking time and user authentication. If both the stages are cleared the vehicle will be allowed to park in the allotted parking area. RFID tag-based screening is performed to check the booked vehicle details, QR code-based screening in the user’s mobile phone is performed to validate whether the proper user is using the car. Hence the user’s mobile becomes an important tool for this parking process to prevent the theft vehicle from entering into the parking area or the vehicle being robbed from the parked area. This enhances the security option in this intelligent parking system [22, 23]. Once the two-stage screening is finished the user can park the vehicle into the parking area. While entering into the parking area, the movement of the car is noted by PIR motion sensor and alerts Arduino to update the count by +1 to the database. The ultrasonic sensor detects the car being parked at the allotted slot [24, 25]. The camera attached at the parking slot will monitor the people with mask and alert them when the mask is not used by the people with the help of the buzzer. The flow diagram for car entry point procedure is depicted in Fig. 11.6. Various steps involved in parking the car is explained below

134

R. S. Krishnan et al.

Fig. 11.5 Parking lot booking procedure

Step1: User checks the parking lot availability through the mobile application and books it by making payment. Once the payment procedure is completed the QR code will be sent to the mobile application. Step 2: First stage of screening takes place. Here the car details like Owner name and Vehicles’ number plate are obtained and cross checked with the booked details. If the details are matched the car will be allowed for next stage of screening process. If the details are not matched the car will not be allowed to enter into the parking area. Step 3: Second stage of screening takes place. When booking the parking lot, QR code will be sent to the mobile application after confirming the payment. The user should open the application in his/her mobile phone and show the QR code to QR code reader. Here the parking slot reservation timings are checked and the car is allowed to enter the parking lot. If the user forgets his mobile phone or if user shows previously generated QR code the car will not be crossing this screening process. Step 4: If the car passes these two stages of screening, then the car will be allowed to park inside the parking area.

11 Machine Learning Based Efficient and Secured Car Parking System

135

Fig. 11.6 Flow Diagram for car entry point procedure

The flow diagram for car exit point procedure is depicted in Fig. 11.7. Various steps involved in car exit procedure is explained below Step 1: User reaches the exit gate along with his car. Step 2: Now the same QR code shown at the entry gate is again processed at the exit gate to check whether the car has exceeded the reserved parking time. Step 3: If the timing is not exceeded then the car is allowed to leave the parking area. If the time is exceeded, then the extra amount should be paid using the penalty payment option available in the mobile application. Upon the successful payment a new QR code will be generated and that should be shown for the verification process at the exit gate. Step 4: Upon successful verification, the exit gate will be opened for the user to leave the parking area as shown in (Fig. 11.8).

136

R. S. Krishnan et al.

Fig. 11.7 Flow diagram for car exit point procedure

The camera is attached at the parking lot will monitor the people through the camera. The face mask screening takes place in 2 phases. During the first phase, hundreds of images with people wearing mask and people without wearing mask are given as the input and it is trained using tensor flow [26, 27]. The second phase is known as the deployment phase which performs the operation of face mask detection using a machine learning classifier technique called CNN. The data set consist of 1486 images out of which 721 images are with mask and remaining images are without mask.

11.4 Result and Discussion Our mobile application based intelligent and secured parking lot booking system proved to be very effective and less time-consuming. Figure 11.9 depicts the home page of the smart drive application using which users can book the nearby parking lot of your visiting place by making the payment. Figure 11.10 explains the mobile application login page where user need to give the user name and password to access

11 Machine Learning Based Efficient and Secured Car Parking System

137

Fig. 11.8 Face mask screening procedure

the application. Figure 11.11 explains the various options supported by this mobile application. Figure 11.12a–c explain the parking lot booking process in detail. First the user must enter the place, date and time of visit in the application to get the various parking places available nearby the visiting area. Next the user must select the flexible parking area based upon the number of parking slots available at each parking area and distance of the parking area from the visiting place. Once these processes get completed, the parking slot confirmation details are displayed before making the payment. Once the parking lot is chosen, the payment can be done by using credit card, debit card or UPI options. Figure 11.13 explains the payment confirmation page. By clicking on get your QR code option, you will get the unique QR code for each payment which can be utilized to verify your payment at the parking area. Figure 11.14 explains about the penalty payment option. This is used when the car is being parked at the parking lot beyond the reserved hours. Figure 11.15 explains the confirmation SMS sent to the user who has booked the parking lot through Smart Drive Application. The details of the parking area and timing are sent to the user mobile number once the payment procedure is completed. Figure 11.16 explains the parking reservation timing expiration details. An SMS is sent to the user intimating that the reservation time is going to end in ten minutes.

138

R. S. Krishnan et al.

Fig. 11.9 Home page

This alerts the user to fasten their activities to prevent them from paying extra amount for the parking. Figure 11.17 explains the unauthorized entry SMS. If the user books the parking lot through his mobile a QR code will be sent to that mobile application. If the user fails to show that exact QR code from his mobile application or the user shows different QR code then the system comes to a conclusion that someone is trying to park the vehicle without the permission of the car owner. Hence it generates the warning SMS with the GPS location of the parking lot to the car owner’s mobile number. Figure 11.18 clearly explains that the number of cars is being parked at a single parking lot is found to be high when pre-booking is done through mobile application than manual parking system. When manual entry method was in existence the total car entry was recorded as 1281 and after implementing the mobile application the car entry was recorded as 1701. Hence on an average after implementing the mobile application for car parking the average number of cars parked per day is found to be 54.87 whereas in manual method it is recorded as 41.32. Hence this application will earn more profit to the parking lot management. This is achieved because the user is capable of viewing the free slots in advance and booking the slots for their cars in advance as shown in (Fig. 11.19). Before the implementation of smart drive mobile application, the average parking time of a vehicle for a month is estimated to be approximately 64 min whereas after implementing our system, the average parking time of a vehicle in a month at one

11 Machine Learning Based Efficient and Secured Car Parking System

139

Fig. 11.10 Login page

Fig. 11.11 App options

particular parking area is estimated to be 57 min approximately. This is achieved because of sending alert SMS 10 min prior to reservation time expiration to remind the user to fasten up their process to avoid penalty payment.

140

R. S. Krishnan et al.

Fig. 11.12 Parking lot booking process. a–c Parking lot payment link

Fig. 11.13 Payment successful screen

Figure 11.20a, b explains the CNN classifier output from Tensor flow. Here the first figure gives the accuracy level of the picture with mask and the second picture gives the accuracy level of the picture without mask.

11 Machine Learning Based Efficient and Secured Car Parking System

Fig. 11.14 Penalty payment screen

Fig. 11.15 SMS to confirm the reservation

Fig. 11.16 SMS to alert time expiry

141

142

Fig. 11.17 SMS regarding unauthorized entry

Fig. 11.18 Day wise car entry assessment

Fig. 11.19 Day wise average parking time assessment for a vehicle

R. S. Krishnan et al.

11 Machine Learning Based Efficient and Secured Car Parking System

143

Fig. 11.20 CNN classifier output for mask detection a, b

11.5 Conclusion A mobile application introduced here facilitates the people to book the parking lot in advance for their pre-planned journey. In order to prevent the theft vehicle being parked at the car parking, a two-way screening process is implemented to let the vehicles enter into the parking lots. This system also let the users to choose the parking slot based on their flexibility. The alerting of user through SMS regarding parking time expiration will also help the user to fasten up the work activity to escape from penalty payment. Analytical results also prove that our mobile application-based parking system is found to be more effective than manual parking scheme.

References 1. https://www.ceicdata.com/en/india/number-of-registered-motor-vehicles/registered-motorvehicles-total 2. https://www.ibef.org/industry/automobiles-presentation 3. https://www.team-bhp.com/forum/indian-car-scene/211449-state-wise-car-sales-data-april2018-march-2019-fy19.html 4. https://www.onelap.in/blog/the-5-most-vehicle-theft-prone-states-in-india-price-of-gps-tra cker-for-vehicle/ 5. Hsu, C.W., Shih, M.H., Huang, H.Y., Shiue, Y.C., Huang, S.C.: Verification of smart guiding system to search for parking space via DSRC communication. In: Proceedings of the 2012 12th International Conference on ITS Telecommunications, pp. 77–81, (2012). https://doi.org/ 10.1109/ITST.2012.6425287 6. Singh, A., Shawl, M.S., Bathla, S., Gaur, N., Mehra, A.: RFID and HDL Based pre-paid car parking system. In: Proceedings of the 2018 Second International Conference on Electronics,

144

7.

8.

9.

10.

11. 12.

13.

14.

15.

16. 17.

18. 19.

20.

21.

22.

23. 24.

R. S. Krishnan et al. Communication and Aerospace Technology (ICECA), pp. 1117–1120, (2018). https://doi.org/ 10.1109/ICECA.2018.8474606 Alshehri, F., Almawgani, A.H.M., Alqahtani, A., Alqahtani, A.: Smart parking system for monitoring cars and wrong parking. In: Proceedings of the 2019 2nd International Conference on Computer Applications & Information Security (ICCAIS), pp. 1–6, (2019). https://doi.org/ 10.1109/CAIS.2019.8769463 Grodi, R., Rawat, D.B., Rios-Gutierrez, F.: Smart parking: Parking occupancy monitoring and visualization system for smart cities. SoutheastCon 2016, 1–5 (2016). https://doi.org/10.1109/ SECON.2016.7506721 Kazi, S., Nuzhat, S., Nashrah, A., Rameeza, Q.: Smart parking system to reduce traffic congestion. International Conference on Smart City and Emerging Technology (ICSCET) 2018, 1–4 (2018). https://doi.org/10.1109/ICSCET.2018.8537367 Mejri, N., Ayari, M., Langar, R., Saidane, L.: Reservation-based multi-objective smart parking approach for smart cities. In: Proceedings of the 2016 IEEE International Smart Cities Conference (ISC2), pp. 1–6, (2016). https://doi.org/10.1109/ISC2.2016.7580840 James, A., Abraham, P.: IOP Conf. Ser.: Mater. Sci. Eng. 396, 012019 (2018) Ramasamy, M., Solanki, S.G., Natarajan, E., Keat, T.M.: IoT Based smart parking system for large parking lot. In: Proceedings of the 2018 IEEE 4th International Symposium in Robotics and Manufacturing Automation (ROMA), pp. 1–4, (2018). https://doi.org/10.1109/ROMA46 407.2018.8986731 M. H. Nushra, Q. A. Rahman, S. M. F. Mursalin, N. B. Asad, M. M. Asif Syeed and M. M. Islam, Smart car parking with the assistance of line following robot. In: Proceedings of the 2019 International Conference on Sustainable Technologies for Industry 4.0 (STI), pp. 1–6, (2019). https://doi.org/10.1109/STI47673.2019.9068046 Krishnan, R.S., Sangeetha, A., Kumar, A., Narayanan, K.L., Robinson, Y.H.: IoT Based smart rationing system. Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) 2021, 300–305 (2021). https://doi.org/10.1109/ICICV5 0876.2021.9388451 Olowolayemo, A., Alarood, A.A., Yap, S.N., Mantoro, T.: SPOT: A low-cost intelligent parking system for urban malls. Third International Conference on Informatics and Computing (ICIC) 2018, 1–5 (2018). https://doi.org/10.1109/IAC.2018.8780413 Hemanth, K., Ramesh, G.P.: Energy efficiency and data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control. Autom. 13(2), 944–95 (2020) Kumar, G.H., Ramesh, G.P.: Energy efficient multi-hop routing techniques for cluster head selection in wireless sensor networks, Springer, Computational Intelligence and Complexity, vol. 193, pp. 297–305 (2021) Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed. Tools Appl. 79(5), 4011–4026 (2020) Barriga, J.J., Sulca, J., León, J.L., Ulloa, A., Portero, D., Andrade, R., Yoo, S.G.: Smart parking: A literature review from the technological perspectived. Appl. Sci. 9(21), 4569 (2019). https:// doi.org/10.3390/app9214569 Yan, G., Yang, W., Rawat, D.B., Olariu, S.: SmartParking: A secure and intelligent parking system. IEEE Intell. Transp. Syst. Mag. 3(1), 18–30, Spring (2011). https://doi.org/10.1109/ MITS.2011.940473 SanthanaKrishnan, R., Julie, E.G., Robinson, Y.H., Raja, S., Kumar, R., Thong, P.H., Son, L.H.: Fuzzy logic based smart irrigation system using internet of things. J. Clean. Prod. 252(10), 119902 (2020) Ram, C.R.S., Ravimaran, S, Krishnan, R.S., Julie, E.G., Robinson, Y.H., Kumar, R., Son, L.H., Thong, P.H., QuangThanh, N., Ismail, M.: Internet of green things with autonomous wireless wheel robots against green houses and farms. Int. J. Distrib. Sens. Netw. 16(6), (2020) Narayanan, D.K.L.,Ramesh, G.P.: Robust and brittle secured video for IoT. Int. J. Eng. Technol. 7(2–20), 93–96, (2018) Robinson Y.H., Krishnan R.S., Raja S.: A comprehensive study for security mechanisms in healthcare information systems using internet of things. In: Balas, V., Solanki, V., Kumar, R.

11 Machine Learning Based Efficient and Secured Car Parking System

25.

26.

27.

28.

145

(eds.) Internet of Things and Big Data Applications. Intelligent Systems Reference Library, vol. 180. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-39119-5_15 Krishnan, R.S., Kannan, A., Manikandan, G., KB, S.S., Sankar, V.K., Narayanan, K.L.: Secured college bus management system using IoT for covid-19 pandemic situation. In: Proceedings of the 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), pp. 376–382, (2021). https://doi.org/10.1109/ICICV50876. 2021.9388378 Narayanan, K.L., Ram, C.R.S., Subramanian, M., Krishnan, R.S., Robinson, Y.H.: IoT Based smart accident detection & insurance claiming system. Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) 2021, 306–311 (2021). https://doi.org/10.1109/ICICV50876.2021.9388430 Thirupathieswaran, R., Suria Prakash, C.R.T., Krishnan, R.S., Narayanan, K.L., Kumar, M.A., Robinson, Y.H.: Zero queue maintenance system using smart medi care application for covid-19 pandemic situation. In: Proceedings of the 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), pp. 1068–1075, (2021). https://doi.org/10.1109/ICICV50876.2021.9388454 Narayanan, K.L., Ganesan, J., Kasinathaguru, E.: Intranet based multiple equipment controller with text panel. Ir. Interdiscip. J. Sci. Res. 4(3), 120–126, (2020)

Chapter 12

Machine Learning Approaches for Smart City Applications: Emergence, Challenges and Opportunities Sonam Mehta, Bharat Bhushan, and Raghvendra Kumar

Abstract Nowadays, smart cities aim to efficiently manage all sectors like growing urbanization, maintaining a green environment, energy consumption and life style of the people. The concept is to increase the capability of people to efficiently adapt and use all modern Information and Communication Technology (ICT) trends. The main effort is to increase the core infrastructure of the cities and give people an improved quality of life. The primary objective of this work is to give detailed background knowledge of Machine Learning (ML) algorithms and explores the role of ML, Deep Reinforcement Learning (DRL) and Artificial Intelligence (AI) in development of the smart city. The paper presents a comprehensive overview of smart city concept and focuses on different privacu solutions in the smart city. Further, the paper highlights the role of ML in various smart city applications such as intelligent transportation system, smart grids, healthcare, cyber security, and supply chain management. Finally, the work enumerates some future research directions to guide further advancements in the area. Keywords Smart city · Machine learning · Artificial intelligence · Healthcare · Intelligent transportation system · Smart grids · Security · Privacy

12.1 Introduction The world population is growing at very fast rate today and according to the reports, by 2050 it is expected to rise by almost 70% [1, 2]. With increase of urban population at a very fast rate, it is becoming relevant and pressing issues to solve the problems regarding efficiency and sustainability of the cities. These world wide concerns of S. Mehta · B. Bhushan (B) School of Engineering and Technology (SET), Sharda University, Greater Noida, India S. Mehta e-mail: [email protected] R. Kumar GIET University, Gunupur, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_12

147

148

S. Mehta et al.

Fig. 12.1 Conceptual framework of smart city

sustainability need actions from all over the world including individual country and every city. Apart from the sustainability concerns, there exist some other issues such as overcrowded people resulting in traffic jam, increase in pollution and noise [3]. Smart technologies like Internet of Things (IoT) combined with Machine Learning (ML) have the potential to reduce the pressure of urbanization by creating smarter and novel approaches for making day to day living of people more comfortable [4–6]. Smart cities refers to the role of Information and Communication Technology (ICT) to enhance or improve all intelligence in smart city areas where smart or intelligence means the ability to make better decisions [7]. A conceptual framework of smart city is depicted in Fig. 12.1. With the usage of various technology streams like Cyber Physical Systems (CPS), IoT and Wireless Sensor Network (WSN), enhancement of smart cities is considered as important [8]. All these technologies developed an environment of data which is available for systems and applications that can be developed or designed to get the different aim of a smart city. Different number of such systems and applications is imagined including smart buildings, smart grids [9], environment sensing, waste management, agriculture [10], smart lightening and health care [11, 12]. IoT is very important part of smart city applications that in result produce large amount of data [13]. With such huge, complex and big data, it makes tough to shortly tell most efficient and accurate actions. The great way to analyze such large amount of data can be done by using advance technology such as Artificial Intelligence (AI), DRL and ML to get a best decision [14]. The precision and the accuracy of the above mentioned methods can be improved. The concept of the smart cities, blockchain, and IoT in different applications are still in developing phase and will surely produce more opportunities in the nearby future [15].

12 Machine Learning Approaches for Smart City …

149

Different sectors such as cyber security, smart grids, intelligent transportation, UAV, 5G and B5G plays an important role in smart cities project. In smart city all the preceding sectors is highly influence by better use of AI, ML and big data analytic and DRL dependent techniques that can improve or increase their scalability and efficiency in a smart city [16]. For instance, it is adopted to ensure security of connected vehicles, realize self driving vehicles, and ensure safe travels and passengers hunt. Numerous surveys have been published in this regard [17–21]. However, none of the existing works can successfully explore the role of ML in various smart city applications. Unlike the previously mentioned reviews, this work brings forth a holistic approach for application of ML and IoT for smart city. The major contribution of this work is a detailed comprehensive discussion on the recent advances in smart grid, health care, transportation, cyber security, supply chain management for IoT system using machine learning. A summary contribution of this work is enumerated as below. • This work gives a detailed description of the background knowledge, various features and requirements of a smart city. • This work discusses various types of ML algorithms including reinforcement learning, dynamic programming, Bayesian methods and many more. • This work explores various privacy violations for smart cities and enumerates the major driving forces for the adoption of smart city concepts. • This works provides a comprehensive review on the recent advances in ML based solutions for varied sectors of smart cities including Intelligent Transportation System (ITS), smart grids, healthcare, cyber security, and supply chain management. • Finally, this work highlights the related research challenges and future research directions. This remainder of the paper is organized as follows. Section 12.2 presents the background knowledge of various ML algorithms. Section 12.3 presents the overview of smart city and highlights the associated privacy violations. Section 12.4 describes ML based solutions in smart city applications such as intelligent transportation system, smart grids, healthcare, cyber security, and supply chain management. Section 12.5 enumerates future research directions followed by the conclusion in Sect. 12.5.

12.2 Background Knowledge of Machine Learning There are three types of ML techniques namely supervised, unsupervised and reinforcement learning. Data set and its values train the network in supervised learning and search for a function called mapping (to map input data with output). Random forest, linear regression and vector machine are some of the famous examples of supervised learning. No support available in unsupervised learning. Only non classified and non labeled input dataset is used to train the AI network. Auto-encoder

150

S. Mehta et al.

and K-means are some of the examples of unsupervised learning models [22, 23]. Various types of learning strategies are discussed in the subsections below.

12.2.1 Reinforcement Learning (RL) The portion of the RL Algorithm that always does interactions and continues learning is called as an agent. The main motive of the reinforcement learning is to enhance the reward period on the basis of environmental interactions. Further, it aims to exploit already known actions and explore new actions at the same time, which may give good reward [24]. Types of Reinforcement learning are Positive and negative RL. When an event happens due to specific behavior and increase the frequency and strength of the behavior is positive reinforcement and negative reinforcement is when there is strengthening of the behavior as a negative scenario or condition is avoided or stopped. There are different practical applications of reinforcement learning. It can be used in data processing, robotics for different industrial automation and to create training system which provides material and instruction according to the requirement. There are two types of RL algorithm namely model free and model based. Function approximator is used by model based RL algorithms and is therefore known as sample efficient [25].

12.2.2 Markov Decision Process (MDP) MDP, a discrete-time stochastic method is derived from the mathematician named as Andrey Markov. Many RL problems are dependent on the MDP. To find the best solutions to one by one decision problems is an objective of an MDP. MDP is incapable of providing better solutions from a set of possible solutions. It is called by a set of actions, a set of states, a reward function and a transition model. Transition and rewards varies according to the chosen action, current state and result. A Markov Decision Process model contains a set of models, a set of possible actions (A), a set of possible states (S), and policy of MDP. The applications of MDPs are Harvesting, agriculture, water resources, inspection and purchase and production [26].

12.2.3 Dynamic Programming (DP) DP is a repeating method which breaks down a complex and complicated work in small and simpler problems. DP needs the full seen and observable knowledge of the environment because it follows model based approach. It is an optimization over plain recursion. It refers to breaking down of complicated problem into simpler sub problems so that their results can be reused. The use of these algorithms is basically

12 Machine Learning Approaches for Smart City …

151

for optimization. Dynamic problems can be used in top down approach as well as in bottom up manner. In stages the problem can be broken into a number of sub problems. A small portion of given problem is a stage. In states the sub problem for which decision has to be taken is indicated. State variables are the variables which are used for taking a decision at each and every stage. Stage decision is the optimal decision taken at each stage and the last optimal policy is the rule which finds the decision at each stage and if it is globally optimal the policy is called optimal policy. Some of the applications of dynamic programming include knapsack problem, longest common subsequence, mathematical optimization problem, time sharing, and reliability design problem.

12.2.4 Deep Q Network (DQN) A neural network is use to find the approximate value of the Q-value function [27]. Q-learning is the most widely adoption TD algorithms as the value is kept in the table or look-up matrix. For example, we store the Q table in a two dimensional array in a Q learning. Visiting and estimating values for all different states for environments becomes problematic with huge actions and associated spaces. It is possible to overcome the issue of generalization using function approximation. To estimate the value function in large state space, DQN uses a Neural Network [28]. Q-learning update rule is used to train the network.

12.2.5 Monte Carlo (MC) It is a broad or large class of computational algorithms that depends on repeated random sampling just to obtain or get numerical values. Generally, it is used in mathematical and physical problem especially when it is difficult to use other approaches. It is mainly used in generating results or drawing inference from a probability distribution; optimization and numerical integration. The main advantage of MC over DP is that MC algorithms are easy and efficient to implement. It can be used with sample models as it learns optimal solutions through direct implementation. It uses random numbers to implement a Monte Carlo method, so it is important to have a source of random numbers. This method is used in development of statistical methods and to compare them. Monte Carlo methods follow some steps. Initially, it finds the statistical properties that help to perform a deterministic calculation and then finally the results are analyzed statistically [29].

152

S. Mehta et al.

12.2.6 Temporal Difference (TD) Methods Waiting for the end of an episode is not needed in TD methods as evident in MC technique. TD helps in prediction or estimation of a quantity which completely depends on the future value of the signal. For the evaluation of the policy it is one of the most widely used methods. It also has received attention in the area of neuroscience. Q learning and SARSA are two TD based algorithms. TD is related to the temporal difference model of the animal learning. The simplest form of the TD learning is TD (0) [30]. Some of the advantages of the TD methods are enumerated as follows. Firstly, It can learn in every step (offline or online). Second, It can learn from the incomplete sequence. Third, it has a lower variance compared to MC and is more efficient than MC as it can also work in non terminating environments. It exploits Markov Property that’s why it is most effective in Markov environments. However, there are certain limitations associated with TD. The major limitation is that the TD method is an estimation scheme which is biased and more sensitive to the initial value.

12.2.7 Bayesian Methods In recent year the Bayesian method has become increasingly popular to analyze geostatistical data. For specifying sophisticated hierarchical models for recent computational advances and complex data, Bayesian method provides a coherent approach. The disadvantage of using Bayesian method is that all joint distributions of parameters and processes have to be specified. Agent gains different rewards and with passage of time it is certain to enhance the reward. To examine and evaluate this, the Bayesian models created an analytical architecture at a sufficient computational cost. Due to power to capture uncertainty in learned symbols Bayesian methods may lead to exploration–exploitation confusion. Myopic and Thompson are the some famous methods which can be used for Bayesian approximations. To solve the problems of exploration–exploitation Thompson sampling can be used [31]. Table 12.1 summarizes the various types so machine learning models discussed in this section.

12.3 Smart City Overview Smart cities are known as the connection of ICT, physical, social and business infrastructure. It aims to enhance and improve all intelligence in a city where intelligence means the ability or technique to make decisions better. Smart cities aim is to manage growing urbanization, maintain and manage a green environment, and adapt the new and modern ICT. In smart cities ICT plays an important role in decision making

12 Machine Learning Approaches for Smart City …

153

Table 12.1 Summary of various ML models ML models

Highlights

Reinforcement learning [24, 25] • Aims to improve and enhance its long period aggregated reward • Aims to exploit already known actions • Used in data processing, robotics for different industrial automation Markov dcision process [26]

• Discrete-time stochastic process in mathematics • Offers better solution • Transition and rewards varies

Dynamic programming

• Repeating method that divides down a complex work into small problems • Follows model based approach • Used in top down approach as well as in bottom up manner

Deep Q network [27, 28]

• Overcome the issue of generalization • Q-learning update rule is used to train the network

Monte Carlo [29]

• • • •

Temporal difference [30]

• Can learn in every step (offline or online) • Lower variance compared to MC

Bayesian methods [31]

• Provides a coherent approach • Exploration- exploitation is balanced using Thompson sampling

Depends on repeated random sampling Can be used with sample models Easy and efficient to implement Uses random numbers to implement a Monte Carlo method

and policy designing. Various dimensions related to smart cities are discussed in the subsections below.

12.3.1 Introduction of Smart City Metropolitan population is expanding at an exceptionally quick rate, because of which taking care of issues in regard to maintainability and proficiency of urban areas is turning out to be extremely important. Along with the manageability development, there are additional issues of overpopulation in urban areas which make gridlock, expansion in contamination and commotion. Urban areas are viewed as the association of data, physical and correspondence innovation, business and social foundation to improve the general insight in a city, where knowledge implies the capacity to settle on better choices equitably [32]. Advancement of brilliant urban communities is acknowledged using wide range of innovation streams, like CPS and WSN [33, 34]. These advances make a biological system of information which can be utilized to accomplish the different objectives of a Smart City. Quite a few such frameworks and application can be envisioned, including keen structures, medical services and

154

S. Mehta et al.

savvy lightning, brilliant matrix, climate detecting and waste administration [35]. It is exceptionally important to distinguish competitor settings for applying this innovation—advanced arrangements, it is likewise foremost to moderate and perceive any dangers made during or in the wake of executing above arrangements [36]. It is imperative to comprehend the protection, prior to exploring the security suggestions in the urban communities. Security can be characterized as a person’s entitlement to not be upset or noticed, with the presentation of progressing and new advances, this perspectives gets deficient [37]. For protection in a smart city, there is a need of more strong definitions for adversaries that can cause malicious activities. Different parts of protection can be uncovered relying upon the ramifications of new advancements. For instance, a person’s actual location can be uncovered with the assistance of worldwide situating frameworks like Global Positioning System (GPS). Seven stage privacy protection schemes are proposed. It include protection of propensities and conduct, security of room, protection of emotions and contemplations, security of individual, protection of picture and information, security of affiliation, protection of correspondence and security of room and area. Ongoing and social security incorporates distinctive individual premium like legislative issues and religion. Correspondence security alludes to telephonic calls, electronic messages or mail. Picture and information security alludes to the need of the individual to have full oversight over any of their own data which can be gathered [38]. The general smart city architecture is shown in Fig. 12.2.

Fig. 12.2 Smart city architecture

12 Machine Learning Approaches for Smart City …

155

12.3.2 Privacy Violations in Smart City There are two different ways for gathered data to be uncovered. First, data can be uncovered through attacks which bring unapproved admittance to private information. Second, data can be distributed by wish as in open information activities. Anyway in both the case, security won’t be essentially disregarded if the data will be taken care of to satisfactorily protect protection. Six measurements for security infringement are identified. These are the estimation scale, the precision of the revelation, the kind of information, the uncovering element, divulgence, and the openness of the divulgence. Various security concerns include blunders in information, unapproved optional utilization of information by an outside association, helpless choices dependent on accessible information, the assortment and capacity of enormous measure of individual information and consolidating individual information from frantic sources. Smart city can undoubtedly prompt protection infringement, without thinking about the hurtful impacts of various exercises including data preparing, data assortment, intrusiveness, data spread. Between every typology, outline and scientific categorization, there are similarities that help to recognize basic subjects in security infringement’s insight. Understanding the constituents and sorts of a security infringement is a significant advancement in contriving an arrangement to ensure data protection [39, 40].

12.3.3 Driving Forces for Smart Cities Prior to social event what components make a city “smart”, it is vital to understand why urban areas are keen on contributing their capital for carrying out new advancements for task improvement. The major driving force for smart city development is the rising urban thickness. By 2030, the number of inhabitants on the planet’s metropolitan territories is projected to consistently increase. As a rate, the projection shows that one out of three individuals with a populace of 500,000 or more noteworthy will live in a metropolitan region. As a measurement for characterizing “huge urban areas”, this populace measurement of 500,000 occupants is likewise utilized. These huge urban areas are contended to have more significant levels of contamination, more gridlock and energize social osmosis. Three contributing variables of fruitful urban communities are effective administration, the capacity to hold gifted inhabitants and the capacity to adapt to change [41, 42]. Furthermore, there are number of restricted and worldwide associations that make targets and norms to encourage cleaner urban communities, equity, uniformity, feasible turn of events and other positive results. Information innovations are fundamental for the objective of these associations. Objective of these association adjust well to the objective of these keen urban communities.

156

S. Mehta et al.

12.4 ML Based Solutions in Smart City ML and AI algorithms have progressively become an important part of different industries. A city is known or recognized as smart city when some kind of ML or IoT is applied. With the help of ML and AI with IoT, a city can operate in a better way. Various ML based smart city solutions are discussed in the subsections below.

12.4.1 Intelligent Transportation System (ITS) It is an application of control systems, ICT and advanced sensors that produce large data and has great impact on the future of ITS and the idea of smart cities [43]. The ML, DRL and AI techniques plays an important role to estimate the real time data flow in the cities or urban areas which is important for ITS [44, 45]. Surya et al. [46] presented a deep study that highlights the role of DRL and ML to different issues such as fleet management, assessing traffic flow, predicting the possibilities of accidents, and passengers hunt. In another work, Shen et al. [47] developed a method in ITS that focuses on issues (e.g. fleet management, cyber-physical security) which plays an important role in the enhancement of smart cities. Bouchelaghem et al. [48] highlighted the driving behavior of human decision-making which is dependent on DRL. Data is converted to hybrid matrix using data converter and a DRL approach is used to attain an optimal policy to extract the important latent features.

12.4.2 Smart Grids (SGs) Voluminous data plays a significant part in changing the operational design of smart grids and in productive use of energy in urban communities. The SG’s depend on IoT gadgets, present day data, correspondence frameworks and voluminous information [49, 50]. In SGs, the various data comes from heterogeneous sources which can be processed and analyzed in an effective way and used for proper management decisions. Large amount of data has the power to increase the power grid performance, the decision making of sharing of power and the safety of power grids. The recent trends shows that SGs are making large use of smart big data for various types of applications such as baseline estimation, load clustering, load assessment and prediction, malicious data deception attacks and demand response [51–53]. Ghorbanian et al. [54] explored a variety of applications assisted by big data in SGs. Hossain et al. [55] comprehensively surveyed the function of DRL techniques and ML in smart grids known applications. Furthermore, their functions in cyber security of smart grids are discussed in detailed manner. Mallikarjuna et al. [56] reviewed different areas or applications of DRL techniques related to transient stability, load forecasting, fault

12 Machine Learning Approaches for Smart City …

157

analysis and power grid control. Morello et al. [57] developed a model that considered ML based techniques and shared energy resources as a joined portion of the SGs system which helps in deciding and finalizing the complex logical decisions based on given data. Mujeeb et al. [58] proposed model known as Deep Long Term Memory (DLSTM) to estimate the demand and price for electricity for a week and a day ahead. Authors also tested it with the help of real electricity market data. In another work, Pallonetto et al. [59] aimed to analyze the effect of demand price policies using a simulation prototype under various time related electricity cost. The ML-based, dual demand price protocols were employed to regulate thermal storage and heat pump.

12.4.3 Health Care With the coming of elite IoT gadgets, progressed sensors, an expansion in information rates and distributed computing have prompted broad utilization of ML, DRL and AI procedures in cutting edge medical care systems which is known as wellbeing intelligence. Ahad et al. [60] created an audit to consider the part of 5G correspondences in the medical care frameworks. The work highlights the necessary equipment, strategies, and engineering goals. In another work, Ngiam et al. [61] examined information utilizing DRL, AI, and ML in medical care frameworks. The work highlights various strategies related to the characterization, infections hazard, complex information examination, patient endurance expectations and best treatment. Anyway the utilization of the above procedures presents such countless difficulties such as clinical issues, exact model preparing, information under examination, and care for characterized contemplations. Sharma et al. [62] highlighted the role of AI in future medical care frameworks in another work, Mak et al. [63] proposed that utilizing AI for drug revelation purposes will reshape the medication innovative work procedures of the current drug industry. Venkatesh et al. [64] examined different potential DRL, ML and AI conventions that can improve the IoT based medical care framework. Tuli et al. [65] proposed a design known as Health Fog that is capable of independently and proficiently examining the coronary illness. Ali et al. [66] proposed edge processing based model to definitely oversee and order the approaching patient’s information. Authors proposed smart health care framework named healthgard. Bruzelius et al. [67] proposed to utilize DRL and ML procedures to map far away networks and satellite symbolism for better help and medical services. Similarly, Cook et al. [68] proposed a ML based strategy to evaluate patient’s odds of endurance after PCI which represents percutaneous coronary mediation.

158

S. Mehta et al.

12.4.4 Cyber Security AI has become now a significant innovation for network protection. Urban communities are defenseless against digital assaults. Huang et al. [69] proposed IoT based secrecy-upgrading scheme that uses exponential number of antennas. By using two different technologies like antenna selection and artificial noise the communication system is improved. Wang et al. [70] proposed data analytics based approach for mitigation of cyber attacks in smart grids. In another work, Li et al. [71] presented a mobile-cloud framework system which is a functioning way to kill the data overassortment. By placing all clients’ information into a cloud, the security of clients’ information can be incredibly improved. Duan et al. [72] proposed an ungraded digital secure energy and information exchange system for the optimal management and operation of the smart city. An improved Directed acyclic graph (DAG) method in this proposed model is presented to improve or upgrade the security of transaction of data inside the smart city. Falco et al. [73] proposed an automated attack generation method that can deliver point to point or detailed, consistent and scalable foundation for cyber security. IoT innovation is defenseless as it is feasible to fix any uncovered territories and launch wide range of malicious activities. Some of the major attacks that can harm brilliant city foundation are enumerated as follows. • Hijacking gadgets: One of the most alarming parts of digital wrongdoing is Device Hijacking. Aggressors can assume responsibility for a gadget utilizing security weaknesses, and use it to disturb an interaction. • Distributed Denial of Service (DDoS) attack. • Asset, information, and wholesale fraud: Data burglary is generally well-known digital wrongdoing. Programmers can penetrate information banks and the city foundation is helpless against this. • Physical interruption: As numerous frameworks rely upon complex criticism and cycles from organizations of sensors, actual harm to any segment can lead to a chain of issues.

12.4.5 Supply Chain Management (SCM) SCM is the treatment of the whole creation stream, beginning from the crude item to conveying the end result to the customer. There are five segments of customary SCM frameworks namely planning, sourcing, manufacturing, delivery and logistics and returning. SCM is significant as it limit cost, time and waste in the creation cycle. SCM makes various advantages that convert to more noteworthy upper hand, better brand picture and higher benefits. These incorporate lower overhead, improve in quality, enhance coordination, and satisfy client need. The clearest face of the business for purchasers and clients are the inventory network. The better an organization’s SCM are, the better it ensures it’s drawn out manageability and business notoriety. Inventory network should use existing and new advancements to adjust

12 Machine Learning Approaches for Smart City …

159

their coordination’s to smart city plans by improving wellbeing, decreasing clog and ensuring the climate. Smart city activities can change the manners in which the business is coordinated. The appropriation of smart city idea presents both imperative and freedoms to SCM.

12.5 Conclusion and Future Research Directions We checked on ongoing smart urban areas research advancement and patterns in regards to various complex applications and issues achieved by industry. A concise investigation of the basics idea of ML, DRL and AI strategies has been created. We investigated the part of the previously mentioned conventions to configuration close to ideal methodologies in regards to different applications that are viewed as essential to brilliant city productivity. We likewise introduced the latest ML, AI and DRL applications in planning savvy administration and need for AI viable and AI helped new guidelines, SGs, energy-effective ITS. We momentarily introduced the job of the referenced methods in shrewd medical care from proficient analysis, the security of wellbeing-arranged IOT gadgets and wellbeing recuperation and the revelations of the most helpful medication. For smart city applications, more information is gathered to ensure that individual protection is the only priority. There are a rundown of innovations and standards for keeping up the security yet it isn’t comprehensive. This work referenced that it isn’t just the obligation of the application planner to keep up the protection yet it is the duty of the client and the director also. In view of key measurements, to gauge the exhibition of security upgrading innovation should remember for the future work without a doubt. Furthermore, carrying out security improving innovations, like memory prerequisites, time to measure and the handling power is the need of the hour.

