Visible Light Communications: Applications and Research Advances 9781685077440, 9781685078287, 9781685076634, 9781685077600, 9781536165562, 9781536165579, 9781536158304, 9781536158311, 9798886975642

Citation preview

Electronics and Telecommunications Research Energy Science, Engineering and Technology

No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

Electronics and Telecommunications Research Recent Trends in Microstrip Antennas for Wireless Applications R. Nagarajan, BE, ME, PhD (Author), S. Kannadhasan, PhD (Authors) 2022. ISBN: 978-1-68507-744-0 (Softcover) 2022. ISBN: 978-1-68507-828-7 (eBook) Types of Photodetectors and their Applications Sunil Singh Kushvaha, PhD, Vidya Nand Singh, PhD (Editors) 2022. ISBN: 978-1-68507-663-4 (Hardcover) 2022. ISBN: 978-1-68507-760-0 (eBook) Recent Advancement in Electronic Devices, Circuit and Materials Suman Lata Tripathi and Sanjeet Kumar Sinha (Editors) 2020. ISBN: 978-1-53616-556-2 (Hardcover) 2020. ISBN: 978-1-53616-557-9 (eBook) Smartphones: Recent Innovations and Applications Paolo Dabove, PhD (Editor) 2019. ISBN: 978-1-53615-830-4 (Hardcover) 2019. ISBN: 978-1-53615-831-1 (eBook)

More information about this series can be found at https://novapublishers.com/product-category/series/electronics-andtelecommunications-research/

Ajit Kumar Nishant Sharan and Swapan Kumar Ghorai

Visible Light Communications Applications and Research Advances

Copyright © 2023 by Nova Science Publishers, Inc. https://doi.org/10.52305/RYER7275 All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. We have partnered with Copyright Clearance Center to make it easy for you to obtain permissions to reuse content from this publication. Please visit copyright.com and search by Title, ISBN, or ISSN. For further questions about using the service on copyright.com, please contact: Copyright Clearance Center Fax: +1-(978) 750-4470

Phone: +1-(978) 750-8400

E-mail: [email protected]

NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works. Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the Publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regards to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS.

Library of Congress Cataloging-in-Publication Data

ISBN: 979-8-88697-564-2

Published by Nova Science Publishers, Inc. † New York

Contents

List of Figures ........................................................................................... vii List of Tables

.......................................................................................... xiii

Preface

........................................................................................... xv

Acknowledgements .................................................................................... xix Abbreviations .......................................................................................... xxi Chapter 1

Introduction ....................................................................... 1

Chapter 2

BER Analysis of a Multipath MIMO-VLC System with Various Source and Receiver Configurations ................................................................. 17

Chapter 3

BER Performance Comparison in a Multipath MIMO-VLC System with Different LED Radiation Patterns ........................................................... 35

Chapter 4

BER Investigation for an OFDM-Based Hybrid PLC-VLC System ............................................... 55

Chapter 5

Downlink Multipath Multi-User NOMA-VLC System Performance Analysis ........................................ 79

Chapter 6

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System Using a Union of Precoder and Companding ........................... 103

Chapter 7

SWIPT Integrated VLC/RF System Performance Evaluation Using the Hybrid-OMA-Cooperative-NOMA Scheme................ 117

Chapter 8

Concave-Convex Lens with High Gain and Volume Efficiency for MIMO-VLC Systems .............. 137

vi

Contents

Chapter 9

Conclusion and Future Scope ....................................... 157

References

......................................................................................... 161

Index

......................................................................................... 173

About the Authors .................................................................................... 177

List of Figures

Figure 1.

Reflected sunlight used in photophone for transmitting signals [2]. ........................................................................... 2

Figure 2.

Photophone receiver [2]....................................................... 3

Figure 3.

Expected device share by 2023............................................ 4

Figure 4.

The various options available to overcome the overloaded RF spectrum. ..................................................... 5

Figure 5.

Electromagnetic spectrum. .................................................. 8

Figure 6.

Block diagram of VLC system. ........................................... 8

Figure 7.

Advantages of VLC system. .............................................. 11

Figure 8.

Application of VLC system. .............................................. 13

Figure 9.

Major challenges encountered in implementation of VLC system. .................................................................. 14

Figure 10.

Indoor 4×4 geometrical model with multipath propagation. ....................................................................... 20

Figure 11.

First reflection light rays captured within the receiver’s field of view. ..................................................... 22

Figure 12.

Case 1 and Case 2 room views. ......................................... 23

Figure 13.

Flow chart for determining impulse response. .................. 24

Figure 14.

BER curves for configuration-1. ....................................... 25

Figure 15.

BER curves for configuration-2. ....................................... 26

Figure 16.

LOS channel impulse response (IR) with ST = 0.2m at Rx3 due to all LEDs. ..................................... 29

viii

List of Figures

Figure 17.

LOS channel impulse response (IR) with ST = 0.4m at Rx3 due to all LEDs. ..................................... 29

Figure 18.

The RMS delay spread for ST = 0.4m is Max = 0.7854ns and Min = 0.1542ns. ............................... 30

Figure 19.

The RMS delay spread for ST = 0.6m is Max = 0.049ns and Min = 0.1724ns. ................................. 31

Figure 20.

BER performance for LOS, L-R1, and L-R1-R2 channels at various LED positions. .................... 32

Figure 21.

Bit rate for ST = 0.4m. Max. = 64.85Mbps & Min. = 12.76Mbps. ............................................................ 32

Figure 22.

Bit rate for ST = 0.6m. Max. = 58Mbps & Min. = 9.53Mbps. .............................................................. 33

Figure 23.

Indoor multipath MIMO-VLC geometrical model. ........... 38

Figure 24.

Radiation pattern of Lambertian, Batwing and Elliptical pattern (a) Cartesian plot (b) Polar plot.............. 41

Figure 25.

VLC-MIMO system considered for BER simulation in ceiling-mounted LED layout. ...................... 44

Figure 26.

Impulse response for LOS signal at centre of room. ......... 44

Figure 27.

Impulse response for LOS signal at corner of room. ......... 45

Figure 28.

Impulse response at the centre of room for first reflection signal. ................................................................ 45

Figure 29.

Impulse response at the corner of room for first reflection signal. ................................................................ 46

Figure 30.

Channel gain distribution for ceiling-mounted layout a) LOS signal b) First reflection signal................... 47

Figure 31.

BER performance for LOS and L-R1 signals using all three radiation patterns at the centre position of the room................................................................................... 48

Figure 32.

BER performance for LOS and L-R1 signals using all three radiation patterns at the corner position of the room. ....................................................................... 49

List of Figures

ix

Figure 33.

VLC-MIMO system considered for BER simulation in wall-mounted LED layout. ............................................ 50

Figure 34.

Impulse response for wall-mounted layout using a) LOS signal b) First reflection signal.............................. 51

Figure 35.

Channel gain for wall-mounted layout a) LOS signal b) First reflection signal. .................................................... 52

Figure 36.

BER performance for wall-mounted scenario (a) LOS signal (b) L-R1 signal. ......................................... 53

Figure 37.

Block diagram of proposed HPMV system. ...................... 59

Figure 38.

Flow chart of PLC-MIMO-VLC transmitter system. ........ 60

Figure 39.

Frequency response of PLC channel. ................................ 63

Figure 40.

Indoor Geometrical model of HPMV system. ................... 63

Figure 41.

BER performance of HPMV system for configuration A a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m). ............................................................. 67

Figure 42.

Frequency response of MIMO-VLC system for configuration A a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m). ........................... 69

Figure 43.

Frequency response of HPMV system for configuration A. ................................................................. 70

Figure 44.

BER performance of HPMV system for configuration B a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m). .................................................................................. 71

Figure 45.

Frequency response of MIMO-VLC system for configuration B a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m). ........................... 72

Figure 46.

Schematic diagram of experimental set up. ....................... 73

Figure 47.

The experimental set up of PLC-VLC system. .................. 74

Figure 48.

Generated signal on pin no. 9 and received signal on pin no. 10 by interfacing Arduino hardware with MATLAB Simulink. .................................. 74

Figure 49.

Transmitter and receiver of PLC. ...................................... 74

x

List of Figures

Figure 50.

Hardware of VLC transmitter circuit using four LEDs. .... 75

Figure 51.

Hardware of VLC receiver circuit using single LDR. ....... 75

Figure 52.

(a) The generated signal taken from output of pin 9 of Arduino (b) FSK modulated signal from output of PLC modem (c) FSK demodulated signal at PLC receiver (d) Received signal from LDR output. ........ 76

Figure 53.

Indoor downlink multipath multi-user NOMA-VLC model. ................................................................................ 84

Figure 54.

Downlink k-user NOMA-VLC system schematic block diagram. ................................................................... 87

Figure 55.

Flow chart for BER and sum rate simulation using LOS and L-R1 signal. ........................................................ 89

Figure 56.

LOS BER performance of a NOMA-VLC system using different values of φ(1/2) for (a) a two-user system, (b) a three-user system, and (c) a gradual increase in BER with increasing LED semi-angle............. 91

Figure 57.

L-R1 BER performance of a NOMA-VLC system using various values of φ1/2 for (a) a two-user system and (b) a three-user system. ............................................... 92

Figure 58.