References 1. O’Dwyer, E., Pan, I., Acha, S., Shah, N.: Smart energy systems for sustainable smart cities: Current developments, trends and future directions. Appl. Energy 237, 581–597 (2019). https:// doi.org/10.1016/j.apenergy.2019.01.024 2. Liu, Y., Yang, C., Jiang, L., Xie, S., Zhang, Y.: Intelligent edge computing for IoT-based energy management in smart cities. IEEE Network 33(2), 111–117 (2019). https://doi.org/10.1109/ mnet.2019.1800254 3. An, J., Le Gall, F., Kim, J., Yun, J., Hwang, J., Bauer, M., Song, J.: Toward global IoTenabled smart cities interworking using adaptive semantic adapter. IEEE Internet Things J. 6(3), 5753–5765 (2019). https://doi.org/10.1109/jiot.2019.2905275 4. Yau, K.A., Peng, S., Qadir, J., Low, Y., Ling, M.H.: Towards smart port infrastructures: Enhancing port activities using information and communications technology. IEEE Access 8, 83387–83404 (2020). https://doi.org/10.1109/access.2020.2990961

160

S. Mehta et al.

5. Goyal, S., Sharma, N., Bhushan, B., Shankar, A., Sagayam, M.: Iot enabled technology in secured healthcare: Applications, challenges and future directions. In: Cognitive Internet of Medical Things for Smart Healthcare, pp. 25–48 (2020). https://doi.org/10.1007/978-3-03055833-8_2 6. Nie, X., Fan, T., Wang, B., Li, Z., Shankar, A., Manickam, A.: Big data analytics and IoT in operation safety management in under water management. Comput. Commun. 154, 188–196 (2020). https://doi.org/10.1016/j.comcom.2020.02.052 7. Alwajeeh, T., Combeau, P., Aveneau, L.: An efficient ray-tracing based model dedicated to wireless sensor network simulators for smart cities environments. IEEE Access 8, 206528– 206547 (2020). https://doi.org/10.1109/access.2020.3037135 8. Masera, M., Bompard, E.F., Profumo, F., Hadjsaid, N.: Smart (electricity) grids for smart cities: Assessing roles and societal impacts. Proc. IEEE 106(4), 613–625 (2018). https://doi.org/10. 1109/jproc.2018.2812212 9. Haque, A.B., Shurid, S., Juha, A.T., Sadique, M.S., Asaduzzaman, A.S.: A novel design of gesture and voice controlled solar-powered smart wheel chair with obstacle detection. In: Proceedings of the 2020 IEEE International Conference on Informatics, IoT, and Enabling Technologies (ICIoT) (2020). https://doi.org/10.1109/iciot48696.2020.9089652 10. Pranto, T.H., Noman, A.A., Mahmud, A., Haque, A.B.: Blockchain and smart contract for IoT enabled smart agriculture. PeerJ Comput. Sci. 7, (2021). https://doi.org/10.7717/peerj-cs.407 11. Indumathi, J., Shankar, A., Ghalib, M.R., Gitanjali, J., Hua, Q., Wen, Z., Qi, X.: Block chain based Internet of medical things for uninterrupted, ubiquitous, user-friendly, unflappable, unblemished, unlimited health care services (BC Iomt U6 HCS). IEEE Access 8, 216856–216872 (2020). https://doi.org/10.1109/access.2020.3040240 12. Goyal, S., Sharma, N., Bhushan, B., Shankar, A., Sagayam, M.: IoT Enabled technology in secured healthcare: Applications, challenges and future directions. In: Cognitive Internet of Medical Things for Smart Healthcare Studies in Systems, Decision and Control, pp. 25–48 (2020). https://doi.org/10.1007/978-3-030-55833-8_2 13. Magaia, N., Fonseca, R., Muhammad, K., Segundo, A.H., Neto, A.V.L., De Albuquerque, V.H.: Industrial internet-of-things security enhanced with deep learning approaches for smart cities. IEEE Internet Things J. 8(8), 6393–6405 (2021). https://doi.org/10.1109/jiot.2020.3042174 14. Mohammadi, M., Al-Fuqaha, A., Guizani, M., Oh, J.: Semisupervised deep reinforcement learning in support of IoT and smart city services. IEEE Internet Things J. 5(2), 624–635 (2018). https://doi.org/10.1109/jiot.2017.2712560 15. Bhushan, B., Khamparia, A., Sagayam, K.M., Sharma, S.K., Ahad, M.A., Debnath, N.C.: Blockchain for smart cities: A review of architectures, integration trends and future research directions. Sustain. Cities Soc. 61, 102360 (2020). https://doi.org/10.1016/j.scs.2020.102360 16. Ramchurn, S.D., Vytelingum, P., Rogers, A., Jennings, N.R.: Putting the ‘smarts’ into the smart grid. Commun. ACM 55(4), 86–97 (2012). https://doi.org/10.1145/2133806.2133825 17. Ullah, Z., Al-Turjman, F., Mostarda, L., Gagliardi, R.: Applications of artificial intelligence and machine learning in smart cities. Comput. Commun. 154, 313–323 (2020). https://doi.org/ 10.1016/j.comcom.2020.02.069 18. Saharan, S., Kumar, N., Bawa, S.: An efficient smart parking pricing system for smart city environment: A machine-learning based approach. Futur. Gener. Comput. Syst. 106, 622–640 (2020). https://doi.org/10.1016/j.future.2020.01.031 19. Sethi, R., Bhushan, B., Sharma, N., Kumar, R., Kaushik, I.: Applicability of industrial IoT in diversified sectors: Evolution, applications and challenges. In: Studies in Big Data Multimedia Technologies in the Internet of Things Environment, pp. 45–67 (2020). https://doi.org/10.1007/ 978-981-15-7965-3_4 20. Majumdar, S., Subhani, M.M., Roullier, B., Anjum, A., Zhu, R.: Congestion prediction for smart sustainable cities using IoT and machine learning approaches Sustain. Cities Soc. 64, 102500 (2021). https://doi.org/10.1016/j.scs.2020.102500 21. Shafiq, M., Tian, Z., Bashir, A.K., Jolfaei, A., Yu, X.: Data mining and machine learning methods for sustainable smart cities traffic classification: A survey Sustain. Cities Soc. 60, 102177 (2020). https://doi.org/10.1016/j.scs.2020.102177

12 Machine Learning Approaches for Smart City …

161

22. Gulyani, V., Dhiman, T., Bhushan, B.: Introducing machine learning to wireless sensor networks. In: Deep Learning Strategies for Security Enhancement in Wireless Sensor Networks Advances in Information Security, Privacy, and Ethics, pp. 1–22 (2020). https://doi.org/10. 4018/978-1-7998-5068-7.ch001 23. Kumar, K.S., Anbarasi, M., Shanmugam, G.S., Shankar, A.: Efficient predictive model for utilization of computing resources using machine learning techniques. In: Proceedings of the 2020 10th International Conference on Cloud Computing, Data Science & Engineering (Confluence), Noida, India, pp. 351–357 (2020). https://doi.org/10.1109/Confluence47617.2020.905 7935 24. Spanò, S., et al.: An efficient hardware implementation of reinforcement learning: The QLearning algorithm. IEEE Access 7, 186340–186351 (2019). https://doi.org/10.1109/ACCESS. 2019.2961174 25. Li, H., Wei, T., Ren, A., Zhu, Q., Wang, Y.: Deep reinforcement learning: Framework, applications, and embedded implementations: Invited paper. In: Proceedings of the 2017 IEEE/ACM International Conference on Computer-Aided Design (ICCAD) (2017). https://doi.org/10. 1109/iccad.2017.8203866 26. Zaheer, K., Othman, M., Rehmani, M.H., Perumal, T.: A survey of decision-theoretic models for cognitive internet of things (ciot). IEEE Access 6, 22489–22512 (2018). https://doi.org/10. 1109/access.2018.2825282 27. Elsaeidy, A.A., Jagannath, N., Sanchis, A.G., Jamalipour, A., Munasinghe, K.S.: Replay attack detection in smart cities using deep learning. IEEE Access 8, 137825–137837 (2020). https:// doi.org/10.1109/access.2020.3012411 28. Smith, H.J., Milberg, S.J., Burke, S.J.: Information privacy: Measuring individuals’ concerns about organizational practices. MIS Q. 20(2), 167 (1996). https://doi.org/10.2307/249477 29. Maasar, M.A., Nordin, N.A.M., Anthonyrajah, M., Zainodin, W.M.W., Yamin, A.M.: Monte Carlo & Quasi-Monte Carlo approach in option pricing. In: Proceedings of the 2012 IEEE Symposium on Humanities Science and Engineering Research, Kuala Lumpur, Malaysia, pp. 1401–1405 (2012). https://doi.org/10.1109/SHUSER.2012.6268822 30. Van den Dries, S., Wiering, M.A.: Neural-Fitted TD-Leaf learning for playing othello with structured neural networks. IEEE Trans. Neural Netw. Learn. Syst. 23(11), 1701–1713, Nov (2012). https://doi.org/10.1109/TNNLS.2012.2210559 31. Hua, F., Shen, X., Chen, Z., Yang, F., Gu, J.: Bayesian DOA estimation method using population monte carlo. In: Proceedings of the 2012 IEEE International Conference on Signal Processing, Communication and Computing (ICSPCC 2012), pp. 158–161 (2012). Hong Kong, China. https://doi.org/10.1109/ICSPCC.2012.6335671 32. Wenge, R., Zhang, X., Dave, C., Chao, L., Hao, S.: Smart city architecture: A technology guide for implementation and design challenges. China Commun. 11(3), 56–69 (2014). https://doi. org/10.1109/cc.2014.6825259 33. Bhushan, B., Sahoo, G.: Recent advances in attacks, technical challenges, vulnerabilities and their countermeasures in wireless sensor networks. Wireless Pers. Commun. 98(2), 2037–2077 (2017). https://doi.org/10.1007/s11277-017-4962-0 34. Sharma, N., Kaushik, I., Agarwal, V.K., Bhushan, B., Khamparia, A.: Attacks and security measures in wireless sensor network. Intell. Data Anal. Terror. Threat. Predict. 237–268 (2021). https://doi.org/10.1002/9781119711629.ch12 35. Lin, J., Yu, W., Zhang, N., Yang, X., Zhang, H., Zhao, W.: A survey on internet of things: Architecture, enabling technologies, security and privacy, and applications. IEEE Internet Things J. 4(5), 1125–1142 (2017). https://doi.org/10.1109/jiot.2017.2683200 36. Bettini, C., Riboni, D.: Privacy protection in pervasive systems: State of the art and technical challenges. Pervasive Mob. Comput. 17, 159–174 (2015). https://doi.org/10.1016/j.pmcj.2014. 09.010 37. Aguilera, U., Peña, O., Belmonte, O., López-de-Ipiña, D.: Citizen-centric data services for smarter cities Futur. Gener. Comput. Syst. 76, 234–247 (2017). https://doi.org/10.1016/j.fut ure.2016.10.031

162

S. Mehta et al.

38. Kotsiopoulos, T., Sarigiannidis, P., Ioannidis, D., Tzovaras, D.: Machine learning and deep learning in smart manufacturing: The smart grid paradigm. Comput. Sci. Rev. 40, 100341 (2021). https://doi.org/10.1016/j.cosrev.2020.100341 39. Hasankhani, A., Hakimi, S. M., Bisheh-Niasar, M., Shafie-khah, M., Asadolahi, H.: Blockchain technology in the future smart grids: A comprehensive review and frameworks. Int. J. Electr. Power Energy Syst. 129, 106811 (2021). https://doi.org/10.1016/j.ijepes.2021.106811 40. Feng, C., Wang, Y., Chen, Q., Ding, Y., Strbac, G., Kang, C.: Smart grid encounters edge computing: opportunities and applications. Adv. Appl. Energy 1, 100006 (2021). https://doi. org/10.1016/j.adapen.2020.100006 41. Sánchez-Vergara, J.I., Ginieis, M., Papaoikonomou, E.: The emergence of the sharing city: A systematic literature review to understand the notion of the sharing city and explore future research paths. J. Clean. Prod. 295, 126448 (2021). https://doi.org/10.1016/j.jclepro.2021. 126448 42. Salehi, H., Burgueño, R., Chakrabartty, S., Lajnef, N., Alavi, A.H.: A comprehensive review of self-powered sensors in civil infrastructure: State-of-the-art and future research trends. Eng. Struct. 234, 111963 (2021). https://doi.org/10.1016/j.engstruct.2021.111963 43. Zhu, F., Li, Z., Chen, S., Xiong, G.: Parallel transportation management and control system and its applications in building smart cities. IEEE Trans. Intell. Transp. Syst. 17(6), 1576–1585 (2016). https://doi.org/10.1109/tits.2015.2506156 44. Li, Z., Al Hassan, R., Shahidehpour, M., Bahramirad, S., Khodaei, A.: A hierarchical framework for intelligent traffic management in smart cities. IEEE Transactions on Smart Grid 10(1), 691–701 (2019). https://doi.org/10.1109/tsg.2017.2750542 45. Zhao, Z., Chen, W., Wu, X., Chen, P.C., Liu, J.: LSTM network: A deep learning approach for short-term traffic forecast. IET Intel. Transport Syst. 11(2), 68–75 (2017). https://doi.org/10. 1049/iet-its.2016.0208 46. Prakash, K.S., Srimannarayana, K., Kaatnam, P.K., Dayal, A.: Precise survey assistance for civil structures using survey assist. Learn. Anal. Intell. Syst. 47–52 (2019). https://doi.org/10. 1007/978-3-030-24314-2_7 47. Shen, T., Hua, K., Liu, J.: Optimized public parking location modelling for green intelligent transportation system using genetic algorithms. IEEE Access 7, 176870–176883 (2019). https:// doi.org/10.1109/access.2019.2957803 48. Bouchelaghem, S., Omar, M.: Reliable and secure distributed smart road pricing system for smart cities. IEEE Trans. Intell. Transp. Syst. 20(5), 1592–1603 (2019). https://doi.org/10. 1109/tits.2018.2842754 49. Bhattarai, B.P., Paudyal, S., Luo, Y., Mohanpurkar, M., Cheung, K., Tonkoski, R., Zhang, X.: Big data analytics in smart grids: State-of-the-art, challenges, opportunities, and future directions. IET Smart Grid 2(2), 141–154 (2019). https://doi.org/10.1049/iet-stg.2018.0261 50. Wang, Y., Kong, T.: Air quality predictive modeling based on an improved decision tree in a weather-smart grid. IEEE Access 7, 172892–172901 (2019). https://doi.org/10.1109/access. 2019.2956599 51. Du, D., Chen, R., Li, X., Wu, L., Zhou, P., Fei, M.: Malicious data deception attacks against power systems: A new case and its detection method. Trans. Inst. Meas. Control. 41(6), 1590– 1599 (2018). https://doi.org/10.1177/0142331217740622 52. Dighriri, M., Lee, GM., Baker, T.: Big data environment for smart healthcare applications over 5g mobile network. In: Applications of Big Data Analytics, pp. 1–29 (2018). https://doi.org/ 10.1007/978-3-319-76472-6_1 53. Fallah, S., Deo, R., Shojafar, M., Conti, M., Shamshirband, S.: Computational intelligence approaches for energy load forecasting in smart energy management grids: State of the art, future challenges, and research directions. Energies 11(3), 596 (2018). https://doi.org/10.3390/ en11030596 54. Ghorbanian, M., Dolatabadi, S.H., Siano, P.: Big data issues in smart grids: A survey. IEEE Syst. J. 13(4), 4158–4168 (2019). https://doi.org/10.1109/jsyst.2019.2931879 55. Hossain, E., Khan, I., Un-Noor, F., Sikander, S.S., Sunny, M.S.: Application of big data and machine learning in smart grid, and associated security concerns: A review. IEEE Access 7, 13960–13988 (2019). https://doi.org/10.1109/access.2019.2894819

12 Machine Learning Approaches for Smart City …

163

56. Mallikarjuna, B., Maddikara, J.B.: Synchrophasor measurement-assisted system integrity protection scheme for smart power grid. J. Control. Autom. Electr. Syst. 31(1), 207–225 (2019). https://doi.org/10.1007/s40313-019-00516-4 57. Morello, R., Mukhopadhyay, S.C., Liu, Z., Slomovitz, D., Samantaray, S.R.: Advances on sensing technologies for smart cities and power grids: A review. IEEE Sens. J. 17(23), 7596– 7610 (2017). https://doi.org/10.1109/jsen.2017.2735539 58. Mujeeb, S., Javaid, N., Ilahi, M., Wadud, Z., Ishmanov, F., Afzal, M.: Deep long short-term memory: A new price and load forecasting scheme for big data in smart cities. Sustainability 11(4), 987 (2019). https://doi.org/10.3390/su11040987 59. Pallonetto, F., De Rosa, M., Milano, F., Finn, D.P.: Demand response algorithms for smartgrid ready residential buildings using machine learning models. Appl. Energy 239, 1265–1282 (2019). https://doi.org/10.1016/j.apenergy.2019.02.020 60. Ahad, A., Tahir, M., Yau, K.A.: 5G-Based smart healthcare network: Architecture, taxonomy, challenges and future research directions. IEEE Access 7, 100747–100762 (2019). https://doi. org/10.1109/access.2019.2930628 61. Ngiam, K.Y., Khor, I.W.: Big data and machine learning algorithms for health-care delivery. Lancet Oncol. 20(5), (2019).https://doi.org/10.1016/s1470-2045(19)30149-4 62. Sharma, D., Bikshandi, B.: Artificial empathy–an artificial intelligence challenge. Artif. Intell. 321–326 (2020). https://doi.org/10.4324/9780429317415-19 63. Mak, K., Pichika, M.R.: Artificial intelligence in drug development: Present status and future prospects. Drug Discov. Today 24(3), 773–780 (2019). https://doi.org/10.1016/j.drudis.2018. 11.014 64. Venkatesh, J., Aksanli, B., Chan, C.S., Akyurek, A.S., Rosing, T.S.: Modular and personalized smart health application design in a smart city environment. IEEE Internet Things J. 5(2), 614–623 (2018). https://doi.org/10.1109/jiot.2017.2712558 65. Tuli, S., Basumatary, N., Gill, S.S., Kahani, M., Arya, R.C., Wander, G.S., Buyya, R.: HealthFog: An ensemble deep learning based smart healthcare system for automatic diagnosis of heart diseases in integrated IoT and FOG computing environments. Futur. Gener. Comput. Syst. 104, 187–200 (2020). https://doi.org/10.1016/j.future.2019.10.043 66. Ali, Z., Muhammad, G., Alhamid, M.F.: An automatic health monitoring system for patients suffering from voice complications in smart cities. IEEE Access 5, 3900–3908 (2017). https:// doi.org/10.1109/access.2017.2680467 67. Bruzelius, E., Le, M., Kenny, A., Downey, J., Danieletto, M., Baum, A., Singh, P.: Satellite images and machine learning can identify remote communities to facilitate access to health services. J. Am. Med. Inform. Assoc. 26(8–9), 806–812 (2019). https://doi.org/10.1093/jamia/ ocz111 68. Cook, D.J., Duncan, G., Sprint, G., Fritz, R.L.: Using smart city technology to make healthcare smarter. Proc. IEEE 106(4), 708–722 (2018). https://doi.org/10.1109/jproc.2017.2787688 69. Huang, Z., Peng, Y., Li, J., Tong, F., Zhu, K., Peng, L.: Secrecy enhancing of SSK systems for IoT applications in smart cities. IEEE Internet Things J. 8(8), 6385–6392, 15 April, (2021). https://doi.org/10.1109/JIOT.2021.3050331 70. Wang, Y., Amin, M.M., Fu, J., Moussa, H.B.: A novel data analytical approach for false data injection cyber-physical attack mitigation in smart grids. IEEE Access 5, 26022–26033 (2017). https://doi.org/10.1109/ACCESS.2017.2769099 71. Li, Y., Dai, W., Ming, Z., Qiu, M.: Privacy protection for preventing data over-collection in smart city. IEEE Trans. Comput. 65(5), 1339–1350, 1 May (2016). https://doi.org/10.1109/TC. 2015.2470247 72. Duan, Q., et al.: Optimal scheduling and management of a smart city within the safe framework. IEEE Access 8, 161847–161861 (2020). https://doi.org/10.1109/ACCESS.2020.3021196 73. Falco, G., Viswanathan, A., Caldera, C., Shrobe, H.: A master attack methodology for an AIbased automated attack planner for smart cities. IEEE Access 6, 48360–48373 (2018). https:// doi.org/10.1109/ACCESS.2018.2867556

Chapter 13

Machine Learning and Deep Learning Models for Privacy Management and Data Analysis in Smart Cites Trisha Bhowmik, Abhishek Bhadwaj, Avinash Kumar, and Bharat Bhushan

Abstract The smart city is a philosophy, that is established to manage the growing population, fulfil all the basic daily requirements of the citizens, maintain a green and healthy environment, manage energy consumption and help the people to adopt the modern technologies which make life easier. Smart city first designs the policy and by taking Information and Communication Technology in account it makes decisions. This paper deliberates concise analysis of the Machine Learning approaches in a smart city. This paper briefly concentrates on intelligent transport systems, smart grid, cybersecurity and healthcare system in the smart city using the applications of Machine Learning and Deep Reinforcement Learning. This paper aims to give a conclusive acquaintance of Artificial Intelligence, Machine Learning and Deep Reinforcement Learning approaches that can perform an essential role in the formation of a smart city. Finally, the paper highlights the complex problems that occur in a smart city and solution of these problems using Internet of Things, Artificial Intelligent, Machine Learning, Deep Reinforcement Learning techniques. Keywords Machine learning (ML) · Internet of things (IoT) · Artificial intelligence (AI) · Smart city · Smart healthcare · Smart grid

13.1 Introduction An urban area which deals with the problems of the residents by giving them a better living environment is known as a smart city. The idea of a smart city has been suggested by many countries to solve the problem of population growth. This is achieved by intelligent collection and analysis of the data received from the city’s regular observations that rely on improved information. The cities will face more traffic slowdown, higher level of population and inspire cultural estimation because of increase in the population. Information technologies are mandatory to fulfil the goals of numerous local and international organizations. The population of the urban city is T. Bhowmik · A. Bhadwaj · A. Kumar · B. Bhushan (B) School of Engineering and Technology (SET), Sharda University, Greater Noida, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_13

165

166

T. Bhowmik et al.

growing rapidly which is a great problem for the urban city. The phenomenon of smart city first came into existence when Los Angeles formed the urban big data project [1]. The first smart city is considered as Amsterdam by forming a virtual digital city. The phenomenon of smart city boosted when corporate companies launched their distinct initiatives. Smart cities are made possible by purposeful and appropriate integration of IoT devices. The devices commissioned include systems like smart city platforms and wireless IoT protocols for energy efficiency in buildings and smart street lightning [2, 3]. All the embedded technologies are evolving by time and hence more densely connected devices and network are assigned [4]. The smart city provides all the facilities to the civilians, in the context of physical, social and business infrastructure. Information and Communication Technology (ICT) implementation in a city is considered as a smart city which increases the overall intellect in a city operation. The utilization of some technology like Cyber Security (CS), Internet of things (IoT), Wireless Sensor Networks (WSN) helps to develop smart cities properly. These techniques provide privacy to the smart city and help them in meeting their requirements. Many advanced techniques such as Artificial Intelligent (AI), Machine Learning (ML), Deep Reinforcement Learning techniques (DRL) helps to carry out the analysis of big data to reach an ideal decision [5, 6]. In the healthcare system, smart cities take care of privacy using ML and Deep learning (DL) [7]. In order to engender an enormous amount of data, the IoT performs a very significant part in the applications of the smart city [8, 9]. The data analysis of Big Data using an advanced method and the idea of smart city consciousness have grown drastically with the time [10]. The goals of a smart city can be accomplished using IoT, Cyber-Physical Security (CPS) and WSN technologies [11, 12]. The prime contribution of this paper is encased as follows: • The philosophy of a smart city and smart healthcare systems is discussed taking security and privacy as a parameter. • Cybersecurity of smart cities, intelligent transportation system and smart grid are presented in detail. • Recently proposed ML and DL based solutions for smart city are reviewed. • Current research challenges in developing smart cities using ML and DL is discussed in detail along with future research trends in the field. The remainder of the paper is organised as follows. Section 2 presents the smart city basics. Section 3 elaborates ML overviews and its subdomains. The vital role of ML in smart city is described in Sect. 4. Section 5 deals with the privacy-enhancing technologies used in the smart cities. The future scope and challenges are analysed in Sect. 6. Finally, Sect. 7 is devoted to the conclusion of this work.

13.2 Smart City Basics The procedure of using ICT to escalate efficiency in reducing the effect of the vast population residing in small areas is termed as smart cities. The populations of

13 Machine Learning and Deep Learning Models …

167

urban areas are going to reach 5 million by 2030, as per estimated data studies [13]. Hence there is a need to model and design smart cities accordingly. Dustdar et al. [14] explained the utilization of ICT as the main aim of smart cities which helps to enhance the quality of life, sustainability and growth of cities. The driving forces of smart cities are the integration of both inner sustainability and global sustainability. Internal sustainability is a driving force concerning the effect of population growth. Global sustainability is a driving force concerning social, environmental and economic factors. Smart cities have multiple application domains. Thirty-one factors of matrices given by Giffinger having six categories were used to derive lists of application domains [15]. Figure 13.1 shows the six categories of Giffinger matrices. Anthopoulos et al. [16] identified that application domain can also be derived from indexes. Some applications in governance, economy, mobility, people, environment and quality of life are included by this domain. Ownership privacy, location, identification of the user, footprint are the privacy concerns of smart cities. The multidisciplinary technologies from the range of computer engineering to social science are used in building smart city. Connecting common devices with the internet is termed as the IoT. This technology helps to share data to enhance decision-making in smart cities. CPS helps to interact with the digital entity on behalf of the physical objects. In order to manage the waste collection, noise around the city, population and controlling street lights, smart city uses a Wireless Sensor Network (WSN).

Fig. 13.1 The six categories of Giffinger matrices

168

T. Bhowmik et al.

13.2.1 Pillars of a Smart City The pillars of a smart city work for safeguarding its residents, irrespective of their gender, age, income levels and social status. The pillars are the backbone of a smart city. The significant pillars of smart cities are mentioned below.

13.2.1.1

Government and Leadership

The evaluation and implementation of the best solutions for the most emerging problems at all levels of the government is highly required to build a smart city [17]. The development of a smart city needs to break down the bureaucratic cellar in order to cooperate the city and communal establishments. There is a need of strong leadership and out of the box thinking for the implementation and engagement of policies, building consent and to set a plan in motion. The accountability and transparency are brought in smart cities by utilizing online services and higher participation of community. At the place of going to the government office, smart city residents are provided with a better option of mobile services and e-forums that reduces the cost of service as well as provides a faster service.

13.2.1.2

Technology and Innovation

ICT must be utilized on a very large scale in a smart city to enhance the city’s work culture, sustainability and living standard. The abundance of smartphones and technologies like 4G and 5G ensures that citizens of a smart city are always connected which leads towards the goal of making all the residents’ connectivity easily accessible [18]. Technology is the most important pillar of smart cities. Analysing the data is the key factor to improve the technology. The existing systems can be integrated and improved by using cutting-edge data analytics. The data analytics include the use of data that has been collected for specific purposes in a way to enhance decisions and routine operations.

13.2.1.3

Cyber and Physical Security

The population rise comes with a lot of privacy and security concerns in urban cities and these concerns are tackled by smart cities by adopting smart security procedures. Smart cities are equipped with many technologies like real-time monitoring systems and many smart safety mechanisms like drone-based alarm systems and video surveillance based on Interconnection Protocol (IP) [19]. In order to monitor a crowd or identifying a criminal in crowd, many current technologies like advanced analytics, AI, ML based image processing are used in smart cities.

13 Machine Learning and Deep Learning Models …

13.2.1.4

169

Smart Energy

Smart cities use a high amount of energy and to optimise their energy consumption they rely on technology. Smart cities have fine networked smart grids that make energy usage more cost effective and reduces power intake [20]. The preservation of energy in smart cities is a complex integration mechanism of IoT and real-time demand response management system. The real-time demand response management includes a wide range array of dispersed energy resources and advanced demand response system.

13.2.1.5

NGO’s and Universities

The development of smart cities not only depends on the technological advancements but, it also depends on the development of human resource of the city. The infrastructure development of Non-Governmental Organisations (NGOs) and educational institutions like universities is a necessity to provide smart cities an intellectual firepower and nourish public trust [21]. Many of the NGOs and universities work as incubators for the upcoming projects for the city. These institutions help in getting better project outcomes by providing training to the required human resource in very efficient manner.

13.2.2 Components A smart city can’t be imagined without some components which make it intelligent, responsive, sustainable, resourceful and connected. Some of the prominent components of smart city are described below.

13.2.2.1

Smart Transportation System

Many components such as vehicle, infrastructure, public transportation and people combinedly work as a share of connected transportation system that enhances mobility and safety [22]. In order to tackle a number of issues in a specified community, smart transportation systems can be used that can enhance commute times and can increase security for riders and pedestrians. Technologies such as connected vehicles and electric vehicles can help with the issues of traffic in a smart city.

13.2.2.2

Health: Building Smart Healthcare System

Smart cities focus on the health and sustainable lifestyle of their residents. The process of improving healthcare includes solving the existing challenges in healthcare system

170

T. Bhowmik et al.

like access of it for every citizen in an affordable way or inefficient or ineffective usage of emergency services and reach of healthcare facilities in remote areas [23]. There are a number of fronts on which work is being done, like fast and reliable ambulance service in all areas, increasing the number of doctors in required hospitals, remote clinics for ease of access and much more.

13.2.2.3

Environment of Smart City

In the changing environmental and economic circumstances smart cities needs to make themselves green, energy efficient and ready for the upcoming changes in all type of situations with the help of innovations [24]. In case of any natural disaster, if it has been already predicted, required preparations can save loss of lives, money and resources. Waste management is one of the biggest obstacles in the path for saving environment. It does not only include reusing, recycling waste but it also focuses on reducing the very origin of waste. Many vital resources such as water and electricity need to be saved or optimised for their best use before they are wasted due to lack of knowledge or infrastructure. Smart automated waste collection process and smart grid are one of kind solutions for these problems.

13.2.2.4

Smart Industrial Environment

The development of applications related with IoT and connected technologies is always greater in industrial environments. The areas like detecting forest fire, air or noise pollution check, snow level monitoring and detection of radiation levels in an area can effectively done using sensors. IoT devices connected with these sensors could execute and produce an accurate result [25].

13.2.2.5

Smart City Services

There are always concerns for public safety in case of emergencies which smart cities tackle using IoT [26]. Smart Kiosk is an example of enhanced IoT solutions that play an important part in providing diverse city services to the people like Wi-Fi services, camera-based IP surveillance and analytics as well as public announcements. Monitoring of risky areas can be done using IoT devices embedded in cameras, street lights and actuators that can help residents to avoid the risk zone temporarily in case these devices record any crime or mishap.

13 Machine Learning and Deep Learning Models …

171

13.2.3 Characteristic The characteristics of a smart city depend on some social framework such as the location of city or the culture of the city. The most important values of a smart city are explained below.

13.2.3.1

Smart Data

Smart cities gather a large amount of data to be evaluated frequently to provide valuable information to the residents. The data collected from various sources includes data from smart cars, camera systems data and environmental sensor data. The big data gathered from these sources is then used to prevent road accidents, enforce traffic rules, surveillance and analysis of air quality of the city. Smart cities use three distinct technologies to store the collected data. The first technology is cloud storage, that is used to store the data remotely and that data can be accessed from anywhere without knowing its actual storage location. Secondly, edge computing that computes the data near to the source. It comprises of AI traffic management that can process the data on the spot and produce output for it [27].

13.2.3.2

Smart Infrastructure

In order to augment economy, empower human resource for social, cultural and urban advancement, smart cites prioritise to develop an optimal infrastructure. Using advanced technologies like connected devise and improved communication channel among all the entities, a most advantageous infrastructure is developed for a smart city to grow at rapid speed [28]. The entities of smart city like big data, needs a specialised infrastructure that can be used to proactively maintain and alter the data. Numerous technologies like IoT, Big Data and much more can be integrated to form an intelligent infrastructure.

13.2.3.3

Connected Mobility

There are many levels through which the data travels to perform the task of administration in a smart city [29]. These levels need to be connected in such a way that data mobilisation can be optimised considering the impact of its latency and reusability in almost all the fields of a smart city. The data flow needs to be free between machines so that it can be considered for use of security and protection of user’s privacy.

172

13.2.3.4

T. Bhowmik et al.

Smart Building and Properties

Smart building is a concept in which a building is considered as smart if it uses diverse technologies to ensure the privacy and safety of the building as well as maintain assets and health of the neighbour. The technologies used in the building comprise of biometrics, remote monitoring, wireless alarm and remote monitoring to condense the unauthorized entry in the building to avoid theft [30]. Smart building also contains advanced heating and ventilation technologies to monitor numerous parameters like temperature, pressure, humidity and much more. WSN is the necessity to ensure suitable heating and ventilation. Automated fire and safety alarms are used in the building to ensure safety of the people and the property at its maximum.

13.3 Machine Learning Overview The study which helps to improve automatically through the experience of a computer algorithm is called ML. There are three categories of ML i.e., Supervised Learning (SL), Unsupervised Learning (UL) and Reinforcement Learning (RL). All the branches of ML have been discussed in Fig. 13.2 [31]. The implementation of the subdomains of ML are discussed in below subsections.

13.3.1 Supervised Learning The task of ML where we provide the data set as input and get the output familiar to us is termed as SL [32]. SL gives the idea of connection between the input and the output. SL is categorized into two different problems i.e., Regression and Classification.

Fig. 13.2 The classification of machine learning

13 Machine Learning and Deep Learning Models …

173

In the problem of regression, the outcomes are predicted in a continuous output. In the classification problem, input variables are mapped into discrete categories. Spam detection, text classification problems, linear regression and random forest are applications of SL. Linear regression identify the connection between variables and forecasting. It is a method to model the relationship among a scalar response and one or more than one dependent and independent variables. It predicts the targeted value that is based on independent variable. Classification and regression both can use Random Forest algorithm. It helps to resolve a critical problem and progress the performance of the model with the process of compounding numerous classifiers.

13.3.2 Unsupervised Learning This is a machine learning technique that is used to train Artificial Intelligent (AI) network for finding patterns that are hidden, answers and distributions [33]. In UL, inputs are both unlabelled and unclassified. The k-means and auto-encoder algorithm are an example of UL problems. The algorithm of k-means allocates each data point to the closest cluster after identifying k- number of centroid and keeps the centroids as minor as possible. It is a technique of vector quantization, that intent to divide n observations into k clusters. Auto-encoder is a category of Artificial Neural Network. Autoencoder has diverse layers that extract features. The middle layer contains the algorithm that signifies the efficiency of a Autoencoder. It learns to represent a set of data for dimensionally reduction. Autoencoder is mainly used in DL to reduce features for each level.

13.3.3 Reinforcement Learning This is the area of ML which mainly focuses on finding stability among exploration and exploitation [34]. The agent is the part of the RL algorithm which mainly does the interactions to learn it. RL agent improves long-term aggregated reward of RL. Model-based and model-free are the two different types of RL algorithms. In order to solve the model-based RL algorithm, there are many different techniques like searching policy, value function, model of transition and output function. There are also two types of model-free RL algorithms; these are Monte Carlo (MC) and Temporal Difference (TD). MC techniques have usage in algorithms that impersonate policy iteration. The evaluation policy step has usage of MC. TD learning model is trained by using bootstrapping from the function’s value present in estimation. The Q-learning (Quality-learning) and SARSA (State-Action-Reward-State-Action) technique is the major TD-based algorithm.

174

T. Bhowmik et al.

13.4 ML for Smart Cities Smart cities maintain a healthier environment with the help of ML. ML advance and automate a large number of activities and operations in smart cities. These advances are discussed in the subsections below.

13.4.1 Intelligent Transportation System The Intelligent Transportation System (ITS) consists of ICT, control systems and advanced sensors, which generates big data. The future concept of smart cities and ITS is effectively impacted by ICT [35]. The key to sustainable ITS is to effectively monitor and evaluate the data flow of traffic in real-time, where AI, ML and DRL plays a vital role [36, 37]. Veres et al. [38] explored the function of ML and DRL for various issues. This was a detailed study that can be utilized properly in a smart city in the area of ITS. The issues that can play a major solving role in developing smart cities, a detailed study based on edge analytics and DRL methods in ITS has been carried out [39]. Based on DRL, an improved driving behaviour for taking a decision, a technique was proposed in a heterogeneous traffic environment, where data converted into a hyper-grid matrix by a data pre-processor [40]. Hyper-grid matrix is a two steam Deep Neural Network (DNN). Through unsupervised learning, various attacking possibilities are learned. This approach is based on DRL, which mainly focuses on the rising security concerns with Mobile Edge Computing (MEC). This technique is 6% more accurate than others [41]. Li et al. [42] give an approach based on DRL to forecast short term flow of traffic on a highway. For passengers hunting in an area, a study of using GPS trajectories data of Taxis was adopted [43]. In order to ensure optimum system performance and predict parameters of a wireless communication, Long Short-Term Memory (LSTM) i.e., a prediction technique, is being used [44]. For vehicle edge computing, a smart offloading system is developed with the technique of DRL [45]. Ye et al. [46] proposed Vehicle to Vehicle (V2V) communication, based on the DRL technique which is a decentralized resource allocation method. The authors proposed a study to learn many aspects of the flow of traffic into the stacked auto-encoder model [47]. A traffic-aware technique is proposed to facilitate Unnamed Air Vehicles (UAV) deployment in a vehicular environment. Liu et al. [48] proposed a study that the users of the handle device in a pre-stated region for the DRL-based decentralized architecture of Arial UAVs that was intended to provide coverage services. Wang et al. [49], investigate the utility of UAVs for the downlink transference of the automobiles taking maximum output into consideration.

13 Machine Learning and Deep Learning Models …

175

13.4.2 Cybersecurity The interconnectivity of several devices has unplugged cybersecurity affairs such as secure and reliable interconnected sensors, actuator and depends on gathering, transmitting and processing data which need to alleviate [50]. The philosophy of smart city architecture for security, transmission and privacy outlook is focused on arising issues that are challenging during the assimilation of current infrastructure, protocols for sensors and communication [51]. Exploring the role of ML and DRL technique, Al-Garadi et al. [52] have done a substantial study of IoT and newly introduced security threats from a progressed security perspective. Miotto et al. [53] look over the achievable opportunities and role by using ML and DRL methods in the Bioinformatic and healthcare area. The authors in [54] proposed an Anomaly Detection-IoT (AD-IoT) system, which is a ML-based architecture that improves the safety of IoT devices in the smart city using an algorithm called Random Forest. It is possible to investigate any kind of incredulous work happening at the dispersed fog using ML. In the Fog-Cloud-IoT, a protected computational offloading framework based on ML to enhance latency and energy consumption has been presented [55]. In order to confirm the data security at the gateway, a Neuro-fuzzy model was proposed. The decision of supportive fog nodes to compute was made by IoT devices via Particle Swarm Optimization (PSO) [55]. The Edge Cognitive Computing (ECC) network architecture has been traversed for providing defence in network [56]. This architecture was proposed for active service shifts and dynamic service shifts. An online offloading method based on DRL has been proposed to progress binary decision offloading abilities from experience [57].

13.4.3 Smart Grids The efficient energy utilization and operational structure of SGs have been revolutionized by bid data, in smart cities [58]. The SGs are going towards voluminous data, IoT devices, communication systems and modern information system [59]. The efficient use of smart meter big data for various applications like estimating the baseline, clustering the load and demand-response, etc. are created by SGs [60, 61]. Bhattarai et al. [62] mainly worked on the Phase Measurement Units (PMUs) big data for calibrating dynamic model, data visualizing, and estimation of state and transmission of grid. Ghorbanian et al. [63] explored the distinct types of big data-assisted applications in SGs. In [64], the author focused on analysing 5G communication in SGs. Hossain et al. [65] specified the performance of ML and DRL techniques in the cybersecurity of SGs and correlated applications. Different applications of DRL techniques, about the computation of new power generation, transient stability, fault analysis of controls of grid and power, and load forecasting have been explored [66]. In order to forecast the demand and price of electricity before a day or week

176

T. Bhowmik et al.

by using real market data, a Deep Long Short-Term Memory (DLSTM) model has been proposed [67]. A well-organized and measured building simulation prototype to understand the effect of Demand Response (DR) policies has been explored [68]. In order to build the decision factors at the time of development of ML-based application, an automated ML- platform has been proposed [69]. The description for using Power Line Communication (PLC) modems and ML as well as intrusion detection, and position finding mechanism in SGs has been explored [70]. A reconfigurable distributed and cross-domain platform has been proposed [71]. It is the design of the ISAAC security testbed for the SGs system. In order to constraint data integrity attacks in AC power system has been proposed [72]. By using UAVs, an innovative (break thought) technique based on DRL has been proposed [73]. Paramanik [74] invented the Pan-Tilt-Zoom (PTZ) the camera in order to increase the efficiency of SGs and remove the possibility of disaster. Gulyani et al. [75] have focused on UAVs based on DRL for wind turbine monitoring.

13.4.4 Healthcare Systems Health intelligence is a healthcare mechanism that uses AI, ML and DRL techniques which are attended by the arrival of advanced sensors, cloud computing, high performance of IoT devices rise in data in data rates. [76–78]. In order to decrease diagnosing, media imaging, cure prediction and social media analytics for a specific disease, the previous technique is playing a very important role [79]. In smart cities, these are the recent researches and activities of healthcare that are carried out. The part of 5G communication in the field of healthcare, hardware, essential techniques, architecture and the main objective has been reviewed [80]. The healthcare system and big data analysis, with the help of AI, ML, and DRL applications have been described [81]. HealthFog is a novel architecture for analysing heart disease efficiently and autonomously [82]. DRL protocols support the edge computing devices to compose the HealthFog framework. In order to analyse the medical image and categorize various disorders related to the abdominal and spleen, a review has been developed to study the impact of DRL protocol [83]. A technique based on ML has been developed to evaluate patient’s odds of enduringness after Percutaneous Coronary Intervention (PCI) [84]. A deep learning model that used CNN to diagnose Glaucoma Disease (GD) diagnosis has been presented [85]. The data used in training the model was acquired from Beijing Tongren Hospital. This model is 81.69% accurate [85]. In a smart city, there is a new trend of the healthcare system that is called smart health which has e- heath as a subpart of it. This healthcare care system uses Electronic Health Record (EHR) and some other different variable which comes from the smart city architecture. The Table 13.1 comprises the comparison among diverse approaches proposed for advancement of smart city.

13 Machine Learning and Deep Learning Models …

177

Table 13.1 Compares different approaches proposed for the development of smart city Reference

Year

Description

Advantages

Awan et al. [36]

2021

AI, ML and DRL based traffic flow data has been evaluated

Solves the problem Ignores small of analysing Big frequency data data

Shortcomings

Zhou et al. [37]

2018

ITS effectively monitor and evaluate the traffic flow data

Management and storage of big data at same time

The dense network system need improvement in reaction time

Ke et al. [40]

2020

A DRL based automated decision-making driving behaviour has been proposed

Facilitates the driver in controlling the vehicle

The features were more dependent in some particular cases

Li et al. [42]

2019

Forecast the short-term traffic flow on highway based on DRL

Produces reliable results that can be used in real time

Irregularity in data can alter results

Ye et al. [46]

2019

A DRL based decentralised resource allocation method in V2V communication has been proposed

Solves the problem of communication among connected devices

Request overload problem arise while communicating with many devices

Al-Garadi et al. [52]

2020

Introduced security Facilitates accurate Focuses more on threats with the help security breach stated regulations at of IoT regulations the place of alterations in them

Hossain et al. [65]

2019

Specified ML and DRL techniques in cyber security of SGs

Guarantees the security of smart grids from cyber-attacks

Direct port-based attacks were ignored

Pallonetto et al. [68]

2019

Measured building simulation prototype to understand the effect of DR policies

Solves the issue of ventilation and excessive energy consumption with increasing population

Cost of building and its maintenance is not considered in optimisation

R et al. [81]

2019

Introduced bigdata analysis in healthcare system using AI, ML and DRL applications

Guarantees the The management prediction and ease technique is static of management in not dynamic healthcare systems

Zhang et al. [83]

2019

Analyse medical images and categorize various abdominal disorders and spleen

Facilitates the classification of various abdominal disorders by technology

The categorization is based heavily on results of a single classifier (continued)

178

T. Bhowmik et al.

Table 13.1 (continued) Reference

Year

Description

Advantages

Viegas et al. [85]

2017

A CNN based deep learning model to diagnose Glaucoma disease

Helps in predicting The computational the early stage of speed of model was glaucoma disease depended on the machine used

Shortcomings

13.5 Privacy-Enhancing Technologies There are many ML technologies for the enhancement of privacy in the smart city. Variance, substitution, encryption, shuffling, k-anonymity, blockchain, onion routing and zero-knowledge proof are some commonly used techniques in the smart cities. Apart from these techniques there are some more Privacy-Enhancing Technologies (PET).

13.5.1 Substitution The value of the data in identification or quasi-identification can be replaced randomly. This random replacement is called substitution [86]. In that case, any data that can be replaced with any string such as “ET6K” can be used for the name “ALEX”. Substitution helps in smart city to trace per-user data which is important to substitute characteristic information. In the sensor layer, the consumed data can be composed by substitution. In order to categorize data, an anonymization method is used. Masking and Nulling out are two different technologies that are similar to substitution.

13.5.2 Shuffling Shuffling is the technology that belongs to the application layer. Shuffling helps to lose the alliance between attributes because it rearranges the values [87]. The connection between sensory attributes and identifying attributes can be removed by shuffling which is the main goal of shuffling. It is a defensible technology that helps to secure privacy in data publications. The categorical and numeric information are types of protected shuffling.

13 Machine Learning and Deep Learning Models …

179

13.5.3 Variance Variance is a type of perturbation method. The numerical data which are connected or unconnected to the classification of number can be distorted by variance [88]. A smart city use variance to create a protected data set which has identical mean value like original set, but altered individual values. There are various akin methods to variance such as Synthetic Data and Differential Privacy.

13.5.4 Encryption The method of hiding information completely by converting the original information into an alternative using cryptography is known as Encryption [89]. Encryption is a part of security that can achieve anonymity. Encryption is very much useful to achieve privacy objectives in smart city architecture. There are different technologies and implementations of encryption like public key, homomorphic, hashing, biometric and secret key generation. All these encryption methods enhance the security and privacy of a smart city.

13.5.5 Blockchain A cryptographic system that can protect and privatize transactions between the networks is known as blockchain [90]. Since smart city is based on connected network, the blockchain could play a very important role for secure communication between the smart devices. It is a certain type of database that stores data in a block. The implementations for blockchain have been explored in IoT because of its decentralized nature. The financial transaction can be secure by using blockchain currencies which can be utilized in smart cities as a secure payment method.

13.5.6 K-Anonymity K-Anonymity avoids reidentifications via quasi-identifiers by making sure that all available entries over the data share values related to the quasi-identifier having k-1 different entries [91]. An attacker can’t find its objective among k-1 different entries if K-anonymity has been applied on the data. Micro aggregation is used by K-anonymity for creating grouping that imparts anonymization perturbation in a cautiously constructed manner. Some related techniques to K-anonymity are lDiversity and t-Closeness.

180

T. Bhowmik et al.

13.5.7 Onion Routing At the time of communication, sender and receiver are provided anonymity by Onion Routing [92]. The anonymous socket relationships in the midst of server and user are also provided by onion routing. This supports various applications data. The concatenation of encryption layer is used by onion routing for every hop and it does not take a direct path from source to destination.

13.5.8 Zero-Knowledge Proof The protocol for allowing involvement in a system while facilitating a technique for covering any information of the participation is known as zero-knowledge proof [93]. In a smart city, without giving the real key to retrieve a system, a user can provide evidence of their verification. This is the work of zero-knowledge proof in a smart city. This method helps to secure categorical and numerical information of a user in a smart city by using encryption. The Table 13.2 shows the advantages and drawbacks of PET. Table 13.2 Advantages and drawbacks of PET used in smart cities Technology

Advantage

Disadvantage

Substitution [86]

Substitution creates an incomprehensible data randomly, that makes the data secure

Substitution techniques like nulling out can sometime remove a vital attribute resulting in loss of information

Shuffling [87]

Shuffling secures data by removing direct association between attributes. It also safeguards data integrity

Shuffling is ineffective in case of small data as it might be guessable for the third party

Variance [88]

Variance facilitates in separation of data with correlation and no correlation

The uncontrollable type of variance stops the user to regulate the separation criteria of the data and it authorises third party

Encryption [89]

Encryption technique can protect the info and communication from unauthorized exposure and access of info

A strongly encrypted, reliable and digitally signed info can be problematic to access even for a genuine user at a critical time of execution (continued)

13 Machine Learning and Deep Learning Models …

181

Table 13.2 (continued) Technology

Advantage

Disadvantage

Blockchain [90]

Blockchain allows the user to store data at several dispersed nodes that prevents system failures and malicious attacks

Modification of data is very difficult in blockchain if once it is deployed

K-anonymity [91]

K-Anonymity averts record correlation by creating large correspondence class

If the records in the correspondence class contain similar values for a particular attribute the attacker can identify a attribute by finding similarity

Onion routing [92]

It adds layer upon layer of encryption while forming an indirect anonymous path between sender and receiver

The channel can be breached if the receiver does not use secure socket layer

Zero-knowledge proof [93]

It allows the user to enter into It needs a substantial a system without sharing its computational power to encrypt data beyond the authentication the data access

13.6 Future Research Trends The recent development in ML and DL techniques in the context of smart cities is escalating with a very high slop each day. Many approaches and technologies of AI, ML and DRL have shown a favourable output that is dispossessed in the given literature review. Hyper grid matrix in DNN, LSTN prediction techniques for ensuring the optimal system performance of the wireless network, edge computing DRL techniques for cybersecurity AD-IoT and Neuro-Fuzzy models, for health sector health fog framework using DRL, etc. are the example of approaches and techniques used in the smart city [94–97]. These approaches can be refined in the future to produce more optimum output. Further work, ML and DL approach models to be more efficient in handling more data account of the growing population in smart cities. In order to assure safety of the residents and sustainability of the city, some more advanced technologies are needed in the smart city.

13.6.1 Connecting Technologies The introduction of technologies in a smart city for completing tasks is the practise of current time. This leads to development of those technologies that can work as a single unit for all the assigned tasks in a city. Connecting or merging technologies means that making a useful connection between them that can be used in situations where single technology or device is not capable enough to give the desired output. IoT devices and embedded systems can work as a stand-alone system and also work as an integrated

182

T. Bhowmik et al.

part for a bigger system [98]. IoT devices have capability of communicating with each other but without a proper algorithm the communication becomes just a mere exchange of binary numbers. A centralised control system that can assign, monitor and evaluate the tasks for the devices in use that can eventually decrease the human intervention in all tasks. Centralised control system can prove to be a vital step towards full automation.

13.6.2 Management of Water and Waste Treating waste water and managing waste are growing concerns for cities, along with the issue of abundant access to clean water. Smart cities have already managed to overcome the above-mentioned concerns with automated solutions [99]. The solutions of smart cities rely both on technology and human intervention. The next step is to fully automate the technologies embedded for water and waste management. Automated devices used in current time need human supervision to operate but that only reduces the work hence it does not eliminate it on any level. Waste management is the emerging issue of current times, smart solutions need to be more connected and independent in their functionalities.

13.6.3 Construction and Building Technologies The physical infrastructure of a smart city comprises mainly of buildings. The buildings need energy for either their operation or if they are under construction then for completion. The energy used in the buildings is not fully utilised or optimised and this is the area of future work in this field. Large buildings regardless of their level of automation need a huge amount of energy and that needs to be managed and optimised. Smart solutions can transmute the ordinary buildings into energy-efficient and sustainable structures at the same time also can automate the manner in which buildings are manged and constructed [100]. Temperature and ventilation can also be monitored and controlled by automating the buildings using actuators, IoT and connected devices.

13.6.4 Use of Renewable Resources Smart cities have a reliable energy management system that can optimise the energy use. The energy used in the city is still generated from non-renewable energy resources. The future of energy saving is in using energy harnessed from renewable resources [101]. Use of renewable resources not only reduces the carbon emission in the environment but it can also reduce the energy dependency of the city on some

13 Machine Learning and Deep Learning Models …

183

other source that is inevitable to extinction in near future. Technologies like solar panel, wind turbines, tidal energy is a few examples that need to be deployed more and more to reduce dependency on non-renewable energy resources.

13.7 Conclusion The world population has grown explosively in recent years joined with rapid urbanisation process inclined to jeopardize the sustainability of cities in terms of economic and environmental ways. A smart city is a concept that has been proposed to solve the problem of handling the immensely grown population and its need. The smart city uses IoT, AI, ML and DL to manage big data efficiently. ML and DL approaches are one of the kind solutions for the regular day problem of a smart city. To this end, from the very origin of the phenomenon of smart city to its exponential growth, the objective of smart city has always been to make life convenient and easier for the resident and to defend the sustainability of the city itself. Technological innovations are adding to the sustainable structure of the city constantly. However, there are continuously increasing security, healthcare and electric grid issues in smart cities. These issues can be effectually conquered by the use of IoT, embedded and connected systems. In this paper, the use and possible benefits of applying smart technologies such as IoT devices and embedded systems to smart cities are conferred through a detailed survey. The paper begins with recent developments in the field of ML and background knowledge of smart cities. Then, the motivation behind applying IoT, ML and AI in the basic architecture of smart cities is discussed. Further, the paper aims to combine technologies with smart city infrastructure in diverse ways such as smart healthcare system, smart grid and ITS. Finally, several future guidelines in the context of smart city are outlined. This work is predicted to assist as an information base and methodical standard for future research in integrating AI, ML, DL, IoT devices and connected technology to smart cities.

References 1. Calzada, I.: Replicating smart city citizenship: city-to-city learning programme. In: Smart City Citizenship, pp. 117–40 (2021). https://doi.org/10.1016/b978-0-12-815300-0.00004-6 2. Yang, F., Xu, J.: Privacy concerns in China’s smart city campaign: the deficit of China’s cybersecurity law. Asia Pac. Policy Stud. 5(3), 533–543 (2018). https://doi.org/10.1002/app 5.246 3. Ševˇcík, P., Žák, S., & Hodoˇn, M.: Wireless sensor network for smart power metering. Concurr. Comput. Pract. Exp. 29(23) (2017). https://doi.org/10.1002/cpe.4247 4. Jabbar, M., Aluvalu, R.: Cyber-security: future challenge for a safer and secure smart city. In: 2nd Smart Cities Symposium (SCS 2019) (2019). https://doi.org/10.1049/cp.2019.0227 5. Pranto, T.H., Noman, A.A., Mahmud, A., Haque, A.B.: Blockchain and smart contract for IoT enabled smart agriculture. PeerJ Comput. Sci. 7, (2017). https://doi.org/10.7717/peerjcs.407

184

T. Bhowmik et al.

6. Allam, Z., Dhunny, Z.A.: On big data, artificial intelligence and smart cities. Cities 89, 80–91 (2019). https://doi.org/10.1016/j.cities.2019.01.032 7. Aldeen, Y.A., Salleh, M.: Privacy preserving data utility mining architecture. In: Smart Cities Cybersecurity and Privacy, pp. 253–268 (2019). https://doi.org/10.1016/b978-0-12-8150320.00018-4 8. Goyal, S., Sharma, N., Kaushik, I., Bhushan, B., Kumar, A.: Blockchain as a Lifesaver of IoT. In: Security and Trust Issues in Internet of Things, pp. 209–237 (2020). https://doi.org/ 10.1201/9781003121664-10 9. Al-Turjman, F.: Information-centric framework for the internet of things (IoT): traffic modeling & optimization. Futur. Gener. Comput. Syst. 80, 63–75 (2018). https://doi.org/ 10.1016/j.future.2017.08.018 10. Sethi, R., Bhushan, B., Sharma, N., Kumar, R., Kaushik, I.: Applicability of industrial IoT in diversified sectors: evolution, applications and challenges. In: Studies in Big Data Multimedia Technologies in the Internet of Things Environment, pp. 45–67 (2020). https://doi.org/10. 1007/978-981-15-7965-3_4 11. Lin, J., Yu, W., Zhang, N., Yang, X., Zhang, H., Zhao, W.: A survey on internet of things: architecture, enabling technologies, security and privacy, and applications. IEEE Internet Things J. 4(5), 1125–1142 (2017). https://doi.org/10.1109/jiot.2017.2683200 12. Dameri, R.P.: Smart city DEFINITION, goals and performance. In: Progress in IS, pp. 1–22 (2016). https://doi.org/10.1007/978-3-319-45766-6_1 13. Zhang, K., Ni, J., Yang, K., Liang, X., Ren, J., Shen, X.S.: Security and privacy in smart city applications: challenges and solutions. IEEE Commun. Mag. 55(1), 122–129 (2017). https:// doi.org/10.1109/mcom.2017.1600267cm 14. Sharma, L.: The rise of internet of things and smart cities. In: Towards Smart World, pp. 3–14 (2020). https://doi.org/10.1201/9781003056751-1 15. Bruzzone, M.: Medium-sized smart cities: a smart vision for urban centralities and buildings from the European case history, to a proposal for the city of Parma, Italy. In: Organizing Smart Buildings and Cities, pp. 99–123 (2021). https://doi.org/10.1007/978-3-030-60607-7_7 16. Anthopoulos, L.G.: Understanding smart cities: A tool for smart government or an industrial trick? In: Public Administration and Information Technology (2017). https://doi.org/10.1007/ 978-3-319-57015-0 17. Paulin, A.: Understanding governance: fundamental principles. In: Smart City Governance, pp. 109–110 (2019). https://doi.org/10.1016/b978-0-12-816224-8.09992-1 18. Petrin, J.: Von der Beteiligung ZUR “Citizen Innovation”. In: Smart City—Made in Germany, pp. 133–140 (2020). Springer, Wiesbaden. https://doi.org/10.1007/978-3-658-27232-6_14 19. Jagannathan, M.: Building cyber physical systems in the context of Smart Cities. In: Smart Cyber Physical Systems, pp. 99–112 (2020). https://doi.org/10.1201/9780429321955-8 20. Energy-efficient smart grid communications.: Smarter Energy: From Smart Metering To The Smart Grid, pp. 461–482 (n.d.). https://doi.org/10.1049/pbpo088e_ch15 21. Entrepreneurial Universities.: Entrepreneurship, Innovation and Smart Cities, pp. 97–113. Routledge, London (2017). https://doi.org/10.4324/9781315407463-7 22. Janos, V., Kriz, M.: Smart urban transport. In: 2018 Smart City Symposium Prague (SCSP) (2018). https://doi.org/10.1109/scsp.2018.8402664 23. Smart Healthcare System.: Int. J. Recent Trends Eng. Res. 3(3), 198–200 (2017). https://doi. org/10.23883/ijrter.2017.3067.gav6j 24. Rachmawati, T., Pertiwi, P.D.: Smart environment PROGRAM, smart way to smart city. Policy Gov. Rev. 1(1), 26 (2017). https://doi.org/10.30589/pgr.v1i1.24 25. Cheng, J., Chen, W., Tao, F., Lin, C.: Industrial IoT in 5g environment towards smart manufacturing. J. Ind. Inf. Integr. 10, 10–19 (2018). https://doi.org/10.1016/j.jii.2018.04.001 26. Delgado, J.C.: Asymmetric interoperability for software services in smart city environments. In: Handbook of Smart Cities, pp. 255–283 (2018). https://doi.org/10.1007/978-3-319-972718_10 27. Toh, C.K.: Security for smart cities. IET Smart Cities 2(2), 95–104 (2020). https://doi.org/10. 1049/iet-smc.2020.0001

13 Machine Learning and Deep Learning Models …

185

28. Smart Nation: Smart Cities, Smart Future, pp. 145–159 (2018). https://doi.org/10.1002/978 1119516224.ch8 29. Trombin, M., Pinna, R., Musso, M., Magnaghi, E., De Marco, M.: Mobility management: from traditional to people-centric approach in the smart city. In: Studies in Systems, Decision and Control, pp. 165–182 (2019). https://doi.org/10.1007/978-3-030-22773-9_11 30. Yi, Y., He, P.: Smart campus building based on big data. In: ISME 2016—Information Science and Management Engineering IV (2016). https://doi.org/10.5220/0006450403580361 31. Trappenberg, T.P.: Machine learning with sklearn. In: Fundamentals of Machine Learning, pp. 38–65 (2019). https://doi.org/10.1093/oso/9780198828044.003.0003 32. Sendhil Kumar K.S., Anbarasi M., Shanmugam G.S., Shankar A.: Efficient predictive model for utilization of computing resources using machine learning techniques. In: 2020 10th International Conference on Cloud Computing, Data Science & Engineering (Confluence), Noida, India, pp. 351–357 (2020). https://doi.org/10.1109/Confluence47617.2020.9057935 33. Yang, Y.: Unsupervised learning via an iteratively constructed clustering ensemble. In: Temporal Data Mining Via Unsupervised Ensemble Learning, pp. 75–92 (2017). https:// doi.org/10.1016/b978-0-12-811654-8.00006-3 34. Nandy, A., Biswas, M.: Reinforcement learning basics. In: Reinforcement Learning, pp. 1–18 (2017). https://doi.org/10.1007/978-1-4842-3285-9_1 35. Fiore, S., Elia, D., Pires, C.E., Mestre, D.G., Cappiello, C., Vitali, M., Aloisio, G.: An integrated big and fast data analytics platform for smart urban transportation management. IEEE Access 7, 117652–117677 (2019). https://doi.org/10.1109/access.2019.2936941 36. Awan, N., Ali, A., Khan, F., Zakarya, M., Alturki, R., Kundi, M., Haleem, M.: Modeling dynamic spatio-temporal correlations for urban traffic flows prediction. IEEE Access 9, 26502–26511 (2021). https://doi.org/10.1109/access.2021.3056926 37. Zhou, L., Chen, X.: Short-term forecasting of traffic flow and speed: A deep learning approach CICTP (2018). https://doi.org/10.1061/9780784481523.217 38. Veres, M., Moussa, M.: Deep learning for intelligent transportation systems: A survey of emerging trends. IEEE Trans. Intell. Transp. Syst. 21(8), 3152–3168 (2020). https://doi.org/ 10.1109/tits.2019.2929020 39. Bhavsar, P., Safro, I., Bouaynaya, N., Polikar, R., Dera, D.: Machine learning in transportation data analytics. Data Analytics for Intelligent Transportation Systems, pp. 283–307 (2017). https://doi.org/10.1016/b978-0-12-809715-1.00012-2 40. Ke, P., Yanxin, Z., Chenkun, Y.: A decision-making method for Self-driving based on deep reinforcement learning. Journal of Physics: Conferenve Series, vol. 1576, pp. 012025 (2020). https://doi.org/10.1088/1742-6596/1576/1/012025 41. Li, W., Song, H., Wei, Y., Zeng, F.: Toward more secure and trustworthy transportation cyberphysical systems. SpringerBriefs in Computer Science, pp. 87–97 (2017). https://doi.org/10. 1007/978-981-10-3892-1_5 42. Li, J., Wang, J.: Short term traffic flow prediction based on deep learning. CICTP (2019). https://doi.org/10.1061/9780784482292.214 43. Huang, Z., Shan, G., Cheng, J., Sun, J.: Trec: An efficient recommendation system for hunting passengers with deep neural networks. Neural Comput. Appl. 31(S1), 209–222 (2018). https:// doi.org/10.1007/s00521-018-3728-2 44. Jiang, H., Chang, L., Li, Q., Chen, D.: Trajectory prediction of vehicles based on deep learning. In: Proceedings of the 2019 4th International Conference on Intelligent Transportation Engineering (ICITE) (2019). https://doi.org/10.1109/icite.2019.8880168 45. Ning, Z., Dong, P., Wang, X., Rodrigues, J.J., Xia, F.: Deep reinforcement learning for vehicular edge computing. ACM Trans. Intell. Syst. Technol. 10(6), 1–24 (2019). https://doi.org/ 10.1145/3317572 46. Ye, H., Li, G.Y., Juang, B.F.: Deep reinforcement learning based resource allocation for V2V communications. IEEE Trans. Veh. Technol. 68(4), 3163–3173 (2019). https://doi.org/10. 1109/tvt.2019.2897134 47. Liu, H.: Characteristics and analysis of urban traffic flow in smart traffic systems. In: Smart Cities: Big Data Prediction Methods and Applications, pp. 125–158 (2020). https://doi.org/ 10.1007/978-981-15-2837-8_5

186

T. Bhowmik et al.

48. Liu, C.H., Ma, X., Gao, X., Tang, J.: Distributed energy-efficient multi-uav navigation for longterm communication coverage by deep reinforcement learning. IEEE Trans. Mob. Comput. 19(6), 1274–1285 (2020). https://doi.org/10.1109/tmc.2019.2908171 49. Wang, H., Ke, H., Sun, W.: Unmanned-aerial-vehicle-assisted computation offloading for mobile edge computing based on deep reinforcement learning. IEEE Access 8, 180784– 180798 (2020). https://doi.org/10.1109/access.2020.3028553 50. Rawat, D.B., Ghafoor, K.Z.: Acknowledgements. Smart Cities Cybersecurity and Privacy, Xvii. (2019). https://doi.org/10.1016/b978-0-12-815032-0.09996-0 51. Habibzadeh, H., Soyata, T., Kantarci, B., Boukerche, A., Kaptan, C.: Sensing, communication and security planes: A new challenge for a smart city system design. Comput. Netw. 144, 163–200 (2018). https://doi.org/10.1016/j.comnet.2018.08.001 52. Al-Garadi, M.A., Mohamed, A., Al-Ali, A.K., Du, X., Ali, I., Guizani, M.: A survey of machine and deep learning methods for Internet of Things (iot) security. IEEE Commun. Surv. & Tutor. 22(3), 1646–1685 (2020). https://doi.org/10.1109/comst.2020.2988293 53. Miotto, R., Wang, F., Wang, S., Jiang, X., Dudley, J.T.: Deep learning for healthcare: Review, opportunities and challenges. Brief. Bioinform. 19(6), 1236–1246 (2017). https://doi.org/10. 1093/bib/bbx044 54. Cui, M., Wang, J., Yue, M.: Machine learning-based anomaly detection for load forecasting under cyberattacks. IEEE Trans. Smart Grid 10(5), 5724–5734 (2019). https://doi.org/10. 1109/tsg.2018.2890809 55. Goyal, S., Sharma, N., Bhushan, B., Shankar, A., Sagayam, M.: IoT enabled technology in secured healthcare: Applications, challenges and future directions. In: Cognitive Internet of Medical Things for Smart Healthcare Studies in Systems, Decision and Control, pp. 25–48 (2020). https://doi.org/10.1007/978-3-030-55833-8_2 56. Chen, M., Li, W., Fortino, G., Hao, Y., Hu, L., Humar, I.: A dynamic service migration mechanism in edge cognitive computing. ACM Trans. Internet Technol. 19(2), 1–15 (2019). https://doi.org/10.1145/3239565 57. Huang, H., Guo, S., Gui, G., Yang, Z., Zhang, J., Sari, H., Adachi, F.: Deep learning for physical-layer 5g wireless techniques: Opportunities, challenges and solutions. IEEE Wirel. Commun. 27(1), 214–222 (2020). https://doi.org/10.1109/mwc.2019.1900027 58. Tureczek, A., Nielsen, P., Madsen, H.: Electricity consumption clustering using smart meter data. Energies 11(4), 859 (2018). https://doi.org/10.3390/en11040859 59. Grover, M., Sharma, N., Bhushan, B., Kaushik, I., Khamparia, A.: Malware threat analysis of IoT devices using deep learning neural network methodologies. In: Security and Trust Issues in Internet of Things, pp. 123–143 (2020). https://doi.org/10.1201/9781003121664-6 60. Wang, Y., Chen, Q., Hong, T., Kang, C.: Review of smart meter data analytics: Applications, methodologies, and challenges. IEEE Trans. Smart Grid 10(3), 3125–3148 (2019). https:// doi.org/10.1109/tsg.2018.2818167 61. Fallah, S., Deo, R., Shojafar, M., Conti, M., Shamshirband, S.: Computational intelligence approaches for energy load forecasting in smart energy management grids: State of the art, future challenges, and research directions. Energies 11(3), 596 (2018). https://doi.org/10. 3390/en11030596 62. Bhattarai, B.P., Paudyal, S., Luo, Y., Mohanpurkar, M., Cheung, K., Tonkoski, R., Zhang, X.: Big data analytics in smart grids: State-of-the-art, challenges, opportunities, and future directions. IET Smart Grid 2(2), 141–154 (2019). https://doi.org/10.1049/iet-stg.2018.0261 63. Ghorbanian, M., Dolatabadi, S.H., Siano, P.: Big data issues in smart grids: A survey. IEEE Syst. J. 13(4), 4158–4168 (2019). https://doi.org/10.1109/jsyst.2019.2931879 64. Dragiˇcevi´c, T., Siano, P., Prabaharan, S.S.: Future generation 5g wireless networks for smart grid: A comprehensive review. Energies 12(11), 2140 (2019). https://doi.org/10.3390/en1211 2140 65. Hossain, E., Khan, I., Un-Noor, F., Sikander, S.S., Sunny, M.S.: Application of big data and machine learning in smart grid, and associated security concerns: A review. IEEE Access 7, 13960–13988 (2019). https://doi.org/10.1109/access.2019.2894819

13 Machine Learning and Deep Learning Models …

187

66. Yongde, G.: A brief analysis of the application of automation in smart power grid. Res. Smart Grid 2(2), 52–56 (2020). https://doi.org/10.35534/rsg.0202009c 67. Mujeeb, S., Javaid, N., Ilahi, M., Wadud, Z., Ishmanov, F., Afzal, M.: Deep long short-term memory: A new price and load forecasting scheme for big data in smart cities. Sustainability 11(4), 987 (2019). https://doi.org/10.3390/su11040987 68. Pallonetto, F., De Rosa, M., Milano, F., Finn, D.P.: Demand response algorithms for smartgrid ready residential buildings using machine learning models. Appl. Energy 239, 1265–1282 (2019). https://doi.org/10.1016/j.apenergy.2019.02.020 69. Lee, K.M., Yoo, J., Kim, S., Lee, J., Hong, J.: Autonomic machine learning platform. Int. J. Inf. Manage. 49, 491–501 (2019). https://doi.org/10.1016/j.ijinfomgt.2019.07.003 70. Almakrami, H.: Intrusion detection system for smart meters. 2016 Saudi Arabia Smart Grid (SASG) (2016). https://doi.org/10.1109/sasg.2016.7849674 71. Zhang, T., Zhong, H., Tan, Z., Yan, X.: Misaka: Interactive SWARM testbed for smart Grid distributed algorithm test and evaluation. 2020 IEEE/IAS Industrial and Commercial Power System Asia (I&CPS Asia) (2020). https://doi.org/10.1109/icpsasia48933.2020.9208421 72. An, D., Yang, Q., Liu, W., Zhang, Y.: Defending against data integrity attacks in smart grid: A deep reinforcement learning-based approach. IEEE Access 7, 110835–110845 (2019). https:// doi.org/10.1109/access.2019.2933020 73. Khamparia, A., Singh, P.K., Rani, P., Samanta, D., Khanna, A., Bhushan, B.: An internet of health things-driven deep learning framework for detection and classification of skin cancer using transfer learning. Trans. Emerg. Telecommun. Technol. (2020). https://doi.org/10.1002/ ett.3963 74. Paramanik, S., Sarkar, P.S., Mondol, K.K., Chakraborty, A., Chakraborty, S., Sarker, K.: Survey of smart grid network using drone & ptz camera. 2019 Devices for Integrated Circuit (DevIC). (2019). https://doi.org/10.1109/devic.2019.8783610 75. Gulyani, V., Dhiman, T., Bhushan, B.: Introducing machine learning to wireless sensor networks. In: Deep Learning Strategies for Security Enhancement in Wireless Sensor Networks Advances in Information Security, Privacy, and Ethics, pp. 1–22 (2020). https:// doi.org/10.4018/978-1-7998-5068-7.ch001 76. Topol, E.J.: High-performance medicine: The convergence of human and artificial intelligence. Nat. Med. 25(1), 44–56 (2019). https://doi.org/10.1038/s41591-018-0300-7 77. Boulos, M.N.K., Peng, G., VoPham, T.: An overview of GeoAI applications in health and healthcare. Int. J. Health Geogr. 18(1), (2019). https://doi.org/10.1186/s12942-019-0171-2 78. Al-Turjman, F., Nawaz, M.H., Ulusar, U.D.: Intelligence in the internet of medical things era: A systematic review of current and future trends. Comput. Commun. 150, 644–660 (2020). https://doi.org/10.1016/j.comcom.2019.12.030 79. Saraswathi, H.S., Rafi, M., Manjunath, K.G., Shankar, A.: Review on computer aided diagnosis of pancreatic cancer using artificial intelligence system. In: Proceedings of the 2020 10th International Conference on Cloud Computing, Data Science & Engineering (Confluence), Noida, India, 623–628 (2020). https://doi.org/10.1109/Confluence47617.2020.9057939 80. Ahad, A., Tahir, M., Yau, K.A.: 5G-Based smart healthcare network: Architecture, taxonomy, challenges and future research directions. IEEE Access 7, 100747–100762 (2019). https:// doi.org/10.1109/access.2019.2930628 81. Tayal, R., Shankar, A.: Learning and predicting diabetes data sets using semi-supervised learning. In: Proceedings of the 2020 10th International Conference on Cloud Computing, Data Science & Engineering (Confluence), Noida, India, 385–389 (2020). https://doi.org/10. 1109/Confluence47617.2020.9058276 82. Tuli, S., Basumatary, N., Gill, S.S., Kahani, M., Arya, R.C., Wander, G.S., Buyya, R.: HealthFog: An ensemble deep learning based smart healthcare system for automatic diagnosis of heart diseases in integrated IoT and FOG computing environments. Futur. Gener. Comput. Syst. 104, 187–200 (2020). https://doi.org/10.1016/j.future.2019.10.043 83. Zhang, Q., Bai, C., Chen, Z., Li, P., Yu, H., Wang, S., Gao, H.: Deep learning models for diagnosing spleen and stomach diseases in smart Chinese medicine with cloud computing. Concurr. Comput.: Pract. Exp. 33(7), 1–1 (2019). https://doi.org/10.1002/cpe.5252

188

T. Bhowmik et al.

84. Zack, C.J., Senecal, C., Kinar, Y., Metzger, Y., Bar-Sinai, Y., Widmer, R.J., Gulati, R.: Leveraging machine learning techniques to forecast patient prognosis after percutaneous coronary intervention. JACC: Cardiovasc. Interv. 12(14), 1304–1311 (2019). https://doi.org/10.1016/j. jcin.2019.02.035 85. Viegas, R., Salgado, C.M., Curto, S., Carvalho, J.P., Vieira, S.M., Finkelstein, S.N.: Daily prediction of ICU readmissions using feature engineering and ensemble fuzzy modeling. Expert Syst. Appl. 79, 244–253 (2017). https://doi.org/10.1016/j.eswa.2017.02.036 86. Such Aparicio, J.M.: Enhancing privacy in multi-agent systems. (n.d.). https://doi.org/10. 4995/thesis/10251/13023 87. Sengupta, P., Paul, S., Mishra, S.: BUDS: Balancing utility and differential privacy by shuffling. In: Proceedings of the 2020 11th International Conference on Computing, Communication and Networking Technologies (ICCCNT). (2020). https://doi.org/10.1109/icccnt49239. 2020.9225470 88. Swanberg, M., Globus-Harris, I., Griffith, I., Ritz, A., Groce, A., Bray, A.: Improved differentially private analysis of variance. Proceedings on Privacy Enhancing Technologies 2019(3), 310–330 (2019). https://doi.org/10.2478/popets-2019-0049 89. Chowdhary, C.L., Patel, P.V., Kathrotia, K.J., Attique, M., Perumal, K., Ijaz, M.F.: Analytical study of hybrid techniques for image encryption and decryption. Sensors 20(18), 5162 (2020). https://doi.org/10.3390/s20185162 90. Bhushan, B., Khamparia, A., Sagayam, K.M., Sharma, S.M., Ahad, M.A., Debnath, N.C.: Blockchain for smart cities: A review of architectures, integration trends and future research directions. Sustain. Cities Soc. 61, 102360 (2020). https://doi.org/10.1016/j.scs.2020.102360 91. Bhushan, B.: Acknowledgment to reviewers of smart cities in 2020. (2021). Smart Cities 4(1), 192–194 (2021). https://doi.org/10.3390/smartcities4010011 92. Kuhn, C., Beck, M., Strufe, T.: Breaking and (partially) fixing provably secure onion routing. In: Proceedings of 2020 IEEE Symposium on Security and Privacy (SP) (2020). https://doi. org/10.1109/sp40000.2020.00039 93. Yang, C., Zhang, M., Jiang, Q., Zhang, J., Li, D., Ma, J., Ren, J.: Zero knowledge-based clientSIDE deduplication for encrypted files of secure cloud storage in smart cities. Pervasive Mob. Comput. 41, 243–258 (2017). https://doi.org/10.1016/j.pmcj.2017.03.014 94. Ijaz, M.F., Attique, M., Son, Y.: Data-driven cervical cancer prediction model with outlier detection and over-sampling methods. Sensors 20(10), 2809 (2020). https://doi.org/10.3390/ s20102809 95. Ijaz, M., Alfian, G., Syafrudin, M., Rhee, J.: Hybrid prediction model for type 2 diabetes and hypertension using DBSCAN-based outlier detection, synthetic minority over sampling technique (SMOTE), and random forest. Appl. Sci. 8(8), 1325 (2018). https://doi.org/10.3390/ app8081325 96. Sharma, P., Shankar, A., Cheng, X.: Reduced papr model predictive control based Fbmc/oqam signal for NB-IoT paradigm. Int. J. Mach. Learn. Cybern. (2021). https://doi.org/10.1007/s13 042-020-01263-8 97. Alfian, G., Syafrudin, M., Ijaz, M., Syaekhoni, M., Fitriyani, N., Rhee, J.: A personalized healthcare monitoring system for diabetic patients by utilizing Ble-based sensors and real-time data processing. Sensors 18(7), 2183 (2018). https://doi.org/10.3390/s18072183 98. Bhushan, B., Sinha, P., Sagayam, K.M., Andrew, J.: Untangling blockchain technology: A survey on state of the art, security threats, privacy services, applications and future research directions. Comput. Electr. Eng. 106897 (2020). https://doi.org/10.1016/j.compeleceng.2020. 106897 99. Schneider, M.: Smart waste management für die smart city. In: Smart city–made in Germany, pp. 373–379 (2020). https://doi.org/10.1007/978-3-658-27232-6_44 100. Yuan, Y., Ye, S., Lin, L.: Process monitoring with support of IoT in prefabricated building construction. Sens.S Mater. 33(4), 1167 (2021). https://doi.org/10.18494/sam.2021.3003 101. Yoon, J., Shin, U.: Effective application of urban renewable energy system for smart energy city: Case study of sejong 5–1 smart energy city. In: 2019 IEEE Transportation Electrification Conference and Expo, Asia-Pacific (ITEC Asia-Pacific) (2019). https://doi.org/10.1109/itecap.2019.8903873

Part III

Applications

Chapter 14

FPGA Based Implementation of Brent Kung Parallel Prefix Adder K. Gunasekaran, D. Muthukumaran, K. Umapathy, and S. A. Yuvaraj

Abstract In addition to evaluating FPGA’s design, this paper aims to achieve the best time reduction possible by enhancing FPGA’s performance and to demonstrate its applicability in reconfigurable high-performance computing. Carry Select Adder is one of the key supplements used in arithmetic operations. A high-speed adder is the VLSI Architecture but at the expense of the area and power. This paper presents VLSI architectures of the chosen adder. The proposed work shows a less time-saving, slightly more areas, and higher-speed adder efficiency between the 16-bit adders with Brent Kung adder. A robust system for machine learning-based optimal adder analysis that connects the prefix adder design synthesis to the final physical design. The work proposed tests are carried out and Xilinx 14.7 simulations are conducted. Keywords Brent Kung adder · Carry select adder · Parallel prefix adder

14.1 Introduction Binary addition has become the main operation in digital systems in most digital systems, such as Digital Signal Processor (DSP), Arithmetic Unit and Microprocessor. The action of the resident adders therefore greatly influences the quick and precise function of a digital system. Due to their widespread use in various core arithmetic operations such as subtraction, multiplication or division, adders are also important factors in digital systems. Therefore, increasing digital adder performance would make the execution of binary activities in a circuit consisting of such blocks significantly advanced. In the past few years, several additional architectures for binary additional features are studied and planned. Parallel Prefix Adder, due to its application in Very Large-Scale Integration chips is highly useful in today’s digital world. In comparison to the traditional adder architectures, improvement K. Gunasekaran (B) · S. A. Yuvaraj GRT Institute of Engineering and Technology, Tiruttani, Tamil Nadu, India D. Muthukumaran · K. Umapathy SCSVMV, Kanchipuram, Tamil Nadu, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_14

191

192

K. Gunasekaran et al.

in PPA performance is observed. Finally, delay, area and power are observed and compared with the Existing adder [1]. The enormous growth of computer and electronic technologies has recently created a part of the today’s integrated systems for different uses, from low-power, low-cost to high-performance coprocessor designs. This growth has increasingly contributed to many key areas including environmental monitoring, automation, transport and logistics, intelligent energy networks, security and monitoring. An integrated system designed to cope with specific functions, including storage, processing and controlling of data in various systems is a kind of multi-computer system. Integrated systems can be found in many standard devices such as mobile phones, payers, digital cameras, etc. Embedded systems can be considered as computer hardware systems with software. This may consist of separate systems or other major system components which carry out certain tasks [2]. Single or multiple processing core systems are controlled by embedded systems. The field programmable door array [3] is one of the most important processing cores. Semiconductor devices consisting of blocks that can be configured and re-established after production (CLBs, also known as LE logic elements). A FPGA logical block can be as straightforward as a transistor or as complex as a microprocessor, and uses multiple combination and sequential logic tasks [4] to be implemented. FPGAs however represent an important development that allows users to implement several functions in a digital hardware design through millions of logic doors and flip-flops. The computer system or digital arithmetic is concerned mainly with the addition, subtraction, multiplication, splitting, square roots, and other arithmetical operations of number systems. Nearly all arithmetical operations are mainly based on additional operations. It can help produce complex computations in practically all processors, for example random generators [5] and the calculation of the largest common divider [6]. Low power applications [7], small scale RSA cryptographer designs [8]. MIMO communications systems and many others, such as the design of the elliptical curve cryptographer (ECC) [9], are also used to build cryptographic coprocessor. Because of this important role, researchers proposed comprehensive options for arithmetic computation and processing. The first fast adder that manipulates the carriage to achieve faster performance to replace ordinary low friction suppliers with more efficient fast suppliers such as CLA [10–12]. In order to finalize distinguishable design alternatives, Parallel Prefix Adders (PEAs), to achieve areas, output and delay efficiencies and also to improve barriers, more precise additional topologies have been proposed [13–17]. PPAs are implemented with high-speed, high-speed arithmetic computing chips (VLSIs), Thus in the current technological architecture PPAs are very useful [18–20]. In practice, various researchers suggested that five PPAs be distributed independently by transmission propagates and signal generations [21]. Cascading multiple adder blocks into a scalable and effective connection allows the implementation of larger adders [22]. Besides the comparison of many art design states, we compare suppliers in the sector (number of logical elements) with the delay.

14 FPGA Based Implementation of Brent Kung Parallel Prefix Adder

193

14.2 Study Work Due to the Adder Carry network, the PPA has advanced architecture over the Adder Carry Look ahead. Parallel prefix suppliers are well known for the best performance in VLSI implementations and are widely used in the high-performance digital circuit operation of Arithmetic Logic Units. The high-speed adjustment operation of PPA, thanks to its advanced carry network, reduces delays and consumption of energy compared to other conventional suppliers. Parallel Prefix Adder is used in a parallel fashion to execute the partial and the final results. The actual stage result at that point depends on the initial input bits. The following consists of three phases to produce the final results in the total structure of the Parallel Prefix Adder. Figure 14.1 explains the steps of the parallel architecture prefix. Pre Processing Block The parallel prefix adder starts with pre-processing and at this stage generates two signals known as signal generation (Gi) and propagation. For every hour of the input signal the generated and propagated signal is calculated, and the following equations represent its operation. Pi = Ai XOR Bi Gi = Ai AND Bi Carry Generation Block

Fig. 14.1 Parallel prefix adder

194

K. Gunasekaran et al.

Fig. 14.2 Block diagram of 8-bit BKA

The main block in the Parallel Prefix Adder since the transport is calculated using a carrying graph before the final results. The transport graph for each adder is different and is therefore calculated on this basis. The diagram is made up of the Black Cell and the Gray cell. Black Cell generates the required signal for the calculation of next phase, which is generated and propagated. Gray cell use is based on previous inputs as shown in Fig. 14.2 to produce only a signal generated. Black Cell In order to generate the signal and spread a set (G, P), the black cell operator receives two sets of signal generation and spread (Gi, Pi) and (Gj, Pj). G = Gi OR (Pi AND Pj) P = Pi AND Pj Gray Cell Two sets of generated and propagated signals are received by the gray operator (Gi, Pi) and one set is computed by (Gj, Pj). G = Gi OR (Pi AND Pj) Post Processing Block This is the last stage of the adder; the total sum is the final result of the adder. One of the parallel adder shapes of the carrying appearance. The BKAs prefix adder is a bit

14 FPGA Based Implementation of Brent Kung Parallel Prefix Adder

195

Fig. 14.3 Logical level for BKA 8-bit carrier network generation

difficult to design because its logic and the O(log2n) gateway depth are the highest. Design uses less transistor numbers and takes less area and speed than other prefix additions. Si = Pi XOR Ci − 1 The delay reduced by the BKA structure without changing adder capacity. BKA is a widely used, popular adder. For simpler construction and less area it avoids explosion of wires. Furthermore, there is a minimum number of fans, where only two are available. However, as shown in Fig. 14.3, it has the highest logical level for BKA 8-bit carrier network generation.

14.3 Results and Discussions In Xilinx ISE 2014.7 RTL synthesis reports, the delays in adder designs are shown in Fig. 14.4. Delay observed in comparison to the proposed Xilinx method for various adders. The delays compared and shown in Table 14.1 were investigated. For Xilinx spartan 3 FPGA, the area of the adder design is measured by the look-up table (LUT). The method proposed is less time-consuming than the previous method in Table 14.1. The simulation result is shown in Fig. 14.5. Out of parallel prefix adders BKA has the best delaying. The synthesis reports show that BKA is better off delayed from the existing parallel prefix adder.

196

K. Gunasekaran et al.

Fig. 14.4 RTL view of proposed method Table 14.1 Comparison of delay and power

Parameter

Power (mW)

Delay (ns)

Utilization of slices LUT’S

Existing method

120

12.28

25

Proposed method

121

10.13

24

Fig. 14.5 Simulation results of proposed method

14 FPGA Based Implementation of Brent Kung Parallel Prefix Adder

197

14.4 Conclusion FPGA’s parallel implementation prefix Brent Kung has been thoroughly investigated in this paper to improve the process of computation. In order to compare the existing method, the performance evaluation of these Parallel Prefix Adders has been studied in terms of time propagation time and area size (in the LUT’S). The most rapid BKA with minimal total FPGA power dissipation among others was discovered, where BKA has optimization in area size. Finally, many other FPGA kits can be synthesized with the proposed PPA implementations. A robust system for machine learning-based optimal adder analysis that connects the prefix adder design synthesis to the final physical design. A prefix adder machine learning based model driven by quasi-random data sampling and featuring structural features and EDA tool settings. In relation to the existing adder, the proposed method is better.