LOS BER vs. Power allocation coefficient (a) Two-user system (b) Three-user system (c) Two-user system at SNR of 123dB for α values ranging from 0.4 to 0.85 [79]. ............................. 93

Figure 59.

Sum rate vs. normalized offset using LOS signal.............. 95

Figure 60.

Avg. sum rate vs. semi-angle of LED................................ 96

Figure 61.

Sum rate vs. normalized offset using blocked-LOS signal. ................................................................................ 96

Figure 62.

Avg. sum rate vs. semi-angle for blocked LOS. ................ 97

Figure 63.

Sum rate Vs. SNR (dB) for two-user & three-user system using different values of φ1/2. ................................ 97

Figure 64.

Outage probability of blocked LOS scenario. ................. 101

List of Figures

xi

Figure 65.

Transceiver of the proposed NOMA DCO-OFDM system. ............................................................................. 106

Figure 66.

PAPR comparison of different NOMA DCO-OFDM systems. ........................................................................... 111

Figure 67.

BER comparison of different NOMA DCO-OFDM systems. ........................................................................... 111

Figure 68.

Mean EVM comparison for different LED Bias Voltages (a) LED Bias = 1.6 V, (b) LED Bias = 1.625 V, (c) LED Bias = 1.675 V and (e) LED Bias = 1.7 V. .................................................................... 112

Figure 69.

SIC performance comparison for 3 user systems. ........... 114

Figure 70.

Geometrical indoor model for proposed system model. .. 123

Figure 71.

Sum rate Vs. semi-angle of the LED for (a) SNR = 40dB (b) SNR = 60dB (c) SNR = 80dB (d) SNR = 100dB (e) SNR = 120dB (f) SNR = 140dB. .. 128

Figure 72.

Concept of TDMA, NOMA, NOMA+TDMA. ............... 128

Figure 73.

Sum rate Vs. power allocation factor for (a) SNR = 40 dB (b) SNR = 60dB (c) SNR = 80dB (d) SNR = 100dB (e) SNR = 120dB (f) SNR = 140dB. .. 130

Figure 74.

(a) FOV of scenario 1 & scenario 2. (b) Sum rate vs. semi-angle of LED for scenario 1 (c) sum rate vs. semi-angle for scenario 2 (d) Comparison of scenario 1 & scenario 2 in same figure. ........................... 133

Figure 75.

Achievable rate for far user and total sum rate for (a) Transmit power = 40dB (b) Transmit power = 60dB (c) Transmit power = 80dB (d) Transmit power = 120dB. .......................................... 134

Figure 76.

Power harvested vs semi-angle of the LED. .................... 135

Figure 77.

3D structure of (a) Hemispherical lens (b) Concave-convex lens (c) CPC lens (d) DTIRC lens designed in AutoCAD. .......................... 139

Figure 78.

Cross-sectional view of concave-convex (OAPB) and hemispherical (OAO’B) lens. ................................... 140

xii

List of Figures

Figure 79.

Geometrical configuration to find the ratio of entrance aperture area of concave-convex lens to the hemispherical lens. ................................................ 141

Figure 80.

Radius of convex aperture of proposed lens is R = 4.77 mm corresponding to radius of hemispherical lens r = 2.5 mm as designed in AutoCAD......................................................................... 142

Figure 81.

The ratio of power received using concave-convex lens and the hemispherical lens as a function of (r/R). .... 143

Figure 82.

Ray trajectories for the scenario when rays are incident parallel to z-axis for (a) Hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC. ................ 144

Figure 83.

Ray trajectories for scenario when rays are incident at 450 w.r.t z-axis (a) hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC. ................ 145

Figure 84.

Close view of ray trajectories inside (a) Hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC. ........ 146

Figure 85.

Ray trajectories using two sources whose light rays are incident at φ = 450 (a) Hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC (e) Rotated version of Figure 85 (d) to show scattered rays from the exit aperture. ............................... 147

Figure 86.

Geometrical model of indoor VLC system with four LEDs and array of four PDs..................................... 149

Figure 87.

Geometrical model showing ray tracing in Concave-convex lens with photodetector array. .............. 150

Figure 88.

Received power using (a) Hemispherical lens [Max. = 1.4682 mW & Min. = 0.5083 mW] (b) Concave-convex lens [Max. = 1.5858 mW & Min. = 0.5490 mW]. ........................................................ 152

Figure 89.

Channel gain for different values of φ and ϕ1/2 using hemispherical lens [151] and concave-convex lens. ........ 154

List of Tables

Table 1.

Difference between Li-Fi and Wi-Fi ................................... 7

Table 2.

MIMO-VLC parameters and their values .......................... 27

Table 3.

Parameters and their values ............................................... 43

Table 4.

Comparison of SNR at BER of 10-5 between scenario 1 and 2 for ceiling-mounted case ........................ 50

Table 5.

Comparison of SNR at BER of 10-5 for wall-mounted case .................................................................................... 53

Table 6.

The parameters of HPMV and their values ....................... 66

Table 7.

Comparison of required SNR at BER of 10-5 between configuration A and B ......................................... 72

Table 8.

Parameters and their values ............................................... 88

Table 9.

Comparison of the benefits and drawbacks of the proposed work with various research articles .................... 99

Table 10.

Comparative analysis of simulated results with previous research work .................................................... 100

Table 11.

Simulation parameters (NOMA DCO-OFDM) ............... 110

Table 12.

EVM comparison at LED bias = 1.6 V............................ 113

Table 13.

EVM comparison at LED bias = 1.65 V.......................... 113

Table 14.

EVM comparison at LED bias = 1.7 V............................ 114

Table 15.

Simulation parameters and their associated values.......... 123

Table 16.

Sum rate comparison using various schemes at various SNR ranges ..................................................... 127

xiv

List of Tables

Table 17.

Sum rate using HOCN and SC-NOMA under different SNR and power allocation coefficient ranges ... 131

Table 18.

Summary for selecting the best option from scenarios 1 and 2 ............................................................................. 133

Table 19.

Parameters and their corresponding values for indoor VLC system.......................................................... 149

Table 20.

Comparison of received power using concave-convex, hemispherical, bare CPC, gradient-index CPC................ 153

Preface

Owing to the unprecedented rise in the demand of the data network, the total number of internet users is forecast to exceed 5 billion by 2023, accounting for approximately 66% of the global population. Machine-to-Machine (M2M) connections will be the fastest-growing device and connection category, increasing nearly 2.4-fold (19% compound annual growth rate-[CAGR]) to 14.7 billion connections by 2023. As a result, the current constrained RF spectrum does not appear to address the bottlenecks of meeting everincreasing demands for low-cost seamless networks, reliable connectivity, and high-speed data rates in 5G networks and beyond. In this context, visible light communication (VLC), which uses visible spectrum to communicate data, appears to be an appealing alternative for RF networks. According to one survey, most people spend 80 percent of their time indoors, and thus VLC, with its dual advantage of illumination and communication, becomes a more promising complementary to the overburdened RF system. LEDs provide additional benefits to the VLC system due to their long life, energy efficiency, low cost, and ease of availability. In spite of many advantages of LEDs, they have one disadvantage of limited modulation bandwidth which can limit the bit rate during their use in communication. This limitation is compensated by using MIMO technique which increases the bit rate through spatial multiplexing, in addition, to increase in illumination inside the room. In indoor VLC system, there are two types of paths through which light ray reaches the photodetector (PD) from the LED: line-of-sight (LOS) and non-line-of-sight (NLOS). Light ray arriving at PD through LOS path has good strength since there is no obstruction, whereas, in case of NLOS path, the light ray suffers multiple reflections through different reflecting surfaces such as walls, floor, ceiling, table etc. and hence its strength is lesser than LOS signal. Moreover, in many practical scenarios, an unobstructed optical path may not be available due to various room configuration, e.g., transmitter may not be placed at sufficient height or because of mobility of receiver as per requirements. Therefore, NLOS signal must be taken into consideration in VLC system.

xvi

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

In order to provide seamless wireless communication with high data rates and low latency, the service must use highly reliable, low-cost, and high-speed technologies. These issues are expected to be addressed by implementing technologies like OFDM and NOMA. However, the non-linear current voltage (I-V) characteristics of LED becomes hindrance for the implementation of OFDM due to high PAPR. So, there is a need of strategies which can mitigate the effect of non-linear distortion caused by the LEDs characteristics. Also, to meet the target of last meter and last mile network coverage, there is need of hybrid combination of VLC system with other wired communication system. No doubt, fibre optic networks which are covering the cities and continents have been beneficial in meeting the broadband subscriber targets. However, it may not be practical to lay down cables in the last mile due to cost and time-consuming factors along with the concern of maintenance and breakage. In this context, the already existing power line infrastructure, due to its multiple advantages like cost-effectiveness, ability to carry both power and data, makes power line communication (PLC) technology a promising and strong alternative. PLC is already being used in home automation, smart grid and telemetry. The VLC system has some limitations, one of which is uneven quality-ofservice (QoS) distribution. This is due to the fact that light intensity varies inversely to the square of the distance between the LED and the users, causing the VLC channel to degrade significantly with distance. As a result, the conic area of the LED light beam is small, and it is easily blocked by obstacles. In a real-world indoor scenario, this causes a noticeable performance gap between the near and far user. Other factors that influence the received signal power include the receiver’s field-of-view (FOV) and the half-power semi-angle of the LEDs. The problem of limited FOV of the receiver can be overcome by designing an optical antenna having large FOV, high gain, and small size. The QoS can be uniform by implementing cooperative NOMA along with hybrid combination of VLC/RF system.