References 1. Gaur, N., Mehra, A., Kumar, P., Kallakuri, S.: 16 bit power efficient carry select adder. In: 2019 6th International Conference on Signal Processing and Integrated Networks (SPIN), pp. 558–561. IEEE, Mar 2019 2. Parmar, S., Singh, K.P.: Design of high speed hybrid carry select adder. In: 2013 3rd IEEE International Advance Computing Conference (IACC), pp. 1656–1663. IEEE, Feb 2013 3. Reddy, A.R.: Multi precision arithmetic adders. In: 2016 International Conference on Computer Communication and Informatics (ICCCI), pp. 1–6. IEEE, Jan 2016 4. Saini, J., Agarwal, S., Kansal, A.: Performance, analysis and comparison of digital adders. In: 2015 International Conference on Advances in Computer Engineering and Applications, pp. 80–83. IEEE, Mar 2015 5. Akila, M., Gowribala, C., Shaby, S.M.: Implementation of high speed Vedic multiplier using modified adder. In: International Conference on Communication and Signal Processing (ICCSP), pp. 2244–2248. IEEE, Apr 2016 6. Gavali, K.R., Kadam, P.: VLSI design of high speed Vedic multiplier for FPGA implementation. In: IEEE International Conference on Engineering and Technology (ICETECH), pp. 936–939. IEEE, Mar 2016 7. Ram, G.C., Lakshmanna, Y.R., Rani, D.S., Sindhuri, K.B.: Area efficient modified Vedic multiplier. In: International Conference on Circuit, Power and Computing Technologies (ICCPCT), pp. 1–5. IEEE, Mar 2016 8. Gaur, N., Tyagi, D., Mehra, A.: Performance comparison of adder architectures on 28 nm FPGA. In: 2nd International Conference on Advances in Computing, Communication, & Automation (ICACCA) (Fall), pp. 1–5. IEEE (2016) 9. Raju, R., Veerakumar, S.: Design and implementation of low power and high performance Vedic multiplier. In: International Conference on Communication and Signal Processing (ICCSP), pp. 0601–0605. IEEE, Apr 2016 10. Suganya, R., Meganathan, D.: High performance VLSI adders. In: 2015 3rd International Conference on Signal Processing, Communication and Networking (ICSCN), pp. 1–7. IEEE, Mar 2015 11. Sendhilkumar, N.C.: Design and implementation of power efficient modified Russian peasant multiplier using ripple carry adder. Int. J. MC Sq. Sci. Res. 9, 154–165 (2017) 12. Sendhilkumar Prasad, N.C., Ramesh, G.P.: Analysis of digital FIR filter using RLS and FT-RLS. In: Solanki, V., Hoang, M., Lu, Z., Pattnaik, P. (eds.) Intelligent Computing in Engineering. Advances in Intelligent Systems and Computing, vol. 1125. Springer, Singapore (2020)

198

K. Gunasekaran et al.

13. Gunasekaran, Ramesh, G.P.: Design of digital FIR filters for low power applications. In: Solanki, V., Hoang, M., Lu, Z., Pattnaik, P. (eds.) Intelligent Computing in Engineering. Advances in Intelligent Systems and Computing, vol. 1125. Springer, Singapore (2020) 14. Gunasekaran, K., Regan, D.: Design of enhanced half ripple carry adder for VLSI implementation of two-dimensional discrete wavelet transform. Int. J. MC Sq. Sci. Res. 8(1), 50–59 (2016) 15. Ramesh, G.P., Prabhu, S.: FPGA implementation of 3D NOC using anti-Hebbian for multicast routing algorithm. In: Sharma, D.K., Son, L.H., Sharma, R., Cengiz, K. (eds.) MicroElectronics and Telecommunication Engineering. Lecture Notes in Networks and Systems, vol. 179. Springer, Singapore (2021) 16. Tapasvi, B., Sinduri, K.B., Lakshmi, B.G.S.S.B., Kumar, N.U.: Implementation of 64-bit Kogge Stone carry select adder with ZFC for efficient area. In: IEEE International Conference on Electrical, Computer and Communication Technologies (ICECCT), pp. 1–6. IEEE, Mar 2015 17. Ramesh, G.P.: Design of digital FIR filters for low power applications. In: Intelligent Computing in Engineering, pp. 433–440. Springer, Singapore (2020) 18. Adilakshmi Siliveru, M.B.: Design of Kogge-Stone and Brent Kung adders using degenerate pass transistor logic. Int. J. Emerg. Sci. Eng. 19. Hemanth Kumar, G., Saravanan, M.: Design and implementation of 10-bit pseudo random sequence generator for 50 MHz. Int. J. Eng. Tech. 2(3), 1–9 (2016) 20. Shamsudheen, S., Mubarakali, A.: Smart agriculture using IOT. Int. J. MC Sq. Sci. Res. 11(4) (2019) 21. Yezerla, S.K., Naik, B.R.: Design and estimation of delay, power and area for Parallel prefix adders. In: Recent Advances in Engineering and Computational Sciences (RAECS), pp. 1–6. IEEE, Mar 2014 22. Saxena, P.: Design of low power and high speed carry select adder using Brent Kung adder. In: International Conference on VLSI Systems, Architecture, Technology and Applications (VLSI-SATA), pp. 1–6. IEEE, Jan 2015

Chapter 15

Vehicle Entry Management System Using Image Processing R. Vallikannu, Krishna kanth, L. SaiPavan Kumar, Monisha, and Karthik

Abstract Many organizations or Institutes aim to implement a paid parking system for stakeholders such as employees, students who possess a vehicle to park inside the campus premises. In Most of the scenarios, the verification of the pass is done manually by the security guards and has increased traffic congestion at peak working hours. Moreover, there is no asset monitoring system prevailing to provide information about the total vehicles inside the campus and optimize the parking slots accordingly. Parking slots are one of the few areas that are uncovered by CCTV surveillance, which requires more security aspects. Therefore, the research context prevails that there is a mere need for an automated system for vehicle entry management in such organizations or public parking areas. Hence, this project aims to implement a system to check the entry status of the vehicle automatically. The verification process is based on the recognition of the number plate of the vehicle by using image processing based on OpenCV and Raspberry pi, comparing with the pre-existing stored information in the database. The entire details of the registered and unregistered vehicles can be monitored through the webpage by respective authorities. Additional reminders such as renewal of entry pass, expiry of license, vehicle pollution control check, vehicle insurance is also provided to the owners of registered vehicle. Also, we can append the data of the new vehicle into the database with a timestamp. The details of the both registered and non-registered vehicles are recorded that includes vehicle number, some personal details of the vehicle owner such as license number, insurance, mobile number, etc. It also intimates the registered vehicle owners regarding the renewal of the pass-through papers such as insurance, driving license renewal, updating of mobile number, Aadhaar number and vehicle number with the GPS tracker using the mobile based text message. This machine learning network helps in the vehicle owner to know each and every movement of their vehicle and acts as personnel reminder. Keywords OpenCV · Pytesseract · Number plate recognition · Webserver R. Vallikannu (B) · K. kanth · L. S. Kumar · Monisha · Karthik Hindustan Institute of Technology and Science, Electronics and Communication Engineering, Chennai, TN, India e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_15

199

200

R. Vallikannu et al.

15.1 Introduction In the modern era, the usage of automobiles has been increasing day by day. With the increase of these vehicles, there is a need for surveillance of these vehicles for the protective purpose. The need for the surveillance is mainly used in colleges, apartments, public places like shopping malls, parking areas etc. where a large gathering of vehicles takes place. There are many ways to monitor these vehicles, one of the best methods is to monitor through the license plate of the vehicle. A license plate is also known as a vehicle registration plate which is attached to the both sides of the vehicle (i.e., front and back). A license plate is an official identification provided to the vehicle by the government. The license plate varies from country to country. The Indian license is provided with a unique code for every vehicle consisting of state code, district code, where the vehicle is registered and an alphanumeric unique code, e.g. AP 23 PV 1381. The first two characters of an Indian license plate are denoting the union territory or state, of which there are 29 in India. The next two digits indicating the regional transport authority within the state followed by one or two letters showing the current series of the registration numbers and the final four numbers are unique to the plate. During the registration of these license plates, the owner is required to provide all the basic details (personal details), so with these license plates (number), the details of the owner of the vehicle can be traced easily. Since the license plate is unique from vehicle to vehicle, tracking of the vehicles through license is an easy task to monitor these without any difficulty. Verification through license Plate mode is opted in this project, to verify the entry status of the Vehicle and also the parking payment by the users. Here we have provided a Registration form for the user in order to register his vehicle during the payment process, autonomous monitoring of the payment status of the vehicle is carried out with the license number in this project as shown in Fig. 15.1. Here a designed webpage is used for registration by the users, to check the validity of the payment, to monitor the vehicles in the premises by the authorities.

15.2 Related Work This works on the implementation of image (jpeg/png) to text conversion and to create separate text file that consists of extracted information from the captured image. It uses Open CV and tesseract for image processing and text extraction. The author deals with extraction of the number from the number plate with various font sizes, length, and width. The limitations of this work include the quality of the original image precariousness and the lighting. This article is mainly for the application of security and surveillance which includes theft vehicles and Road Traffic monitoring [1]. It compares the extracted data with the pre-existing data in the database to find out whether the vehicle belongs to that person or not. The detection of the number plate is done by Optical Character

15 Vehicle Entry Management System Using Image …

201

Fig. 15.1 Proposed methodology

Recognition (OCR) using MATLAB. The limitation of this work is that the number plate images of Vehicles with white colored plates should be crystal clear. This system deals with the gate automation based on the Number Plate Recognition with Raspberrypi [2]. It uses OpenCV and Optical Character Recognition (OCR) platforms. It also uses ultraviolet sensor for calculating the distance between cameras to the vehicle. The limitations of this article is temporary storage of vehicle data. This work is designed a system of vehicular number plate recognition which is used for the purpose of security system [3]. It considered the detection of number plate of a vehicle using Image processing technique and further used to store the data, allows the entry of the vehicles. The work incorporated template matching algorithm and is implemented at the entrance of highly restricted areas. This work focused on design of an efficient automatic authorized number plate identification system [4]. The number plate detection was done by using image and video segmentation and OCR for text extraction. This system was designed using technologies like OpenCV Tensor Flow, mongo DB. This article has not concentrated on the live implementation process with digital advanced cameras and asset system management. By deployment of intelligent systems huge amount of data is collected and the data collected can be analyzed in the fog. The results of the analysis can be sent to cloud. Various parameters can be monitored and shared by people across the globe as discussed in [5]. Moreover various algorithms [6, 7] have been suggested to effectively manage and handle the data with minimal cost. A hierarchical approach of mobile agent based layers to optimize load balancing was developed for a large scale network, which plays a momentous role in reduced

202

R. Vallikannu et al.

energy [8]. The network lifetime can be increased by implementing energy efficient algorithms for routing [9]. Image comparison is essential in image processing [10– 12]. The multi-stage power conversion system is reduced through introduction of a nearly z-source switched-inductor that produces the same step-up voltage as an NFSBB converter [11]. This article dealt with the design and development of effective image processing techniques and algorithms to localize the license plate in the captured image and to divide the characters from it and identify each character of the segment [9]. It was implemented using OpenCV library and KNN Algorithm through Python Programming Language. This system also included various operations such as taking pictures, localizing the number pad, truncating characters and OCR from alphanumeric characters. This system has considered various applications like security, highway speed detection, theft vehicles etc. [12–14]. However, this article fails to detect exact number plate area by using the shape analysis and the detection of number plate is not working efficiently in bright light conditions.

15.3 Proposed Method A system is implemented for the entry of the vehicle. First, a database has been created which consists of the details of the students. Here the vehicles are divided into two categories. The vehicles which have paid the parking fee is considered as registered vehicles and those didn’t pay the fee and the outside vehicles (cabs, school buses, water tankers etc.) are considered as non-registered vehicles. When a vehicle enters the campus through the main gate where an optical camera is attached, detects the number plate of the vehicle. With the help of Image Processing steps, the obtained input is processed and the text from the license plate image is extracted. The extracted number is compared with the pre-existing data which is stored in the database. Here the comparison is for the license plate. If the license number of the extracted plate matches with the pre-existing data, then the system gives a green signal indicating that the vehicle is a registered vehicle or else it gives a red indication which is an unregistered vehicle. The detection of a license plate from the image of a vehicle is based on a general approach which is nothing but step by step procedure. It is an effective object detection approach. To implement this, OpenCV library is used. The text conversion from the detected number plate image is done using PYTESSERACT. If a student wants to register his/her vehicle, the registration access is given to the security guards. New vehicle registrations can be done by the security guards in a designed website. The website can be accessed by the students, security guards and the higher authorities. Through this website each vehicle details can be monitored with their time stamps.

15 Vehicle Entry Management System Using Image …

203

15.4 Implementation The proposed methodology considers an input image taken from the camera. The image must be clear to extract the vehicle number from the image which is taken as input. The brightness and the contrast should be clear and the number plate must be in the Indian government format. The following steps are used in the proposed methodology as shown in Fig. 15.1.

15.4.1 Image Capture from Camera Firstly, an image from a camera is captured which is in RGB format. The following image processing techniques will be performed on this RGB image to get the vehicle number plate from the image. The function used to read the image is given as follows.

15.4.2 RGB to Grey Conversion The RGB is nothing but the true colour of an image. It refers to the ‘Red, Green, and Blue’ is a combination of three different shades of light that mix together that generates different colours. A grey scale image consists of only the shades of Gray. It does not have any other colour information. There is a mere need for converting an RGB into a GRAY image in image processing. An RGB image contains a lot of information that may not be required for the processing. During this process, there is no loss of information, i.e. if this conversion takes place, there will be only loss in information related to colour but not in the actual information.

15.4.3 Blurring The concept of blurring is to make image looks sharper or more detailed that come to the knowledge that all the objects and their shapes to fix correctly in it. Edges define the shapes of the objects; Blurring makes the transition from one colour to the other very smooth. a bilateral filter is used, which is a non-linear filter (i.e. the change of output is independent of change in input), edge-preserving (it is an image processing technique which soothes noise while continue to have sharp edges), and a noise reducing (the process of removing various noises from a signal, these kind of algorithms change the signals to greater or lesser degree). The bilateral filter dissolves the noise and also soothes away the edges.

204

R. Vallikannu et al.

Fig. 15.2 Flow chart

15.4.4 Edge Detection An edge is a curve that contains a rapid change in the image intensity; these are associated with the boundaries of objects. They occur on the boundary between two different regions in an image. Edge detection is one of the image processing techniques. This technique helps to find the boundaries of objects within the images. This method uses two different thresholds, to detect strong and weak edges and includes the weak edges in the output if and on condition that they are connected to the strong edges. Here, there is a less chances that it is affected by noises and more likely to detect weak edges. The canny edge detection algorithm is executed with 5 steps as shown in Fig. 15.2.

15.4.5 Finding and Drawing Contours Contours are explained as the curve that joins all the continuous points including the boundaries that are having the same colour or intensities. There is not much difference between contours and edges. Contours are obtained from edges, but they need to be closed curves. Contours play an important role in image processing. To

15 Vehicle Entry Management System Using Image …

205

get better accuracy in finding contours, binary images are used. It is essential to apply threshold or canny edge detection to the image before finding contours for an image. The contours are the integer values that represent the boundary points of the object. Since these contour points are invisible, to make sense, these boundary points are drawn to identify the contour part.

15.4.6 License Plate Detection and Text Extraction The next task is to crop the contoured part. It is accomplished by drawing the suitable shape across the contoured part and then finding the coordinates of their corners and then crop the main image with these Corners as coordinates. Now the license plate image from the main image is extracted. To extract text from that image, pytesseract is used. It is an optical character recognition technique to identify text in an image. It produces output in a text format.

15.4.7 Comparing Vehicle Number with Database The proposed methodology is based on the vehicle number that is being extracted from the previous steps. Here, the system checks for the vehicle number in the data base. Upon vehicle number found in the database, it serves as an indication that the vehicle has been registered or else there is an option given to create that vehicle number database. This facility is only accessed by the security guards only.

15.4.8 Website Development It is a set of webpages and the related content with a selected domain and published on a minimum of one web server, the website can be accessed by any of the students in the campus, security guards and the higher officials. Each one of them have given separate access to login into their respective accounts. The Home page of this site consists of three login blocks. One for students, one for security guards and the other for admins as shown in Fig. 15.3. An account has to be created for the security guards with suitable username and password before logging in. For signup the accounts, a separate registration page is provided for every category.

206 Fig. 15.3 Home page and student page

Fig. 15.4 Admin page

R. Vallikannu et al.

15 Vehicle Entry Management System Using Image …

207

Fig. 15.5 Hardware setup

15.4.9 Hardware Setup The interfacing of hardware includes connection with the Raspberry pi module with the Raspberry pi camera and with the Relay module. The camera is connected to the zip 15 connector in the raspberry pi and the relay is connected to the GPIO pins of the raspberry pi. The hardware setup includes the Raspberry pi module, Relays, and the Raspberry pi camera. The Raspberry pi camera is connected to the raspberry pi module. The camera consists of flat ribbon cable known as CSI cable used for the connectivity. One end of the CSI cable looks in the blue color and the other end is of silver notches. The cable is plugged to the zip 15 connector in the raspberry pi module facing the blue part towards the audio jack. Once the interface is done, the camera module is get installed into the raspberry pi as shown in Fig. 15.5. To connect Relay to the Raspberry pi module, first the configuration of GPIO pins is to be done and then connect the GND and 5 V pin of the Raspberry pi to the GND and 5 V pin of the Relay Module respectively, then connect IN1-IN4.With GPIO pins that are set in the GPIO configuration.

15.4.9.1

Flow Chart

15.5 Results The Vehicle Number plate gray scale image is shown in Fig. 15.6, edge detected image with high resolution is shown in Fig. 15.7. The blurred image is shown in Fig. 15.8 and the colored image is shown in Fig. 15.9 with appropriate algorithm, the number plate is detected as shown in Fig. 15.10.

208 Fig. 15.6 Gray-scale image

Fig. 15.7 Edge detection

Fig. 15.8 Blurring

R. Vallikannu et al.

15 Vehicle Entry Management System Using Image …

209

Fig. 15.9 Finding color

Fig. 15.10 Number plate detection. OUTPUT: HR26DA2330

15.6 Conclusion In this project, the automatic vehicle identification system using vehicle license plate is presented that aids the machine learning based embedded system using a series of image processing techniques. The vehicle management system helps in identifying the vehicle movement using the license plate number it can detect the registered vehicle or yet to be registered data from the database stored in the webpage. It uses OpenCV software and Raspberry pi for the identification and recognition of the number plate and its performance is tested on real images. This also helps preventing the vehicle threat and tracking the vehicle in any undescriminable situations using the mobile or computer (IoT) based security alert and also for administering the vehicle database. OpenCV gives the best result in vehicle plate detection consists of more function in computer vision than Matlab. For the text recognition we have used Pytesseract which gives us the better results.

15.7 Future Works This system can be implemented in public places like shopping malls, IT companies, private organizations, Movie theatres etc., for the number plate detection Convolution Neural Networks (CNN), K-Nearest Neighbour (KNN) can be used through Deep

210

R. Vallikannu et al.

Learning can be used which gives more accurate and less complex time results. With the own trained classifiers in the Deep Learning algorithms the number plate of the bikes can also be detected. Cloud Computing can also be used for the permanent storage of the data.

References 1. Singh, A., Gupta, A.K., Singh, A., Gupta, A., Johri, S.: Number plate detection using image processing. Int. Res. J. Eng. Technol. (IRJET) 5(3) (2018). ISSN: 2395-0072 2. Kumthekar, A.V., Owhal, S., Supekar, S., Tupe, B.: Recognition of vehicle number plate using raspberry pi. Int. Res. J. Eng. Technol. (IRJET), 5(4) (2018). ISSN: 2395-0072 3. Sutar, G.T., Shah, A.V.: Number plate recognition using an improved segmentation. Int. J. Innov. Res. Sci., Eng. Technol. 3(5) (2016). ISSN: 2319-8753 4. Santoshmanojkumar, B., Prasad, M.V.K., Sripath Roy, K.: University campus number plate logging system. Int. J. Innov. Technol. Explor. Eng. (IJITEE). 8(7) (2019). ISSN: 2278-3075 5. Kavitha. B.C., Vallikannu. R.: IoT Based intelligent industry monitoring system using Raspberry-Pi. In: Proceedings of the 6th IEEE International Conference on Signal Processing and Integrated Networks SPIN (2019), 7th and 8th March, Amity University, Noida (2019) 6. Kavitha, B.C., Vallikannu, R.: Delay-aware concurrent data management method for IoT collaborative mobile edge computing environment, Microprocessors and microsystems, Elsevier (2020) 7. Rajan, A., Ramesh, G.P.: Glaucoma detection in optical coherence tomography images using undecimated wavelet transform. 7, 878–885 (2016) 8. Vallikannu, A.G., Srivatsa, S.K.: Dynamic and secure joint routing and charging scheme with mobile power back ferry nodes in mobile adhoc networks. Indian J. Sci. Technol. 9(33) (2016) 9. Vallikannu, R., George, A., Srivatsa, S.K.: A novel energy consumption model using Residual Energy Based Mobile Agent Selection Scheme (REMA) in MANETs. In: Proceedings of the 2nd IEEE International Conference on Signal Processing and Integrated Networks (SPIN), New Delhi, pp. 334–339 (2015). https://doi.org/10.1109/SPIN.2015.7095410 10. Kaur, S., Kaur, S.: An efficient approach for automatic number plate recognition system under image processing. Int. J. Adv. Res. Comput. Sci. 5(6) (2014). ISSN:0975-9646 11. Satpathy, R.B., Ramesh, G.P.: Advance approach for effective EEG artefacts removal. In: Balas, V., Kumar, R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent Systems Reference Library, vol. 172. Springer, Cham (2020) 12. Narayanan, K.L., Ramesh, G.P.: VLSI Architecture for multi-band wavelet transform based image compression and image reconstruction. J Eng Appl Sci 12, 6281–6285 (2017) 13. Karthik, S., Annapoorani, V., Dineshkumar, S.: Recognition and tracking of moving object in underwtaer sonar images. Int. J. MC Sq. Sci. Res. 8(1), 93–98 (2016) 14. Narayanan, K., Ramesh G.: Discrete wavelet transform based image compression using frequency band suppression and throughput enhancement. vol. 9, No. 2, pp. 176–182 (2017)

Chapter 16

A Non-negative Matrix Factorization for IVUS Image Classification Using Various Kernels of SVM S. P. Vimal, M. Vadivel, V. Vijaya Baskar, and V. G. Sivakumar

Abstract Intravascular ultrasound (IVUS) is a medical methodology and it is a specially constructed catheter with a miniaturized Ultrasound Probe attached to the distal end of the catheter is a medical imaging technique. An efficient method for IVUS image classification using Non-Negative Matrix Factorization (NNMF) and various Support Vector Machine (SVM) kernels are presented in this study. The input IVUS images are given to NNMF for feature extraction and stored in feature database. Finally, SVM kernels like linear, polynomial, quadratic and Radial Basis Function (RBF) are used for prediction and classification of coronary artery lesions and IVUSbased ML algorithms shows good diagnostic performance for identifying ischemiaproducing lesions. An IoT based alert is given to the patient’s database cloud that has information of self or blood relation to alert messages in case of emergency artery disease using wearable sensors. The system produces the classification accuracy of 94% by using NNMF and different SVM kernels. Keywords IVUS image classification · Non-negative matrix factorization · Support vector machine · Kernels

S. P. Vimal (B) Department of Electronics and Communication Engineering, Sri Ramakrishna Engineering College, Coimbatore, Tamil Nadu 641022, India e-mail: [email protected] M. Vadivel Vidya Jyothi Institute of Technology, Hyderabad, India V. V. Baskar School of EEE, Sathyabama Institute of Science and Technology, Chennai, Tamil Nadu, India V. G. Sivakumar Department of Electronics and Communication Engineering, Vidya Jyothi Institute of Technology, Hyderabad, Telangana, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_16

211

212

S. P. Vimal et al.

16.1 Introduction A computerized ultrasound is connected to the proximal end of the catheter. IVUSbased atherosclerotic plaque characterization with feature selection and SVM classification [1]. The features are determined using multiple window sizes to change the values of the different patterns in the region. Assessment of IVUS images identification techniques [2]. Classification of diseases using the determination of artery cross section layers, which is the adventiveness, media, and lumen layer. Automatic classification and distinguish of IVUS and texture characteristics of atherosclerotic lesions in swines [3]. Texture steps have been used to minimize measurements, followed by a main component analysis. The research dataset was evaluated by two independent experts and the findings were compared. Genetic, IVUS tissue characterisation systems rule-based classification schemes [4]. Increase class discrimination, a rich array of textural features comes at various scales that include first-order statistics, co-occurrence matrices of gray rates, run lengths, waves, local binary patterns. Automatic identification in IVUS image using a Cascade of Classification Stents [5, 6]. GentleBoost Cascade for identification of stent struts using structural features in order to code the details on the different sub-regions of struts. The IVUS images are fitted with a frost filter to eliminate the noise generated by ultrasound waves in the imaging technology. Automated coronary stent identification in IVUS pictures by using the classificator cascade [7–12]. Cascade of Gentle Boost classifiers for stent struts to identify the separate sub-regions of strutes with structural features. A non-negative matrix factorization for IVUS image classification using various kernels of SVM is described. Section 16.2 describes the methods and materials used for IVUS image classification. Section 16.3 describes the experimental result and discussion. The last section concludes the IVUS image classification.

16.2 Methods and Materials Initially, the input IVUS images are given to NNMF for feature extraction. Then different kernels in SVM like linear, polynomial, quadratic and RBF for prediction. The work flow of proposed system is shown in Fig. 16.1.

Input IVUS images

Fig. 16.1 Workflow of proposed system

NNMF feature extraction

Different SVM kernel classification

16 A Non-Negative Matrix Factorization …

213

Fig. 16.2 NNMF matrix order

16.2.1 NNMF Feature Extraction NNMF is a category of multivariate algorithms and linear algebra with a factor of matrix V in (usually) two matrices W and H, the property of which is the absence of all three matrixes. This negative effect promotes inspection of the resulting matrices. Non-negativity is often fundamental to the study of data in applications such as audio spectrograms processing or muscular activity. As the problem is not necessarily resolvable precisely, it is normally numerically approximated. Figure 16.2 shows the matrix order of NNMF.

16.2.2 SVM Kernels Classification SVM algorithms use a set of kernel defined math functions. The kernel’s job is to input data and convert it into the appropriate form. Different SVM algorithms are using different kernel functional forms. There can be various kinds of functions [7, 8]. The function of kernel is to take data as input and transform it into the required form. Figure 16.3 shows the SVM kernel functions.

16.3 Results and Discussions The performance of IVUS image classification is measured in terms of accuracy, sensitivity and specificity. Table 16.1 shows the performance of IVUS image classification system using NNMF and SVM kernels. From Table 16.1 it is observed that the overall classification accuracy is 94% obtained by the SVM-RBF kernel by using NNMF factor P value is 72 and its computation time is 5.07.

214

S. P. Vimal et al.

Fig. 16.3 SVM kernel function

Table 16.1 Shows the performance of IVUS image classification system using NNMF and SVM kernels

P

NMFF computation time (seconds)

Classification accuracy (%)

16

5.14

77

30 44

8.01 5.04

83 88

58

5.35

91

72

5.07

94

16.4 Conclusion An efficient method for IVUS image processing and classification of any infected coronary artery lesion using NNMF and different SVM kernels using ML algorithm helps in early study of the disease infected artery. The IoT used with this processing method helps in alerting the patients in case of any artery infections are detected is described in this study. Initially, the NNMF is given for feature extraction with the Pfactor values and computation time. At last, different SVM kernels like linear, RBF, quadratic and polynomial kernels are used for early prediction and classification of coronary disease aiding appropriate treatment for the cure of disease. The system yields the overall classification accuracy of 94% by using NNMF and different SVM kernels.

References 1. Giannoglou V.G., Stavrakoudis D.G., Theocharis J.B.: IVUS-based characterization of atherosclerotic plaques using feature selection and SVM classification. In: 2012 IEEE 12th International Conference on Bioinformatics & Bioengineering (BIBE), pp. 715–720. IEEE (2012)

16 A Non-Negative Matrix Factorization …

215

2. Sridevi S., Sundaresan M.: Evaluation of classification techniques for IVUS images. In: 2019 6th International Conference on Computing for Sustainable Global Development (INDIACom), pp. 998–1002. IEEE (2019) 3. PBrathwaite P., Nagaraj A., Kane B., McPherson D.D., Dove E.L.: Automatic classification and differentiation of atherosclerotic lesions in swine using IVUS and texture features. In: Computers in Cardiology, pp. 109–112. IEEE (2002) 4. Giannoglou V.G., Stavrakoudis D.G., Theocharis J.B., Petridis V.: Genetic fuzzy rule-based classification systems for tissue characterization of intravascular ultrasound images. In: 2012 IEEE International Conference on Fuzzy Systems, pp. 1–8. IEEE (2012) 5. Rajan, A.: Classification of intravascular ultrasound images based on non-negative matrix factorization features and maximum likelihood classifier. Int. J. Adv. Signal Image Sci. 4(1), 16–22 (2018) 6. Rotger D., Radeva P., Bruining N.: Automatic detection of bioabsorbable coronary stents in ivus images using a cascade of classifiers. IEEE Trans. Inf. Technol. Biomed. 14(2), 535–537 (2009) 7. Rajan A., Ramesh G.P.: Automated early detection of glaucoma in wavelet domain using optical coherence tomography images. Biomed. Pharmacol. J. 8(2) (2015) 8. Kumarapandian, S.: Melanoma classification using multiwavelet transform and support vector machine. Int. J. MC Sq. Sci. Res. 10(3), 01–07 (2018) 9. Narayanan K.L., Ramesh G.P.: Discrete wavelet transform based image compression using frequency band suppression and throughput enhancement. Int. J. MC Sq. Sci. Res. 9(2), 176– 182 (2017) 10. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of IOT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 11. Hemalatha R.J., Vijaybaskar V., Thamizhvani T.R.: Automatic localization of anatomical regions in medical ultrasound images of rheumatoid arthritis using deep learning. In: Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine (2019) 12. Flashy A.M., Ramesh G.P.: Multi band antenna system for quality evaluation application of apple fruit. In: Balas, V., Kumar, R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent Systems Reference Library, vol. 172. Springer, Cham (2020)

Chapter 17

Novel Approach to Monitor the Respiratory Rate for Asthma Patients V. G. Sivakumar, S. P. Vimal, M. Vadivel, and V. Vijaya Baskar

Abstract In the smart health care sector the exponentially increasing Internet of Things (IoT) technology with increasing patients interest have played an important role. It is found to be the fact that, India has 18% of the world population and an rising tension of chronic respiratory diseases. There is no proper understanding about the wide spreading respiratory diseases and their rapid updates are not available for states across the border of modern India. So, in order to hamper the death toll I have proposed this system which continuously monitors the Asthma patients using WSN. This proposed system is based on flex sensor with a controller. This circuit installed in a waist belt, which can monitor the breathing pattern of the patient continuously. Due to the variation in the value of the flex sensor the serial monitor displays the live status of the patients. At times of any abnormalities a SMS is triggered to the kin and friends of the patients. The suggested Respiratory Rate Monitoring System was checked and assessed on satisfactory findings. Keywords Respiratory rate · Flex sensor · Arduino · Asthma patients

17.1 Introduction Asthma affects up to 334 million people around the world, and its been rising incidence for all the past three decades. It effects all genders, ethnic groups, but there is broad difference inside the same nation in various countries and in different communities. This is the most prevalent genetic disease in kids, and is more serious in un-affluent kids [1]. Drowsiness is considered to be expressed in the behavior not V. G. Sivakumar (B) · M. Vadivel Vidya Jyothi Institute of Technology, Hyderabad, Telangana, India S. P. Vimal Sri Ramakrishna Engineering College, Coimbatore 641022, India e-mail: [email protected] V. V. Baskar Sathyabama Institute of Science and Technology, Chennai, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_17

217

218

V. G. Sivakumar et al.

only of the nervous system but also of the autonomic nervous system (ANS). Therefore, by analyzing the ANS activity, it is considerable to estimate somnolence [2]. Prior to this, the PRC was used as a way of determining HRV. [I] A powerful method for HRV research, because it demonstrates changes in heart rate and rhythm during ventilation [3, 4]. The respiration rate (BR) is described as the total breaths taken during a one-minute span and calculated when an individual is at rest. BR and HR are important signs to identify symptoms of cardiovascular diseases such as heart attack or asthma [5–9]. The dissemination of breath analysis as a screening and monitoring method is decelerated by the device’s expense, that can be managed by experienced physicians, and the shortage of systematic protocols for breath lab analysis. The International Association for Breath Testing (IABR) is working closely towards the concept of uniform breath sampling and analytical procedures [10–13]. Respiratory rate is among the main indicators of symptoms of a persons health. That is the respiratory rate of a human at ease and measured in respirations [14–16]. Detector is centered on a plastic optical fiber (POF) in conjunction with the sufferer’s torso and properly calibrated to be sensitive to normal breathing activity-induced malformations [17, 18]. This article describes a way to measure a person’s respiration levels through thermography for Respiratory Alkalosis detection. In the spectral images we measure inhalation and exhalation to monitor the level of breathing [18–20]. In the above article we have used both an embedded device and a Camera to develop an embedded body breath detection monitoring system (EMSFBBD) that tracks, tracks the sufferer’s breath and transmits the data on the internet to a particular database (Figs. 17.1 and 17.2). POWER SUPPLY

FLEX SENSO R

ARDUINO UNO GPI

SERIAL MONITOR

BUZZER

GSM MODULE UAR

Fig. 17.1 Block diagram of the proposed method

17 Novel Approach to Monitor the Respiratory Rate for Asthma Patients Fig. 17.2 Flow chart of the proposed method

219

START

READ FLEX VALUE

Print the value in Serial Monitor

Yes Normal Value

No

Triggers GSM Alert

17.2 Flow Chart 17.3 Proposed System As shown in the above block diagram Arduino UNO, Flex Sensor, GSM Module and Buzzers are used. Flex Sensor is placed in the abdomen of a patient who is suffering from a respiratory problem like Asthma or immobile patients. There are two set of values obtained from the flex sensor. One set of values obtained with the patients with a good breathing pattern. And the other one with the patients who struggles to make out a proper respiration. So once the proposed prototype is attached with the people with above mentioned problems it starts monitoring their breathing patterns continuously. If a person breathes normally his monitored value will be printed in the serial screen and again controller starts looking for the value. Whereas if a person doesn’t breathe or his pattern looks somewhat similar to the second set of pattern then the value is monitored and the controller triggers a SMS alert from the GSM module. Here the controller commands the GSM module to go through a certain set of at commands which sends the message to the user. Finally, one can continuously monitor the respiratory rate of an Asthma patients using the proposed method (Fig. 17.3).

220

V. G. Sivakumar et al.

Fig. 17.3 Shows breathing pattern of humans

17.4 Results and Discussions Figure 17.4 shows the flex sensor placement in the body which can be placed in waist as well as ribcage in which respiration movement can be widely seen. Figure 17.5 shows the Slow Breathing Pattern of the patient monitored through this flex sensor and we can quite clearly see breathe per minute is found to be 10. It is plotted between ribcage and volume. Figure 17.6 shows the Medium Breathing Pattern of the patient monitored through this flex sensor and we can quite clearly see breathe per minute is found to be 20. It is plotted between ribcage and volume. Figure 17.7 shows the Fast Breathing Pattern of the patient monitored through this flex sensor and we can quite clearly see breathe per minute is found to be 40. It is plotted between ribcage and volume.

17 Novel Approach to Monitor the Respiratory Rate for Asthma Patients

221

Fig. 17.4 Sensor placement

Fig. 17.5 Slow breathing pattern

Fig. 17.6 Medium breathing pattern

17.5 Conclusion This proposed prototype will reliably calculate the breathing pattern or respiratory rate of a person as opposed to portable respiratory rate monitoring devices based on

222

V. G. Sivakumar et al.

Fig. 17.7 Fast breathing pattern

the temperature probe and MEMS sensor. This device will monitor the temperature and respiratory rate of the body and the data it has received from the IOT network to android applications wirelessly. Emerging technology that will revolutionize the health sector is IOT-based health surveillance. Patient health status is tracked through various procedures and technologies within the hospital-centric healthcare service. It is important to remember that medical gadgets are very costly and these gadgets rarely have functions for monitoring the respiratory rate. And suggested systems can be used in emergency care units in hospitals. As tested so far, gadgets or devices based on accelerometer and thermistor provide inaccurate readings, but this issue has been significantly reduced by the model examined in this article. Therefore, the proposed program meets all of the above-mentioned objectives.

References 1. Igasaki T., Kagasawa N., Murayama N., Hu Z.: Drowsiness estimation under driving environment by heart rate variability and/or breathing rate variability with logistic regression analysis. In: 2015 8th International Conference on Biomedical Engineering and Informatics (BMEI), pp. 189–193. IEEE (2015) 2. Cordero D., Tapp W., Bang P., Reisman S.: Phase response curve analysis of heart rate variability with differing breathing rates. 0-7803-0925-l/93W3. IEEE (1993) 3. Rosmina J., Rozali M.A.A.: Estimation of breathing rate and heart rate from photoplethysmogram. In: 2017 6th International Conference on Electrical Engineering and Informatics (ICEEI), pp. 1–4. IEEE (2017) 4. Noguchi Y., Dwyer G., Szeto H.H.: System function between breathing rate and heart rate in fetal sheep. In: Images of the Twenty-First Century. Proceedings of the Annual International Engineering in Medicine and Biology Society, pp. 731–732. IEEE (1989) 5. Lomonaco T., Salvo P., Ghimenti S., Biagini D., Bellagambi F., Fuoco R., Di Francesco F.: A breath sampling system assessing the influence of respiratory rate on exhaled breath composition. In: 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 7618–7621. IEEE (2015) 6. Alam M., Hussain M., Amin A.: A novel design of a respiratory rate monitoring system using a push switch circuit and Arduino Micocontroller. In: 2019 International Conference on Robotics, Electrical and Signal Processing Techniques (ICREST), pp. 470–473. IEEE (2019)

17 Novel Approach to Monitor the Respiratory Rate for Asthma Patients

223

7. Alberto V., Carullo A., Casalicchio M.L., Penna A., Perrone G., De Vietro N., Milella A., Fracassi F.: A plasma modified fiber sensor for breath rate monitoring. In: 2014 IEEE International Symposium on Medical Measurements and Applications (MeMeA), pp. 1–5. IEEE (2014) 8. Anushree B., Dasgupta A., Routray A.: A thermographic method for detecting respiratory alkalosis by monitoring breath patterns. In: 2016 International Conference on Systems in Medicine and Biology (ICSMB), pp. 26–30. IEEE (2016) 9. Kumar, G.H., Ramesh, G.P.: Novel gateway free device to device communication technique for IoT to enable direct communication between homogeneous devices. Int. J. Pure Appl. Math. 118(16), 565–578 (2018) 10. Hemanth K., Ramesh G.P.: Energy efficiency and data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–95 (2020) 11. Dam Q.B., Nguyen L.T., Nguyen S.T., Vu N.H., Pham C.: E-breath: breath detection and monitoring using frequency cepstral feature fusion. In: 2019 International Conference on Multimedia Analysis and Pattern Recognition (MAPR), pp. 1–6. IEEE (2019) 12. Ravichandran, S.: Internet connected high tech street lighting system using RTOS. Int. J. MC Sq. Sci. Res. 9(1), 331–334 (2017) 13. Shamsudheen, S.: Smart agriculture using IoT. Int. J. MC Sq. Sci. Res. 11(4), 25–33 (2019) 14. G.K. et al.: Reducing power feasting and extend network life time of IoT devices through localization. IJAST 28(12), 297–305 (2019) 15. Zeleke B., Demissie M.: IoT based lawn cutter. Int. J. MC Sq. Sci. Res. 11(2), 13–21 (2019) 16. Fahad, A.A.A.: Design and implementation of blood bank system using web services in cloud environment. Int. J. MC Sq. Sci. Res. 11(3), 09–16 (2019) 17. Shahada S.A.A., Hreiji S.M., Shamsudheen S.: IoT based garbage clearance alert system with GPS location using Arduino. Int. J. MC Sq. Sci. Res. 11(1), 1–8 (2019) 18. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of Iot devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 19. Swarnalatha A., Manikandan M. Intravascular ultrasound image classification using wavelet energy features and random forest classifier. In: Advances in Intelligent Systems and Computing (2020) 20. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed. Tools Appl. 79(5), 4011–4026 (2020)

Chapter 18

Representation of Boolean Function as a Planar Graph to Reduce the Cost of a Circuit Vinitha Navis Varuvel, A. Kanchana, and D. Samundeeswari

Abstract Minimized Boolean function is used in many practical situations especially in the creation of a circuit. A new approach of reducing the cost of a circuit is dealt with by representing the circuit as a non overlapping graph called the planar graph. Boolean function representation as a planar graph to bring down the cost of IOT circuits. Keywords Boolean function · Java script software · Planar graph

18.1 Introduction In 1854, George Boole introduced the concept of Boolean function [1]. This concept has wide application in the field of Computer Science. The switching circuits was designed by Claude Shannon which used this concept. Many methods introduced in this regard were explained in papers [2, 3]. The concept mentioned above did not result in the reduction of variables. However, the circuits solved in paper [4] could be used in few problems. In that case, for cost minimization of circuits, the number of variables used in the Boolean function [5–8] must be minimized. Furthermore, in the problem of designing circuits, where crossing of two or more connections make the problem tedious, a graph structure could be implemented [9–11]. To avoid additional expenses, the communication lines must be placed one above the other with some gap between them. Circuits are easier to manufacture if their connections are in fewer layers without crossing [12, 13]. These highlights are integrated through the concept of non-overlapping graph called as a planar graph.

V. N. Varuvel (B) · A. Kanchana · D. Samundeeswari Humanities and Science, Rajalakshmi Institute of Technology, Kutambakkam, Chennai, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_18

225

226

V. N. Varuvel et al.

18.2 Material and Methods 18.2.1 A. Boolean Function A Boolean function consists of the binary variables, either True(1) or False(0), and the logic symbols connected by means of the logical connectives AND, OR and NOT which are a part of a statement (proposition). They are utilized mostly in logical and switching networks. There are several methods of specifying Boolean functions: tables, formulas and normal forms.

18.2.2 Planar Graph The In Graph theory, vertices represent nodes or points and edges represent links or lines. A planar graph is a graph in which there is no overlapping of edges. Graph planarity is an inherent property of a graph.

18.2.3 C. Java Scripts This JavaScript was inspired by JAVA and created by Brendan Eich. Both are similar with respect to the language name, syntax and respective standard libraries. They differ only in terms of their design. JavaScript is a widely used programming language having first-class functions used especially in creating web pages.

18.3 A Three Variable Boolean Function A three variable Boolean function is considered in order to check whether the corresponding circuit can be represented as a planar graph. The three variables are treated as three different end points and the cable attached is considered as an edge connecting these points. With the help of a Boolean circuit, it is ensured that overlapping does not occur resulting in a planar graph. Let us assume that the function is defined based on the current which flows to at least one area. The JAVA scripts software has been used to minimize the equation. The software can be activated by using https://s3.amazonaws.com/kanchana-boolean/index.html. Using the software mentioned above, the operations are performed, and the results executed are shown in Table 18.1. From Table 18.1 it is clear that the equation in terms of numbers is

18 Representation of Boolean Function as a Planar …

227

Table 18.1 Execution of result

(continued)

228

V. N. Varuvel et al.

Table 18.1 (continued)

(1, 4, 5, 7) + (2, 6) + 3 and the equation in terms of variable is, x + x  y + x  y z

(18.1)

18 Representation of Boolean Function as a Planar …

229

x

Fig. 18.1 Planar graph of Three variables

y

z

In the above equation if we substitute x  = 1 − x and y  = 1 − y, we get x + x  y + x  y z = x + (1 − x)y + (1 − x)(1 − y)z = x + y − x y + (1 − x − y + x y)z = x + y − x y + z − x z − x z − yz + x yz The minimized equation can be expressed as below format x + y + z + x yz = x y + x z + yz The right-hand side and left-hand side of the above equation will represent the planar graph as given below (Fig. 18.1). Conclusion Since the above graph exhibits the planar graph, the Eq. (18.1) will give a nonintersecting circuit. The above procedure can be verified for four variables.

18.4 A Four Variable Boolean Function A four variable Boolean function is considered to check whether the circuit represents a planar graph in which overlapping does not occur. Again, the function is defined based on the current which flows to at least one area. Similar to the previous case, JAVA scripts software has been used to minimize the equation. From Table 18.2 it is clear that the equation in terms of number is (6, 10, 11, 13, 9, 14, 12, 15) + (1, 3, 2, 4) + (5, 7) + 8 and the equation in terms of variable is, y + y  w + x y  w  + x  y  zw 

(18.2)

230

V. N. Varuvel et al.

Table 18.2 Execution of result

(continued)

18 Representation of Boolean Function as a Planar …

231

Table 18.2 (continued)

(continued)

232

V. N. Varuvel et al.

Table 18.2 (continued)

In the above equation if we substitute x  = 1 − x, y  = 1 − y and z  = 1 − z, we get

18 Representation of Boolean Function as a Planar …

233

y + y  w + x y  w  + x  y  zw  = y + (1 − y)w + x(1 − y)(1 − w) + (1 − x)(1 − y)z(1 − w) = y + w − yw + x(1 − y − w + yw) + (1 − x − y + x y)(z − zw) = y + w − yw + x − x y − xw + x yw + z − zx − zy + x yz− zw + xwz + zyw − zyzw The minimized equation can be expressed as follows: y + w + z + x yw + z + x yz + xwz + zyw =yw + x y + xw +zx + zy + zw + x yzw Both sides of the above equation will represent a planar graph as given below:

Conclusion Hence non-overlapping of circuits is verified for four variables. In general, this result can be verified for more number of variables. The above applet can be used to minimize the Boolean equation up to 15 variables.

18.5 Application The concept of representing a Boolean function as a planar graph has several practical applications. In the twenty first century, subway tunnels, oil/gas pipelines and metro lines are essential because underground routes, in the current scenario, reduce traffic to a great extent. There could be great mishaps due to overlapping or crossing. Moreover, the cost is high in underground routes which involves crossing. To avoid this, it is better to create non-overlapping communication lines (planar graph representation).

234

V. N. Varuvel et al.

18.6 Conclusion Using the Java Script Software, the variables involved in the Boolean function has been minimized. So that the Boolean function representation as a planar graph which reduces the cost of IOT circuits further, the function was represented in the form of a planar graph. It can be further optimized using machine learning technique.

References 1. Shannon, C.E.: A symbolic analysis of relay and switching circuits. Trans. AIEE (1938) 2. Kumar, V.D.A., Amuthan, S.G.: Static structure simplification of Boolean function for ‘N’ variables–A novel approach. J. Microelectron. 1(4), 160–167 (2016) 3. Tomaszewski, S.P., Celik, I.U., Antoniou, G.E.: WWW- Base boolean function minimization. Int. J. Appl. Math. Comput. Sci. 13(4), 577–583 (2003) 4. Kanchana, A., SrinivasaRao, K.: Distinct minimized equation from unique boolean function in simple truth table logic. Int. J. Pure Appl. Math. 116(23), 115–120 (2017) 5. Guillermo Ochoa de Aspuru: java applet software for FuzzyCognitiveMaps. www.ochoadeas puru.com/fuzcogmap/index.php 6. Kanchana, A., SrinivasaRao, K.: Software approach to minimize boolean function as ‘n’ distinct functions. J. Adv. Res. Dyn. & Control. Syst. 10(7) (2018) 7. Hemanth, K., Ramesh, G.P.: Energy efficiency and data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control. Autom. 13(2), 944–954 (2020) 8. Samanta, S., Pal, M.: Fuzzy planar graph. IEEE Trans. Fuzzy Syst. 23(6), Dec. (2015) 9. Ramesh, G.P., Parasuraman, S.: Design and implementation of U-Shape microstrip patch antenna for bio-medical application. Int. J. Adv. Sci. Technol. 28(12), 364–374 (2019) 10. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 11. Vasudevan, V., Balaji, K.: Performance of Cuk-KY converter fed multilevel inverter for hybrid sources. Indones. J. Electr. Eng. Comput. Sci. 10(2), 436–445 (2018) 12. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed Tools Appl 79, 4011–4026 (2020) 13. Rebinth, A., Kumar, S.M: A deep learning approach to computer aided glaucoma diagnosis. In: Proceedings of the IEEE International Conference on recent Advances in Energy-efficient Computing and Computation at St. Xaviers Catholic College of Engineering, Nagercoil on 7th and 8th March (2019)

Chapter 19

A Man Power Model Forthree Grade System with Univariate Policy of Recruitment Using Geometric Process for Inter Decision Times D. Samundeeswari, A. Kanchana, and Vinitha Navis Varuvel Abstract An association with n-grades which takes strategy choices at arbitrary age is thought of. At each dynamic age, an arbitrary number of people quit the association. There is a related loss of worker hours to the association if the individual stops. In this paper, for an marketing association comprising of multi grades subject to the consumption of manpower (wastages) because of strategy choices with high or low wearing down rate is thought of. The time to hire is evaluated in this paper using univariate max recruiting policy for a third-level program with waste created by its policy decision. The distribution of the threshold for all three degrees is considered to be linear and the inter-decision periods are geometrical. Numerical diagrams help the theoretical findings. Keywords Ordinary Renewal process · Univariate max policy of recruitment · Geometric Process · Threshold · Expected Mean · Manpower Planning

19.1 Introduction When initiatives involving changes to wages, benefits and updated revenue goals are revealed in companies, personnel departure happens. In addition, this refers to a lack of jobs, which can be measured in terms of hours of service. It will be unfair to be hired regularly. That explains when the organization, as the total reduction of employees, crosses a arbitrary threshold point on consecutive times. In [1–3], the writers examined workforce strategy structures and analyzed the mean and variation of employment policies in [4, 5]. In [6], the writers spoke with geometrical method regarding the mean and variance. In [7, 8] it was estimated in separate epochs for decision-making and leaving and in the same geometric phase [9–12] that the planned period to hire in one degree method was estimated. For an exponential random variable of three degrees the average and deviation is determined in [13–15]. The aim

D. Samundeeswari (B) · A. Kanchana · V. N. Varuvel Humanities and Science, Panimalar Engineering College, Chennai, Tamilnadu, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_19

235

236

D. Samundeeswari et al.

of this paper is to obtain the required period for recruiting by means of a three-grade geometrical method using a univariate Max recruitment strategy. Model Description Indicate that a three-tier entity A, B and C decides arbitrarily so that a small number of entities depart the group on all decision-making stages. A associated reduction of workers (depletion) is continuous and accumulated. For I = 1,2,3 … let, be a continuous random variable suggesting a decrease in the method induced by the departure of people in grade A, B and C. Let Z kA =

∞  i=1

WiA , Z kB =

∞ 

WiB , and Z kC =

i=1

∞ 

WiC

i=1

be the cumulative amount of depletion caused in grade A, B and C in the first k decision respectively. The period between the decisions shape a “a” parameter geometrical cycle. The threshold point in every class is its own. Let T be an ongoing random variable that shows time of recruiting in the enterprise with the L.) (and L.) (feature of scale. Recruiting policy in the enterprise is as follows: The revised management measures shall extend before the level of the net gross reduction of human capital is exceeded. A exact (0,t) k- judgment is anticipated to be Vk(t). Vk(t) = Renewal Principle Gk(t)-Gk + 1(t) G0(t) = 1. The recruitment cycle is to be estimated by E(T). Z A , Z B, Z C . Notations Wi ZA , ZB , ZC z G(.) Gk (.) H(.) Uk Vk (t) T L(.) E(T) V(T)

Loss of labour to the client by the I (i=1, 2, 3 ……) Exponential random variable, representing respectively the degree of threshold for grades A, B and C. Continuous random variable shows the operational threshold point. Density function of Wi . k-fold convolution of g(.) with itself. The distribution of ZA , ZB and ZC respectively. The constant random variable is the time between decisions (k-1)th and kth, k = 1,2, …with f.) (and F.). It is possible that the k- decision-making cycles are exactly in (0,t). Continuous random variable which represents the period for organizational recruitment. Function of cumulative T distribution. Think Training Time. Difference of recruiting time.

Result MODEL-I. The threshold level for the organization is defined as,

19 A Man Power Model Forthree Grade System …

237

Z = max(min(ZA , ZB ), ZC ). It is known that p[min(Z A , Z B ) > z] = P[Z A > z] · P[Z B > z] and p[max(Z A , Z B ) > z] = 1 − |P[Z A ≤ z]P[Z B ≤ z] ZA , ZB , ZC follows exponential distribution with parameters θ1 , θ2 , θ3 respectively, we have P[max(min(Z A , Z B ), Z C > z)] = 1 − P[min(Z A , Z B ) ≤ z]P[Z C ≤ z] = 1 − (1 − P(Z A > z)P(Z B > z))P[Z C ≤ z]    = 1 − 1 − e−θ1 z e−θ2 z 1 − e−θ3 z    = 1 − 1 − e−(θ1 +θ2 z) 1 − e−θ3 z = e−θ3 z + e−(θ1 +θ2 )z − e−(θ1 +θ2 +θ3 )z Distribution function of Z is,   H (z) = 1 − e−θ3 z + e−(θ1 +θ2 )z − e−(θ1 +θ2 +θ3 )z

(19.1)

⎡

⎤ exactly k decision epochs in (0,t] and ∞ ⎢ the threshold level z is not crossed by ⎥  ⎢ ⎥ P(T > t) = P⎢ ⎥ ⎣ the total loss of manhours in these k ⎦ k=0

decisions  

∞ k P(T > t) = V (t)P W < z By the equation of full chance, k i k=0 i=1      k k k ∞    Wi < Z = P Z> Wi / Wi = z gk (z)dz P i=1

0



=

i=1 ∞

i=1

gk (z)(1 − H (z))dz

 0∞

gk (z)(e−θ3 z + e−(θ1 +θ2 )z − e−(θ1 +θ2 +θ3 )z )dz .  ∞  ∞ −θ3 z = gk (z)e dz + gk (z)e−(θ1 +θ2 )z dz 0 0  ∞ − gk (z)e−(θ1 +θ2 +θ3 )z dz =

0

0

From renewal theory Vk (t) = Fk (t) −− Fk+1 (t) with F0 (t) = 1. By using convolution theorem on Laplace transform,

238

D. Samundeeswari et al.

 P

k 

 Wi < Z

= (g(θ3 ))k + (g(θ1 + θ2 ))k

i=1

− (g(θ1 + θ2 + θ3 ))k P(T > t) =

∞  

 Fk (t) − Fk+1 (t) (g(θ3 ))k

k=0 ∞ 

+



 Fk (t) − Fk+1 (t) (g(θ1 + θ2 ))k

k=0

−

∞  

 Fk (t) − Fk+1 (t) (g(θ1 + θ2 + θ3 ))k

k=0

= 1 − [1 − g(θ3 )]

∞ 

[Fk (t)](g(θ3 ))k−1

k=0

− 1 − [1 − g(θ1 + θ2 )]

∞ 

[Fk (t)](g(θ1 + θ2 ))k−1

k=0 ∞ 

+ [1 − g(θ1 + θ2 + θ3 )]

[Fk (t)](g( θ1 + θ2 + θ3 ))k−1

k=0

L(t) = 1 − P(T > t) = [1 − g(θ3 )]

∞ 

Fk (t)[g(θ3 )]k−1

k=1

+ [1 − g(θ1 + θ2 )]

∞ 

Fk (t)[g(θ1 + θ2 )]k−1

k=1

− [1 − g(θ1 + θ2 + θ3 )]

∞ 

Fk (t)[g(θ1 + θ2 + θ3 )]k−1

k=1

l(t) = [1 − g(θ3 )]

∞ 

f k (t)[g(θ3 )]k−1

k=1

+ [1 − g(θ1 + θ2 )]

∞ 

f k (t)[g(θ1 + θ2 )]k−1

k=1

− [1 − g(θ1 + θ2 + θ3 )]

∞ 

f k (t)[g(θ1 + θ2 + θ3 )]k−1

k=1

Since, {F k }is a geometric  s  process with rate “a” it is known that f k (s) = kn=1 f a n−1

19 A Man Power Model Forthree Grade System …

l(s) =[1 − g(θ3 )]

239

∞  k  s   f n−1 (g(θ3 ))k−1 a k=1 n=1

+ [1 − g(θ1 + θ2 )]

k ∞   s   f n−1 (g(θ1 + θ2 ))k−1 a k=1 n=1

∞  k  s   f n−1 (g(θ1 + θ2 + θ3 ))k−1 a k=1 n=1   −d E(T ) = l(s) ds s=0 ⎤ ⎡ 1 1 + ⎢ a − g(θ3 ) a − g(θ1 + θ2 ) ⎥ ⎥ E(T ) = a E(U1 )⎢ ⎦ ⎣ 1 − a − g(θ1 + θ2 + θ3 )

− [1 − g(θ1 + θ2 + θ3 )]

Let g(t) follows exponential distribution with parameter μ > 0, g(s) = and E(U1 ) is the mean of U1 with parameter λ1 .

μ . μ+s

⎤ μ + θ1 + θ2 μ + θ3 + ⎢ a (μ + θ3 ) − μ a (μ + θ1 + θ2 ) − μ ⎥ ⎥ E(T ) = a E(U1 )⎢ ⎦ ⎣ μ + θ1 + θ2 + θ3 − a (μ + θ1 + θ2 + θ3 ) − μ ⎡

This gives the mean time to recruitment Numerical Illustrations: Table 19.1: The intermediate recruiting duration is seen numerically by various parameters and by setting the other parameters. θ1 = 0.3, θ2 = 0.1, θ3 = 0.4 and a = 4. Table 19.2: Table 19.1 .

a

E(T)

2

1.72650

3

1.39773

4

1.27349

5

1.20833

6

1.16823

7

1.14107

8

1.12145

9

1.10662

240

D. Samundeeswari et al.

Table 19.2 .

μ

λ

E(T)

0.5

1.2

1.01349

1.0

1.2

1.06125

1.5

1.2

1.08082

2.0

1.2

1.09076

1.3

0.2

6.44811

1.3

0.4

3.22410

1.3

1.6

0.80603

1.3

2.0

0.64482

Effect of ”a” on the Expected Time to Recruitment (μ = 1.0; λ = 1.0; θ1 = 0.3; θ2 = 0.1; θ3 = 0.4).

19.2 Conclusion The following results can be included in the aforementioned tables: (i)

(ii)

(iii)

If μ is raised and other parameters are kept constant, the mean recruiting period should increase. That is to add, since μ decreases the total work losses, the period for recruiting is decreased. When t decreases and certain parameters are retained, the mean recruiting period will decrease. When μincreases, on average choices are always made, which, in effect, gives flexibility for promotion. If a > 1 and keeping other parameters fixed, the mean time for recruitment decreases. If a > 1, the interdisciplinary geometric cycle decreases stochastically, whereas the hiring period.

It takes the mean time to hire from a geometrical phase, given that the periods of decision are interdecision. The influence of nodal parameters on the overall recruiting period is also evaluated numerically. This paper demonstrates that the model described is practical and empirical review indicates that the findings are in line with fact.

References 1. Baratholomew, D.J.: Stochastic model for social processes. John Wiley and Sons, NewYork (1973) 2. Grinold, R.C., Marshall, K.T.: Manpower planning models. North–Holland, New York (1977) 3. Medhi, J.: Stochastic processes, 2nd edn. Wiley Eastern, New Delhi (1994) 4. Uma, K.P.: A study on manpower models with univariate and bivariate policies of recruitment (2010)

19 A Man Power Model Forthree Grade System …

241

5. Esther Clara, J.B.: Contributions to the study on some stochastic models in manpower planning, doctoral diss, Bharathidasan University, Tiruchirappalli (2012) 6. Sridharan, J., Saivarajan, A., Srinivasan, A.: Expected time to recruitment in a single grade manpower system with two thresholds. J. Adv. Math. 7(1), 1147 -1157 (2014) 7. Devi, A., Srinivasan, A.: Variance of time to recruitment for single grade Manpower system with different epochs for decision and exists. Int. J. Res. Math. Comput. 2, 23–27 (2014) 8. Ravichandran, G., Srinivasan, A.: Time to recruitment for a single grade manpower system with two thresholds, different epochs for exists and inter-decisions. IOSR J. Math. II(2), 29–32 (2015) 9. Vidhya, S.: A study on some stochastic models for a multigraded manpower system. Vancouver (2015) 10. Ramesh, G.P.: Design and implementation of U-Shape microstrip patch antenna for bio-medical application. Int. J. Adv. Sci. Technol. 28(12), 364–374 (2019) 11. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of iot devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 12. Vasudevan, V., Balaji, K.: Performance of Cuk-KY converter fed multilevel inverter for hybrid sources. Indonesian J. Electr. Eng. Comput. Sci. 10(2), 436–445 (2018) 13. Ramesh, G.P., Prabhu, S.: FPGA implementation of 3D NOC using anti-hebbian for multicast routing algorithm. In: Sharma, D.K., Son, L.H., Sharma, R.., Cengiz, K. (eds.) Microelectronics and Telecommunication Engineering. Lecture Notes in Networks and Systems, vol 179. Springer, Singapore (2021) 14. Gunasekaran, R.G.P.: Design of digital FIR filters for low power applications. In: Solanki, V., Hoang, M., Lu, Z., Pattnaik, P. (eds.) Intelligent Computing in Engineering. Advances in Intelligent Systems and Computing, vol 1125. Springer, Singapore (2020) 15. Rebinth, A., Mohan Kumar, S.: A deep learning approach to computer aided glaucoma diagnosis. In: IEEE International Conference on Recent Advances in Energy-Efficient Computing and Computation at St. Xaviers Catholic College of Engineering, Nagercoil on 7th and 8th Mar (2019)

Chapter 20

Denial of Service Attack in Wireless Sensor Networks D. Jeyamani Latha, P. Akshaya, M. M. Nilavarasi, N. J. Raghel, V. Utharapathi, V. M. Sanjaykumar, and P. Kishore

Abstract It is significant to protect the manufacturing internet of things devices since potentially devasting effects in case of an attack. Malware attacks are getting increased now a day rapidly because of the increased number of users. In public networks most of the devices are connected in a common network. The chances of insertion of malwares will happen frequently and create a miscellaneous attacks on devices connected. In the existing system dos attack is discussed in which the cyber attack vulnerabilities is being discussed. In the proposed system, design of dos attack detection through machine learning algorithm is evaluated. Machine learning is the prevailing purchases for examining as well as ensuring the internet of things technology. In this approach, machine learning is a random forest classification algorithm that increasing the predictive capacity and it produces a great effect mostly the time even lacking the hyper-parameter tuning. Here the malware in web applications is evaluated through a test website. In this approach, the real-world that constructed to perform cyber-attacks. Here, the machine learning based abnormality recognition approach can execute well in identifying these attacks. By extending these algorithms can assists to enhance the security of the system also. In order to measure the efficacy we measured the results using the respective criteria. Keywords Cyber attack · Internet of things industrial · Machine learning

20.1 Introduction A number of sensor nodes that automatically create a lack of personal interference the wireless sensor network (WSN). These network applications are military, urgent situation response, observation, as well as scientific examination of injurious surrounds. These sensor nodes are mainly utilized in the agriculture thus to do smart agriculture. This sensor monitors the agriculture surrounding as well as offers the data regarding D. Jeyamani Latha (B) · P. Akshaya · M. M. Nilavarasi · N. J. Raghel · V. Utharapathi · V. M. Sanjaykumar · P. Kishore Electronics and Communication Engineering, Velammal Institute of Technology, Chennai 601204, India © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_20

243

244

D. Jeyamani Latha et al.

the soil areas [1]. Every node has the capability to determine its neighbors as well as to build routes to attain other nodes in the collection. However, WSNs[ are susceptible to malicious attack. Malware attacks are getting increased now a day because of the increased number of users [2, 3]. In public networks most of the devices are connected in a common network. The chances of insertion of malwares will happen. Frequently and create a miscellaneous attacks on devices connected. In the proposed system, design of dos attack detection through machine learning algorithm is evaluated. The hardware effortlessness of these devices builds security method planned for conventional networks impracticable. This thesis explores the Denial-of -Service (DOS) attack by targeting a sensor node. T It aims at developing networks dedicated to legal customers and makes for quick launch as well as hard-to-stop DOS assaults with present network structure. These assaults are irresistible for resource services in particular and are efficient in providing highly skillful protection.

20.2 Related Works Cyber threat taxonomy named AVOIDIT to assist the detection and security against cyber threats by five primary classificators (Threat Path, Tactical Impact, Defense, Information Impact and Target). Between five, the security designation is used to notify the network administrator how to prevent or resolve an attack [4]. Two systems under development which propose to enhance cyber security learning. First, they have created taxonomy of cyber security. Second, they have builder a gateway which does as stand for users to talk about the safety of websites. These origins can be associated mutually [5]. Matthew Peacock et al. describe a proof-of-concept protocol attack on a BACnet system and examine the potential of modeling the basis of the attack [6]. Priya et al. introduced an improve data security protection approach for cloud applying two elements. Here, the source transmits an encrypted message to a destination with the assistance of cloud system. The source necessitates significant individuality of destination but no require of other information for example public key. To decrypt the cipher text, destination necessitates two parts. The significant thing is that the safety device missing or stolen, next cipher text can’t be decrypted as well as hardware device is vacated or struck down to decrypt cipher text [7– 9]. Simon M. Lucas et al. proposed a completely unique evolutionary approach for education DFA which develops only the alteration matrix as well as uses an simple settled process to optimally allot state labels [10]. Attribute based encryption plus data Reduplication is the optimal method to evade the seclusion troubles [2, 11–14].

20 Denial of Service Attack in Wireless …

245

Fig. 20.1 Current levels detected

20.3 Proposed Method 20.3.1 Malware Data Visualization With the IoT explosion worldwide, there is a growing threat from malware attackers which requires effective supervising of susceptible methods. Great amounts of information obtained from computer networks, servers as well as mobile devices have been examined for the spread of malware [15]. To match the size as well as difficulty of this data-intensive surrounding, effective analytics methods are required. Visualization methods in today’s Big Data environments can help malware researchers deeply explore the prolonged procedure of evaluating doubtful behavior. This work goes one step additional to contribute to the growing field of visualization technique (Figs. 20.1, 20.2, 20.3 and 20.4).

20.3.2 Random Forest Algorithm The fashionable approach for a choice of machine learning functions is the arbitrary forest algorithm, as it produces a great effect mostly the time even lacking the hyperparameter tuning. It has better performance than a neural network [16, 17]. Usually, neural networks are organized in layers. Layers consist of several interconnected’ nodes’ containing an’ activation operation.‘ Patterns are introduced to the network through the’ input layer’ that transmits to one or more’ hidden layers’ where the real dealing out is performed through a weighted’ associations method. Then, the hidden layers connect to an’ output layer’ where the answer is generated as shown in the graph (Fig. 20.5).

246

D. Jeyamani Latha et al.

Fig. 20.2 Correlated voltage levels

Fig. 20.3 Power spectrum

“While Tree learning” is closest to fulfilling the requirements to operate as an off-the-store data mining tool, “Hastie et al. says,” since it is invariant under scaling and many other transformations, feature values are robustly included and inspectable models are generated. In specific trees that are developing rapidly appear to develop very odd patterns: their preparation kits are over fit, i.e. High sensitivity, but incredibly high variance. Random forests integrate many deep decision-making bodies that have been taught in various areas of the same educational curriculum, in order to minimize discrepancies. It is at the detriment of a slight increase in bias and a certain loss of interpretability, but the performance is usually greatly improved in the final model. Random forests

20 Denial of Service Attack in Wireless …

247

Fig. 20.4 Correlated voltage levels

Fig. 20.5 Weighted connections method

build and combine different decision-making bodies to produce a simpler and more accurate forecast. A big benefit of random woods is that they can be used in the classification and regression problems, which comprise the bulk of the present method of machine learning. Let’s include random forests in classification, since classification is often known as the machine learning building bloc Fig. 20.6. Show how two trees will look like a random forest. The process involves the following.

248

D. Jeyamani Latha et al.

Fig. 20.6 Random forest with two trees

By logging in with the user I’d the user must open the web app. When it is opened it shows the user’s information along with the user’s E-Consumption info. The specifics needed shown in front end alone. But the rear end involves multiple procedures. The admin must login and view the user’s E-consumption information. Now, the malware test starts in the back end. It will then pre-process the test. Then the features will be read, evaluated with the database. if any malware is found, the pop up will be shown and the malware will be fixed. Random forest hyper parameters are either used for increasing the predictive capacity of the model or for speeding up the process.

20.4 Integration of Web App The Web app integration (Fig. 20.7) involves the mechanism at the front end and back end. The combination of both detections and solves the problem of the malware. The front end of this web app requires some process that is different from back end process. Web app integration involves the mechanism at the front end and back. The combination of both detection and solves the problems of malware. The front end of this web app requires some process that is different from back end process. By logging in with the ID the user must open the web app. When it is opened it shows the user’s information along with the user’s E consumption info. The specifies needed shown in front end alone. But the rear end involves multiple procedures. The admin must login and view the E consumption information. Now the malware test in the back end. It will then pre-process the test. Then the features will be read, evaluated with the database. If any malware is found, then the pop up will be shown and the malware will be fixed.

20 Denial of Service Attack in Wireless …

249

Fig. 20.7 Block diagram of Web app integration

20.5 Results and Discussion Machine learning technologies have been commonly used to guarantee a stable device interface. Via studies and practical assessment, the utility of machine learning for systems safety have been shown. The Random Forest Algorithm Scheme has been proposed. Our framework is still very scalable and can be expanded for more search queries.

20.6 Conclusion We present a system in this paper to detect and stop denial of service in networks with wireless sensor. The device’s cyber protection is important. The appropriate protection of these devices remains a major void. We believe that it provides an important new building block for the construction of stable infrastructure. Attack is discussed in which the cyber attack vulnerabilities are being discussed. Here, machine learning algorithm is used to detect the dos attack. In this approach, the malware in web applications is evaluated through a test website. Machine learning is the prevailing purchases for examining as well as ensuring the internet of things technology. Here, the machine learning based abnormality recognition approach can execute well in identifying these attacks. A machine learning function is the arbitrary forest algorithm that it produces a great effect mostly the time even lacking the hyperparameter tuning. It has better performance than a neural network. The key benefits of our strategies are provably effective, they enable searching and question isolation that are managed and concealed. This system can mainly use to enhance the WSN security.

250

D. Jeyamani Latha et al.

References 1. Karthik, R., Menaka, R.: A multi-scale approach for detection of ischemic stroke from brain MR images using discrete curvelet transformation. Measurement 100, 223–232 (2017) 2. Hemanth, K., Ramesh, G.P.: Energy efficiency and Data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–95 (2020) 3. Hemanth Kumar, G., Ramesh, G.P.: Energy efficient multi-hop routing techniques for cluster head selection in wireless sensor networks. Springer, Computational Intelligence and Complexity, vol. 193, pp. 297–305 (2021) 4. Selkar, R.G., Thakare, M.: Brain tumor detection and segmentation by using thresholding and watershed algorithm. Int. J. Adv. Inf. Commun. Technol. 1, 321–324 (2014) 5. Alam, M.S., Rahman, M.M., Hossai, M.A.: Automatic human brain tumor detection in MRI image using template-based K means and improved fuzzy c means clustering algorithm. Big Data Cogn. Comput. 3(27) (2019) 6. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of iot devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019) 7. Prastawa, M., Bullitt, E., Ho, S., Gerig, G.: A brain tumor segmentation framework based on outlier detection. Med. Image Anal. 8, 275–283 (2004) 8. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed Tools Appl. 79, 4011–4026 (2020) 9. Nithya, F.N.V. et al.: EAR-EEG signal transmission in WSN to evaluate co-channel interference of OLSR routing through residual of Fractional spline wavelet for auditory hallucination diagnosis. Int. J. Adv. Sci. Technol. 28(9), 271–279 (2019) 10. Devkotaa, B., Alsadoona, A., Prasada, P.W.C., Singhb, A.K., Elchouemic, A.: Image segmentation for early stage brain tumor detection using mathematical morphological reconstruction. Proc. Comput. Sci. 125, 115–123 (2018) 11. Wu, Y., Zhu, M., Li, D., Zhang Y., Wang, Y.: Brain stroke localization by using microwavebased signal classification. In: IEEE International Conference on Electromagnetics in Advanced Applications, 828–831 (2016) 12. Ay, H., Furie, K.L., Singhal, A., Smith, W.S., Sorensen, A.G., Koroshetz, W.J.: An evidencebased causative classification system for acute ischemic stroke’. Ann. Neurol. 58(5), 688–697 (2005) 13. Database: Brainweb http://www.bic.mni.mcgill.ca/brainweb/. Last accessed 10 Sept 2018 14. Fiebach, J.B., Schellinger, P.D., Geletneky, K., Wilde, P., Meyer, M., Hacke, W., Sartor, K.: MRI in acute subarachnoid haemorrhage; findings with a standardised stroke protocol’. Neuroradiology 46(1), 44–48 (2004) 15. Mathur, N., Meena, Y.K., Mathur, S., Mathur, D. :Detection of brain tumor in MRI Image through fuzzy-based approach in high-resolution neuroimaging-basic physical principles and clinical applications. InTech: Rijeka, Croatia (2018) 16. Arzoo, M., Prof, A.,Rathod, K.: K-Means algorithm with different distance metrics in spatial data mining with uses of NetBeans IDE 8.2. Int. Res. J. Eng. Technol. 4 2363–2368 (2017) 17. Fiebach, J.B., Schellinger, P.D., Jansen, O., Meyer, M., Wilde, P., Bender, J., Hähnel, S.: CT and diffusion-weighted MR imaging in randomized order. Stroke 33(9), 2206–2210 (2002)

Chapter 21

Android Application for Business Expense Management S. Surekha, G. Swetha, Sri Ram Gowd Vuppala, Arya Vishnu Thotakura, Tejasvi Dasari, R. Imayavaramban, and T. C. Jermin Jeaunita

Abstract Mobile devices stood top among accessibility and comfort for consumers. There are many solutions in the industry for the management of personal and company expenses. In this project, we are developing a smartphone application that traces users ‘ personal expenses and their personal contribution to community costs. Recording and reporting all the company expenditures is a cloud-based expense management program. It makes it easier to keep track of any expenditure for more workers in the company where traditional approaches can be of no use and often time consuming. Our application was developed for the needs of businesses and organizations where expenditures must be made by the workers and accepted by the members of the business with the access and authority to approve it. Using our service, customers can more efficiently handle its expenses. Not only can this program help users handle their expenditures, but it will also help marketing executives organize campaigns according to user needs. With advanced features such as auto scan for receipts, it stands out as a quick but effective solution for controlling and maximizing the expenses of the company. This makes the analysis and cost flow smoother, with the recording of expenses offline without internet will not be a burden. Keywords Expenses · Generating Report · Advances · Policies · Approvals · Users · Roles and Departments

21.1 Introduction TBecause of the growing usage of cell phones, people choose to use a smartphone device instead of sitting in front of the computers to get their job one. The Expense Manager is a software program designed to run on smartphones, i.e. on android S. Surekha · G. Swetha · S. R. G. Vuppala · A. V. Thotakura · T. Dasari · R. Imayavaramban Computer Science Engineering, R.M.K. Engineering College, Chennai, India T. C. Jermin Jeaunita (B) Member Technical Staff, Zoho Corporation, Chennai, India e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_21

251

252

S. Surekha et al.

devices. Expense Manager is designed to handle costs relevant to the company effectively. The goal is to use better solutions to help users quickly report their expenses. This latest framework includes different categories such as Automobile, Internet, Restaurant, Menu, Parking, Transportation, Telephone etc. This program falls under the category of finance and serves the main function of handling finance and is a very important part of an organization. As part of the lifecycle of software development the software product went through the design, production and testing process. The application interface is designed using custom art elements, the functionality is implemented using Android SDK (Software Development Kit), and successfully completed the product testing process. Users may build expenses with the information entered, and handle user roles etc. They are explained in all the topics in depth in their respective chapters. The one and only goal of our project is to make easy recording, managing and reporting the expenses done for business purposes either by the company manager or employee of a company with any role. We have brought some of the advanced technologies like AI for automatic scanning of receipts and to retrieve the spent amount. A businessman with multiple firms can manage all his expenses using a single account.

21.2 Background Study The innovation to build this framework in a forum comes from the daily problems people in companies face when tracking their expenditures and producing reports. Many of the questions around separating expenses are that having a personal expense is a BIG issue, it’s difficult to break the expenses into categories. Some typical approaches for dealing with this problem in ordinary circumstances are like sticky user notes, professional people manage this sort of problem by using tablets to record expenses and a database to retain significant volumes of information, particularly by experts. As this illustrates, different groups use vector strategies. This uses this detail incoherently. Issues exist in fields such as no quality controls, there is a possibility of losing vital inputs and manual mistakes. There are also issues. The Data Recorder is not always convenient so getting an aggregate view of those expenses may be a hectic operation. We believe in the development of a convenient mobile application that handles these issues. This app is capable of tracking expenses and providing an easy-to-use user interface with a detailed view, and this app is smart enough to answer: ‘Who spent? And how big is it?’.

21 Android Application for Business Expense Management

253

21.3 Related Work The mobile devices on the market make their lives easier for Smartphone users. One of those programmers, which has a great deal to give in daily life, is also the cost planner. We still have innovative tools in place that are special, simple to use and powerful in our application, when many related apps are now accessible. In addition to adding unique features such as integrating a person’s group expenditure and organizational expenses into one framework, we also introduced objectives such as report generation, analysis and estimates. We hit upon a new concept here to make use of an application for survey purposes in the area of consumer spending for business purposes. The concept serves as the research project’s main objective. The study also includes the integration of the software with other social networks and emails [1, 2]. Md. Rashedul Islam, Md. Rofiqul Islam and TahidulArafhinMazumder (2010). The use and effects of mobile use in individuals, businesses and communities have been identified. Mobile applications are among the most impacted and fastdeveloping sectors in the real world. This study showed that the mobile app and its success are facilitated by individual mobile users. The researcher Pramana Research Journal Volume 9, Issue 3, 2019 ISSN NO: 2249- 2976 563 https://pramanaresearch. org/ explains the consequence of mobile application in the business sector. In this research, statistical data of the past and present situation of mobile application have been presented to express the impact. The research concludes some effect of mobile application on society from the ethical perspective. Track all of your costs and sales in order to keep you updated by week, month, and year and divisions of financial health and track costs and profits. From this the daily expense tracking idea has been picturized. • Mint. • YNAB. • QuickBooks Online. The above are some of the most popular mobile applications for expense management which is used in today’s business world. Almost all the main features from these different Applications are available in our Expense manager. Different types of expenses can be viewed with different filters. Analyzing expenses monthly and yearly are also made available along with the existing features.

21.4 Proposed Work The program is modularized based on the individual and the criteria needed for its processing. The program is structured to keep the expenditure under control when an employee submits an expenditure and must be approved by someone with the authority to authorize a specific expenditure. They also require auto scanning of

254

S. Surekha et al.

receipts where one can take a photo of the receipt and add it to the expenditure. This expenditure includes car, internet, lodging, food, parking, transport, telephone, etc. The consumer may choose the form of expenditure and request it for approval where he is allowed to choose the approver. This expenditure is approved by the approver, or if he has no right to do so, the expenditure is forwarded to the individual entitled to. Expenditures can be clubbed together and submitted as a report, for example, during a business trip various expenses are incurred such as meals, hotel charges, travel expenses. Travel expenses can be reported in terms of mileage, and this is measured depending on the country as the fuel cost varies and also the vehicle mileage. These reports can be submitted to the approver for approval. Then Approver approves or rejects the reports submitted to them. The approved expense reports can be submitted for reimbursement. Both the approved and rejected expense reports are stored in the Database. The expense data can be used to track the financial status of an organization (Fig. 21.1). Fig. 21.1 Expense manager activity diagram

21 Android Application for Business Expense Management

255

21.5 Modules This application is for managing expenditure and maintaining accounts in an organization. In the earlier days, the expense management was done by recording them manually in accounting books, which was later on enhanced by the usage of electronic mails. People sent their expenses as emails to the corresponding Approvers. But it was difficult to track the expenses made by the respective employees under certain period, location and various other criteria. This disadvantage led to the emergence of expense management applications. The following are the modules available in this application.

21.5.1 Users, Roles and Departments The brain MRI images used in this paper for evaluating the stroke pixel region detection system are accessed from BrainWeb open access dataset [3]. This dataset contains the huge number brain MRI images in the category of both stroke regions and normal pixels alone. This dataset is categorized into 156 stroke free brain MRI images and 125 stroke affected brain MRI images. This dataset also contains annotations of all brain stroke MRI images by expert radiologist.

21.5.2 Recording of Expenditure The expenditure made by the employee for the company can be recorded to claim reimbursement. The expenses can be recorded under different categories such as automobile expense, food expense, travel expense and so on. There are two different types of expenses. The expenses recorded under the mileage expense category are the mileage expenses. The admin can configure a default rate for a particular distance in kilometers. While recording the mileage expense, the user has to enter only the distance travelled in kilometers for which the corresponding amount will be calculated automatically using the default rate which was configured. The expenses recorded under other expense categories fall under the normal expense type for which the amount will be entered manually while recording them.

21.5.3 Creation of Reports The recorded expenses should be clubbed into an expense report before being submitted to the respective manager for approval. This clubbing is required to identify the expenses made on different scenarios such as expenses made on a site visit,

256

S. Surekha et al.

expenses related to a specific project for a specific customer and so on. The created reports are then passed on for approval to the higher authorities who in turn approve or reject the reports based on the company policies. The approved reports are then reimbursed to the users who submitted them.

21.5.4 Creation of Trips Employees are provided with the facility to raise trip requests before going on business trips. The trip details such as departure and arrival dates, hotel and flight reservations should be filled in the trip requests. These trip requests are either approved or rejected by the travel management team in the company after which the booking of tickets for the submitters of the corresponding trip requests is done. The expenditures made during the trip can be recorded and clubbed into a required number of reports which are then associated with the corresponding trip for accounting purposes.

21.5.5 Advance Payments The employees can be given advance amounts before going on a trip, before visiting a customer and various other scenarios. These advance payments can be applied to the reports before sending them for approval which in turn has an effect on the reimbursement of the reports. If the advance amount is the same as the report’s total amount, then the report will be marked as reimbursed once it gets approved. If the applied advance is lesser when compared to the associated report’s amount, then the rest of the amount has to be reimbursed by the company in order to mark the report as reimbursed. If the advance seems to be greater than the report’s amount, then the employee who is responsible for reimbursing the report will be given two options regarding the reimbursement. He/She can either choose to obtain the extra advance amount back from the submitter of the report or can let them use it for future expenditures. The remaining advance amount which is being returned by the employee to the company will be maintained as an advance refund amount.

21.5.6 Types of Approval There are two types of approvals which are customisable for each user. One is Single Level Approval while the other is Multi Level Approval. In the former, the approver of every employee will be configured in his/her user information by the higher officials. So, once the configured approver approves, the report will be sent for reimbursement. In the latter, the approval flow will be configured for each user. Only when all the

21 Android Application for Business Expense Management

257

approvers approve the report in the specified sequence, it will be sent for reimbursement. The above-mentioned types of approval are applicable for advance payments and trip requests also.

21.5.7 Types of Currencies Three different types of currencies are supported in this application known as company currency, expense currency and reimbursement currency. The currency which is being treated as the common currency across various branches of the company is the company currency. The expense currency is the currency in which the expense was made. The reimbursement currency is the currency type in which the employee wants the report to be reimbursed. Let’s consider the scenario in which all the three types of currencies play a role. Let the main branch of the company be run in a country with currency type called ‘A’ (company currency) and an employee working in a different branch of the company, which is being run in a different country of currency type ‘C’, be gone for a business trip to a different country with currency type ‘B’ where all his expenditures were made with the currency type ‘B’ (expense currency). After the business trip, the employee would want the report to be reimbursed in the currency type ‘C’ (reimbursement currency), as he works in the country whose currency type is ‘C’. Considering the above case, three different types of currencies were taken into account.

21.5.8 Policy Configuration This application facilitates the creation of different policies that can be associated with different employees of the company. The mileage rate configuration can be different under each policy. The allowed categories under which the expenses can be recorded can also be configured in the policy. The main functionality of policy creation is to assign rules to it and verify if the employees abide by the rules. The rules are customisable such as setting up an expense limit for a particular category and so on. If the policy rules are violated by the employees to whom it is associated, then warnings and other alerts will be given to the group of users involved.

21.5.9 Automation Using Templates Since we notify the user for main actions like Approved, Rejected or Reimbursed, we can also automate the way of notifying users for certain actions. We can create an Email template to be sent for any particular action (for example, we can create an Email template with the documents to be carried while travelling to Country A

258

S. Surekha et al.

and trigger an Email with this template when the user creates a Trip to County A). We can also create a Web App Notification template (for example, we can create a Notification template warning the user to minimise his expenditure and trigger that Notification if a user submits a report with an expense amount greater than the specified amount in the template).

21.5.10 Customised Fields Other than the default set of fields for all entities, we’ll allow the user to create his own set of fields based on their company requirement (for example, in Users module, the company might want to store the Passport and VISA details of an employee). Also, we can set whether the field has to be filled mandatorily while creating an entry for the module. We’ll support the fields with short text, paragraph, check box and dropdown select data types.

21.5.11 Providing Statistical Data of a Company Users with Superior roles can view all of the company’s data in a customised format. (For example, we can generate a statistical report which comprises the expenses made in a particular month / year or a report which includes the expenses made by a particular employee). By this way, a person analysing the financial status of the company, can identify where his company is going wrong in spending and can evaluate the company’s total expenditure. We also provide a report which summarises the time taken in each stage of an expense reporting process with which they can find where the process is getting delayed (approval or reimbursement). Also they can generate a summary report of the performance of Approvers (Managers) which gives data of the time taken by each approver to approve a report.

21.5.12 Customised Status Tracking We’ll allow the users to track the Reports and Trips in a deeper way by creating userdefined sub-statuses for a Status. (For example, a Trip may have different stages after it gets approved like Booking in Progress, Booked, Tickets Attached, Trip Completed which may vary depending upon the Company. By creating all these as sub-statuses and marking a trip with the corresponding sub-status allows the user to filter all his completed/ticket booked trips).

21 Android Application for Business Expense Management

259

21.5.13 Budget Maintenance Budget plays a vital role in managing a company’s finances. We can create a Budget for a month or a year which can include the overall budget amount and the allotted amount for particular categories. At any point of time, we shall see the comparison between our budget planned, the actual expenditure and the status (budget overflow or in control).

21.5.14 Procurement Purchasing of goods in large numbers is a tedious process for larger companies. Procurement solves it in an easy way. An employee can raise a request for Purchase of Goods (with quantity) to his Manager and get it approved. The Request will then be forwarded to the Procurement team who’ll check for the availability of Goods in the Inventory. If the Goods are not available in the Inventory, they’ll do the background analysis regarding the purchase and request a quote from the vendors. After getting the quotes, they’ll decide on the Vendor to buy from and forward the Purchase Request to the corresponding Vendors. After receiving the Goods, they’ll pay the Vendors and raise that payment as an Expense against the Purchase Request and mark the Purchase Request as Purchased (Fig. 21.2).

21.6 Technical Implementation In this application, we made use of SQLite, an open source database that stores data to a text file on a device. Android comes with the implementation of SQLite database. SQLiteOpenHelper class is responsible for facilitating the usage of the SQLite database. This class is used for database creation and version management. The Android SDK (software development kit) is a set of development tools used to develop applications for Android platforms. It is freely available under the Apache License 2.0. Here we have used Android version 6.0 for development.

21.7 Future Scope 21.7.1 Forecasting With all the data available for a company, it is possible to predict the future spend of a company. We’re working on a Machine Learning algorithm which predicts the future expense based on the past expense history.

260

S. Surekha et al.

Fig. 21.2 Workflow of expense manager

21.7.2 Receipt Scanning Instead of an employee creating an expense by filling all the necessary fields like Merchant, Amount, Tax details and attaching a receipt to it, we can allow the user to upload the receipt alone and extract necessary details from the receipt via Image Processing. This will reduce the manual work of filling the fields for an Expense creation.

21.8 Conclusion Cell phones are a basic piece of our lives what’s more, it seems like all that we have is inside them behind the screen. It has no more been the lone mean for associating and talking with individuals who are far away from us. It is our individual partner who awakens us in the morning, helps us to remember significant gatherings, catches minutes with our friends and family and lovely pictures, controls the apparatuses in our home, oversees accounts and business, discovers all that we care about and helps us present it in proper structure. We present an efficient mobile application for

21 Android Application for Business Expense Management

261

business expense managing. This Expense manager app is user friendly. It can be customized depending on the user requirements. The data stored will be maintained with high security. We allow handling multiple organizations by a single user in case he has multiple organizations and wishes to maintain all the expenses in a single place. We provide options to track expenses weekly, monthly and yearly as well as their categories. Hence by analyzing the expense reports, an organization can become financially stable.