Prerequisite Though the book has been written to appeal to a wide range of readers, the primary target audience is the research scholar whose field of interest corresponds to the subject matter of this book. Many academic institutions offer a course called “Optical Wireless Communication System,” in PG & there it would be beneficial to those pursuing higher education in the

Preface

xvii

engineering field. The book covers a wide range of 5G topics that will pique the interest of both academics and industry. The analysis presented in the book will provide many insights into future problem solving. The presentation language is simple to understand and reach a conclusion. Furthermore, the book is written coherently so that the reader does not lose interest in the middle.

Approach With all the above discussed issues related to visible light communication system, this book provides a thorough examination of an indoor visible light communication system and covers four major aspects: (i) Transmitters (ii) Multipath channel (iii) Receiver (iv) Multiple access technologies. Thus, all the major systems of VLC system have been addressed in the book.

Organization of the Book The book is divided into total 9 chapters as described below: The chapter 1 provides a brief history of communication systems before progressing to the VLC system. This chapter has discussed the benefits, applications, and challenges associated with the VLC system. The chapter 2 discusses the BER analysis of a 4×4 MIMO-VLC system using LOS and LOS plus NLOS signals with different spatial separations between LEDs and PDs. The effect of LED spatial separation on impulse response, RMS delay spread, and bit rate inside the room has been studied. The significance of higher-order reflected signals in the MIMO-VLC system has been stressed. Chapter 3 presents the impact of different types of radiation profile of LEDs on the BER performance of the 4×4 MIMO-VLC system using LOS and LOS plus diffused signals. The BER analysis has been considered for different positions of the room using different radiation patterns of LED. The aim of this chapter is to show that radiation pattern of LED plays a vital role in BER performance for indoor MIMO-VLC system. In Chapter 4, the BER performance of OFDM-based integrated power line communication-MIMO-visible light communication is investigated. In addition, a low-cost experimental setup of an integrated PLC-MISO-VLC system has been demonstrated.

xviii

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Chapter 5 outlines the performance of multipath multi-user nonorthogonal multiple access (NOMA) system using BER, sum rate and outage probability. The optimal value of the semi-angle of the LED and power allocation coefficient have been examined for two-user and three-user system for the minimum BER. Also, the effect of shadowing on the sum rate has been examined. Chapter 6 offers a hybrid scheme using a blend of precoder with compander to reduce the PAPR of NOMA based DC-biased optical OFDM system. The error vector magnitude (EVM) using proposed scheme has been compared with precoded and companded NOMA DCO-OFDM systems. The proposed system shows improvement in nonlinearity and BER performance and hence proved to be a good choice for the high speed 5G indoor mobile data network. Chapter 7 proposes a hybrid OMA-Cooperative-NOMA (HOCN) scheme that integrate VLC and radio frequency (RF) system. Also, an energy harvesting scheme has been utilized to ensure that near user have enough power to relay data to far users. Further, the various FOV combinations of near and far users with varying semi-angles has been studied. The sum rate using proposed HOCN system with different user combinations has been compared with the conventional schemes. The impact of power harvesting efficiency on the far user rate as well as the sum rate has been investigated with different transmit power. Chapter 8 introduces a novel lens that has higher light collection efficiency than conventional lenses for the MIMO-VLC receiver system. This has been demonstrated by simulating ray trajectories in Comsol software. The received power has been measured in a 5m×5m×3m room using the proposed lens and four LEDs. The proposed lens is also volume efficient, making it ideal for compact MIMO-VLC receivers. Chapter 9 includes the conclusion and future scope.

Contact Your feedback and suggestions for the improvement of this book are welcome. While every attempt has been made to eliminate errors in this book, a few may still have managed to creep in. Kindly point them out to me – my email id is [email protected]

Acknowledgements

The work presented in the book has only been possible due to the immense help rendered by people over the duration of its execution. I take this opportunity to convey my heartfelt gratitude to everyone who has been part of this journey. I first thank Nadya S. Columbus, president of Nova Science Publishers, Inc., who invited me to write this book. I hereby acknowledge the contribution, impetus, and knowledge provided by my co-authors, Dr. S. K. Ghorai, Professor, dept. of ECE, BIT, Mesra & Dr. Nishant Sharan, Assistant professor, dept. of ECE, Madanapalle Institute of Technology & Science, Angallu, Madanapalle. Without their input, continued support and judicious advice, I would not have been able to bring a book of this scope to such a conclusion. My parents (Foudi Rana & Champa Devi), brothers (Ranjit Kumar & Sujit Kumar), wife (Kajal Rana) and son (Arnav Kumar) have been greatest source of inspiration and the driving force behind this book. Without their help, support and kind understanding, I would never have been able to undertake a research work of this magnitude. The support provided by them, to produce this outcome cannot be quantified in any manner. I also acknowledge the roles played by my senior research fellows Dr. Mainak Basu and Dr. Somnath Sengupta along with my colleagues Dr. Amresh Kumar, Dr. Ashutosh Anand for their moral support and much needed advice during this venture. These people provided a congenial atmosphere and helped me every step up the ladder to the successful completion of this book. My heartfelt gratitude goes to University Grant Commission (ministry of HRD) for financially supporting me from 2015 to 2020 for my research work. Last, but not the least, I thank to Presidency University, under whose umbrella I have been able to write this book. Ajit Kumar, PhD

Abbreviations

OWC VLC BER MIMO RF FSO UVC IrDA IROW Li-Fi WLAN LED PD LOS NLOS ISI RMS FET FOV SV FWHM AR AWGN dB SNR NRZ OOK P/S PSD LCD SMPSs

Optical Wireless Communication Visible Light Communication Bit Error Rate Multiple-input Multiple-output Radio Frequency Free-space Optical Ultraviolet Communication Infrared Data Association Infrared Wireless Communication Light Fidelity Wireless Local Area Network Light Emitting Diode Photodetector Line-of-sight Non-Line-of-sight Intersymbol Interference Root Mean Square Field-effect Transistor Field of View Singular Value Full width at Half Maximum Active Area Additive White Gaussian Noise Decibel Signal-to-noise Ratio Non-return-to-zero On-off Keying Parallel to serial Power Spectral Density Liquid Crystal Display Switch-Mode Power Supplies

xxii

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

MATLAB AutoCAD CPC DTIRC L-R1 L-R1-R2 S/P DC ZF D/A IFFT FFT CP OFDM PLC DSO HPMV PDF P/S PSD

Matrix Laboratory Automatic Computer-aided Design Compound Parabolic Concentrator Dielectric Total Internal Reflecting Concentrator LOS plus first Reflection LOS plus first reflection along with second reflection Serial to Parallel Direct Current Zero-Forcing Digital to Analog Inverse Fast Fourier Transform Fast Fourier Transform Cyclic Prefix Orthogonal Frequency Modulation Power Line Communication Digital Storage Oscilloscope Hybrid Power Line Communication Multiple-Input Multiple-Output Visible Light Communication Probability density function Parallel to serial Power Spectral Density

Chapter 1

Introduction 1.1. Brief History of Communication System Light communication is not a new concept, but it is attributed to the ancient Greeks and Romans. They used polished shields to transmit signals using reflected sunlight during battles. Optical communication systems based on semaphore lines were developed in the 1790s. In 1792, Claude Chappe invented the first visual telegraphy system. In 1810, new inventions such as the heliograph, which used a pair of mirrors to send a controlled beam of light to a distant station, extended communication range. As a result of the demand for such instruments for military purposes in the late nineteenth and early twentieth centuries, Graham Bell invented the photophone in 1880, which transmitted speech over a distance of about 200 metres [1]. The Photophone transmitter and receiver are depicted in Figures 1 and 2, respectively [2]. The disadvantage of this device was that it did not work well in cloudy conditions. The German army modified this photophone in 1935 by using a tungsten filament lamp with an IR transmitting filter as a light source. In 1962, the scientists of MIT Lincoln lab used a light-emitting GaAs diode to establish an OWC link and transmit TV signals over a distance of 30 miles. F. R. Gfeller and G. Bapst demonstrated the technical feasibility of indoor optical wireless communication using infrared light emitting diodes (LEDs) in 1979 [3]. To achieve low data rate communication, a transceiver system based on fluorescent lamps was used [4]. The fast switching characteristic of visible light LEDs encouraged active researchers on high-speed VLC as the LED illumination industry advanced. Pang et al. [5] proposed the first optical signal transmitter concept using traffic light LEDs in 1999. S. Haruyama and M. Nakagawa later conducted a series of fundamental studies at Keio University in Japan [6-7]. They investigated the feasibility of providing illumination and communication for VLC systems at the same time using white LEDs. Meanwhile, the effects of light reflection and shadowing on system performance was examined [8-9]. Not only individual research groups, but large-scale organisations and research teams on a global scale have made significant contributions to the

2

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

development and standardisation of the VLC system. In 2008, the HOME Gigabit Access (OMEGA) project was launched in order to develop a novel wireless indoor network capable of providing gigabit data rates to users [10]. The University of Oxford, University of Cambridge, France Telecom, Siemens, and other companies and universities were project members who were finally able to demonstrate a real time VLC system with 100 Mbps data rate using sixteen white LEDs on the ceiling of the room. Visible Light Communication Consortium (VLCC), which was founded in Japan (2003), was renamed Visible Light Communication Association (VLCA) in 2014 to better collaborate with various industries in realising the visible light communication infrastructure and providing social infrastructure, Internet, computer, semiconductor, and so on.