References 1. Cordova, C.R., Inga, E., Yaguache, M.F.: Accounting software and profitability in SMEs: the case of Ecuador. In: 12th Iberian Conference on Information Systems and Technologies (CISTI), Lisbon, pp. 1–4 (2017) 2. Sandoval-Mora, K.S., Quezada-Sarmiento, P.A., Mayorga-Diaz, M.P.: Importance of the adoption and application of international financial reporting standards IFRS in the business context. In: 2019 14th Iberian Conference on Information Systems and Technologies (CISTI), Coimbra, Portugal, pp. 1–6 (2019) 3. Hemanth Kumar, G., Ramesh, G.P.: Energy Efficient Multi-hop Routing Techniques for Cluster Head Selection in Wireless Sensor Networks. Springer, Computational Intelligence and Complexity, vol. 193, pp. 297–305,2021 4. Yadav, S.S., Malhotra, R., Tripathi, J.: Smart expense management model for smart homes. In: 2016 International Conference on Computational Techniques in Information and Communication Technologies (ICCTICT), New Delhi, pp. 544–551 (2016) 5. Yuqing, P., Jiuli, W.: Design and implementation of Accounting e-Business platform for source documents. In: International Conference on Computer Application and System Modeling (ICCASM 2010), Taiyuan, pp. V10-5–V10-8 (2010) 6. Bozkus, Z., Bisson, C., Arsan, T.: Analytical expense management system. In: 2009 First International Conference on Networked Digital Technologies, Ostrava, pp. 527–53 (2009) 7. Sabab, S.A., Islam, S.S., Rana, M.J., Hossain, M.: Expense: a smart approach to track everyday expense. In: 2018 4th International Conference on Electrical Engineering and Information & Communication Technology (iCEEiCT), Dhaka, Bangladesh, pp. 136–141 (2018) 8. Hemanth, K., Ramesh, G.P.: Energy efficiency and Data packet security for wireless sensor networks using African Buffalo Optimization. Int. J. Control Autom. 13(2), 944–95 (2020) 9. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimedia Tools Appl. 79(5), 4011–4026 (2020) 10. Bouwman, H., Dihal, S., Reuver, M., Warnier, M., Carlsson, C.: Mobile cloud computing: state of the art and outlook. Q Emerald Group Publish Limit 15(1), 4–16 (2013) 11. Bendovschi A., Ionescu B., Ionescu I., Tudoran L. (2013) Traditional accounting vs. cloud accounting. In: Proceedings of the 8th International Conference Accounting and Management Information Systems, pp. 106–125. AMIS (2013) 12. Manikandan, G., Anand, M.: Radix-2/4 FFT multiplierless architecture using MBSLS in OFDM applications. In: Intelligent Computing in Engineering (pp. 553–559). Springer, Singapore (2020) 13. Sundaram, A., Ramesh, G.P.: Investigation of solar based SL-QZSI fed sensorless control of BLDC motor. In: Intelligent Computing in Engineering (pp. 779–787). Springer, Singapore (2020) 14. Shamsudheen, S., Azathmubarakali: Smart agriculture using IOT. Int. J. MC Square Sci. Res. 11(4) (2019)

Chapter 22

Student Perception Regarding Chatbot for Counselling in Higher Education Shivani Agarwal, Nguyen Thi Dieu Linh, and Gloria Jeanette Rincón Aponte

Abstract The present research discusses about the issue of student’s counselling which is very much based on one’s own cognitive judgment. This chapter empirically investigates what effect chatbots have on student’s perception for counselling. The current study discussed about the perception of students regarding the counseling app (chatbot) which can be implemented in college/universities to deal with three aspects of life, namely: Personal life with chatbot: Professional life with chatbot and Genera life with chatbot. A Questionnaire was developed which measure the effect of chatbot while counseling the students on the above-mentioned aspects of life. The results are measured and well presented in the study. So, it is recommended to the college and university to implement chatbots in their premises for the wellbeing of their students. Keywords Chatbots · Counselling · Students · Higher education · Research · Life · Satisfaction

22.1 Introduction India consists of 935 universities in which 50 are central universities, 409 state universities, 127 deemed universities, 349 private universities (UGC 2020) [1]. Among the established institutions five are functioning under the State Act, and 75 Institutes of National importance which include IIMs, AIIMS, IITS, IIEST and NITS among them. In the above universities and colleges, approximately 19.9 million students

S. Agarwal (B) KIET School of Management, KIET Group of Institutions, Delhi-NCR, Ghaziabad 201206, India N. T. D. Linh Hanoi University of Industry, Hanoi, Vietnam G. J. R. Aponte Facultad de Ingeniería, Universidad Cooperativa de Colombia, Bogotá, Colombia e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_22

263

264

S. Agarwal et al.

have attended in 2019. The notable shift in admission process over a past few years have seen in universities across globe, their staff, and their students. In twenty-first century, the demands from the community of students are very high in terms of personal life, professional life and in general life. In the era of cutthroat competition, these kinds of increasing demands like to get good marks, placement and finally able to locate/prove themselves in the front of the family members. The communities of students are facing a lot of stress, depression, anxiety and low level of morale, low trust, etc. To subjugate over the above-mentioned psychological construct, the emergences of counselor play its role. In the digital era, the role of recruiter, waiter, and analyst is very well played by Artificial Intelligence (AI). AI bots is being advertised as the need of hour in all the aspects of business and also considered as the future of higher education. In a nutshell, though it’s because AI bots contribute towards making these colleges and future ready. There are numerous reasons why colleges and universities around the world have chosen to integrate online chat bots for dealing with admission enquires with their websites and social media pages. 1. Better Communication For any college or universities, better and timely communication is the solution to converting potential applicants into enrolled students. The potential candidate belongs to young Millennial who can handle multiple devices at a single point of time. They need or prefer instant responses to the queries. 2. Campus guidance and troubleshooting after admission Numerous researchers have shown that delayed in reply to queries are among the key reasons why students drop out of colleges or universities. Timely and frequent communications with the students, joined with proper identifications of information gaps, can help quash this trend. The best chat bots for higher education websites are able to deliver on both these fronts. 3. A detailed and up-to-date repository of information for universities The function of online chat bots in higher education goes beyond serving to students’ requests. AI bots can also serve as an asset for the colleges or universities that use them on their website. University can gather large amount of data with the help of Chabot over a period of time. This data is of paramount importance as it informs the colleges or university about the behavioral pattern of students what they can or cannot do.

22.1.1 Background About Chatbot Chat bots are a form of human–computer dialog system which operates through natural language via text or speech. Chatbots are typically used in dialog systems

22 Student Perception Regarding Chatbot for Counselling …

265

for various practical purposes including customers service or information acquisition. While some chatbots applications use extensive word classifications processes, Natural Language processors and sophisticated AI, others simply scan for general keywords and generate responses using common phrases obtained from an association library or databases. Today, most chatbots are accessed via virtual assistants such as Google Assistant and Amazon Alexa, via messaging apps such as Facebook Messenger or WeChat. Chatbots are typically classified into usage categories that include commerce, education, entertainment, finance, health, news and productivity.

22.1.2 Application of Chatbot in Business The chatbot now-a-days plays a vital role in conducting business activities efficiently. There are certain applications of chatbots induced in business which are as follows: (i) Accessible Anytime On an average people spend around 7 minutes until they are assigned to a person. Gone are the frustrating days of waiting in a queue for the next available operatives. They are replacing live chat and other forms of slower contact methods such as emails and phone calls. (ii) Multicatch Unlike humans who can only communicate with one human at a time. Chatbots can simultaneously have conversation with thousands of people. No matter what time of the day it is how many people we are contacting every single one of them answered immediately. (iii) Flexible Attribute Chat bots is flexible in nature, it can be implemented in any industry. Researcher need to convey/command the right structure of conversation and it can easily adapt with that particular industry. (iv) Cost Efficient Chat bot never asks or expect for salary, benefits or any sort of reward. Moreover, it will never leave you. So, it is considered to be cost effective in comparison of Humans. Hiring a human for a job is expensive if our revenues are not high or sales targets are not met and would create havoc in the business. (v) Work Automation Chatbot can easily complete monotonous recurring task assigned to it without delay. On the contrary if same set of instruction have to be followed by human being they easily get tired, bored and sometimes shirk from the work has been initiated. Chatbot facilitate people save time and more effective.

266

S. Agarwal et al.

(vi) Personal Assistant Humans can use chat bots as financial advisor, Fashion advisor travel agent and so on. Chatbot has inbuilt feature to mark our choices and provide relevant solutions the next time we need.

22.1.3 Scope of the Study The current study discussed about the perception of students regarding the counseling app (chatbot) which can be implemented in college/universities to deal with three aspects of life, namely: a. Personal life with chatbot According to sharing information/thoughts about personal life with chatbot means that students really find it ease at sharing our personal life with chatbot. We have observed that most of us do not find easy while sharing family problems, financial problems with chatbots. They also do not share the relationship status as well. b. Professional life with chatbot Professional life with chatbot has been classified into about our learning habits, reading habits, writing habits, attendance issues, performance in external exams. Also, most of us “sometimes” share their issues according to our responses. c. General life with chatbot General life with chatbot has been classified it into about our happiness, sorrows, stress level, level of motivation and morale.

22.2 Literature Review Prasad et al. (2020) has suggested that people suffering from depression can obtain much-needed help without the feeling of being judged for having such mental illness [2]. It was clear from this research that it has been a testament to know that depression can be mitigated by means of technology. It also help the people in choosing the best doctor for them who can guide them and can answer all their queries without judging them. Xiaojing (Romy) Wang (2020) suggested that subjective rules, attitudes, and perceived behavioral control predicted participants’ intentions to seek counseling [3]. The study evaluated the kinds of students which may be more interested in seeking counselling and it was concluded that students stress levels and habits have a greater impact towards seeking help.

22 Student Perception Regarding Chatbot for Counselling …

267

Tan Yigitcanlar (2020) suggested the contribution and risks of AI in building smarter cities [4]. The research not only shows the importance of this technology that plays a key role in several applications within smart cities. The study also gives a heads up for urban policymakers, planners, and scholars for them to prepare for the disruptions that AI will cause in our cities, societies and business. Abidah Elcholiqi (2020) presents chatbot in Bahasa Indonesia using NLP to provide banking information [5]. It aims NLP to make the interactions between computers and human feel like communications between two people. They fill the gap wherever they fall. It has concluded that chatbot was developed to provide user queries with a response rate of 84%. Eklund and Isaksson (2019) designed, constructed and tested an AI-powered chatbot, Ava, for identifying and evaluating human trust in AI-Automated service encounters [6]. More specifically, it highlighted how cognitive trust is decreased when system opacity is increased. The thesis highlighted the importance of transparency in the AI and Counselling field and how Subjects put more faith in the system when they are given assurity of the safety of their data. Heidi Sturk et al. (2019) in the study 43 health practitioners explored key features of the Head to Health site and provided feedback via a post-workshop survey [7]. Practitioner feedback highlighted that although many were unaware of the website, overall perceptions were positive with 79% stating they will recommend the site to clients in the future. The research went on to put light on the need and benefits of implementing a mental health app in rural areas where people find it inconvenient to share mental health problems. The same can be said about students who often are uncomfortable to share their problems. Pablo D. Armas Sánchez (2019) highlighted the changing nature of human life, quoting “no one has a clear idea of what skills they are going to need in 5 years’ time [8].” This unpredictability is fueled by the astonishing speed in which technology is changing the nature and form of work, among other things. He concluded with the need of proper guidance for students to tell them of the demands of future thus Counseling app can help fulfill that gap. Kram Mahmoud Kamel Marwan (2019) found out that AI not only reduced bias in the recruitment process while in universities or with employers but also enabled more engagement, personalization and assistance through chatbot [9], Augmented reality and adaptive learning solutions. The study highlighted the unlimited potential that Artificial Intelligence has in the field of education and employment, which can be used extensively in the field of counselling app. Sabine Wilhelm (2020) discusses a number of technology-enabled solutions to our field’s challenges, including internet based and smartphone-based cognitive behavioral therapy [10]. In exploring obstacles to receiving mental healthcare, the researchers identified many patient-level barriers. For instance, individuals often report logistical barriers, including a lack of transportation to attend appointments, an inability to take time off from work to attend appointments, and/or a lack of childcare. Moreover, many persons cannot afford the high costs of treatment. Thus giving a clear shot towards the development of online counselling applications.

268

S. Agarwal et al.

Lee (2019) came with the aim to develop the scenario for implementation of the rule based counselling chatbot and to verify its usefulness [11]. And as a result clear counselling structuring, appropriate use of selective and narrative responses also active listening were identified as important. Meenal Sharma (2019) aims to propose a new framework of learning called phygital learning which attempts to explore the convergence of personalized education content delivery through extensive use of advanced digital technologies like AI. Thus, gap cannot be bridged by only setting up or building new universities and schools by leveraging technology and building new models. Also there are unique characteristics of individuals learners which change at different stages of education. Grant Blashki (2019) suggested AI and mental health [12]. The study revealed high accuracies and provided excellent AI potential in mental healthcare. AI has some useful applications in the mental health fields. It is especially of value inearly detection, diagnosis and treatment also assessing prognosis. Gillian Cameron (2018) has suggested that Chatbots are increasingly used in the mental health care specially with the emergence of Virtual therapist [13], and People enjoy interacting with chatbots as they find it easy to use. There is a consistent personality throughout the conversation, and the chatbot performs well at onboarding. Keeheon Lee (2019) he suggested that there was striking decrease in the number of inquiries received by email once the chatbot was introduced [14]. The subjective evaluation of the difference in workload before and after the introduction of chatbot and also concluded that introduction of chatbot can reduce workers workload in the context of a college administrative office. Saiah Awidi (2018) His study shows high levels of student satisfaction and an enhanced learning experience in response to specific learning designs and it can be concluded from the study that a student-centred model is more likely to guide better learning designs decisions and higher levels of satisfaction for students, than instructor-centred models. Sunil Kumar Srivastava (2018) focused on two different aspects, firstly the cut in future job opportunities due to rise in Artificial Intelligence and secondly the rising opportunities of AR in the fields of health care [15], finance, law, information browsing etc. He put focus on the importance of policy framework and research in AI and its use in various spheres of life, giving an opportunity to use AI in the field of human Counselling. Akihiro Yorita (2018) showed that the conversation model and Sense of Coherence model are capable of measuring stress and can be used by the Peer Support model to successfully select appropriate support actions [16]. The study highlighted that careful and coordinated efforts can help measure the actual stress level of individuals and thus a well-designed counselling app can understand the feeling of individuals. Timothy Bird (2018) presents the results of a web based randomized controlled trial of manage your life online (MYLO) [17]. The study reveals that gap cannot be bridged by only setting up or building new universities and schools by leveraging technology and building new models which cannot improve efficiency. It has concluded that findings are consistent with previous smaller, laboratory-based trial

22 Student Perception Regarding Chatbot for Counselling …

269

and provide support for the acceptability and effectiveness of MYLO delivered over the internet for a non-clinical sample. Gillian Cameron (2017) came with the aim to outline the design of a chatbot to be used within mental health counselling [18]. And, in the study it has bridge the gap between the demand for mental health care and the resources that NHS can provide is likely to widen. Therefore it thus concluded using chatbot for mental health counseling can provide many benefit for the user. Research has shown users find chatbot safe and easy to talk to. Ataliia Mozre (2017) he suggested that the most commonly used app by the students was telegram and use a bot that can be similar to this app called telegram. And also added to this that there were many staff who were not willing to install new software on their own mobile devices, then you need to use those applications that are already installed. After analyzing they found telegram was used mostly by many staff. Hsin-Hung Chou (2017) presents to design a Facebook messenger chatbot for course counselling came with the aim to relieve the loads while providing questions and answers services with no limits in time and space. It thus included core algorithm, a depiction of the system architecture. It has concluded that chatbot can answer students Q&A via Facebook messenger. Nataliia Mozre (2017) he suggested that the most commonly used app by the students was telegram and use a bot that can be similar to this app called telegram. And also added to this that there were many staff who were not willing to install new software on their own mobile devices, then you need to use those applications that are already installed. After analyzing they found telegram was used mostly by many staff. Mghweno Richard, Baguma Peter (2014) concluded that accessing guidance and counseling services has an effect in shaping students’ attitude towards studies and career choice [19]. Thus, with the intervention of guidance and counselling service the performance of students improved and thus online counselling can help focus the students to perform better. Adriaan de Vos (2015) presents conversational agent based behavioural change support system have different features and implementation and aims to treat many disorders gives an overview of the different possibilities and techniques [20]. It thus concluded that BCCSc have shown great improvements over the past few years, but the true strength will lie in introducing industry standards and combining technological advancements. N. Banu Priya, I. Juvanna (2014) said that students’ mental health has a direct impact on their academic performance, on their experience at an institution and on the relationships they foster during this time [21]. It is not only in the best interests of an institution to invest in a student’s mental wellbeing, but it is in fact their responsibility. The study found out that, with increasing costs institutions can effectively and economically help their student with a 24*7 available online counselling application. Cicco (2014) discussed the responsibilities of the counselor educator and counseling student in the online internship course to ensure that the supervisory relationship supported, rather than hindered the growth and development of the future

270

S. Agarwal et al.

professional counselor [22]. Thus, with the emergence and boom in the field of online counsellors and their trainers, it can be assumed that there exists a positive atmosphere in the society towards online counselling. Roy Bhakta et al. (2014) found that students disclosed additional information to the chatbot on more sensitive topics when the length of engagement was increased, but that this effect could be negated by the inclusion of the depth of engagement questions [23]. It can be concluded that subjects provide more reliable and true information to the chatbot with increase in length of interaction. Furthermore, questions related to Sensitive topics require more focus and attention to get correct answers. Uyoto, Tri Prasetyaningrum (2011) presented a design application “ NingBK: Social GC” as a means of providing assistance field of social counseling services that run on mobile devices [24]. They looked into the design and layout of counselling app and provides an example of counselling app to make it easier to comprehend the functioning of online counselling applications. Patrick Biietal (2013) indicated that students are positively disposed to use of chatbots during instruction. It can be said that students are positive towards the use of chatbots in the education and other fields and would try and enjoy the use of career counselling app. Richards D, Viganó N (2013) highlighted that online counseling can have a similar impact and is capable of replicating the facilitative conditions as face-toface encounters [25]. Thus, with online counselling replicating the same results as face-to-face counselling, more attention can be put into the use of chatbots and artificial intelligence in counselling. Suyoto et al. (2013) proposed the design and implementation an application of mobile leadership with Interactive Multimedia Approach-called “m-Leadership” [26]. This application is used to indirect service of Guidance and Counseling that runs on mobile devices i.e. mobile phones. The researchers explored the possibility of one unanimous counselling application for the entire country thus, providing motivation for more work in the field of online counselling. JA Coughlan (2007) presents moving face to face communication to web-based systems it aims with the special issue is to address this challenge and contribution to ideas in the theory and practice of moving face to face communication [27]. It has highlighted some of the areas like consumer interface, in the design and evaluation of web interfaces that can stimulate social responses in consumers and become more reflective of the real-life information consumer experience. Young KS (2005) suggested that Caucasian, middle-aged males, with at least a four-year bachelor’s degree were most likely to use counselling [28]. It could be concluded that subjects found it more suitable to provide information and seek guidance keeping their anonymity intact.

22 Student Perception Regarding Chatbot for Counselling …

271

Table 22.1 Demographical details Demographic n = 327

No. of respondents

Percentage (%)

Age (in years)

Young (18–23)

281

85.93

46

14.07

Gender

Male

198

60.55

Female

129

39.45

Graduation

147

44.95

Post-graduation

180

55.05

Middle age (23–27)

Education

22.3 Methodology 22.3.1 Participants and Procedure For this research, the author developed a questionnaire which has consisted of 10 questions and having three dimensions, namely: Personal life, Professional life and General life. Through, Google form data was collected from the students from different colleges located in NCR, India. The filled questionnaires of 356 responses were received. Out of which 327 were found to be suitable for further analysis. The reliability of questionnaire was measured and reported as 0.875 and for checking the validity of questionnaire; it was sent to the experts and revised twice as per the suggestion given by them. Demographical profile of respondents is shown in Table 22.1.

22.3.2 Measures The name of the questionnaire is SHIVBOT to analyze the perception of students to implement Shivbot in universities and colleges. The Questionnaire was developed by author having 10 number of questions. The questions were scaled on 5-point Likert scale. The Reliability of questionnaire is 0.875.

22.3.3 Objectives of the Study The main objective of the study is to analyze the perception of the students about implementation of the chat bots (Counseling App) in the college and university. The objectives were further divided into three dimensions: a.

To analyze that student were comfortable in sharing their personal life with chatbot.

272

b. c.

S. Agarwal et al.

To analyze that student were comfortable in sharing their professional life with chatbot. To analyze that student were comfortable in sharing their general life with chatbot.

22.3.4 Hypothesis of the Study The main hypothesis of the study is that students feel comfortable while sharing the information/thoughts with chatbots The hypotheses were further divided into three dimensions: a. b. c.

Students feel comfortable in sharing their personal life with chatbot. Students feel comfortable in sharing their professional life with chatbot. Students feel comfortable in sharing their general life with chatbot.

22.4 Result and Discussion The findings of the data were represented in the form of bar graph and pie charts for easily understanding to the large number of audiences. To differentiate between the comfort level of sharing the information with chatbot or human, question was raised that “Are you comfortable in sharing your personal information to counselor (human interaction)?” The first finding of the result was shown in Fig. 22.1. A majority of 52.8% Students are sometimes comfortable in sharing the personal information with counselor. Out of 327 students, there are 17.3% students who are rarely comfortable sharing any personal information to counselor, 12.6% students are undecided about comfortably sharing personal information, 52.8% students are sometimes comfortable in sharing personal information, 10.2% students are almost always comfortable sharing personal information with counselor, 7.1% students are not at all comfortable in sharing any personal Information. Fig. 22.1 Comfortable in sharing your personal information to counselor

22 Student Perception Regarding Chatbot for Counselling …

273

The second findings about “Are you comfortable in sharing your personal information to chatbot/counseling App (Artificial intelligence)?” shows that a majority of 29.1% students feel comfortable while sharing your personal information with a Chatbot (Counselling App) which is shown in Fig. 22.2. Out of 327 students, there are 22.8% students who are rarely comfortable sharing any personal information to Chatbot, 20.5% students are undecided about comfortably sharing personal information, 29.1% students are sometimes comfortable in sharing personal information, 6.3% students are almost always comfortable sharing personal information to Chatbot, 21.3% students are not at all comfortable in sharing any personal Information to Chatbot. The third findings about Students ever chatted with a Chatbot is shown in Fig. 22.3. The findings shows that a majority 84.3% students have never chatted with a Chatbot (Counselling App) and 15.7% have chatted with a chatbot on marketing website, recruitment website etc. The findings of the results in Fig. 22.1 shows that students are more comfortable in sharing their personal information to counselor (human interaction) which is near about 10.2% while on the other hand, students are more comfortable in sharing their personal information to chatbot (AI) shown in Fig. 22.2 which is near about 6.3% which clearly shows that students are not able to share their thoughts Fig. 22.2 Comfortable in sharing your personal information to chatbot/counseling App (Artificial intelligence)

Fig. 22.3 Chatted with any chatbot

274

S. Agarwal et al.

and feelings with any one of source because in India, we hardly talks about psychological ailment such as depression, stress, anxiety, negative emotions etc. which consequently leads to the increasing death toll among millennial. So, it is suggested to the universities and colleges to implement Chatbot in the colleges and provides them proper knowledge about the secrecy of the data. So that students feel free awhile chatting with chatbots which is lacking in human interaction (Counselling). The fourth findings shows that a majority of 44.1% Students feel chatbots can empathize little with them because AI technology is booming day in and day out is represented in Fig. 22.4. Out of 327 Students, 20.5% students feel chatbots cannot at all empathize with them, 22.8% Students feel chatbots can empathize very little with them, 10.2% Students feel chatbots can empathize very much with them and only 2.4% Students feel chatbots can completely empathize with them. The fifth findings of the result depicted that a majority of 42.5% Students trust Chatbots (Counselling Apps) little for communication is clearly represented in Fig. 22.5. Out of 327 Students, 13.4% of students did not trust Chatbots for communication at all, 33.9% Students trust Chatbots very little for communication, 10.2% Students trust Chatbots very much for communication and 0% Students completely trust Chatbots for communication as without practically using any app for counseling, Fig. 22.4 Chatbot can empathise with you

Fig. 22.5 Trust chatbot for communication

22 Student Perception Regarding Chatbot for Counselling …

275

Fig. 22.6 Feel relieved or happy after conversing with chatbots

and it’s a human tendency to trust any app. Thus, we can suggest that approximately 89.8% of students trust on the chatbots for communication which varies from very little to very much. Therefore, to reduce the psychological distress among community of students, colleges and universities should implement Chatbot in their premises. The sixth findings of the results show that students feel relieved or happy after conversing with chatbots is represented in Fig. 22.6. A majority 55.1% does not feel relieved or happy after conversing with chatbots and 44.9% students feel relieved or happy after conversing with chatbots. The seventh findings depict that students level of convenient sharing following information about your personal life with a chatbots is represented in Fig. 22.7. Personal Life consists of about your parents, about your siblings, about your cousins, about your Relationship status if any, about your friends, about your family Problems, if any, about your financial Condition which is discussed one by one in detail: a. About Sibling Out of 327 students, 22% students are never convenient sharing information about sibling, 31% students are rarely convenient sharing information about sibling, 9% students are convenient once in a while sharing information about sibling, 26% students are sometimes convenient sharing information about sibling, 11% students are almost always convenient sharing information about sibling.

Fig. 22.7 Converse with the chatbot/counseling app regarding your personal life

276

S. Agarwal et al.

b. About Parents Out of 327 students, 19.6% students are never convenient sharing information about Parents, 25% students are rarely convenient sharing information about Parents, 11% students are convenient once in a while sharing information about parents, 24.4% students are sometimes convenient sharing information about parents, 18.8% students are almost always convenient sharing information about parents. c. About Cousins Out of 327 students, 22% students are never convenient sharing information about cousins, 28.3% students are rarely convenient sharing information about cousins, 3.1% students are convenient once in a while sharing information about cousins, 25.9% students are sometimes convenient sharing information about cousins, 16.5% students are almost always convenient sharing information about cousins. d. About Relationship Status Out of 327 students, 18.11% students are never convenient sharing information about relationship status, 27.5% students are rarely convenient sharing information about relationship status, 9.4% students are convenient once in a while sharing information about relationship status, 22% students are sometimes convenient sharing information about relationship status, 22.4% students are almost always convenient sharing information about relationship status. e. About Friends Out of 327 students, 44% students are never convenient sharing information about Friends, 22.8% students are rarely convenient sharing information about Friends 10.2% students are convenient once in a while sharing information about Friends. 15.7% students are sometimes convenient sharing information about Friends 7.08% students are almost always convenient sharing information about Friends. f. About Family Problems Out of 327 students, 44.09% students are never convenient sharing information about Family problems 25.1% students are rarely convenient sharing information about Family problems, 7.8% students are convenient once in a while sharing information about Family problems, 15.7% students are sometimes convenient sharing information about Family problems, 7% students are almost always convenient sharing information about Family problems. g. About Financial Condition Out of 327 students, 45.6% students are never convenient sharing information about financial condition, 23.6% students are rarely convenient sharing information about financial condition, 7.8% students are convenient once in a while sharing information about financial condition, 19.6% students are sometimes convenient sharing information about financial condition, 3.1% students are almost always convenient sharing information about financial condition.

22 Student Perception Regarding Chatbot for Counselling …

277

The eighth findings of the results depict that student’s level of convenient sharing following information about your Professional life with a Chatbot (Counselling App) in Fig. 22.8. The professional life was measured in terms of about your learning habits, about your reading habits, about your writing habits, about your attendance issues, about your performance in internal exams, about your performance in external exams, about your day to day activities at workplace which is discussed one by one in detail: a. About Learning Habits Out of 327 students, 11% students are never convenient sharing information about Learning habits, 25.9% students are rarely convenient sharing information about Learning habits, 7.8% students are convenient once in a while sharing information about Learning habits, 31.4% students are sometimes convenient sharing information about Learning habits, and 23.6% students are almost always convenient sharing information about learning habits. b. About Reading Habits Out of 327 students, 8.6% students are never convenient sharing information about Reading habits, 22.8% students are rarely convenient sharing information about Reading habits, 13.3% students are convenient once in a while sharing information about Reading habits, 31.4% students are sometimes convenient sharing information about Reading habits, 23.6% students are almost always convenient sharing information about Reading habits. c. About Writing Habits Out of 327 students, 12.5% students are never convenient sharing information about writing habits, 21.2% students are rarely convenient sharing information about writing habits, 11% students are convenient once in a while sharing information about writing habits, 31.4% students are sometimes convenient sharing information about writing habits, and 23.6% students are almost always convenient sharing information about writing habits. d. About Attendance Issue Out of 327 students 17.3% students are never convenient sharing information about attendance issue, 21.2% students are rarely convenient sharing information about attendance issue, 14.9% students are convenient once in a while sharing information

Fig. 22.8 Converse with the chatbot/counseling App regarding your professional life

278

S. Agarwal et al.

about attendance issue, 25.9 % students are sometimes convenient sharing information about attendance issue, and 20.4% students are almost always convenient sharing information about attendance issue. e. About Performance in Internal Exams Out of 327 students, 18.1% students are never convenient sharing information about performance in internal exams, 23.6% students are rarely convenient sharing information about performance in internal exams, 14.9% students are convenient once in a while sharing information about performance in internal exams, 23.6% students are sometimes convenient sharing information about performance in internal exams, 21.2% students are almost always convenient sharing information about performance in internal exams. f. About Performance in External Exams Out of 327 students, 18.8% students are never convenient sharing information about performance in external exams, 19.6% students are rarely convenient sharing information about performance in external exams, 14.1% students are convenient once in a while sharing information about performance in external exams, 22.8% students are sometimes convenient sharing information about performance in external exams, 24.4% students are almost always convenient sharing information about performance in external exams. g. About Day to Day Activities at Workplace Out of 327 students, 16.5% students are never convenient sharing information about day to day activities at workplace, 20.4% students are rarely convenient sharing information about day to day activities at workplace, 18.1% students are convenient once in a while sharing information about day to day activities at workplace, 28.3% students are sometimes convenient sharing information about day to day activities at workplace, 14.9% students are almost always convenient sharing information about day to day activities at workplace. The last but not the least findings of the study depicted in Fig. 22.9 about how much students are ready to converse with the chatbot regarding their general life. The general life was further bifurcated into about your happiness, about your sorrows,

Fig. 22.9 Converse with the chatbot regarding your general life

22 Student Perception Regarding Chatbot for Counselling …

279

about your stress level, about your level of motivation which is discussed one by one in detail: a. About Happiness Out of 327 students, 23.6% students are never convenient sharing information about happiness, 28.3% students are rarely convenient sharing information about happiness, 13.3% students are convenient once in a while sharing information about happiness, 19.6% students are sometimes convenient sharing information about happiness, and 14.9% students are almost always convenient sharing information about happiness. b. About Sorrow Out of 327 students, 25.9% students are never convenient sharing information about sorrow, 33% students are rarely convenient sharing information about sorrow, 13.3% students are convenient once in a while sharing information about sorrow, 20.4% students are sometimes convenient sharing information about sorrow, and 7% students are almost always convenient sharing information about sorrow. c. About Stress Level Out of 327 students, 27.5% students are never convenient sharing information about stress level 23.6% students are rarely convenient sharing information about stress level 22% students are convenient once in a while sharing information about stress level 18.8% students are sometimes convenient sharing information about stress level 6.2% students are almost always convenient sharing information about stress level. d. About Level of Motivation Out of 327 students, 20.4% students are never convenient sharing information about level of motivation, 20.4% students are rarely convenient sharing information about level of motivation, 16.5% students are convenient once in a while sharing information about level of motivation, 25.1% students are sometimes convenient sharing information about level of motivation, 17.3% students are almost always convenient sharing information about level of motivation.

22.5 Conclusion The main purpose of this study was to find out the perception of students about chatbots [28]. Numerous studies have suggested the use of chatbots in guidance and counselling [16]. Chatbots have been helpful in reducing stress level of individuals and help them deal with both the work related and personal life wary situations in a better way. For a long time, human interaction has been the sole way of sharing personal thoughts and feelings, but with giant leaps in Artificial Intelligence and other Cognitive behavior monitoring systems and tools, now there lies a golden opportunity to test the implementation and benefits of using advanced technology in the field of Guidance and Counselling.

280

S. Agarwal et al.

The study intended to observe and analyze the comfort level of students while sharing their personal, professional and general information with a chatbot. Students generally refrain themselves from sharing their personal information with human counselors; the study suggests the use of chatbot technology in higher education to capture a better perspective of the various problems faced by the students in their dayto-day life; and with the use of advanced technology, work out enhanced solutions to help students deal with various situations and events in their student life. Modern Chatbots are packed with various features like 24 h availability, attitude sensing, conversational maturity, emotional support and SOS alarms which observe and track the behavior of subjects and help them in the best possible ways. With time Chatbot Technology can surely be a milestone in field of Guidance and Counselling.

22.6 Limitations The present study has several limitations which deserve to be mentioned. First, for the purpose of this research, a small sample size of 327 respondents has been used. Significant increase in the sample size can provide an enhanced outcome. Second, in this research, data was collected from people residing in Delhi/NCR region. A larger demographic area under consideration can provide more variation and diversity in data. Lastly, data was collected only through online survey; thus a Quantitative Approach was used for collection of data. A more detailed analysis can be done with the help of Qualitative Approach of data collection.

References 1. UGC: University-UGC. 16/5/2020, (2020) from https://www.ugc.ac.in/oldpdf/Consolidated% 20list%20of%20All%20Universities.pdf 2. Prasad, N.A., Prthaban, L., Rana, M.E.: Use of recommendation engine and chatbot for depression consultancy platforms. J. Crit.Al Rev. 7(3), 2020 (2019) 3. Wang, X., Joyce, N., Namkoong, K.: Investigating college students’ intentions to seek online counseling services. Commun. Stud. 1–18 4. Yigitcanlar, T., Desouza, K.C., Butler, L., Roozkhosh, F.: Contributions and risks of artificial intelligence (AI) in building smarter cities: Insights from a systematic review of the literature. Energies 13(6), 1473 (2020) 5. Elcholiqi, A., Musdholifah, A.: Chatbot in Bahasa Indonesia using NLP to provide banking information. IJCCS Indones. J. Comput. Cybern. Syst. 14(1), 91–102 6. Eklund, J., Isaksson, F.: Identifying & evaluating system components for cognitive trustin AI-Automated service encounters: Trusting a study-& vocational chatbot (2019) 7. Sturk, H., Crowther, R., Kavanagh, D.J.: Head to health: Practitioner perceptions of the new digital mental health gateway. Aust. J. Rural Health 27(5), 448–453 (2019) 8. Sánchez, P.D.A.: Perspectives from university graduates facing AI and automation in Ireland: How do Irish higher education’s graduates from Maynooth University perceive AI is going to impact them? Doctoral dissertation, National University of Ireland Maynooth (2019) 9. Marwan, A.: Impact of artificial intelligence on education for employment:(learning and employability Framework) (2019)

22 Student Perception Regarding Chatbot for Counselling …

281

10. Wilhelm, S., Weingarden, H., Ladis, I., Braddick, V., Shin, J., Jacobson, N.C.: Cognitivebehavioral therapy in the digital age: presidential address. Behav. Ther. 51(1), 1–14 (2020) 11. Lee, J.W., Yang, H., Kim, J.G.: Developing scenario for implementation of counseling chatbot and verifying usefulness. J. Korea Contents Assoc. 19(4), 12–29 (2019) 12. Blashki, G., Lock, S.: Artificial intelligence and mental health 13. Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., O’Neill, S., McTear, M.: Assessing the usability of a chatbot for mental health care. In: Proceedings of the International Conference on Internet Science, pp. 121–132. Springer, Cham (2018) 14. Lee, K., Jo, J., Kim, J., Kang, Y.: Can chatbots help reduce the workload of administrative officers? -Implementing and deploying FAQ chatbot service in a university. In: Proceedings of the International Conference on Human-Computer Interaction pp. 348–354. Springer, Cham (2019) 15. Srivastava, S.K.: Artificial intelligence: way forward for India. JISTEM-J. Inf. Syst. Technol. Manag. 15 (2018) 16. Yorita, A., Egerton, S., Oakman, J., Chan, C., Kubota, N.: A robot assisted stress management framework: Using conversation to measure occupational stress. In: Proceedings of the 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC) pp. 3761–3767. IEEE 17. Bird, T., Mansell, W., Wright, J., Gaffney, H., Tai, S.: Manage your life online: a web-based randomized controlled trial evaluating the effectiveness of a problem-solving intervention in a student sample. Behav. Cogn. Psychother. 46(5), 570–582 (2018) 18. Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., O’Neill, S., McTear, M.: Towards a chatbot for digital counselling. In: Proceedings of the 31st British Computer Society Human Computer Interaction Conference, p. 24. BCS Learning & Development Ltd (2017) 19. Eliamani, M.P., Richard, M.L., Peter, B.: Access to guidance and counseling services and its influence on Students” school life and career choice. Afr. J. Guid. Couns. 1(1), 007–015 (2014) 20. De Vos, A., Nieuwdorp, J., Keulemans, R.: Conversational agent-based behaviour change support systems: Comparing the technical aspects of recent applications 21. Priya, N.B., Juvanna, I.: An android application for university online counseling. Int. J. Comput. Sci. Mob. Comput. 3(2), 261–266 (2014) 22. Cicco, G.: Building effective supervisory relationships in the online counseling course: Faculty and student responsibilities. J. Sch. Educ. Technol. 10(2), 1–8 (2014) 23. Bhakta, R., Savin-Baden, M., Tombs, G.: Sharing secrets with robots? In: EdMedia+ Innovate Learning, pp. 2295–2301. Association for the Advancement of Computing in Education (AACE) (2014) 24. Prasetyaningrum, T., Gregorius, R.M.: Design and implementation of mobile leadership with interactive multimedia approach. In: Proceedings of the International Conference on Multimedia, Computer Graphics, and Broadcasting, pp. 217–226. Springer, Berlin, Heidelberg. (2011) 25. Richards, D., Viganó, N.: Online counseling: A narrative and critical review of the literature. J. Clin. Psychol. 69(9), 994–1011 (2013) 26. Suyoto, P.: Development of mobile application social guidance and counseling for junior high school. Int. J. Comput. Electr. Autom. Control Inf. Eng. 7, 1398–1404 (2013) 27. Coughlan, J., Macredie, R., Patel, N.: Moving face-to-face communication to web-based systems (2007) 28. Young, K.S.: An empirical examination of client attitudes towards online counseling. Cyberpsychol. Behav. 8(2), 172–177 (2005) 29. Agarwal, S.: Trust or No trust in Chat bots: A dilemma of millennial, published in book entitled cognitive computing for human-robot Interaction: Principles and practices by Elsevier Inc (2021) 30. Agarwal, S., Linh N.T.D.: A study of student’s subjective well-being through chatbot in higher education. In: Balas V.E., Solanki V.K., Kumar R. (eds) Further advances in internet of things in biomedical and cyber physical systems. Intell. Syst. Ref. Libr. 193, (2021). Springer, Cham.

Chapter 23

Empirical Performance Evaluation of Machine Learning based DDoS Attack Detections Bao-Sam Tran, Thi-Huyen Ho, Thanh-Xuan Do, and Kim-Hung Le

Abstract A distributed denial-of-service attack (DDoS) is a critical attack-type that strongly damages the Quality of Service (QoE). Although various novel security technologies have been continually developing, completely preventing DDoS threats is still unreached. Hence, applying deep learning to detect DDoS attacks effectively is high interest. However, comprehensively analyzing these techniques remains unobservant. In this paper, we present a solid architecture supporting evaluating machine-learning-based DDoS detection techniques from both public and self-generated datasets. A high-accuracy ensemble DDoS detection method is proposed from the evaluation results. Furthermore, we expect that these results could be essential resources for later DDoS researches. Furthermore, the study also provides an overview of the features, labels from which there is a basis for creating a complete dataset used for DDoS attack detection methods. Keywords Evaluation framework · Intrusion Detection System (IDS) · Machine learning · DDos Detection

B.-S. Tran · T.-H. Ho · T.-X. Do · K.-H. Le (B) University of Information Technology, Ho Chi Minh City, Vietnam e-mail: [email protected] B.-S. Tran e-mail: [email protected] T.-H. Ho e-mail: [email protected] T.-X. Do e-mail: [email protected] Vietnam National University, Ho Chi Minh City, Vietnam © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_23

283

284

B.-S. Tran et al.

23.1 Introduction In recent years, smart technologies have been raising the fourth industrial revolution (e.g., Internet of Things, Artificial Intelligence, Big Data). Along with the increasing demand for humans technology, information security and system protection are beware [1]. One of the most dangerous network attacks committed to a computer system is the Distributed Denial-of-Service (DDoS) attacks. Such attack is a malicious attempt to disrupt the normal traffic of a targeted server, service, or network by a flood of Internet traffic. Since service providers place growing importance on delivering cloud-based services to enterprises and consumers, it is no wonder that attackers are increasingly targeting these services with DDoS attacks [2]. This has caused significant financial losses to industry and governments worldwide, as shown in information security reports. The largest DDoS attack to date took place in February of 2020. This attack saw incoming traffic at its peak at a rate of 2.3 Terabits per second (Tbps). Amazon Web Services reported that it mitigated this massive attack but did not disclose which customer is targeted by the attack. Returns 2018, before this 2.3 Tbps attack, the largest verifiable DDoS attack on record targeted GitHub, a popular online code management service used by millions of developers. This attack reached 1.3 Tbps, sending packets at a rate of 126.9 million per second [3]. Another largest ever at the time attack is in October 2016, targeting systems operated by Domain Name System (DNS) provider Dyn. The Dyn DDoS attack set a record at 1.2 Tbps. This attack is made using malware called Mirai. Mirai creates a botnet out of compromised Internet of Things (IoT) devices such as cameras, smart TVs, radios, printers, and even baby monitors [4]. These compromised devices create attack traffic by sending requests to a single victim. This attack is devastating and disrupted many significant sites, including Airbnb, Netflix, PayPal, Visa, Amazon, The New York Times, Reddit, GitHub, CNN, and many others in Europe and the US. In the digital age, the security of applications and networks is of paramount importance. DDoS detection and mitigation have been under study in both the scientific community and industry for several years. To prevent heavy damage caused by DDoS attacks, attack detection mechanisms with fast real-time computation are among the top concerns [5]. On the other hand, the evaluation of DDoS attack detection techniques is highly recommended. Several DDoS detection mechanisms using machine learning techniques are recently proposed. However, these mechanisms mainly rely on the dataset collected from the network traffic when the DDoS attack occurs [6]. This makes them only efficient for a specific context. In this paper, we have designed, implemented, and evaluated various models of detecting DDoS attacks. These models are trained through self-generated and public datasets. More in detail, we focus on the following objectives: • We build an DDoS evaluation framework that effectively collects network traffic and use these network traffic to evaluate the performance of several machine learning based DDoS detection. • Based on the evaluation results, we propose an ensemble approach from different DDoS detection that is able to increase the detection accuracy.

23 Empirical Performance Evaluation of Machine Learning …

285

• We use test parameters to analyze and evaluate the performance of DDoS detection algorithms on both servers and edge devices (emulated by Raspberry Pi). Based on this analytical result, we produce usage suggestions to effectively deploy machine learning based IDSs to different context. The remainder of the paper is organized as follows: Sect. 23.2, we present our motivation and the systematic overview of artificial intelligence. Section 23.3 describes building a system to collect data and train the DDoS attack detection model. Section 23.4, delves into the design and the implementation details of the proposed models to detect DDoS attacks. Finally, in Sect. 23.5 we draw our conclusions.