Figure 1. Reflected sunlight used in photophone for transmitting signals [2].

The Ubiquitous Communication by Light Center (UC-Light), Center on Optical Wireless Applications (COWA), and Smart Lighting Engineering Research Center are prominent research groups in the VLC system in the United States (ERC). UC-mission Light’s is to develop new technological inventions, economic activities, and energy-saving benefits. COWA focuses on optical wireless communications, networking, imaging, positioning, and remote sensing applications. ERC focuses on LED communication systems and networks, supporting materials and lighting devices, and biotic and

Introduction

3

biomedical threat detection applications. The Chinese Visible Light Communications Alliance (CVLCA) was founded in 2014 to promote the mutual commercialization of VLC technologies, attracting the attention of various universities and industries in lighting, telecommunications, energy, user electronics, and funding agencies. In 2013, two large teams were formed in China to focus on OWC research across a broad spectrum, including VLC. One team was sponsored by China’s National Key Basic Research Program, and it included approximately 30 researchers from top universities and research institutes. Another team was supported by China’s National High Technology Research and Development Program. Both project teams have worked tirelessly to advance theory, develop technology, and demonstrate real-time VLC system demonstrations. They were able to provide multimedia services at a real-time data rate of 1.145 Gbps over a distance of 2.5 m, and a maximum off-line data rate of 50 Gbps was achieved over a shorter distance.

Figure 2. Photophone receiver [2].

1.2. Introduction to VLC In the last two years, there has been an unprecedented increase in the use of high-speed indoor data communication networks for various services such as work from home (WFH), online education, internet of things (IoT), cloud

4

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

storage, and so on, as well as ubiquitous internet connectivity, resulting in a massive increase in global data traffic. The total number of internet users is forecast to exceed 5 billion by 2023, accounting for approximately 66% of the global population. Machine-to-Machine (M2M) connections will be the fastest-growing device and connection category, increasing nearly 2.4-fold (19% compound annual growth rate-[CAGR]) to 14.7 billion connections by 2023 [11]. As a result, the M2M connection will account for half or half of the total device connections. Smartphones will grow at the second-fastest rate, at a CAGR of 7%. (increasing by a factor of 1.4). Connected TVs (flat-panel TVs, set-top boxes, digital media adapters [DMAs], Blu-ray disc players, and gaming consoles) will grow the fastest (at slightly less than a 6% CAGR) to 3.2 billion by 2023. PCs will continue to fall (by 2.3 percent) over the forecast period. Throughout the forecast period, however, PCs will outnumber tablets, and by the end of 2023, PCs will outnumber tablets (1.2 billion PCs vs. 840 million tablets). Figure 3 shows the expected device distribution by 2023.

Figure 3. Expected device share by 2023.

Mobile devices are transitioning from second-generation (2G) network connectivity to higher-generation connectivity (3G, 3.5G, 4G or LTE and now also 5G). Combining device capabilities with faster, higher bandwidth, and more intelligent networks will enable widespread testing and adoption of advanced multimedia applications, which will contribute to increased mobile and Wi-Fi traffic. To sustain a maturing mobile industry, there is a need for optimised bandwidth management and new network monetization models due

Introduction

5

to the explosion of mobile applications and the expanded reach of mobile connectivity to a growing number of end users. We have seen an increase in global 4G deployments as well as early-stage 5G implementations in a highly competitive mobile market. By 2023, 4G connections will account for 46% of total mobile connections. As mobile IoT grows, 5G connectivity is emerging as a strong contender for mobile connectivity, 11% of devices and connections will be 5G capable by 2023.

Figure 4. The various options available to overcome the overloaded RF spectrum.

Copper and coaxial cables, wireless internet access, broadband RF/microwave, and optical fibre are currently used in access networks. These technologies, particularly copper and coaxial cables and broadband RF/microwave, have limitations such as a crowded spectrum, and a lower data rate. Thus, the more pressing question is whether the telecommunications infrastructure can keep up with the pace and service providers can increase transmission capacity if all of the channels are fully occupied by the signal spectrum. These technologies, particularly copper and coaxial cables and broadband RF/microwave, have limitations such as a crowded spectrum, a

6

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

lower data rate, costly licencing, security concerns, and a high cost of installation and accessibility to all. Some other alternative options are shown in Figure 4. The option of acquiring new spectrum is not ideal among the options highlighted in the figure because it is expensive. Increasing cell density by reducing cell size (beyond the limit) is also not a viable option. Though wireless technology’s spectral efficiency has improved in recent years, its progress has been limited to 20% per year. In this context, visible light communication (VLC), which uses visible spectrum to communicate data, appears to be an appealing alternative for RF networks. According to one survey, most people spend 80 percent of their time indoors, and thus VLC, with its dual advantage of illumination and communication, becomes a more promising complementary to the overburdened RF system [12]. The VLC contributes to the green aspect of next-generation 5G/6G wireless networks with energy-efficient data transmission due to the vast unlicensed spectrum, robust security, low cost, low latency, immunity to electromagnetic interference (EMI), and so on. In addition to providing low-cost illumination, LED-based VLC systems have recently been used to provide extended radio services and medical applications. The low 3-dB bandwidth (~20 MHz) of LEDs used as transmitters limits the achievable system capacity in an optical wireless system. Modern communication has seen exponential growth in application and research over the last few decades. Wireless communication based on radio frequency (RF) has undoubtedly met some of the demand. However, with the advent of broadband access, the demand for data-intensive applications in the home has skyrocketed. The number of devices connected to the IP network is expected to exceed three times the global population by 2022. Aside from spectrum deficiency, there are other issues such as interference, security (because RF can penetrate walls), human health risk (transmission power exceeding a certain limit), power inefficiency, and so on. To overcome the limitations of the RF communication system, new communication technologies must be developed. Several technology candidates have entered the race to provide users with high-speed wireless communication systems, including OW systems for indoor and outdoor use [13-14], ultra-wideband wireless system (UWB), 60 GHz band for local wireless multimedia access, and terahertz (THz), which has been studied as a new alternative in this race. The invention of the laser in 1960 catapulted modern optical wireless communication (OWC) into the spotlight. However, due to the high cost of

Introduction

7

optical transmitter and receiver components, this technology was not implemented until the 1990s. Outdoor systems such as free-space optical (FSO), ultraviolet communication (UVC), and underwater communications (uFSO) are examples of OWC systems, while indoor systems include infrared data association (IrDA), infrared optical wireless communication (IROW), and visible light communication (VLC) systems. The difference between LIFi and Wi-Fi is shown in Table 1. Table 1. Difference between Li-Fi and Wi-Fi Features Medium for data transmission Interference

Li-Fi The LED transmitter uses light as a communication medium.

Wi-Fi Wi-Fi routers use radio waves to transmit data.

No issue of interference

Technology

IrDA compliant devices

Applications

Airlines, undersea explorations, hospitals, internet in office, mall, etc. Can pass through salty seawater and is also useful in densely populated areas. Because they cannot penetrate walls, they are more secure. Low

Suffers from interference due to nearby routers WLAN 802.11a/b/g/n/ac/ad standard compliant devices Internet browsing using Wi-Fi hotspot

Advantages

Privacy Environment impact Efficiency Data rate Frequency of operation Coverage distance System components

LEDs use less power and thus are more efficient. ~ 1 Gbps Ten thousand mores frequency spectrum as compared to RF ~ 10m LED, LED driver, Photodetector

Cannot pass through seawater and is useless in dense areas. Can penetrate walls, making it less secure Medium Base stations consume more power and thus have lower efficiency. WLAN-11n offers 150 Mbps, ~ 1-2 Gbps is achievable using Wi-Gig/Giga-IR 2.4 GHz, 4.9 GHz, and 5 GHz ~32 meters (WLAN 802.11b/11g), vary based on transmit power and antenna type Installation of routers is required, subscriber devices (laptops, PDAs, desktops) are referred as stations

Visible light communication is a method of transmitting signals in the form of light rays with wavelengths ranging from 380 nm to 780 nm and frequencies ranging from 430 THz to 780 THz, as illustrated in Figure 5. Because of the large bandwidth, the constraints of the low bandwidth problem in RF are thus resolved in VLC (350 THz).

8

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 5. Electromagnetic spectrum.

Because light cannot pass through walls, it is far more secure than RF. The VLC receiver only receives signals if it is in the same room as the transmitter, receivers outside the VLC source’s room will not be able to receive the signals, and thus it is immune to security issues that occur in RF communication systems. Furthermore, visible light source can be used for both illumination and communication. Moreover, the VLC system requires less power than RF. VLC is one of the promising candidates because of the additional benefit of a non-licensed channel.

1.3. Architecture of VLC Model The block diagram of the VLC system is shown in Figure 6. The major blocks are explained below:

Figure 6. Block diagram of VLC system.

Transmitter The LEDs and LASERs are used as transmitters in the VLC system. The LEDs are used when communication as well as illumination is required. The most common method for producing white light is trichromatic (red, green,

Introduction

9

blue), abbreviated as RGB. The RGB based LED provides high bandwidth and hence high data rate. However, the modulation complexity is relatively high, and some glitches, like how to control the three chips to avoid flickering and maintaining the mixed colour stability, need to be further studied. Another method is based on phosphor-based LEDs and its principle uses blue light to inspire the yellow phosphor to produce white light. This type of LED has a simple structure, lower cost, and a relatively lower modulation complexity, but the modulation bandwidth is very low, which makes the spectrum utilization rate also low. VLC systems based on two types of LEDs have their respective advantages. The system based on phosphor LEDs can be achieved easily and has a lower cost, while the other based on RGB-LEDs can achieve a higher transmission rate.

Channel In indoor VLC system, there are two types of paths through which light ray reaches the photodetector (PD) from the LED: line-of-sight (LOS) and nonline-of-sight (NLOS). Light ray arriving at PD through LOS path has good strength since there is no obstruction, whereas, in case of NLOS path, the light ray suffers multiple reflections through different reflecting surfaces such as walls, floor, ceiling, table etc. and hence its strength is lesser than LOS signal. Inter-symbol interference (ISI) is caused due to multiple reflections for NLOS signal, which, becomes a hindrance for high data rate. In many practical scenarios, an unobstructed optical path may not be available due to various room configuration, e.g., transmitter may not be placed at sufficient height or because of mobility of receiver as per requirements. The multipath dispersion for fixed transmitter and receiver is described by impulse response which can be considered as static for a specific room configuration.

Receiver The receiver of VLC system consists of an amplifying circuit, optical filter and optical concentrator. The optical concentrator is used capture light and focus it on the photodetector. Thus, optical concentrator compensates the attenuation caused due to beam divergence from LED source. The different types of photodetectors used in VLC system are silicon photodiode, PIN diode and avalanche photodiode. The avalanche photodiode has a higher gain than a

10

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

PIN photodiode but at the expense of high cost. Photo diode is generally used for stationary receiver. The optical filter is used to mitigate the ambient noise caused due interference from other sources such as sunlight and other unwanted illumination. The information is transmitted in the VLC system by modulating the instantaneous optical intensity in response to an input electrical signal. The information thus transmitted on the VLC channel is not contained in the amplitude, phase, or frequency of the transmitted optical waveform, but rather in the signal intensity. LEDs perform optical intensity modulation; however, modulation should be done in a way to avoid flickering. An LED driver is required to ensure that the LEDs work within their operating range. This driver provides a constant amount of power to the LEDs. On the transmitter side, a bias Tee is used to combine the DC driving current with the message signal. A photodiode on the receiver side converts the received light signals to electrical signals. This opto-electrical conversion is accomplished through direct detection of incident optical intensity.

1.4. Advantages of VLC system Because of various advantages associated with the VLC system, researchers around the world are devoting a great deal of enthusiasm to it and are making rapid advances in VLC technology. Some of the advantages are displayed in Figure 7 and are listed below: i.

Interference immunity: Because light does not cause electromagnetic interference, VLC is suitable for use in EMI environments such as hospitals, nuclear power plants, and airplanes. ii. Security: Because light cannot penetrate the walls, the VLC system ensures more secure communication. iii. Economical: Because VLC reuses the ubiquitous lighting infrastructure, only a few additional modules must be added to the lighting system. In addition, as the LED industry grows, the cost of VLC transceivers falls. iv. Energy-efficient: LEDs are green lighting devices that can reduce traditional lighting sources’ energy consumption by up to 80%. If all traditional lighting sources are replaced with LEDs, global electricity consumption is expected to fall by half. Laser diodes (LDs) have recently emerged as potential candidates for illumination, with the

Introduction

11

potential to provide higher data rates for VLC. However, because of the high coherence of the sources, laser-based lighting produces speckle patterns, which can be harmful to one’s health. When occupants are exposed to laser speckles for an extended period of time, they may experience eye irritation and headaches. Furthermore, LDs are currently more expensive than LEDs with comparable output power. Furthermore, LD has a small aperture and can only be used for point-to-point communication. v. Huge unlicensed bandwidth: Unlike RF, which has a spectrum limited to 3 KHz to 300 GHZ, VLC has a spectrum 1,000 times larger, ranging from 430 THz to 780 THz. vi. Health safety: Lighting LED is a diffusive light source, as opposed to infrared LED and laser, whose optical power is concentrated in a narrow beam. As a result, the high emitted optical power poses no health risk. Furthermore, because LED does not emit radiation like radio waves or microwave devices, it is not associated with health problems. vii. Ease of implementation: VLC modules are small and compact, making them simple to integrate into existing lighting infrastructure. LEDs can incorporate transmitter units such as modulation units, digital-to-analog converters, and driving circuits. Similarly, an external portable receiver unit comprised of photodiode, analog-todigital, and other signal processing units is available.

Figure 7. Advantages of VLC system.

12

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The aforementioned features contribute to a variety of indoor applications, the most desirable of which is high-speed internet access for smart phones and computers. People spend the majority of their time indoors rather than outdoors for study, work, entertainment, and so on, so access to the internet via LED lighting would be most convenient.

1.5. Applications of VLC System VLC has a wide range of applications (Figure 8) that could gradually penetrate different markets for various services, a few of which are described below: i. ii.

iii.

iv.

v.

Educational system: Li-Fi can be used in libraries, laboratories, auditorium halls, and other educational facilities. Medical application: Because mobile phones and Wi-Fi are not permitted in hospitals, particularly MRI scanners and operating rooms, VLC is appropriate for medical applications. Underwater application: While RF does not work underwater, visible light can be used for short-range communication. Radio frequencies are severely attenuated due to the conductive nature of water. When EM waves travel through a conductive medium, they are reflected. Any medium’s conductivity is determined by the number of ions present. When we consider water as a conductive medium of propagation, it will reflect EM waves. Though VLC is significantly attenuated by water, it still has a transmission distance of 10m as opposed to 1m when using an RF signal [15]. Furthermore, the bandwidth of a VLC signal is 1GHz as opposed to 1 MHz in an RF signal. Disaster management: VLC can be used for emergency communication during disasters such as earthquakes, hurricanes, or subway stations or tunnels, which are typically dead zones for most emergency communications. Traffic management: VLC is used in traffic management due to the presence of vehicle lights and existing traffic light infrastructure. Forward collision warning, pre-crash sensing, emergency electronic brake lights, lane change warning, stop sign movement assistant, traffic signal violation warning, and curve speed warning can all be used to accomplish this.

Introduction

13

vi. Mobile connectivity: By directing light from one device to another, a very high data link can be established. This method of mobile connectivity has a much higher data rate than Bluetooth or Wi-Fi.

Figure 8. Application of VLC system.

Other potential applications for VLC include Li-Fi, vehicle-to-vehicle communication, hospital robots, and information displayed on sign boards. The Li-Fi uses visible light to communicate and can provide internet speeds of up to 10Gbps. VLC can be used for cooperative forward collision warning, pre-crash sensing, emergency electronic brake lights, lane change warning, stop sign movement assistant, left turn assistant, traffic signal violation warning, and curve speed warning in vehicular communication. However, outdoor VLCs face challenges such as interference with ambient light sources, interference between VLC devices, and so on.

1.6. Challenging aspects of VLC system The following are some of the difficulties (Figure 9) associated with an indoor VLC system: i. The data rate is limited by the LED’s limited modulation bandwidth. ii. The LED’s non-linearity deteriorates the performance of the VLC system.

14

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

iii. VLC cannot be the sole source of information. iv. The multipath reflection signal causes intersymbol interference (ISI). v. Small field of view (FOV) of the receivers reduces the light collection efficiency.

Figure 9. Major challenges encountered in implementation of VLC system.

The limited modulation bandwidth of the LED (up to 20 MHz) can limit the bit rate. However, this limitation is compensated by using multiple-input multiple-output (MIMO) scheme along with spatial multiplexing technique. The multiple LEDs not only increases the bit rate, it also increases the uniform illumination inside the room. Another problem is non-linear voltage-current (V-I) characteristics of LED which becomes major stumbling block in implementing orthogonal frequency division multiplexing (OFDM) scheme in VLC system. The high peak-to-average power ratio (PAPR) causes non-linear distortion and necessitates increasing amplifier’s power for signal transmission. In a high PAPR Optical-OFDM system, in-band and out-of-band interferences are caused by the LEDs non-linearity. The non-linearity of the LEDs can’t be reduced but it is possible to reduce high PAPR by several strategies. Added problem is the implementation of VLC in isolated and rural areas where fibre optic networks are not covered. Also, it may not be practical to lay

Introduction

15

down cables in the last mile due to cost and time-consuming factors along with the concern of maintenance and breakage. So, there is a need of cost-effective scheme to counter this problem. Further, the multipath signal in the indoor VLC system leads to intersymbol interference (ISI). The multipath propagation depends on room size, order of reflection, reflection coefficient of the room. The ISI affects the quality of the transmission, reduces data rate and increases BER. Therefore, various parameters and factors that can mitigate this ISI becomes important case study. The spatial separation between individual LEDs and each PDs plays important role in mitigating the ISI. The limited FOV of the receiver is also a major problem in the performance of VLC system. To collect light from different types of layout of the LEDs, the FOV of the receiver needs to be large for an indoor MIMOVLC system. The size of PD can’t be increased as it reduces the receiver’s bandwidth and therefore an optical concentrator having large FOV, high gain, and small size is required.