23.2 Related Works The detection of DDoS attacks contributing to computer networks’ protection is very much interested and included in recent studies. Many articles have access to this topic. The input dataset requirement is very important, which gets the most accurate results of detecting attacks. Iman Sharafaldin et al. [7] evaluated publicly available DDoS attack datasets from 2007 to 2018, CAIDA UCSD, and DARPA 2000 [8]. Moreover, propose a new classification for DDoS attacks, broken down into reflection-based DDoS and Exploitation-based attacks. The authors generated a new dataset, namely CICDDoS2019, which is also the dataset we used in our study. The dataset is labeled with over 80 features, fixing all the shortcomings of the others they have looked. The authors here also propose a DDoS attack detection method through building models to capture patterns by training data using four popular machine learning algorithms: ID3, Random Forest, Naive Bayes, and Logistic Regression. The authors provide some of the most critical feature sets for detecting different types of DDoS attacks with their respective weights from the results obtained. Judging objectively, the feature set of the author’s Iman Sharafaldin et al. [7] has a lot to do with the most intuitive evaluation of data, with features that do not play a role in the detection of DDoS attacks. Because for a well-explored field of research like the IDS, the performance of a perfect IDS is heavily dependent on the data features. Adel Binbusayyis and Thavel Vaiyapuri are IEEE members [9], which have performed research to identify and benchmark the set of potential features that can characterize network traffic for intrusion detection. The authors used four different performance evaluation measures, including correlation, consistency, information, and distance, tested on four sets of intrusion detection evaluation datasets, namely KDDCup’99, NSL-KDD, UNSW-NB15, and CICIDS2017. The results achieve increased detection efficiency with a detection rate of 3.2%, a false alarm rate of 38%, and 12% detection time. These results demonstrate that the proposed feature-selective approach contributes more potential features than modern approaches, leading to achieving promising performance gains, contributing as a standard. In the future, to build an effective IDS.

286

B.-S. Tran et al.

Habibi et al. has researched whitelisting IDS, or Heimdall, for protection against DDoS attacks [10]. Heimdall queries each URL (or DNS response) to VirusTotal [11] to mark it as malicious or mild, and the URL (or DNS response) is then added to the whitelist or blacklist. Hence only network traffic from whitelisted destinations is allowed to/from devices behind the gateway. Many researchers have commented that this approach’s limitation is that it relies on VirusTotal for security, meaning that other attacks that have not been analyzed by VirusTotal will go undetected. Furthermore, another limitation has also been indicated that this system keeps traffic to a new server in a queue to be analyzed by a remote service (i.e., VirusTotal). As DDoS attacks become more complex and sophisticated, IoT edge devices may also become the target of these attacks. Many researchers direct research on the IDS approach to DDoS attacks at IoT devices. In particular, with some recent approaches, many researchers have proposed solving this problem by running IDS out of IoT devices [12] to analyze the computability of some IDS tools. The familiar open-source is Snort and Bro. With the Raspberry Pi 2 device scenario, the authors conducted three network attacks: SYN Flood, Address Resolution Protocol Spoofing (ARP), and port scanning. The results obtained based on their report claim that both Snort and Bro can detect all network attacks that initiate while, relative to the total cost measured, Snort performs a little better than Bro. DDoS attacks have become a severe problem, and most research related to this issue revolves around how to approach the topic effectively. One of the closer approaches is to come up with a classification for these attacks. Unlike the classification of the author’s Iman Sharafaldin of the dataset CICDDoS2019 above, Yizhen Jia et al. [13] proposed a classification focusing on IoT DDoS attacks in their study. Specifically, two types of victim resource exhaustion and network bandwidth: Flooding-based Attacks exploit vulnerabilities that create disguised large network packets and Slow Request/Response Attacks by inducing spoofing high- workload requests or responses such as attackers establish fake HTTP connections to the victim’s device and render the victim unavailable. Once DDoS attacks are classified, researchers will turn to defense techniques for their projects. There are several techniques, such as IP traceback, packet marking, entropy variations, or intrusion detection and prevention. Research by Roschke et al. in 2009 [14] launched an Intrusion Detection System architecture using a cloud computing model to collect alerts from central cloud management units and sensors and analyze them. Nychis et al. [15] analyzed the possibilities and correlation of various entropy-based metrics such as stream header features including IP address, port and stream size, and behavior related to packet count in the process of touch between buttons. Furthermore, in this way, researchers can detect DDoS attacks based on metrics with those features.

23 Empirical Performance Evaluation of Machine Learning …

287

23.3 The proposed framework architecture As shown in Fig. 23.1, we first build an attack model to generate our data. After the datasets are available, they are divided into two parts: 80% for training and 20% for testing. In the training section to create a model machine learning, it has available features, and the column contains label information. Then run the test features through the model taken from the train, and prediction results, compared with the test label. The final result is the model’s accuracy compared to the test, including the important parameters: F1-score, Accuracy, Precision. Besides, for other datasets from articles, we also perform the same steps as in the model. Some of the tools that we use to build the data collection model in our research include: • Bonesi [16]—DDoS Botnet Simulator is a Botnet traffic simulation tool, generating ICMP, UDP, TCP (HTTP) flood attacks from a source IP address, URL. This simulator is a network traffic generator for different types of protocols. The properties of the packets and connections created can be controlled by several parameters such as the sending rate or packet size or determined by chance. It spoofs source IP addresses even when generating TCP traffic. Hence, Bonesi includes a simple TCP stack for handling TCP connections in promiscuous mode. To function correctly, one must ensure that response packets are routed to the host on which the Bonesi is running. Therefore, Bonesi cannot be used in arbitrary network infrastructures. The most advanced types of traffic that can be generated are HTTP requests. • CICFlowMeter [17]—is a network traffic generator and analyzer. It can create bidirectional streams, where the first packet specifies the forward (source to destination) and backward (source to source) directions, thus having more than 80

Fig. 23.1 The overall architecture of DDoS evaluation framework

288

B.-S. Tran et al.

Table 23.1 Hosts’s information in proposed evaluation architecture Machine OS IPs Attacker machine Normal machine Victim machine

Ubuntu 16.04 Windows 8.1 Kali linux

192.168.243.144 192.168.243.143 192.168.243.142

network traffic features. Numbers such as Duration, Number of packets, Number of bytes, Length of packets. It can be calculated individually in the forward and backward directions. It creates Bidirectional Flows (Biflow), where the first packet specifies the forward direction (source to destination) and vice versa (destination to source). After that, more than 80 features will be listed as Duration, Number of packets, Number of bytes, Length of packets, which are also calculated separately in the transition direction, and the reverse direction. The output of CICFlowMeter is a CSV file with six columns labeled for each stream, namely FlowID, SourceIP, DestinationIP, SourcePort, DestinationPort, and Protocol with more than 80 network traffic features. TCP flows are usually terminated by FIN packet, while UDP streams are terminated by flow timeout. The flow timeout position can be arbitrarily specified, e.g., 600s for both TCP and UDP. • HTTP—ping is a small, free, easy-to-use Windows command-line utility that probes a given URL and displays relevant statistics. It is similar to the popular ping utility, works over HTTP/S instead of ICMP, and a URL instead of a computer name/IP address. Due to a weak machine configuration, we decided to set up a minimum number of three machines to build this model. Each machine in the model plays a different role and in the same LAN, are described in Table 23.1. The first server acts as the attacker: We install and configure the DDoS Botnet Simulator to send out DDoS irregular traffic, with the Protocol we choose TCP playing the primary role; A Normal Traffic role server: We use the HTTP-ping tool to generate normal HTTP traffic; And a victim target machine: We install Wireshark software to capture packets sent to track the attack. Attack Context: We let attackers attack for a specified period, not always attack. That period, we will re-write as a label time for our feature Label. We spend the first 30-45 minutes modeling time for normal traffic. We run HTTP-ping on the machine playing the role of Normal Traffic. Every 10 seconds, HTTP requests will be sent to the Victim machine on port 80 (Note, here we have tested that all three machines in the model have port 80 in Listening state) until we stop the outgoing traffic. Then we went from the Attacker machine and running the Bonesi tool, causing DDoS traffic to the victim machine. Our attack scale is as follows: The TCP protocol we use, our input file contains a list of 50,000 different generated IP addresses, along with browser and URL options. Finally, with different label times (DDoS attack intervals) in each file for at least eight runs, about 14 million packets are captured in the pcap file format.

23 Empirical Performance Evaluation of Machine Learning …

289

After running the captured attack model, our final dataset includes network traffic (in .pcap file format) and a log of events that occurred on the victim machine. To extract features from raw data, we use the CICFlowMeter tool to extract more than 80 traffic features and save them as a CSV file. This is how we collect our dataset— self-created dataset in sections of this study. And then from this dataset, we use AI techniques for analysis.

23.4 Evaluation Results This section evaluates several of our DDoS attack datasets and publicly available DDoS attack datasets (KDD, CIC-DDoS2019). We explain the need for a comprehensive and reliable dataset to test and validate DDoS attack detection systems. We implemented proposed models using an anaconda environment with python version 3 and some libraries of python like NumPy, Pandas, Scikit-learn. That is running Windows 10 OS, x64-based processor, PC with Intel ®, CORE(TM) i5-6600K CPU @ 3.50GHz, 3.50 GHz, 8 GB RAM.

23.4.1 Dataset First, the self-created dataset is in the CSV format, and details attack traffic and normal traffic to the victim and responses to the attack from the victim. They include 2079351 entries from 0 to 2079250 and 80 columns from 0 to 79. Packets with TCP protocol make up 99.92% of the dataset. The rest are packets with UDP and HOPORT protocol. This dataset has two labels, DDoS and Normal, in which DDoS packets account for 13.87%. The second dataset we use in this paper is NSL-KDD [18]. The KDD data part used for training includes 125973 rows and 43 columns. Meanwhile, the data used for testing has 22544 rows and 43 columns. Features of the data have been reduced from 85 to 43 features. DDoS packets account for about 36.46% of the data file. The NSL-KDD is also a dataset used in training and testing on raspberry pi because the size conforms to the raspberry pi’s hardware restrictions. CICDDoS20191 is the third dataset used in our training and evaluation. This dataset published in 2019 by Iman Sharafaldin and colleagues, includes 12 small datasets containing different types of attacks (12 DDoS attacks include NTP, DNS, LDAP, MSSQL, NetBIOS, SNMP, SSDP, UDP, UDP-Lag, WebDDoS, SYN, and TFTP). In this article, we focus on two datasets DrDoS_UDP and DrDoS_SNMP.

1

Dataset is publicly available at http://www.unb.ca/cic/datasets/CICDDoS2019.

290

B.-S. Tran et al.

• The DrDoS_UDP dataset is extracted, including 200,000 lines, 80 features. The proportions of DDoS and Normal packets in the dataset are 1998157 and 1843, respectively. • The DrDoS_UDP dataset is extracted, consisting of 100,000 lines, 80 features. The rates of DDoS and Normal packets in the dataset are 999296 and 704, respectively. In the section labeling packets, we proceed to manually label 0 for packets normal and label 1 for packets of DDoS. In each dataset, we extract 20% of records for testing purposes. 20% of this data is taken randomly in the dataset and did not participate in the training process. The dimensions of the testing data portion of each dataset are as follows: • • • •

Self-created dataset: Test dataset has size 415 870 entries and 80 columns. NSL-KDD dataset: Test dataset has size 22544 entries and 43 columns. DrDoS_UDP: Test dataset has 400000 entries and 80 columns. DrDoS_SNMP: Test dataset has 200000 entries and 80 columns.

23.4.2 Models We use datasets to train and test machine learning models such as Decision Tree, KNearest Neighbors (KNN), Naive Bayes, Random Forest, Neural Network, Boosted Trees Classifier, Logistic Classifier, and Stochastic Gradient Descent. In addition to the above classification models, we tried to combine the models together to try to find the optimal model for each data set. Merging models is deployed through the sklearn.ensemble library and uses the VotingClassifier method. • Ensemble Model [19] includes different classification models that operate in parallel. Each classifier builds a different data model based on the training dataset provided in advance. • A majority vote is one of the traditional and popular ways to combine classifiers. The outputs of the predictors are combined and used by majority vote to get the combined model’s final output. • Each association model category will predict the nominal class label for the sample. The most anticipated label is then selected as the output of the ensemble model. For each dataset, based on the test results of individual models, we will try to combine different basic machine learning classifiers to increase the efficiency of the classification model.

23.4.3 Results In each dataset, we use 80% records to train the model and 20% records to evaluate the training model. The testing is evaluated through 9 parameters as follows:

23 Empirical Performance Evaluation of Machine Learning …

291

• True Positive: the data point is predicted to be positive, and the label is actually positive. • True Negative: the data point that is predicted to be negative and is actually negative. • False Negative: The data point is predicted to be negative, and the label is actually positive. • Accuracy: The ratio between the correctly predicted number of data points over total data points in the test set. Accuracy has the following formula: Accuracy =

T PR + T NR T PR + FPR + T NR + FNR

• Precision: is defined as the ratio of the number of True Positive points on data points that are predicted to be Positive (True Positive + False Positive). High precision means the accuracy of the predicted data points is high. Precision has the following formula: TP Pr ecision(P) = T P + FP • Recall: the ratio of the number of true positive points to the number of actually positive points (True Positive + False Negative). A high recall means a high True Positive Rate, which means a low rate of omitting positives. Recall(R) =

TP T P + FN

• F1-score: is the harmonic mean of precision and recall (assuming these two quantities are nonzero). The higher the F1-score represents, the better the classifier. F1 − scor e =

2 ∗ Pr ecition ∗ Recall Pr ecition + Recall

• Mis-Classifier: number of data points predicted incorrectly. Table 23.2 shows the evaluation results of the models trained by the self-generated dataset. We see that most of the classification models are highly accurate, especially the Decision Tree, Random Forest, and Boosted Tree Classifier (99.99%). In addition, the classification of the models (as assessed by the f1-score) is approximately 99.99%, meaning that the proportion of DDoS packets found is mostly correct. In contrast, the Naive Bayes and Stochastic Gradient Descent models show weaker classification than the other models. To evaluate DDoS attack detection models more intuitively, we continue to use NSL-KDD data to train and test these models. Table 23.3 is a breakdown of the test parameters of the models on the NSL-KDD dataset. As can be seen, the models still perform well on NSL-KDD data, although the accuracy and classification level are

292

B.-S. Tran et al.

Table 23.2 The evaluation results of the DDoS detection models trained by the self-generated datasets Classifier Accuracy Precision Recall F1-Score Mis-Classifier Decision Tree KNN Naïve Baye Random Forest Neural Network Logistic Classifier Stochastic Gradient Descent

1.00 1.00 0.95 1.00 0.99 1.00 0.87

1.00 1.00 0.94 1.00 0.87 1.00 0.87

1.00 0.99 0.82 1.00 1.00 0.96 1.00

1.00 1.00 0.87 1.00 0.93 0.98 0.93

16 63 18178 25 6801 2015 89031

Table 23.3 The evaluation parameters for detection models on NSL - KDD datasets Classifier Accuracy Precision Recall F1-Score Mis-Classifier Decision Tree (DT) KNN Naïve Baye (NB) Random Forest (RF) Neural Network Logistic Classifier Stochastic Gradient Descent

0.96 0.96 0.87 0.97 0.96 0.95 0.87

0.95 0.96 0.72 0.97 0.97 0.86 0.73

0.91 0.98 0.81 0.92 0.93 0.95 0.81

0.99 0.97 0.76 0.94 0.96 0.90 0.77

799 804 2859 654 879 1206 2829

Table 23.4 The evaluation results of the Ensemble models are trained by the NSL-KDD dataset Classifier(Ensemble Model) Accuracy Precision Recall F1-Score Mis-Classifier RF + KNN + DT + NB 0.96 RF + KNN + DT 0.97 RF + KNN + DT + NB + NN 0.97

0.99 0.96 0.97

0.87 0.92 0.91

0.93 0.94 0.94

879 646 719

not as high as on the self-generated data. This can be explained by our dataset having more records and features (80 features), so the training of the model will be provided with more data than NSL-KDD. At here, we decided to combine classifiers with good results on the NSL-KDD dataset together into different ensemble models. Each classifier in the ensemble model will work individually to provide predictive results for the inputs. The prediction result which occurs the most is then selected as the output of the ensemble model. Based on this principle, we investigate which combination is most effective. Table 23.4 shows the evaluation results of the Ensemble models trained by the NSL-KDD dataset.

23 Empirical Performance Evaluation of Machine Learning …

293

Not all ensemble models give better results than single models. Based on the results shown in the statistics in Table 23.4, we can see that the ensemble model has very high accuracy and f1-score results. However, only the model combining Random Forest, KNN, and Decision Tree provided better performance than the single highest performance model (Random Forest). The difference is about 0.04% in accuracy and 0.11% at F1-Score. Meanwhile, the time to train the Random Forest model and the associated model on the NSL-KDD set is 2.96s and 15.9s, respectively, about 5 times. The very small performance difference while the training time difference is very large shows that in this dataset, the Random Forest model gives more optimal results than the combined models. We used the NSL-KDD dataset to train and test the model on a Raspberry Pi with a Raspberry Pi 4 Model B, 4GB of LPDDR4 SDRAM configuration. As shown in Figs. 23.2 and 23.3, the results of training and model testing on Raspberry Pi are equal to the model results on PC. The only difference is the training and testing time of the predictive model. Due to some hardware limitations, the Raspberry Pi has many times greater model testing and training times than a PC. In the self-created and NSL-KDD dataset, TCP packets accounted for about 99.93% and 87.26% of the total packets, respectively. So, to survey DDoS attack packets using UDP protocol, we used DrDoS_UDP and DrDoS_SNMP dataset of CIC-DDoS2019 dataset to train and test classification models. UDP packets account for about 99.89% in the DrDoS_UDP set and 99.9% in the DrDoS_SNMP set. Table 23.5 shows the results of the test model on the DrDoS_UDP test set. As can be seen, the models running on the DrDoS_UDP dataset have good performance, high accuracy, and a classification level of up to 99.99% for each model.

Fig. 23.2 Comparing data training interval between PC and Raspberry Pi

294

B.-S. Tran et al.

Fig. 23.3 Comparing the label prediction interval of prediction models between PC and Raspberry Pi Table 23.5 Test parameters of models on DrDoS_UDP dataset Classifier Accuracy Precision Recall Decision Tree (DT) KNN Naïve Baye (NB) Random Forest (RF) Logistic Regression Stochastic Gradient Descent

1.00 1.00 1.00 1.00 1.00 1.00

1.00 1.00 1.00 1.00 1.00 1.00

1.00 1.00 1.00 1.00 1.00 1.00

F1-Score

Mis-Classifier

1.00 1.00 1.00 1.00 1.00 1.00

35 90 394 14 467 850

This means that most of the DDoS packets are detected by the models. From here, we tried to use the models trained on the DrDoS_UDP dataset for testing on our dataset. Evaluation results are listed in Table 23.6. It can be seen that the accuracy of the model is greatly reduced. Points are misclassified up to more than 50% of the dataset in Decision Tree, Logistic Regression, and Stochastic Gradient Descent models. Especially in KNN, more than 75% of data points are predicted incorrectly. In the Naive Bayes model, the precision is 0.860986 but the f1-score is very low (0.00), the precision-score is 0.02 and a recall of 0.00 shows that most of the packets that the model predicts is DDoS but the actual label is normal. This will give false attack warning when this model is put to practical use. So we tried to use the models in combination with the desire to improve the performance of these models, but the results are not so good. This is shown in Table 23.7.

23 Empirical Performance Evaluation of Machine Learning …

295

Table 23.6 The evaluation results of models on a self-generated dataset Classifier Accuracy Precision Recall F1-Score Decision Tree (DT) KNN Naïve Baye (NB) Random Forest (RF) Logistic Regression Stochastic Gradient Descent

0.52 0.19 0.86 0.89 0.44 0.48

0.05 0.14 0.02 0.89 0.02 0.01

0.13 0.98 0.00 0.22 0.08 0.02

0.07 0.25 0.00 0.35 0.04 0.01

Mis-Classifier 1007379 1674732 289057 232432 1165116 1169720

Table 23.7 The evaluation results when using the ensemble model on self-created dataset Classifier Accuracy Precision Recall F1-Score Mis-Classifier Ensemble Model (NB + RF + DT) Ensemble Model (RF + KNN)

0.86

0.75

0.02

0.05

283844

0.89

0.92

0.21

0.35

232331

Table 23.8 The evaluation results on DrDoS_SNMP test dataset Classifier Accuracy Precision Recall F1-Score Decision Tree (DT) KNN Naïve Baye (NB) Random Forest (RF) Logistic Regression Support Vector Machine Stochastic Gradient Descent

Mis-Classifier

1.00 1.00 1.00 1.00 1.00 1.00

0.97 0.94 0.53 0.97 0.38 1.00

0.95 0.84 1.00 0.98 0.27 0.00

0.96 0.89 0.70 0.97 0.32 0.01

20 52 212 11 284 242

1.00

0.36

0.37

0.36

314

As for the DDoS_SNMP dataset, we perform statistics on the test parameters for Normal packets (labeled Negative—0) shown in Table 23.8 to examine the results of data imbalance when Normal packets only account for 0.0704%. this dataset. In the DrDoS_SNMP dataset, we can see that there are some models such as Naive Bayes, Logistic Regression, Support Machine Learning, and Stochastic Gradient Descent that give quite high accuracy, but the classification parameters such as precision, recall, and f1-scores are not good. For example, the Support Vector Machine model has a precision of 1.00 but recall only 0.01, most normal packets are classified as DDoS, the rate of ignoring Normal packets is very high. From there, it shows the effect of data imbalance on the predicted results of the model.

296

B.-S. Tran et al.

Through examining three datasets and classification models of machine learning, we can see that each dataset and models have their own advantages and disadvantages. On the dataset side, we draw the following conclude: In the self-created dataset, TCP packets dominate and lack the characteristics of UDP attacks. The DrDoS_UDP and DrDoS_SNMP datasets are the opposite. The model trained in these three datasets all have high evaluation results on the test set, but when using data from other datasets, the results are low. From that, we realize that these models are being overfitting, which is the phenomenon that the model is found to be too consistent with the training dataset. This over-matching can lead to the prediction of noisy data points and the model quality is no longer good on the actual data. This phenomenon occurs on three datasets because the number of training datasets is unbalanced, and the data complexity is high (85 features). On the part of the training models, it can be seen that the Random Forest model shows quite good results in different network datasets, with high accuracy, the rate of incorrectly predicted points is lower than the others. Due to some data imbalance limitations, we decided to combine the self-generated dataset and the DrDoS_UDP dataset to create a dataset full of samples of DDoS attacks on both protocols. TCP and UDP also have a number of DDoS and Normal packets of about 0.6: 0.4. This dataset has 200,000 entries, 80 features, including 1000823 packets with UDP protocol, the rest are TCP, 821515 packets of DDoS. We will train machine learning classification models on this aggregate dataset. And test the models obtained with the test dataset from the custom dataset and DrDoS_UDP so that the testing part is not involved in the modeling process. Table 23.9 is the results of the model testing trained with the aggregated dataset on the self-generated dataset. In addition to the evaluation results on UDP dataset from Table 23.10, we can see that the results of most models have been significantly improved on both the TCP and UDP packets. From that we draw the conclusion that the datasets that balance the number of labels in each class are better for data training. And the Machine Learning model that yielded the best results among the models surveyed for network datasets is the Decision Tree and Random Forest tree house models (Fig. 23.4).

Table 23.9 The evaluation results on self-generated dataset Classifier Accuracy Precision Recall Decision Tree KNN Naïve Baye Random Forest Logistic Regression Stochastic Gradient Descent

1.00 0.99 0.98 1.00 0.69 0.98

0.96 0.92 0.64 0.98 0.00 0.68

1.00 0.08 0.08 1.00 0.03 0.08

F1-Score

Mis-Classifier

0.98 0.15 0.14 0.99 0.00 0.15

648 12940 13493 270 273660 13390

23 Empirical Performance Evaluation of Machine Learning … Table 23.10 The evaluation results on UDP dataset Classifier Accuracy Precision Recall Decision Tree KNN Naïve Bayes Random Forest Logistic Regression Stochastic Gradient Descent

1.00 1.00 1.00 1.00 0.98 0.97

1.00 1.00 1.00 1.00 1.00 1.00

1.00 1.00 1.00 1.00 0.99 0.97

297

F1-Score

Mis-Classifier

1.00 1.00 1.00 1.00 0.99 0.98

28 135 510 149 13528 28151

Fig. 23.4 The evaluation results on DrDos_UDP test dataset and Self-generated dataset

23.5 Conclusion Our main contribution is to present an evaluation framework for comparing the machine learning models’ ability to detect DDoS attacks. Our framework is able to create several DDoS attack scenarios by using BoNeSi and HTTP-ping. After having input, we used a total of eight machine learning models and deep learning models, including Decision Tree, Naive Bayes, Neural Network, to support our analysis process. Also, we use two other available DDoS datasets, CICDDoS2019 and KDD, on our models to draw our conclusions. As a result, with the parameters obtained on the models we use, we conclude that Random Forest is the model with the faster prediction time prediction time with the highest accuracy probability. Besides, we also propose an ensemble approach by combining a set of DDos mechanism to increase the detection performance. In future work, we will complete the best features to detect different types of DDoS attacks accurately. More importantly, using raw features (more than 80 features from the CICFlowMeter tool) significantly increases running time. New analytical techniques will also be the next research direction that we need to learn more deeply and use more artificial intelligence techniques. We will build more scenarios, larger attack models to obtain more perfect datasets because the dataset is one of the most

298

B.-S. Tran et al.

critical conditions for accurate analysis best results. Besides, new types of attacks and methods that attackers are constantly evolving to take advantage of will also be explored and updated regularly. This update helps the data creation process overcome the previous shortcomings, contributing to building a more accurate and perfect analysis foundation. Acknowledgements This research was funded by Vietnam National University Ho Chi Minh City (VNU-HCM) under Grant no. DSC2021-26-04.

References 1. Popovi´c, K., Hocenski, Ž: Cloud computing security issues and challenges. In The 33rd International Convention Mipro, pp. 344–349. IEEE (2010) 2. Somani, G., Gaur, M.S., Sanghi, D., Conti, M., Buyya, R.: Ddos attacks in cloud computing: Issues, taxonomy, and future directions. Comput. Commun. 107, 30–48 (2017) 3. Stallings, W.: Foundations of modern networking: SDN, NFV. IoT, and Cloud. Addison-Wesley Professional, QoE (2015) 4. Mansfield-Devine, Steve: Ddos goes mainstream: how headline-grabbing attacks could make this threat an organisation’s biggest nightmare. Netw. Secur. 2016(11), 7–13 (2016) 5. Douligeris, Christos, Mitrokotsa, Aikaterini: Ddos attacks and defense mechanisms: classification and state-of-the-art. Comput. Netw. 44(5), 643–666 (2004) 6. Yuan, X., Li, C., Li, X.: Deepdefense: identifying ddos attack via deep learning. In: 2017 IEEE International Conference on Smart Computing (SMARTCOMP), pp. 1–8. IEEE (2017) 7. Sharafaldin, I., Lashkari, A.H., Hakak, S., Ghorbani, A.A.: Developing realistic distributed denial of service (ddos) attack dataset and taxonomy. In: 2019 International Carnahan Conference on Security Technology (ICCST), pp. 1–8 (2019) 8. DARPA: Instruction detection scenario specific data sets, how published. https://www.ll.mit. edu/r-d/datasets/2000-darpa-intrusion-detection-scenario-specific-datasets (2000) 9. Binbusayyis, Adel, Vaiyapuri, Thavavel: Identifying and benchmarking key features for cyber intrusion detection: an ensemble approach. IEEE Access 7, 106495–106513 (2019) 10. Habibi, Javid, Midi, Daniele, Mudgerikar, Anand, Bertino, Elisa: Heimdall: Mitigating the internet of insecure things. IEEE Internet Things J. 4(4), 968–978 (2017) 11. The virustotal homepage: https://www.virustotal.com (2020). Last accessed 28 Jan 2020 12. Santos, L., Rabadao, C., Gonçalves, R.: Intrusion detection systems in internet of things: a literature review. In: 2018 13th Iberian Conference on Information Systems and Technologies (CISTI), pp. 1–7. IEEE (2018) 13. Jia, Y., Zhong, F., Alrawais, A., Gong, B., Cheng, X.: Flowguard: an intelligent edge defense mechanism against iot ddos attacks. IEEE Internet Things J. (2020) 14. Roschke, S., Cheng, F., Meinel, C.: Intrusion detection in the cloud. In: 2009 Eighth IEEE International Conference on Dependable, Autonomic and Secure Computing, pp. 729–734. IEEE (2009) 15. Nychis, G., Sekar, V., Andersen, D.G., Kim, H., Zhang, H.: An empirical evaluation of entropybased traffic anomaly detection. In: Proceedings of the 8th ACM SIGCOMM conference on Internet measurement, pp. 151–156 (2008) 16. Markus-Go: Bonesi—the ddos botnet simulator, how published. https://github.com/MarkusGo/bonesi (2015) 17. Ahlashkari: Cicflowmeter https://github.com/ahlashkari/CICFlowMeter (2017)

23 Empirical Performance Evaluation of Machine Learning …

299

18. Tavallaee, M., Bagheri, E., Lu, W., Ghorbani, A.A.: A detailed analysis of the kdd cup 99 data set. In:2009 IEEE Symposium on Computational Intelligence for Security and Defense Applications, pp. 1–6 (2009) 19. Das, S., Mahfouz, A.M., Venugopal, D., Shiva, S.: Ddos intrusion detection through machine learning ensemble. In: 2019 IEEE 19th International Conference on Software Quality, Reliability and Security Companion (QRS-C), pp. 471–477 (2019)

Chapter 24

Towards Remote Deployment for Intrusion Detection System to IoT Edge Devices Xuan-Thanh Do and Kim-Hung Le

Abstract Recently, low latency in data transmission has become one of the most critical requirements in developing the Internet of Things (IoT) applications. It triggers a novel network architecture, namely edge computing, that aims to move computing units close to data sources. This transformation emerges several security issues about designing and implementing security applications. An intrusion detection system (IDS), a well-designed system for detecting abnormal behaviors, needs to be transformed into modern system architectures. This article presents an edge-based architecture to quickly deploy a deep learning-based IDS to edge network devices regardless of the heterogeneity in hardware and deep learning model configurations. To demonstrate the effectiveness of our proposal, we also analyze various performance indicators of the architecture, deployment process, and deep-learning models. Keywords Deep Learning-based IDS · Intrusion Detection System · IDS Deployment Architecture · Network Security

24.1 Introduction With the rapid development of IoT solutions, a massive amount of data is generated by smart objects. One of the world’s leading information technology research and consulting firms, Garner, Inc., has predicted that 5.8 billion IoT devices are put into use by 2020 [1]. Although the cloud computing model provides an easily scalable infrastructure, reducing design costs, maintaining resources, and storage capacity, it is not suitable for real-time applications requiring low latency and geographic proximity. Cisco researcher creates the fog computing architecture to solve X.-T. Do · K.-H. Le (B) University of Information Technology, Ho Chi Minh City, Vietnam e-mail: [email protected] X.-T. Do e-mail: [email protected] Vietnam National University Ho Chi Minh City, Ho Chi Minh City, Vietnam © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_24

301

302

X.-T. Do and K.-H. Le

this problem [2]. The architecture of fog computing in IoT systems is divided into three layers, including edge, fog, and cloud layers. The fog architecture allows various services and applications to process packets in real-time by extending the cloud to the network’s edges. It offers physical computation, processing, and storage closer to where the end-user generates the data. As a result, fog computing partly solves cloud computing problems (e.g., low latency, location awareness, management, and processing of large numbers of nodes, real-time application support, and wireless network access) [3]. Since moving services from the cloud to edge, various networking, security, and privacy issues are also inherited from the Cloud networking. Several attacking methods (e.g., Man-in-the-middle (MITM), Denial-of-Service (DoS), Distributed Denial-of-Service(DDoS), Remote Code Execution (RCE), Data Leak) are contextualized to adapt with the edge architecture in IoT system [4]. To mitigate the threats from these attacks, the Intrusion Detection Systems (IDSs) are deployed to edge devices to monitor and analyze the incoming networking traffic to early recognize and prevent abnormal network behaviors that reflect cyberthreats [5]. In recent IDSs, machine learning and deep learning models are integrated to increase detection accuracy, namely machine learning-based or deep learningbased IDS. These models are trained with normal observations. Then, the welltrained models are plugged into IDSs to detect abnormal behaviors by calculating the similarity between normal and abnormal traffic data [6]. The major drawback of this IDS type is that its model is only trained with normal traffic patterns of a specific networking system generated by several data sources. However, the data sources of each networking system may be different. Applying this model to different networking systems may reduce the detection performance. To the best of our knowledge, there are no IDS that could accurately detect all attacks. Selecting appropriate IDS for a specific context is challenging due to the variety in IDS types (e.g., Host-IDS, Network-IDS) and detection models (e.g., machine learning, deep learning, rules) [7]. In addition, the detection accuracy strongly depends on the strategic point used to place the IDS within the edge architecture, including three layers. Therefore, effectively applying IDS to edge computing is much interested in the research community [8]. Our work is motivated by an observed assumption: Deploying deep-learning models of IDSs on various edge network devices requires a massive effort, and these models need to change to adapt to the changes of networking architecture frequently. Indeed, these devices are deployed in a wide area, maybe in different data centers. Physically accessing is difficult or maybe impossible. Moreover, they probably have different hardware and operating system that require specific running environments of IDS. In this paper, we propose a novel fog-based architecture supporting the remote deployment of deep learning IDS to edge devices. This proposal supports the adjustment, configuration update, and deployment of various deep-learning models (6 models) acting as IDS to detect abnormal network traffic running on edge network devices. Consequently, system administrators could quickly adapt IDS configuration to the changes of network conditions and shorten attack detection time. The proposed architecture also supports “no-down-time deployment” that maintains the system running stably when IDS configurations (attack detection models) are

24 Towards Remote Deployment for Intrusion Detection …

303

updated. This means that the system still runs with the previous model until the new model is fully loaded and updated. The main contributions of this study are summarized as follows: 1. Proposing a fog-based architecture for IDS supporting the adjustment, configuration update, and deployment of various deep-learning models to detect abnormal networking traffic. 2. Analyzing performance indicator of the proposed architecture, including IDS deployment time, time to update, deploy a new model, comparison of the time and performance of deploying different machine learning and deep learning models on edge network devices. 3. Evaluating the ability of deployed model about detecting DDoS attacks on edge network devices. For the remainder of the article, we are organized into the following sections: Sect. 24.2 reviews the work involved in implementing the system, such as existing problems, limitations of the current system, and proposed solutions. Section 24.3 describes the attack detection framework in detail, the process of deploying the system to the edge device, and getting the characteristics of the network traffic. Section 24.4 explains the experimentation and evaluation of the results obtained. Section 24.5 presents the conclusion and future development work.

24.2 Related Works In the last ten years, cyber attacks have increased immensely due to the rapid growth of services and applications running on IoT devices [9]. Taking advantage of multiple IoT devices, attackers focus on constrained IoT devices that are globally accessible without comprehensive safety and security mechanisms. For instance, in 2016, a DDoS attack happens against several websites using DNS by Dyn provider involving some IoT devices that executed some malicious botnet software on IoT devices [9]. Therefore, researchers have proposed different approaches to solve the above problems. As shown in [10], the authors introduce an anomaly detection system in the supervised layer that uses Hybrid algorithms to improve the accuracy of the attack detection system and implement the system in cloud computing. This proposal exploits DARPA’s Knowledge Discovery and Data Mining (KDD) cup datasets for training. The trained prediction model gives higher accuracy and lowers false alarm rates; this rate outperforms traditional Native Bayes grading machine learning models and classic ANN techniques. In [11], the authors introduce a new framework for detecting distributed denial of service attacks on cloud computing frequency. These IDSs are deployed to the cloud layer and perform data exchange and warnings with each other through agents that determine whether those alerts are evaluated as attacks from other IDS or not. The IDS’s deployment to the cloud leads to an increase in the latency and high response times of attack detection. Fog networking is developed to solve the above

304

X.-T. Do and K.-H. Le

problem [12]. The fog computing system is increasingly adopted and widely accepted in recent times; many scientific researchers have proposed different solutions to integrate IDSs into the fog computing model effectively. In [13], the author introduced an IDS inspired by the biological immune system, Artificial Immune System (AIS), a system that can fight diseases to identify external pathogens based on the cells and molecules inside the body. This system is proposed and developed over three layers in the fog networking system. Attack detection time is one of the problems to be solved. In [14], researchers proposed a lightweight IDS model that uses a MultiLayer Perceptron (MLP), hidden class, based on vector spatial representation in order to improve the predictive time of data analyzed. Authors exploit datasets based on system calls that contain application layer attack and exploit data of both the Linux operating system with the ADFA-LD and the windows operating system with the ADFA-WD Defense of Australia data set. In the proposed IDS, the IDS is deployed on the fog layer device and installed on the Raspberry Pi device. For resource-limited devices, computational power, memory, and power, [15], the authors propose an IDS. The testbed model achieves a balance between energy consumption and detects accuracy by relying on the anomaly identified through the intended, previously-stored signatures is going to happen on devices running IDS deploying Nash equilibrium-based attack detection techniques with performance and viability analyzed through wireless sensor network (WSN) simulation using TOSSIM for simulation. The increase of malicious attacks such as Scanning, DoS attack, Man In the Middle increasing with the development of IoT, some systems such as SCADA, Cloud, and Smart Grid against those attacks are based on IDS as the first line of defense [16]. So that researchers began to study the application of IDS and in this fog computing model [17]. The IDS has a characteristic that it strongly depends on the layers on which it is deployed. To increase the security level, IDSs should be deployed on multiple layers of the architecture. The multilayer deployment presents significant challenges that need to be addressed [18]. The deployment of IDSs on network devices faces many challenges from the design of the infrastructure architecture and IDS deployment in the fog networking [19]. The first is large-scale networking, heterogeneous devices, and IDS must be equipped with a good source of hardware to support work and implement algorithms on the devices to perform effective display of requests. The next is the problem of geospatial, and the fog computation is very complex and geographically distributed worldwide depending on the purpose of the design [2]. The issue that demands to be addressed next is real-time resolution, an effective IDS, when the ability to detect system threats early is done immediately, affecting latency affects real-time packet analysis, increases notification time. Deployments on layer 2 of fog computing systems present challenges such as detecting an attacker in the local network that are difficult to detect or encrypting data on the transmission path leading to the inability to open packets to analyzing the content of the packet makes detecting malicious packets more difficult [20].

24 Towards Remote Deployment for Intrusion Detection …

305

24.3 The Deployment Framework In this section, we explain how the proposed framework supports training various attack detection models (including machine learning and deep learning models), and then deploying them to edge networks devices effectively.

24.3.1 Overview To accommodate edge networking architecture, we separate our system model into three layers: cloud layer, fog layer, and edge layer. The top layer is a cloud computing layer that provides flexible, reliable, and scalable resources with computing services over the Internet. Cloud computing allows data transmission, statistics, and analysis through the Internet. This layer is used to train the detection model with the observed events from the fog and edge layers. The next layer is the middle layer, namely the fog layer; it contains thousands of router, switch, and server devices managed by the service distributors. In our vision, these devices could turn into an IDS to enhance the security of systems. They support many interfaces and services that allow applications to interact and communicate with each other. On the other hand, collected data are analyzed and processed in the fog computing layer before arriving at the cloud layer. As a result, the workload of cloud servers and data transmitting over the Internet are significantly reduced. The third layer is the edge network layer, which contains billions of IoT devices such as cameras, television, sensors, security cameras, and low memory and limited device resources. These IoT devices act as data sources and generate large amounts of data that demand to be analyzed. These data are pushed through internal devices in fog computing through gateways. The overall proposed system is shown in Fig. 24.1. According to gateways, we group IoT devices into clusters, and these clusters are connected to the Internet through gateways. At the gateway, the collected data from IoT devices are converted into features by using CICFlowmeter. These features are then pushed through IDS models in real-time to detect abnormal behaviors (networking attacks). At the final step, the data is pushed from the fog computing layer to the cloud computing layer. The administrator can monitor the device’s health and predict whether the devices managed by the gateway are infected with botnets or launch a DDoS attack. Base on this information, system administrators could select and deploy the appropriate model to the gateway to improve the attack detection efficiency.

306

X.-T. Do and K.-H. Le

Fig. 24.1 Architecture of the proposed fog-based attack detection framework

24.3.2 Deployment Process In order to deploy the IDS from the server in the cloud to the gateway in the fog networking, we first install the Secure Shell (SSH)1 server on the gateway devices, then add a public SSH key of the control server and these gateways. Then, Ansible2 is installed onto the device control server. Next, we configure the gateways’ IPs to the control server’s configuration in Ansible, then we put IDS on the control server and build deployment scripts, system updates, and deploy updates. Machine learning and deep learning training models are updated to deliver the controller server models to gateways. Once the Ansible playbook is fully installed, the deployment process is started. Firstly, the Ansible command is pushed to the gateways to trigger the scripts that establish running environments, such as installing necessary packages, updating new 1

SSH: is a protocol for secure remote login and other secure network services over an insecure network. 2 Ansible: An open-source software provisioning, configuration management, and applicationdeployment tool enabling infrastructure as code.

24 Towards Remote Deployment for Intrusion Detection …

307

packages, and then delivering the IDS to the gateways. Secondly, the IDS is run as a background service on the gateway under the operating system’s management. After running the command start service up, IDS starts listening on the network card on which it has been configured and proceeds to capture traffic on it. It analyzes and predicts the traffic flow of the IoT devices connected to it, and the predicted results are then pushed to the server so that the administrator can monitor the activity of the IoT devices. When the system is operated correctly, we can update and reconfigure the serverside attack detection models by running the model update playbook with the corresponding model name. Models and configurations are pushed to gateways. Thereby, the system can change and adapt flexibly, and the operator can deploy and update as a model for the IDS quickly and efficiently. This system helps deploy IDSs to be automated and fast, increasing server administrators’ performance and reducing time and latency during attack detection at nodes in the edge network. The workflow of deploying the system and updating the model is shown in Fig. 24.2

24.3.3 Feature Extractor In our proposal, raw data conversion is critical because it helps the model operate more efficiently. All packets passed by the IoT go through the management gateway before going to the cloud. Pcap4J3 is a tool developed to capture real-time packets, analyze the packets, and send them across the network at the edge network layer. Pcap4J is run on network equipment to capture raw data, which is then passed through CICFlowmeter - a network traffic analysis tool to convert raw network traffic into structured data with over 80 features derived from captured data. CICFlowmeter offers a wide variety of features and can choose which one to match the pattern, context, and type of attack to be detected. Once CICFlowmeter’s data streams are analyzed, these data streams are streamed through the attack detection model deployed from the previous control server in real-time. These streams of events are pushed through trained machine learning and deep learning attack detection models, and predictive analysis results are stored and sent back to the server in the cloud, where the administrator values can observe and decide appropriate decisions based on predictive labels and information obtained by CICFlowmeter from raw data. Due to CICFlowmeter extracting necessary information, the ability of IDS to detect attacks is significantly enhanced.

3

Pcap4J: a Java library for capturing, crafting, and sending packets.