Chapter 2

BER Analysis of a Multipath MIMO-VLC System with Various Source and Receiver Configurations 2.1. Introduction There are two types of paths through which light rays from LEDs reach the photodetector (PD) in an indoor VLC system: line-of-sight (LOS) and nonline-of-sight (NLOS). The light ray arriving at PD via LOS path has good strength because there is no obstruction, whereas the light ray arriving via NLOS path suffers multiple reflections through different reflecting surfaces such as walls, floor, ceiling, table, etc., and thus its strength is less than the LOS signal. As a result, most BER simulation research ignores the NLOS signal. However, inter-symbol interference (ISI) is caused by multiple reflections for NLOS signals, posing a challenge for high bit rate systems. Furthermore, in many practical scenarios, an unobstructed optical path may not be available due to various room configurations, such as the transmitter not being placed at a sufficient height or the receiver’s mobility as required. As a result, the NLOS signal must be considered in the VLC system. The impulse response, which can be considered static for a specific room configuration, describes the multipath dispersion for fixed transmitter and receiver. Several methods for approximating the impulse response of LOS and NLOS indoor-based VLC systems have been reported. Transmitters were assumed to cause homogeneous illumination over reflecting surfaces such as walls, ceilings, and floors in the method reported [3], so that reflected power received at the photodetector is obtained by integrating over the entire reflecting surfaces. Because the method is limited to the first reflection, it is only suitable for link budget analysis. Later, an algorithm for calculating the impulse response for multiple reflections was reported [16]. The method divides the reflecting surfaces into N number of numerous reflecting elements. These elements function as both elemental receivers and elemental sources. Summing the impulse responses from each of these reflecting elements yields

18

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

the total impulse response. Because this algorithm is recursive and time consuming for higher order reflection, the authors only simulated up to third order and compared the results to experimental results. The Monte Carlo ray tracing algorithm was introduced and is applicable to Lambertian and Non-Lamberttian sources [17]. The method is based on ray generation from a source, followed by determining whether the ray’s impact point is on the receiver or on a wall using a rotation matrix. If the ray is intercepted by the FOV of the receiver, the power reaching the photodetector is calculated. There is no limit to the number of reflections in the method, which leads to time waste in ray tracing because not all rays reach the photodetector. To address this issue, the modified Monte Carlo (MMC) algorithm was proposed, in which the power received at the photodetector from the impact point of each reflection is calculated. Though the method improves accuracy and reduces computational time, it has some flaws due to the random nature of ray direction. Gu et al. [18] investigated an indoor VLC positioning system using combined deterministic and modified Monte Carlo (CDMMC) for computing impulse response of the channel up to third order of reflection. Farshad Mirakirkhani et al. [19] calculated detected power and path length from source to detector to obtain channel impulse response (CIR) using Zemax ray tracing feature. The impulse response of the light components described above aids in estimating the SNR, which determines the BER. It has temporal dispersion as a result of inter-source interference caused by multipath reflections. The system’s BER performance varies significantly across receiver locations due to temporal distortion. Several researchers have looked into the BER performance analysis for indoor VLC systems. Azhar et al. [20] compared SISO-OFDM and MIMOOFDM transmission systems based on BER versus bit rate using only the LOS component. SISO systems outperform MIMO systems on a per channel basis due to less crosstalk in SISO systems. They demonstrated that changing the separation of LEDs from 15cm to 9.5cm increased coverage area from 25cm2 to 144cm2 while maintaining the target BER of 2×10-3. Andrew et al. [21] used the ZF, MMSE, and V-BLAST algorithms to simulate the BER performance of a 4×4 MIMO imaging system. An adaptive MIMO-OFDM VLC system with angular diversity receivers has been proposed by Yang Hong et al. [22]. They discovered an optimum polar angle for the 4×4 MIMO-OFDM VLC system by assuming five reflections for each ray in their simulation result. Asanka Nuwanpriya et al. [23] investigated the BER performance of a MIMOVLC system using an angle diversity receiver. Thilo Fath et al. [24] used repetition coding (RC), spatial multiplexing (SMP), and spatial modulation to

BER Analysis of a Multipath MIMO-VLC System …

19

simulate BER for MIMO with vertically fixed detectors separated by 0.2m, 0.4m, and 0.6m (SM). BER performance for the indoor MIMO-VLC system has been investigated in all of those works, either for LOS only or LOS with other components as a whole. However, a comparative analysis of the BER performance of LOS and LOS+NLOS signals with different spatial separation of transmitters and receivers is urgently required. BER performance has been simulated in this chapter using both LOS and LOS+NLOS signals with different spatial separation of transmitter and receiver. It has been demonstrated that not all of the reflecting elements of the walls contribute to the impulse response of the NLOS signal at the receiver. This is due to the receiver’s limited field of view. Two cases have been considered in the work. In case 1, the distance between adjacent receivers is set to 10cm in configuration-1 and 20cm in configuration-2, with the distance between adjacent LEDs fixed at 0.2m. In case 2, the distance between LEDs is set at 0.2m, 0.4m, and 0.6m, respectively, and the distance between adjacent receivers is set at 10cm.

2.2. LC Model for Indoors Figure 10 depicts a typical room model with a size of 5×5×3 m3. Four LEDs were used in our simulation, which is described further below. The receiver is positioned 0.85m above the floor. The arrival of the LOS, LOS plus first reflection (L-R1), and LOS with first and second reflection (L-R1-R2) signals at the photodetector are displayed in the figure. Path d1 is for the LOS channel, path d2-d3 is for the first reflection channel, and path d2-d4-d5 is for the second reflection channel (long dash + round dash). The first reflection is on wall-1, and the second on wall-2. Lambertian pattern radiant intensity is a cosine function with a semi-half power angle of ϕ1/2 𝐼(𝜙) =

𝑚+1 2𝜋

𝑐𝑜𝑠 𝑚 (𝜙)

(2.1)

where ϕ is the spherical polar angle w.r.t the normal axis, m is the order of Lambertian pattern, which is a function of ϕ1/2 given as 𝑚 = ln⁡(2)⁄ln⁡(𝑐𝑜𝑠∅1/2 ).

20

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 10. Indoor 4×4 geometrical model with multipath propagation.

The LOS signal arrives at the receiver R from the source S through path d1. The impulse response due to LOS signal is expressed as [25] 1 𝑑12

𝐴𝑅 𝐼(𝜃) cos(𝜑0 )

ℎ0 (𝑡; 𝑆, 𝑅) = cos(𝜓 )𝛿 (𝑡 − 𝑑1 ) , 0 < 𝜓 < 𝐹𝑂𝑉 0 0

(2.2)

𝑐

0,

{

𝜓0 ≥ 𝐹𝑂𝑉

where 𝜙0 and Ѱ0 are the irradiance angle of source and incidence angle at receiver respectively, d1 is the distance between the source and the receiver, Ar is the active area of the photodetector, c is the speed of the light and 𝛿(t) is the Dirac function. Assuming Lambertian profile of the reflected radiation, the impulse responses for the first reflection and the second reflection signals can be written as 1 𝜋𝑑22 𝑑32

ℎ(1) (𝑡; 𝑆, 𝑅) =

𝐴𝜀𝑙 :1 𝜌1 𝐴𝑅 𝐼(𝜃)

cos(𝜑1) cos(𝛼) cos(𝛽) cos(𝜓1 )𝛿 (𝑡 − {

0,

𝑑2 +𝑑3 𝑐

),

(2.3) 0≤

𝜓1 ≤ 𝐹𝑂𝑉

𝜓1 ≥ 𝐹𝑂𝑉

BER Analysis of a Multipath MIMO-VLC System …

21

ℎ(2) (𝑡; , 𝜀𝑙:1 , 𝜀𝑙:2 , 𝑆, 𝑅) = 1

𝜋𝑑22 𝑑42 𝑑52

𝐴𝜀𝑙 :1 𝐴𝜀𝑙 :2 𝜌1 𝜌2 𝐴𝑅

𝐼(𝜃)cos(𝜑1 ) cos(𝛼) cos(𝜔) cos(𝛾) cos(𝜓2)𝛿 (𝑡 − {

𝑑2 +𝑑4 +𝑑5 𝑐

(2.4)

) , 0 ≤ 𝜓2 ≤ 𝐹𝑂𝑉 𝜓2 ≥ 𝐹𝑂𝑉

0,

where 𝜀𝑙:1 and 𝜀𝑙:2 are the lth grid on wall 1 and wall 2, 𝐴𝜀𝑙:1 and 𝐴𝜀𝑙 :2 are the areas of grid on wall 1 and wall 2, ρ1 and ρ2 are the reflection coefficients of wall 1 and wall 2, 𝜑1 is the irradiance angle of LED, 𝛼 and 𝛾 are the angles of irradiance on wall 1 and wall 2. 𝛽, 𝜔 and 𝛿 are the angles of irradiance from the grid of wall 1 and wall 2 respectively, d2 is the distance from source to lth element of wall 1, d4 is the distance between lth elements of wall 1 and wall 2, d5 is the distance between lth element of wall 2 and receiver, Ѱ1 and Ѱ2 are the angles of irradiance of first and second reflection signal respectively. Receiver’s noise consists of shot noise and thermal noise, which is written as 2 2 2 𝜎𝑛𝑜𝑖𝑠𝑒 = 𝜎𝑠ℎ𝑜𝑡 + 𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙

= (2𝑞𝑅𝑃𝑟 𝐵) + (

8𝜋𝐾𝑇 𝐺

𝜂𝐴𝑟 𝐼2 𝐵2 +

16𝜋2 𝐾𝑇𝛤 𝑔𝑚

𝜂2 𝐴2𝑟 𝐼3 𝐵3 )

(2.5)

where q represents electronic charge, R-responsitivity of the receiver, Bequivalent noise bandwidth, K-the Boltzmann constant, T -absolute temperature, G-open loop voltage gain, η-input capacitance of PD, Γ-FET channel noise factor, and gm-FET transconductance. The time-dispersive properties of multipath channel depend on the root mean square delay spread Drms which is defined as the second central moment of the magnitude squared of channel impulse response (h)

Drms =

1⁄ ∫(t-μ)2 h2 (t)dt 2 [ h2 (t)dt ] ∫

(2.6)

where 𝜇 is the mean delay given by

𝜇=

𝑡ℎ2 (𝑡)𝑑𝑡 ℎ2 (𝑡)𝑑𝑡

(2.7)

22

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The maximum bit rate is decided by the RMS delay of the light, if channel equalizer is not used. The relation between maximum bit rate and RMS delay 1

is 𝑅𝑏 ≤ 10×𝐷

𝑟𝑚𝑠

.

2.3. Results and Discussion The reflected light from the wall surfaces will only reach the receiver from grids that are above the receiver plane because the receiver plane is kept at a height of 0.85m in our experiment. The first reflection light rays depicted by the long-dashed arrow line are caught within the FOV of the receiver located in the middle of the room, which would further minimize this limitation for the reasons given in Figure 11. In our simulation, the receiver’s half-angle FOV has been considered as 600. The length BO in ∆AOB of Figure 11 is 2.5m and AOB = 300, so the length AB(Hs) is 1.44m. Light rays reflected from the wall surface beneath point A are not captured within the FOV of the receiver. Only light rays striking all four wall surfaces above point A contribute to the channel gain corresponding to the first reflection as well as the higher order reflection signal. The receiver’s FOV is critical in higher order multipath reflection for indoor VLC systems. The simulation in this chapter considers the aforementioned logic.

Figure 11. First reflection light rays captured within the receiver’s field of view.

BER Analysis of a Multipath MIMO-VLC System …

23

2.3.1. Case 1: The Separation between PDs Is Altered Using Fixed LEDs In our work, we investigate BER performance in a typical 5×5×3 m3 room using four LEDs mounted on the ceiling (downward vertically) and four PDs kept at a height of 0.85m above the floor, as shown in Figure 12. The coordinate of the ceiling’s centre is (0, 0, 3). The four LEDs and four PDs are arranged in four quadrants. Figure 13 depicts the flowchart for determining the impulse response, which has been used up to the second reflection. In this case, ST is fixed at 0.2m and SR has been increased from 10cm (configuration1) to 20cm (configuration-2).

Figure 12. Case 1 and Case 2 room views.

Figure 14 shows BER performance using LOS channel, L-R1 channel and L-R1-R2 channel for configuarion-1 in which the coordinates of four PDs are (0.05, 0.0.5, 0.85), (-0.05, 0.05, 0.85), (-0.05, -0.05, 0.85) and (0.05, -0.05, 0.85). From Figure 14, it may be seen that there is a large difference in the BER performance of L-R1-R2 signal with that of LOS and L-R1 signal. This is due to ill-conditioned channel matrix obtained after the second reflection. It is also evident from the high condition number which is defined as the ratio of the largest singular value of channel matrix to the smallest. The channel matrix, singular value vector (obtained by singular value decomposition) and condition number for LOS, first reflection and second reflection, are given below: 0.3553 0.3381 H_LOS=[ 0.3222 0.3381

0.3381 0.3553 0.3381 0.3222

0.3222 0.3381 0.3553 0.3381

0.3381 0.3222 ] × 10−5 0.3381 0.3553

(2.8)

24

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 13. Flow chart for determining impulse response.

BER Analysis of a Multipath MIMO-VLC System …

25

Figure 14. BER curves for configuration-1.

SV =[0.1354 0.0033 0.0033 0.0001] × 10−4.

(2.9)

Condition number of H_LOS = 1.0623×103.

(2.10)

0.4904 H-LR1=[0.4683 0.4478 0.4683

(2.11)

0.4683 0.4904 0.4683 0.4478

0.4478 0.4683 0.4904⁡ 0.4683

0.4683 0.4478 ] × 10−5 0.4683 0.4904

SV=[0.1875⁡ 0.0043 0.0043 0.0002] × 10−4.

(2.12)

Condition number of H_LR1 = 1.1678×103.

(2.13)

0.0939 H_LR1R2=[0.0882 0.0900 0.0956

(2.14)

0.2097 0.1989 0.2105 0.2215

0.0671 0.0677 0.0722 0.0715

0.0824 0.0776 ] × 10−4 0.0828 0.0878

SV=[0.5074⁡ 0.0047 0.0029 0.0000] × 10−4.

(2.15)

Condition number of H_LR1R2 = 1.3567×105.

(2.16)

where H_LOS, H_LR1, and H_LR1R2 are the channel matrices for the LOS, LOS+first reflection, and LOS+first reflection+second reflection signals, and SV is the singular value vector. It is clear that the channel correlation for the

26

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

H_LR1R2 matrix is high. Because the first element singular value of the H_LR1R2 matrix is very high in comparison to the other three, only one data stream may be obtained at the receiver. As a result, the BER performance of the L-R1-R2 signal is very poor.

Figure 15. BER curves for configuration-2.

Figure 15 depicts BER performance for configuration-2 with receiver coordinates of (0.1, 0.1, 0.85), (-0.1, 0.1, 0.85), (-0.1, -0.1, 0.85) and (0.1, 0.1, 0.85). As a result, the distance between adjacent receivers is 20cm. The results show that the BER of the LOS and L-R1 signals is lower in this configuration than in configuration-1, with a gain margin of 10dB and 12dB in SNR, respectively. However, the BER of the L-R1-R2 signal has been improved by a gain margin of 30dB when compared to the BER of the L-R1R2 signal in configuration-1. Thus, by increasing the distance between receivers by just 10cm, there is a significant difference in BER performance improvement for the L-R1-R2 signal as compared to improvement in BER performance for the LOS and L-R1 signal. Similarly, for configuration-2, the simulated channel matrix, singular vector and condition number, are given below:

0.3727 H_LOS=[0.3371 0.3066 0.3371

0.3371 0.3727 0.3371 0.3066

0.3066 0.3371 0.3727 0.3371

0.3371 0.3066 ] × 10−5 0.3371 0.3727

(2.17)

BER Analysis of a Multipath MIMO-VLC System …

27

Table 2. MIMO-VLC parameters and their values

Room

Source

Parameters Size ρNorth = ρSouth = ρWest = ρEast Number of grids on the ceiling Number of grids on floor Number of grids on wall Coordinates Configuration-1 (4 LEDs) Configuration-2

Channel model Receiver

Semi-angle at half power (FWHM) Barry model Height of receiver plane above the floor Active area (AR) Coordinates (4 PDs) Coordinates (4 PDs) Coordinates (4 PDs)

Noise

Half-angle FOV Δt Ambient light noise Spectral irradiance of ambient light noise Pbn Receiver preamplifier noise Electronic charge q Noise bandwidth factors I2, I3 Receiver area Ar Responsitivity of Photodiode R Boltzman’s constant K Absolute temperature T FET channel noise factor Γ Feedback resistor of FET based transimpedance amplifier Rf Transconductance gm Total Input capacitance η

value 5 × 5 × 3 m3 0.8 35×35 35×35 35×15 (0.05, 0.05, 3), (-0.05, 0.05, 3), (-0.05, -0.05, 3), (0.05, -0.05, 3) (0.1, 0.1, 3), (-0.1, 0.1, 3), (-0.1, -0.1, 3), (0.1, -0.1, 3), 70

SR=10 cm SR=20 cm

0.85 m 1Cm2 (0.1, 0.1, 3),(-0.1, 0.1, 3), (-0.1, -0.1, 3), (0.1, -0.1, 3) (0.2, 0.2, 3), (-0.2, 0.2, 3), (-0.2, -0.2, 3), (0.2, -0.2, 3) (0.3, 0.3, 3), (-0.3, 0.3, 3), (-0.3, -0.3, 3), (0.3, -0.3, 3) 60 0.5 ns

ST=0.2 m ST=0.4 m ST=0.6 m

5.8uw/cm2.nm

1.6×10-19 0.562, 0.0868 10-6 m2 0.5 A/W 1.38×10-23 J/k 300 K 0.82 1.4 KΩ 14 mS 2 pF

SV=[0.1353 0.0066 0.0066 0.0005] × 10−4

(2.18)

Condition number of H_LOS=266.2756.

(2.19)

28

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

0.5130 0.4672 H_LR1=[ 0.4278 0.4672

0.4672 0.5130 0.4672 0.4278

0.4278 0.4672 0.5130 0.4672

0.4672 0.4278 ] × 10−5 0.4672 0.5130

(2.20)

SV=[0.1875 0.0085 0.0085 0.0006] × 10−4

(2.21)

Condition number of H_LR1=292.6701.