308

Fig. 24.2 The workflow of deploying the system and updating the model

X.-T. Do and K.-H. Le

24 Towards Remote Deployment for Intrusion Detection …

309

24.4 Evaluation 24.4.1 Running Environment As for hardware, our testbed with the above system is performed on LattePanda gateway devices with 2GB DDR3L RAM, 4 CPUs, Intel (R) Atom (TM) x5-Z8350 CPU @ 1.44GHz, 32 GB memory[GW ] . The IDS is deployed from the control server with 8 CPUs, 8GB DDR3 RAM, Intel (R) Core (TM) CPU @ 3.60 GHz[SV C L] . The edge network’s testbed node is Raspberry Pi 4, with 4GB LPDDR4 RAM, Cortex-A72 CPU, 32GB of memory[R P] . Besides, IoT devices join the network, such as Smartphone, iPhone 7[S P] , Smart Television[T V ] , Laptop[L T ] . A security information and event management (SIEM) server on the cloud computing layer to receive prediction results from the gateway on the fog networking layer with 4G RAM configuration, 4 CPUs, Intel (R) Xeon (R) Gold 6140 CPU @ 2.30GHz CPU[SV M T ] . For the software, Fig. 24.3 shows the framework elements and requires software, OS corresponding. On the server-side, we use Ubuntu 18.04 with Python 3.6; on this

Fig. 24.3 Elements of the proposed fog-based attack detection framework

310

X.-T. Do and K.-H. Le

management server, we install Ansible and build the system’s deployment scripts to the gateway. On the gateways, Ubuntu 18.04 and Python 3.6 are installed. In the process of deploying the IDS automatically through Ansible, on the LattePanda gateway device, the script proceeds to install java OpenJDK to run Pcap4J and CICFlowmeter, install python3-pip to install attack detection library packages, install libpcap-dev library package to capture the packets on the network card required in the package Pcap4J. After, the script installs the necessary packages such as NumPy, Sklearn, TensorFlow, Keras to detect the attack. In the scenario, we deploy the IDS from the control server to the devices and convert the running system into service on the gateway, the emulator node running on Raspberry Pi 4, connecting to a device running Ubuntu 18.04. This emulator using open-source Bonesi4 the emulator for a DDoS attack, and installing curl5 to simulate the normal requests. After the data is analyzed at the gateway, the results are pushed back to the SIEM system installed to receive the log and observe the management interface’s logs.

24.4.2 Results This section covers resource usage in the system. We analyze the time it takes to first deploy the system to the gateway devices and the time and resources spent in updating a new model on the gateway. System resources consumed when running the IDS on gateway devices in the fog networking layer are also reported. To demonstrate the practical aspect of our proposed framework, we present the IDS detection performance for abnormal event detection generated by CICFlowmeter, the accuracy of the models deployed from the management server to the gateways.

24.4.2.1

Resource Utilization

First, we present the resources used, the timing of the first deployment of the system, and the process of updating attack detection models for the IDS. Resources used for first-time system deployment from the management server. Table 24.1 shows the details of the time it takes to deploy our IDS to the gateway device in the fog layer, in which we have a total deployment time of around 278 s with remote sync source time of 102 s, installation time, and update the required packages is 168 s, the time to create service for the IDS on the gateway is 5s, and the time to start the service is approximately 1.5 s. During the system deployment time from the management server to the gateway devices, the devices take up a certain amount of resources for the system deployment process. Table 24.2 describes the detailed resources used at the gateway for the IDS 4

Bonesi: DDoS Botnet Simulator is a Tool to simulate Botnet Traffic in a testbed environment on the wire. 5 Curl: is used in command lines or scripts to transfer data.

24 Towards Remote Deployment for Intrusion Detection …

311

Table 24.1 Time to deploy from control-server to the gateway Time to deploy IDS source(s) 102 Time to install required packages(s) Time to put the system into service(s) Service start time(s) Total deployment time(s)

168 5 1.5 278

Table 24.2 Resources are used for first-time system deployment Usable hardware size (MB) 37 RAM usage (MB) CPU usage (%)

139.33 35

deployment. We have the source code of the system accounting for around 37 MB, the RAM used for the deployment process is nearly 140 MB, the CPU used for the deployment process accounts for about 35 %. After the system is successfully deployed, the IDS service is started in the gateway’s system, and the resource is used to run the IDS on the gateway, in which the RAM resource used to run the service is about 138 MB and CPU accounts for approximately 2 %. After the IDS runs on the gateway, we update the attack detection models to the gateway system. Table 24.3 shows details about the deployment time and resources used in modeling from the management server to the gateway devices. Six models include Decision Tree, Neural Network, Ensemble Decision Tree and Naive Bayes, Ensemble, Naive Bayes, and Random Forest. A decision tree is a tree-like model of decisions and possible consequences, including chance event outcomes, resource costs, and utility. The decision tree model is installed with a max depth of five and not the random state. A neural network is a network or circuit of neurons, and the neuron connections are modeled as weights. Neural networks are used for predictive modeling, adaptive control, and applications to be trained via a dataset. Ensemble models

Table 24.3 Cost of implementing and updating new models for the IDS at the gateway Model Deployment Updating time(s) RAM CPU usage(%) time(s) usage(MB) DecisionTree NeuralNetwork EnsembleDTandNB Ensemble NaiveBayes RandomForest

10 12 10.12 8.05 10.12 10.36

3.68 3.67 3.55 3.73 3.58 3.69

9.68 9.95 12 14 11.21 12.36

28.2 34 33.6 24.5 25.3 25.5

312

X.-T. Do and K.-H. Le

Table 24.4 Resources are used when the system detects a DoS attack from the node Model RAM usage (MB) CPU usage (%) DecisionTree NeuralNetwork EnsembleDTandNB Ensemble NaiveBayes RandomForest

154.92 131.02 155.57 122.10 173.58 135.45

63.4 66.1 71.1 74.5 55.9 51.5

use multiple learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms alone. Ensemble Decision Tree and Naive Bayes is a model using Decision Tree and Naive Bayes learning algorithms. Naive Bayes classifiers are a family of simple probabilistic classifiers based on applying Bayes’ theorem with strong independence assumptions between the features. Random forests [21] or random decision forests are an ensemble learning method for classification, regression, and other tasks that operate by constructing a multitude of decision trees at training time and outputting the class that is the mode of the individual classification or regression trees. For gateway deployment models, the average time to update an attack detection model is 10.11 s, the average service start time is about 3.65 s, and the resource usage is for the update process models average around 11.53 MB, and the average CPU usage for this process is around 28.52%. Note that the service runs without service downtime; the update time is the time to deploy the model and configuration from the control server to the gateway while the old model is still running during this time, then the service gets restarted and applies the new model to the IDS. Later deploying the IDS to the gateway devices in the fog networking, we run the attack simulator and emulate the system to run normally on the Raspberry Pi 4 node connected to the LattePanda gateway through the wireless network. We measured performance when the system was attacked by node Raspberry Pi 4 by sending an average of 500–700 requests in roughly 1 s to the hacked server via the Bonesi tool. Table 24.4 summarizes the models and resources consumed by each model when attacked by a node in the IoT device connected to the gateway. We can see that RAM’s average resource is about 145.44 MB and CPU usage is about 63.75 %.

24.4.2.2

Detection Performance

Next, simulating attack requests and simulating normal requests on the node, data is collected on the network card and analyzed at the gateway through deployed models and updated to the settings get gateway from the management server. The analytics results are then pushed back to the monitoring server in the cloud. We simulated 4000 generated CICFlowmeter events, of which 2000 events were for

24 Towards Remote Deployment for Intrusion Detection …

313

Table 24.5 The accuracy of the models and the prediction time of each model Model Accuracy Precision Recall F1-Measure NeuralNetwork [22] DecisionTree [23] EnsembleDTandNB [24] Ensemble [25] NativeBayse [26] RandomForest [26]

0.93 0.88 0.68 0.63 0.63 0.99

0.97 0.99 0.73 0.57 0.57 0.99

Table 24.6 The prediction time of each model Model NeuralNetwork DecisionTree EnsembleDTandNB Ensemble NativeBayse RandomForest

0.88 0.77 0.56 0.99 0.99 1.00

0.92 0.87 0.64 0.72 0.72 0.99

Time(ms) 4.80 1.08 4.08 161.27 1.94 212.36

DDoS attack and 2000 events for normal request. Table 24.5 and table 24.6 shows the data evaluating the accuracy of the models tested to the IDS and the time it takes to detect each captured and analyzed event by the IDS. Here, we use metrics to evaluate the accuracy of attack detection models deployed to the system, including accuracy, precision, recall, f1-measure, and time to classify one event as normal or abnormal. According to these tables, we can see that for a deep learning model that has the best accuracy and time to detect each event with an accuracy of 0.93, precision of 0.97, recall of 0.88, f1-measure of 0.92, and time to predict an event is about 4.8 ms. While RandomForest has higher accuracy with accuracy, precision, recall, f1measure, respectively: 0.99, 0.99, 1.00, and 0.99, but the trade-off is time to predict a fairly large event of time of 212.36 ms. Between the models, we have a DecisionTree that gives pretty good results and a fast prediction time with an accuracy of 0.88, precision of 0.99, recall of 0.77, f1-measure of 0.8, and a prediction time for an event of about 1.08ms. As for EnsembleDTandNB model, the experimental result is quite low with accuracy: 0.68, precision: 0.73, recall: 0.56, f1-measure: 0.64 and prediction time is 4.08 ms. As for Ensemble, the results are not good both in terms of accuracy and predictive time with the values accuracy with 0.63, precision at 0.57, recall at rate 0.99, f1-measure with at 0.72, the prediction time was quite slow at 161.27 ms. The NaiveBayes model also has a quite low performance like Ensemble Model but the prediction time is much improved with accuracy is 0.63, precision

314

X.-T. Do and K.-H. Le

is roughly 0.57, recall is about 0.99, f1-measure is nearly 0.72, and predict time one event is approximately 1.94 ms. Through here, we can see that the model when running the actual Random Forest test has the highest accuracy, but the trade-off in predicting time for an event is quite long. While with the deep learning network, Neural Network gives us a good predictive result with shorter prediction time. The above results and tables show IDS deployment on edge devices from controlservers is run with reasonable time, using limited RAM and CPU resources. Machine learning and deep learning models can run on device edges with good performance, predict events time are relatively short, and high rate accuracy in detect attacks. The models can update from server to gateway quickly, and service can always run on gateway.

24.5 Conclusion In this article, the proposed fog-based IDS architecture has proposed being able to update attack models for the IDS to make predictions, detecting attacks on the fog layer instead of detecting attacks on the cloud computing layer. We research deploying a script-based automated intrusion detection system to the gateways to analyze data near the edge layer to minimize latency and present how to implement the IDS on gateways located in the fog compute layer. The remarkable thing is that our system proposes real-time operation, which listens on the gateway’s wireless card and extracts features, then pushes data through models to make predictions for that data in real-time. Our experiments showed the ability to deploy the IDS, update, and predict with machine learning models efficiently, where the deep learning model gives a decent classification reasonable rate and highly accurate. This proposal overcomes the ability to deploy various types of attack detection models directly to low-capacity and low-configuration gateway devices. It can detect attacks quickly, with high accuracy, automatic and rapid change, and update of new models to detect new attacks. In the future, we will improve the predictive time of an event that our framework captures to meet the performance of our attacks detection system in realtime, as well as improve the level of performance, accuracy of models, and propose solutions to automatically deploy models down gateways to detect many different types of attacks. Acknowledgements This research is funded by University of Information Technology-Vietnam National University HoChiMinh City under grant number D1-2019-20. This research was funded by Vietnam National University Ho Chi Minh City (VNU-HCM) under Grant no. DSC2021-26-04.

24 Towards Remote Deployment for Intrusion Detection …

315

References 1. Goasduff, L.: Gartner says 5.8 billion enterprise and automotive iot endpoints will be in use in 2020. https://gtnr.it/35hq94q. Accessed 28 Sept 2019 2. Bonomi, F., Milito, R.: Fog computing and its role in the internet of things. In: Proceedings of the MCC workshop on Mobile Cloud Computing, vol 08 (2012) 3. Yi, S., Li, C., Li, Q.: A survey of fog computing: concepts, applications and issues. In: Proceedings of the 2015 Workshop on Mobile Big Data, pp. 37–42 (2015) 4. Esposito, C., Castiglione, A., Pop, F., Choo, K.K.R.: Challenges of connecting edge and cloud computing: A security and forensic perspective. IEEE Cloud Comput. 4(2), 13–17 (2017) 5. Liu, L., De Vel, O., Han, Q.-L., Zhang, J., Xiang, Y.: Detecting and preventing cyber insider threats: A survey. IEEE Commun. Surveys Tutor. 20(2), 1397–1417 (2018) 6. Deka, R.K., Kalita, K.P., Bhattacharya, D.K., Kalita, J.K.: Network defense: approaches, methods and techniques. J. Netw. Comput. Appl. 57, 71–84 (2015) 7. Shiravi, A., Shiravi, H., Tavallaee, M., Ghorbani, A.A.: Toward developing a systematic approach to generate benchmark datasets for intrusion detection. Comput. Secur. 31(3), 357– 374 (2012) 8. Khraisat, A., Gondal, I., Vamplew, P., Kamruzzaman, J.: Survey of intrusion detection systems: techniques, datasets and challenges. Cybersecurity 2(1), 1–22 (2019) 9. Zhou, W., Jia, Y., Peng, A., Zhang, Y., Liu, P.: The effect of iot new features on security and privacy: New threats, existing solutions, and challenges yet to be solved. IEEE Internet Things J. 6(2), 1606–1616 (2019) 10. Pandeeswari, N., Kumar, G.: Anomaly detection system in cloud environment using fuzzy clustering based ann. Mobile Netw. Appl. 21, 494–505 (2016) 11. Lo, C., Huang, C., Ku, J.: A cooperative intrusion detection system framework for cloud computing networks. In: 2010 39th International Conference on Parallel Processing Workshops, pp. 280–284 (2010) 12. Sudqi Khater, B., Abdul Wahab, A.W.B., Idris, M.Y.I.B., Abdulla Hussain, M., Ahmed Ibrahim, A.: A lightweight perceptron-based intrusion detection system for fog computing. Appl. Sci. 9(1), 178 (2019) 13. Hosseinpour, F., Amoli, P., Plosila, J., Hämäläinen, T., Tenhunen, H.: An intrusion detection system for fog computing and iot based logistic systems using a smart data approach. Int. J. Digital Content Technol. Appl. 10, 12 (2016) 14. Khater, B., Wahid, A., Idris, M., Hussain, M., Ibrahim, A.: A lightweight perceptron-based intrusion detection system for fog computing. Appl. Sci. 9, 178, 01 (2019) 15. Sedjelmaci, H., Senouci, S.M., Al-Bahri, M.: A lightweight anomaly detection technique for low-resource iot devices: A game-theoretic methodology. In: 2016 IEEE International Conference on Communications (ICC), pp. 1–6 (2016) 16. Mukherjee, M., Matam, R., Shu, L., Maglaras, L., Ferrag, M.A., Choudhury, N., Kumar, V.: Security and privacy in fog computing: Challenges. IEEE Access 5, 19293–19304 (2017) 17. Sohal, A.S., Sandhu, R., Sood, S.K., Chang, V.: A cybersecurity framework to identify malicious edge device in fog computing and cloud-of-things environments. Comput. Secur. 74, 340–354 (2018) 18. Kunal, S., Saha, A., Amin, R.: An overview of cloud-fog computing: Architectures, applications with security challenges. Secur. Privacy 2(4), e72 (2019) 19. Simone Raponi, Maurantonio Caprolu, and Roberto Pietro. Intrusion detection at the network edge: Solutions, limitations, and future directions. pages 59–75, 06 2019 20. Tian, Z., Luo, C., Qiu, J., Xiaojiang, D., Guizani, M.: A distributed deep learning system for web attack detection on edge devices. IEEE Trans. Industr. Inf. 16(3), 1963–1971 (2019) 21. Rashmi Agrawal. Predictive analysis of breast cancer using machine learning techniques. Ingeniería Solidaria, 15:1–23, 11 2019 22. Ondrej Linda, Todd Vollmer, and Milos Manic. Neural network based intrusion detection system for critical infrastructures. In 2009 international joint conference on neural networks, pages 1827–1834. IEEE, 2009

316

X.-T. Do and K.-H. Le

23. Nkiama, H., Said, S.Z.M., Saidu, M.: A subset feature elimination mechanism for intrusion detection system. Int. J. Adv. Comput. Sci. Appl. 7(4), 148–157 (2016) 24. Gao, X., Shan, C., Changzhen, H., Niu, Z., Liu, Z.: An adaptive ensemble machine learning model for intrusion detection. IEEE Access 7, 82512–82521 (2019) 25. Anazida Zainal, Mohd Aizaini Maarof, Siti Mariyam Shamsuddin, et al. Ensemble classifiers for network intrusion detection system. Journal of Information Assurance and Security, 4(3):217–225, 2009 26. Bo Dong and Xue Wang. Comparison deep learning method to traditional methods using for network intrusion detection. In 2016 8th IEEE International Conference on Communication Software and Networks (ICCSN), pages 581–585. IEEE, 2016

Chapter 25

A Real-Time Evaluation Framework For Machine Learning-Based IDS Anh-Hao Vu, Minh-Quan Nguyen-Khac, Xuan-Thanh Do, and Kim-Hung Le

Abstract With the rapid evolution of internal and external cyber threats, building a reliable security management system has become an urgent demand to mitigate system risks. In such systems, the Intrusion Detection System (IDSs) and Intrusion Prevention Systems (IPSs) are central components widely deployed to prevent malicious traffic from attackers. Most of the research target to enhance the performance of IDSs and IPSs. One problem that affects the performance is training datasets, and the solution to resolve this problem use benchmark datasets. However, there are many problems with that solution. Firstly, many valuable datasets used for evaluating the IDS model are internal and cannot be shared due to privacy issues. Secondly, opensource datasets such as DEFCON, KDD, CAIDA have its limitation and do not reflect the current world trends. In this paper, we introduce a framework used for practically evaluating the IDS models in real-time. The proposed framework also supports quickly generating different networking attacks that are similar to real scenarios. Indexes of performance (resource consumption, throughput, detection performance) of eight attacking scenarios are recorded and analyzed to demonstrate our proposal’s effectiveness. Keywords Machine learning-based IDS · Evaluation framework · Security management system

A.-H. Vu · M.-Q. Nguyen-Khac · X.-T. Do · K.-H. Le (B) University of Information Technology, Ho Chi Minh City, Vietnam e-mail: [email protected] A.-H. Vu e-mail: [email protected] M.-Q. Nguyen-Khac e-mail: [email protected] X.-T. Do e-mail: [email protected] Vietnam National University, Ho Chi Minh City, Vietnam © The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 V. E. Balas et al. (eds.), Recent Advances in Internet of Things and Machine Learning, Intelligent Systems Reference Library 215, https://doi.org/10.1007/978-3-030-90119-6_25

317

318

A.-H. Vu et al.

25.1 Introduction Recently, many organizations required employees to work from a distance due to the COVID-19 epidemic, make network traffic rapidly increasing, resulting in network overloading. So the security issues being a significant problem to be addressed. Hence, there is a lot more data demanding to manage and protect against cyber attacks. In this scenario, monitoring network assets and detecting anomalous behavior play a vital role in preserving the network system to achieve the CIA triad [1]. An Intrusion Detection System (IDS), which is being widely implemented in many organizations, acts as a monitoring and detection system. Intrusion Detection System (IDS) is a device or software application that monitors network behavior or system activities for malicious activities or policy violations and produces reports to a Management Station [2]. Many IDS types, such as Network Intrusion System (NIDS), analyze incoming networks; HIDS is used to monitor the operating system file. Besides, SIDS detects malicious behaviors by looking for specific patterns, using machine learning to create a defined mode of trustworthy activity, and then comparing new activity against this trust model. However, many IDSs still have some weaknesses, such as generating many true negative alerts that distract security analysts’ attention from incredibly harmful attacks and cannot quickly detect unknown patterns [3]. Thus Machine Learning based IDS appears to address the above problems. Machine learning is an application of artificial intelligence (AI) that can automatically learn and discover useful information. There are three common types of learning for a machine, including Supervised Learning: the model is getting trained on a labeled dataset with both input and output parameters; however, making labeled data manual is expensive and time-consuming. In contrast, unsupervised machine learning uses neither classified nor labeled information, making it much easier to obtain training data [4]. Finally, the semi-supervised machine falls somewhere in between supervised and unsupervised since both previous approaches are found. Some examples of Supervised Learning Algorithms are Linear Regression (LR), Nearest Neighbor (NN), Support Vector Machine (SVM). While Generative Adversarial Net (GAN), K-Means Clustering are examples of unsupervised learning [3]. Nowadays, Artificial Intelligence (AI) based techniques play a crucial role in developing IDS and have many efficiencies over other techniques. We can apply ML-based IDS to detect variant attacks that cannot be identified by legacy signaturebased IDS [5]. Besides, ML models for IDS depends heavily on input data to detect unknown network behaviors and patterns. Without data, ML-based IDS cannot operate correctly. Nevertheless, it is challenging to select a good dataset because many datasets are internal and cannot be revealed for privacy issues [6]. Therefore, the available datasets do not keep up on date with instruction evolution. To address the above problem, we would introduce a real-time evaluation framework comprising all necessary steps to evaluate the IDS model, including generating normal and abnormal traffic to capturing and sending traffic features to the IDS models. This framework provides a method to evaluate the IDS model without datasets and helps us to have insight understand our IDS model to develop it.

25 A Real-Time Evaluation Framework …

319

Finally, the rest of the paper is organized as follows. Section 25.2 would overview some available datasets and their limitations. Section 25.3 give detail of our framework. The evaluation of this framework is presented in Sect. 25.4. Section 25.5 discuss related work by introducing gaps. Section 25.5 conclude the paper and present future work.

25.2 Available Datasets and Its Limitations In this section, we outline several publicly available datasets for evaluating the IDS model and explain each limitation. DEFCON was generated by collecting the normal and abnormal traffic while conducting CTF competitions, which is observed, a collection upon a restrictive environment [7].DEFCON contains intensive attacks such as port scans and sweeps, buffer overflow, bad packets, administrative privilege and FTP by Telnet protocol attacks. However, DEFCON datasets are not compatible with the real world network traffic since DEFCON is produced during CTF competition. CAIDA (Center of Applied Internet Data Analysis 2002-2016): CAIDA consists of the traffic of flooding DDoS attack and anonymized passive traffic traces taken at west coast OC48 peering link and CAIDA’s equinox-Chicago. However, the CAIDA datasets do not contain a diversity of the attacks, mostly anonymized with their payload, protocol information, and destination [8–10]. DARPA (Lincoln Laboratory 1998, 1999): These IDEAVAL datasets were generated by simulation of a typical US Air Force Local Area Network (LAN) operation with multiple attacks. These attacks send and receive mail, browse websites, send and receive files using FTP, use telnet to log into remote computers, perform work, send and receive IRC messages, and monitor the router remotely using SNMP to perform other tasks. However, researchers criticized DARPA because their simulated normal network traffic is unrealistic and contains the absence of false positives. Finally, DARPA is outdated because of attack types and network infrastructure and lack of the actual attack data records [11, 12]. KDD’99: This dataset is the subset of the 1998 DARPA dataset generated by simulation of the operation of a typical US Air Force Local Area Network (LAN) with multiple attacks classified into four categories: denial of service, probe, a user to root, and remote to local [13]. Therefore, some of the existing problems in DARPA’98 remain in KDD’99. One of the KDD dataset limits is the significant number of redundant records, which leads to skewed testing results and prevents them from learning infrequent records [14]. CDX (United States Military Academy 2009): The CDX 2009 dataset was created during the network warfare competition to generate a labeled dataset. In these datasets, attackers used tools such as Nessus, WebScarab, and Nikto to automate and attack [15]. The traffic includes web, email, DNS lookups, and other required services. CDX does not contain the diversity of the attacks and volume in production networks [16].

320

A.-H. Vu et al.

Kyoto (Kyoto University, 2009): Honeypots are a mechanism to create this dataset. There are 24 statistical features, consisting of 14 conventional features like the KDD’99 dataset and ten additional features, which are very useful for analysis and evaluation. However, there is always abnormal traffic because the honeypot captured only attack activity. Therefore, it does not present the real situation of the internet [17]. Twente (University of Twente, 2009): The dataset was presented to be the first labeled data set for flow-based IDS, which had been created by collecting from the honeypot. It is much more realistic than the Tokyo dataset but still has some limitations—lack of labeling data set. Some unknown traffic and uncorrected alerts also collect but are not labeled. Because it mainly consists of malicious traffic, which allows detecting false negative, but not normal traffic for false-positive situations [18]. UMASS (University of Massachusetts, 2011): Tracing wireless packets to create the dataset is the main idea. This dataset can be used for testing purposes, but not enough to train IDS models because of the lack of information. ISCX2012 (University of New Brunswick, 2012): Using a systematic approach to create the dataset, ISCX consists of many advantage characteristics likes the realistic network and traffic, which include both normal and anomalous traffic, labeled dataset. However, it also has a few limits. ISCX captured popular attack types in the past, but nowadays, it lacks some protocol, which was used mostly, such as HTTPS. Its characteristics set a timeline for generating traffic, so there can not be called realworld-traffic [19]. ADFA (University of New South Wales, 2013): The Linux version’s Ubuntu Linux as the host OS with Apache ran the Tiki-Wiki website (written in PHP) to provide vulnerabilities for the attacks. The penetration tester used multiple attack types to get many attack data, such as web-based exploitation, simulated social engineering, and poisoned executables. Since it uses a tester to simulate an attack, the dataset is not realistic and lacks some attack type. Moreover, because the normal and malicious were not separated, it also has some mistakes in labeling [20].

25.3 The Framework In this section, we present the real-time evaluation framework: the network traffic generator, the feature extractor, and provide a table evaluating some of the IDS models.

25.3.1 Overview The real-time evaluating framework operates by creating botnets to create normal and abnormal network traffic, snipping and extracting features from regular and DDoS

25 A Real-Time Evaluation Framework …

321

Fig. 25.1 Framework overview

traffics, sending extracted features to IDS Model, and finally comparing detection results IDS model with labeled abnormal data created by the administrator. Figure 25.1 shows components of framework including: • Network Traffic Generator: The group of open-source tools responsible for simulate Botnet Traffic such as normal traffic and DDoS traffic. Example tools: Bonesi (Markus-Go), TrafficGenerator (HKUST-SING). By using Bonesi,1 we can generate DDos traffic with different protocol types. The attributes of the created packets and connection can be controlled by several parameters like send rate or payload size or they are determined by chance. There are 3 important components of this tools: The server listens for incoming requests and replies with a flow with the requested size; The client establishes persistent TCP connections to a list of servers and randomly generates requests over TCP connections according to the client configuration file; In the client configuration file, the user can specify the list of destination servers, the request size distribution, the Differentiated Services Code Point (DSCP) value distribution, the sending rate distribution and the request fanout distribution. • Traffic Scheduler: The component responsible for scheduling working time with Network Traffic Generator. It was created by crontab in each docker container of bots. There is specific time to enable each bot. When cron became active in time, a new process to generate traffic of each bot was started. Another time, pkill stops this process. Depending on the timestamp of flows catched later, we can determine it is benign or malicious. • Packet Capture & Feature Extractor: The component responsible for snipping the packet and extracting features from captured packets. In the framework, we are using CICFlowMeter [21] for that purpose. It is a useful tool, but has some limitations. Therefore, insteading of using the available pcap library, we used another pcap to capture packets, usually called pcap4j. • IDS Model: The component responsible for selecting features sent to the machine learning model for prediction. There are available models, which were trained and tested, so we just used the feature selector of these models into our framework. 1

You can find all option and more information at https://github.com/Markus-Go/bonesi. HKUSTSING/TrafficGenerator.

322

A.-H. Vu et al.

Fig. 25.2 The workflow of proposed framework

• Model Evaluating: The component compares the results of prediction with original labeled features for evaluating. Written in python, with an IDS model, the result of accuracy could be printed in real-time. Figure 25.2 illustrates the process in the framework works’ detail. Firstly, traffic scheduler schedules working time with Network Traffic Generator and create a label abnormal data. Then, Network Traffic Generator generates normal or DDos network traffic in the time scheduled by Traffic Scheduler. In the next step, Packet Capture acquires new packet and passes it to the Feature Extractor that uses to extract features from the captured packet. These features from Feature Extractor are the input of the IDS model. IDS model would analyze this feature to detect whether or not it is abnormal flows. Finally, the detection result of the IDS model is compared to abnormal label data created by Admin at first to get the accuracy in real-time.

25 A Real-Time Evaluation Framework …

323

25.3.2 Network Traffic Generator Network Traffic Generator is the process of injecting traffic into a network. With Traffic generator tools, we can mimic actual network traffic with different types of the protocol such as UDP, TCP, and ICMP for testing and training IDS model without realistic datasets. It is very challenging to create an artificial generation that generates both attack and benign traffic with different protocol types. For this reason, we would make use of 2 different open-source traffic generator tools: Bonesi, which servers as DDos traffic generator and generates ICMP, UDP, TCP flooding attacks from different IP addresses; and HKUST-SING/TrafficGenerator, which serves as a normal traffic generator which establishes a persistent TCP connection to servers in this framework. At the first step, we must enter a number of each type required when running the framework. The framework is able to create a docker-compose.yaml file and deploy docker containers by docker-compose tool later. The docker-compose.yaml file has many services, which are DDoS or normal bot. Each bot has a unique IP with the same subnet for identification. The number of them is entered in the beginning. The crontab configuration file and their scripts for scheduling is also copied to containers by Dockerfile and executed by the schedule method that we used in the framework. Next step, time-based job scheduler services are executed automatically when all bots are deployed. If some of them fall into an error state, it restarts the docker container and reruns the time-based job scheduler service. At this time, packets are generated and are ready for capture.

25.3.3 Feature Extractor Feature Extraction is a vital part of every machine learning process. In this framework, we also use features to run model predict algorithms. However, we do not dive deeply into the topic of machine learning or feature extraction. We use the available feature extraction of IDS models based on machine learning, which wants to be evaluated. However, before we do that, the generated traffic has been captured. There are many tools for capturing network packets, for example, Wireshark, Tshark, TCPDump. Nevertheless, it is a raw packet with common information; the DDoS detection process can not use that information of packet because the behavior of DDoS depends on a sequence of packets called flow. Therefore, we use CICFlowMeter, which has some modifications for capturing packets and extracting flows as features. The results acquired are written to a CSV file because this is a common file format for machine learning models. When the flow is received, the outputs are appended to the file in real-time.

324

A.-H. Vu et al.

25.3.4 IDS Model Evaluating Evaluating models is a significant part of the framework after generating a dataset. Based on the data created and extracted from the raw packet, called flows, we run models and measure their accuracy from the ratio of correctly malicious flow and benign flows overall flows. Most machine learning models were built in python, so we choose python to write a model evaluation. Firstly, we receive a CSV file path, which is output in the before process as input. In the script, there is a loop each second for checking the changeable in the input file.input CSV file. When it changes, we catch new flows and send them to prediction. Secondly, the machine learning model is loaded as a predicting module. Receiving the first step’s result, the predicted module run for checking the flow is normal or abnormal. Then, it compares the label with the original by timestamp in the schedule. Finally, calculate the accuracy, which is the percent of correct flow per total flow to predict.

25.4 Evaluation In this section, we would illustrate our running environment where the framework is deployed, explain the indexes of performance, and finally show the evaluation results running on multiple multiple scenarios.

25.4.1 Running Environment The framework was deployed on a personal computer with an Intel Core i5 processor (2 cores, four threads) and 12GB DDR3 memory, which ran on 64 bits Ubuntu 20.04 operating system. We used Bonesi (Markus-Go) to generate DDoS attack flows and TrafficGenerator (HKUST SING) to receive normal traffic flows. All of them were run on docker containers, which were remote from a python script. CICFlowMeter was used for snipping and extracting features as flow from both normal traffic and DDoS traffics. The results were saved in a CSV file, and the python script with the machine learning predict function could catch up with new flows to get accuracy in real-time. For clearly, the python script is a central part of the framework. There are deploying bot’s docker container, running a server to receive traffic, executing CICFlowMeter to get features, and sending it to a machine learning predict function for evaluation.

25 A Real-Time Evaluation Framework …

325

25.4.2 Index of Performance Accuracy: Accuracy is a ratio that calculates by place correct predictions flow (malicious as true positive—TP, benign as true negative—TN) onto total flows. Besides TP and TN, we also have false positive (FP) that false detection of benign flows and false negative (FN), which falsely detects malicious flows. TP +TN T P + T N + FP + FN

(25.1)

Scenario: The situation to test our framework with the difference in the number, the protocol of each group (malicious or benign bot), and testing duration. In the runtime, we collect system metrics of each group of bots and the random forest model’s accuracy every minute. By the end, we received the accuracy of the model in total time. Moreover, other models calculated by a dataset had been generated before that no need to execute the scenario again. For each group of the bot: • Max Memory (in megabyte) and Max CPU (in percentage): The metrics which we collect in the docker stats panel every minute and choose the maximum of them at the end. The memory or the CPU we mention is not a metric of each bot, and it is a group (for example, total memory of benign bots or malicious bots). • Requests/s was known as the total number request of the framework was generated in the runtime. For the system: • Max Memory (in megabyte) and Max CPU (in percentage): The metrics which we collect in the system monitor the host machine every minute and choose the maximum of them at the end. The CPU rate is taken directly, but the memory is the difference between before and after framework execution. • Duration: a running time. • Flows/Data Points: the number of flows in the generated dataset after running that we received.

25.4.3 Results About the attacker, we limited docker container memory to 1.914GiB. Trying 30 times with different protocols in one DDoS bot, and we have a result table. In our environment, Table 25.1 presents each DDoS that bot reached approximately 100% CPU. Depending on the protocol type, the memory used may be different. UDP and ICMP have lower memory than TCP and HTTP. Some knowledge about network packets can explain the situation. ICMP is a data link layer protocol in the OSI model. It requires less memory than TCP or UDP (transport layer). Both

326

A.-H. Vu et al.

Table 25.1 DDos bots performance through each protocol Protocol CPU% Mem usage(MiB) UDP TCP ICMP HTTP

83.59% − 95.48% 90.11% − 95.82% 80.10% − 95.59% 88.67% − 93.08%

2.801MiB 513.7MiB 2.574MiB 516.1MiB

Table 25.2 Normal bots performance through each rate RX bandwidth CPU% Mem usage(MiB) and Rate 30 Mbits/sec 1237 requests/595s 1000 Mbits/sec 41265 requests/595s 5000 Mbits/sec 206327 requests/595s

0.06% −0.82% 8.37% −18.98% 40.02% 60.02%

Mem% 0.14% 26.31% 0.13% 26.34%

Mem%

18.2MiB

0.93%

19.14MiB

0.98%

62.02MiB

3.16%

TCP and UDP are in the transport layer, but UDP memory is less than TCP memory. Because UDP is a non-handshake protocol, it just sends a packet and does not care about the response packet. On the other side, TCP is connection-oriented and has a three-way handshake. It requires more memory than UDP to handle packets. The remainder HTTP-is a TCP protocol. It is the same as TCP but better because of its application layer. We must deploy about 638 DDoS bots in UDP or ICMP to reach maximum memory in our calculation. And four bots in TCP or HTTP for the same result. Though, it is just theory, an expensive way to reach the highest performance. In reality, it depends on the CPU, so we had made performance tests in the next section and had some results. About the normal bot, we use limited as the same as the attacker. However, keeping the protocol is TCP; we only change RX Bandwidth to test performance. Table 25.2 presents normal bots performance through each rate. The CPU and memory must not be high rate. Because we demand normal traffic, it is merely to generate a few requests along the time. In our experiments, we create eight scenarios for evaluating the framework. We change the quantity, protocol, and running time of normal and malicious bots in each situation. After that, we measure some characteristics like max CPU (in percent), memory usage (in megabyte) of each bot group, the number of flows that the system received, and the accuracy of available machine learning models. In this case, there are three models: Random Forest, Decision Tree, Naive Bayes, that was trained by our friend in another project. We do not create the models because our purpose is to make the framework to evaluate them. Our result is shown in Table 25.3.

Accuracy

System

DDOS Bot

Normal Bot

Quantity Protocol Memory (MB) Max CPU Requests/s Quantity Protocol Memory (MB) Max CPU Requests/s CPU Memory (MB) Duration Flows/Data Points Random Forest Decision Tree Naive Bayes 46,30% 46,29%

00:34:34 94.386

00:30:00 72.549

40,45% 59,54%

192,00% 500 100,00% 600,00

100,00% 500 100,00% 1.300,00

46,29%

3,04% 100 1 TCP 517

1,45% 100 1 UDP 4,2

40,50%

1 TCP 19,66

1 TCP 19,96

Table 25.3 The evaluation results of six scenarios Scenario 1 Scenario 2 Scenario 3

42,92% 61,32%

58,96%

00:34:13 212

101,00% 1000 100,00% 500,00

2,62% 100 1 ICMP 4

1 TCP 19,97

41,92% 58,05%

41,95%

00:32:10 141.718

111,00% 1000 100,00% 1.100,00

3,00% 200 2 UDP 8,1

2 TCP 40

Scenario 4

21,84% 21,90%

21,80%

00:30:00 76.819

230,00% 1000 100,00% 900,00

2,60% 200 2 TCP 1024

2 TCP 39,6

Scenario 5

56,18% 54,72%

56,60%

00:31:28 478

191,00% 2000 100,00% 700,00

3,20% 200 2 ICMP 8

2 TCP 40,1

Scenario 6

85,47% 14,62%

85,50%

00:31:13 87.606

270,00% 2300 100,00% 2.200,00

7,20% 800 10 UDP 40

10 TCP 197

Scenario 7

75,45% 24,61%

75,43%

03:00:00 403.852

100,00% 500 100,00% 800,00

1,60% 100 1 UDP 6,3

1 TCP 20,2

Scenario 8

25 A Real-Time Evaluation Framework … 327

328

A.-H. Vu et al.

In the first scenario, we have the framework with one normal bot with TCP protocol, one malicious bot with UDP protocol, running in 30 minutes. The second and the third, we change the malicious bot protocol to TCP and ICMP. Normal bots always have the CPU percentage lower than malicious bots. Our malicious bots simulate DDoS traffic, which has a large amount of traffic generated. As the last explained, each scenario’s memory usage is different, highest in TCP, then UDP, and final is ICMP. At the flows received, the framework can catch TCP flows better than UDP and the worst is ICMP. One reason for this is that it can not detect the signature that UDP flows were ended. CIC depends on the timeout to close the flows of UDP. On the other side, TCP connections have a clear way to check the end state (closed connection) because it is a connection protocol. Another side, CIC, was not designed to capture ICMP packets. In this situation, the flows captured are only benign. The fourth, fifth, and sixth have the same scenarios as the three scenarios before, but the quantity of each bot is two. The performance of the framework does not surprise us; all metrics are calculated in theory. The CPU and memory of normal bots are duplicated. The CPU reaches 200%because bots run on docker containers that use a core of the host CPU, but we have four. Regarding the detection accuracy, scenario seven has the highest accuracy in the random forest and decision tree algorithms reported about 85.5% and 85.47%, respectively. This means that these algorithms are suitable to detect UDP DDoS attacks from several attackers (scenario seven simulates 10 DDoS Bot and Normal Bot).

25.5 Conclusion In this paper, we present the limitation of some available datasets. Based on such study, we introduce a novel evaluation framework supporting real-time evaluating DDoS detection algorithms by quickly creating various attack scenarios. The efficiency and practicality of our proposal are demonstrated by generating eight real attacking scenarios. In the future, we would develop our Network Traffic Generator to generate both normal and abnormal traffic with various protocols. Acknowledgements This research was funded by Vietnam National University Ho Chi Minh City (VNU-HCM) under Grant no. DSC2021-26-04.

References 1. Gunduz, M.Z., Das, R.: Cyber-security on smart grid: Threats and potential solutions. Comput. Netw. 169, 107094 (2020) 2. Borkar, A., Donode, A., Kumari, A.: A survey on intrusion detection system (ids) and internal intrusion detection and protection system (iidps). In: 2017 International Conference on Inventive Computing and Informatics (ICICI), pp. 949–953. IEEE (2017)

25 A Real-Time Evaluation Framework …

329

3. Jean-Philippe, R.: Enhancing Computer Network Defense Technologies with Machine Learning and Artificial Intelligence. PhD thesis, Utica College (2018) 4. van Oorschot, P.C.: Intrusion detection and network-based attacks. In: Computer security and the internet, pp. 309–338. Springer (2020) 5. Nazir, A., Khan, R.A.: Network intrusion detection: taxonomy and machine learning applications. In: Machine Intelligence and Big Data Analytics for Cybersecurity Applications, pp. 3– 28, Springer (2021) 6. Meena, G., Choudhary, R.R.: A review paper on ids classification using kdd 99 and nsl kdd dataset in weka. In: 2017 International Conference on Computer, Communications and Electronics (Comptelix), pp. 553–558. IEEE (2017) 7. Defcon 8, 10 and 11: http://cctf.shmoo.com/. Last accessed 30 Sept 2020 8. Caida data set oc48 link a: https://www.caida.org/data/passive/passiveoc48dataset.xml. Last accessed 30 Sept 2020 9. Caida ddos attack dataset. https://www.caida.org/data/passive/ddos-20070804dataset.xml. Last accessed 30 Sept 2020 10. Caida anonymized internet traces 2016 dataset. https://www.caida.org/data/passive/ passive2016dataset.xml. Last accessed 30 Sept 2020 11. Brown, C., Cowperthwaite, A., Hijazi, A., Somayaji, A.: Analysis of the 1999 darpa/lincoln laboratory ids evaluation data with netadhict, pp. 1 – 7, vol 08 (2009) 12. McHugh, J.: Testing intrusion detection systems: A critique of the 1998 and 1999 darpa intrusion detection system evaluations as performed by lincoln laboratory. ACM Trans. Inf. Syst. Secur. 3, 262–294, 11 (2000) 13. Tavallaee, M., Bagheri, E., Lu, W., Ghorbani, A.: “A detailed analysis of the kdd cup 99 data set. In: IEEE Symposium. Computational Intelligence for Security and Defense Applications, CISDA 2, 07 (2009) 14. Dhanabal, L., Shantharajah, S.P.: A study on nsl-kdd dataset for intrusion detection system based on classification algorithms (2015) 15. Sharafaldin, I., Habibi Lashkari, A., Ghorbani, A.: Toward generating a new intrusion detection dataset and intrusion traffic characterization 01, 108–116 (2018) 16. Asnm-cdx-2009 dataset: http://www.fit.vutbr.cz/~ihomoliak/asnm/ASNM-CDX-2009.html. Last accessed 30 Sept 2020 17. Description of kyoto university benchmark data: https://www.takakura.com/Kyoto_data/ BenchmarkData-Description-v5.pdf. Last accessed 30 Sept 2020 18. Sperotto, A., Sadre, R., Vliet, F., Pras, A.: A labeled data set for flow-based intrusion detection 10, 39–50 (2009) 19. Intrusion detection evaluation dataset (iscxids2012). https://www.unb.ca/cic/datasets/ids.html. Last accessed 30 Sept 2020 20. Creech, G., Hu, J.: A semantic approach to host-based intrusion detection systems using contiguousand discontiguous system call patterns. Comput. IEEE Trans. 63(04), 807–819 (2014) 21. Lashkari, A., Zang, Y., Owhuo, G., Mamun, M., Gil, G.: Cicflowmeter-a network traffic biflow generator and analyzer. Acessado em, 15–08 (2018)