(2.22)

0.0972 0.0863 H_LR1R2=[ 0.0893 0.0995

0.2134 0.1946 0.2144 0.2336

0.0654 0.0667 0.0754 0.0734

0.0830 0.0742 ] × 10−4 0.0838 0.0932

(2.23)

SV=[0.5162 0.0092 0.0053 0.0001] × 10−4

(2.24)

Condition number of H_LR1R2=4.7895×103.

(2.25)

When compared to configuration-1, the condition number of the H_LR1R2 matrix has been reduced by 28.32 times, while the condition number of the H_LR1 matrix and the LOS matrix has been reduced by 3.99 and 3.98 times, respectively. This large difference in H_LR1R2 matrix condition number demonstrates the previously mentioned effect of increased separation on BER performance. Table 2 display the simulation parameters and their corresponding values.

2.3.2. Case 2: The Separation between LEDs Is Varied While the PDs Remain Constant 2.3.2.1. Effect of LED Separation on the Impulse Response Because the light rays from the four different LEDs arrive at the receiver via different LOS paths, there will be time differences in signal arrival. As a result, the impulse responses from the four LEDs are separated in time and have different peaks. Figures 16 and 17 show the impulse responses for the LOS channel at the receiver (Rx3) caused by transmitters (LEDs) Tx1, Tx2, Tx3, and Tx4 for spatial separation of 0.2m and 0.4m respectively. It can be seen that the impulse responses caused by Tx2 and Tx4 overlap, whereas the rest LEDs separate them.

BER Analysis of a Multipath MIMO-VLC System …

29

Figure 16. LOS channel impulse response (IR) with ST = 0.2m at Rx3 due to all LEDs.

Figure 17. LOS channel impulse response (IR) with ST = 0.4m at Rx3 due to all LEDs.

The overlapping occurs due to the symmetrical position of Tx2 and Tx4 with respect to Rx3. Also, as the distance between adjacent LEDs is increased from 0.2m to 0.4m, the impulse responses from all LEDs are separated, with the exception of Tx2 and Tx4. In other words, the time difference between LEDs grows as they are separated. In a MIMO system, separating the impulse response reduces inter-source interference. It can also be seen that the magnitude difference between impulse responses obtained with different LEDs increases with the distance between adjacent LEDs. The distinction

30

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

reduces channel correlation, which aids in signal separation at the receiver. This means that increasing the spatial separation between adjacent LEDs improves the impulse response of the MIMO-VLC LOS system.

2.3.2.2. Effect of LED Separation on RMS Delay Spread The root mean square delay is commonly used to characterise the timedispersive properties of a multipath channel. Intersymbol interference (ISI) will degrade the performance of an indoor MIMO-VLC system if the RMS delay of the received signal exceeds the symbol duration and no equaliser is used (ISI). As a result, a simulation of a 4×4 MIMO system was performed to determine the RMS delay distribution in the room using impulse response. Figures 18 and 19 show the RMS delay spread at each receiver location using 0.4m and 0.6m separation for the L-R1 signal, respectively. The shortest RMS delay is observed when the distance between the source and the receiver is the shortest, i.e., in the centre of the room. The results show that the RMS delay at ST of 0.4m varies between 0.1542ns and 0.7854ns inside the room, whereas at ST of 0.6m, the RMS delay varies between 0.1724ns and 1.049ns, indicating that the RMS delay spread has increased as the separation between LEDs has increased. Though this reduces the coherence bandwidth of the channel, ISI will occur if and only if the signal bandwidth exceeds the coherence bandwidth of the channel. Though this reduces the coherence bandwidth of the channel, ISI will occur if and only if the signal bandwidth exceeds the coherence bandwidth of the channel.

Figure 18. The RMS delay spread for ST = 0.4m is Max = 0.7854ns and Min = 0.1542ns.

BER Analysis of a Multipath MIMO-VLC System …

31

Figure 19. The RMS delay spread for ST = 0.6m is Max = 0.049ns and Min = 0.1724ns.

2.3.2.3. Effect of LED Separation on BER Performance To demonstrate the system’s performance, BER simulation was performed for the above 4×4 MIMO system at various LED separations. The error ratio was calculated by comparing the transmitted energy to the power spectral density of AWGN. When the LOS channel gain is in the 10-5 range, the electrical path loss at the receiver is around -100dB. As a result, the BER curve in Figure 14 has a 100dB offset with respect to the received Es/No. The simulation employs a spatial multiplexing transmission scheme in which each of the four LEDs transmits a different stream of data. The results show that as the distance between adjacent LEDs increases, the bit error rate for LOS, L-R1, and L-R1R2 channels decreases. The difference in BER performance between LOS and L-R1 signal decreases as the separation increases. For example, at 0.6m separation, for BER of 10-5, the required SNR is 172dB using LOS channel and 196dB using L-R1 channel, whereas at 0.2m separation, the same BER requires 190dB for LOS and 216dB for L-R1 channel. Thus, at 0.6m separation, the difference in required SNR between the LOS and L-R1 channels is only 14dB, whereas at 0.2m separation, the difference is 26dB. Because commercial LEDs have limited power, the SNR at the receiver cannot be increased beyond a certain point. It indicates that when using a MIMO system, spatial separation of LEDs is beneficial in situations where signal transmission relies on a diffused channel in addition to a LOS channel. However, too much separation can reduce coherent bandwidth, resulting in signal distortion and ISI. As a result, optimal LED separation should be chosen.

32

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 20. BER performance for LOS, L-R1, and L-R1-R2 channels at various LED positions.

2.3.2.4. Effect of LED Separation on the Bit Rate Figures 21 and 22 show the bit rate variation at each receiver location within the room, with ST of 0.4m and 0.6m for the L-R1 signal, respectively. The highest bit rate is available when the receiver is directly beneath the source. When Figures 21 and 22 are compared, it can be seen that the maximum bit rate has decreased from 64.85Mbps to 58Mbps and the minimum bit rate has decreased from 12.76Mbps to 9.53Mbps as the separation between MIMO LEDs has increased. The result is consistent with the BER performance discussed in section 2.2.2.3, namely that BER performance improves at the expense of reduced bit rate.

Figure 21. Bit rate for ST = 0.4m. Max. = 64.85Mbps & Min. = 12.76Mbps.

BER Analysis of a Multipath MIMO-VLC System …

33

Figure 22. Bit rate for ST = 0.6m. Max. = 58Mbps & Min. = 9.53Mbps.

2.4. Conclusion The current chapter investigates the BER performance of the LOS, L-R1, and L-R1-R2 signals in a 4×4 MIMO-based indoor VLC system. When the distance between receivers is increased from 10cm to 20cm, there is a gain margin of 30dB, 12dB, and 10dB in SNR for L-R1-R2, L-R1, and LOS signals, respectively, in BER performance. Higher order reflection exhibits more BER performance changes than lower order reflected signal and LOS signal. As a result, not only the LOS signal but also the higher order reflection signal should be considered, because they have a significant impact on the BER performance of the indoor-based MIMO-VLC system. Since the impulse responses using individual transmitters at any one receiver of a MIMO system become separated, the separation between transmitters (LEDs) improves BER performance. The magnitude difference between the impulse responses of four LEDs has been increased, reducing channel correlation of the channel matrix. The effect of transmitter spatial separation on bit rate and RMS delay spread inside the room has also been studied, and it has been discovered that an optimal separation should be chosen for the MIMO-VLC system to prevent ISI, which may occur due to a reduction in coherent bandwidth of the channel. However, in this chapter, the BER analysis is limited to the Lambertian radiation pattern of the LED. Commercially available LEDs, on the other hand, may not have a Lambertian radiation pattern. In Chapter 3, we included other types of LED radiation patterns and demonstrated how BER varies with receiver position as well as LED radiation pattern.

Chapter 3

BER Performance Comparison in a Multipath MIMO-VLC System with Different LED Radiation Patterns 3.1. Introduction Commercially available high-power white-light LEDs can meet the requirements for proper illumination and high-speed data transmission, which can be increased further by using the MIMO technique with multiple LEDs [26-28]. The most important feature of an LED is its radiation pattern, which can be symmetrical or asymmetrical, such as Lambertian, Batwing, elliptical, and so on. The Lambertian pattern has been used in the majority of VLC system research. However, as previously mentioned, commercially available LEDs may have a different radiation pattern. The intensity of light received at the photodetector (PD) is primarily determined by three factors: transmitter/receiver position, LED radiation patterns, and LOS/NLOS (lineof-sight) signals. Because BER performance is determined by signal-to-noise ratio (SNR), which is governed by the intensity of light at the receiver, it is necessary to study BER performance analysis taking the above three factors into account. Several researchers in the indoor MIMO-VLC system have done extensive work on the BER performance analysis. A minimum bit error rate (MBER) combiner has been proposed for the MIMO-VLC system to combat impairments such as user mobility and imperfect channel estimation information (CSI) [29]. BER simulation has revealed that the proposed combiner provides nearly equivalent BER performance to the maximum ratio combining (MRC) method with perfect CSI. To reduce the condition number of the overall channel matrix, an adaptive precoder with minimum symbol error rate (AP-MSER) based on singular value decomposition has been proposed for massive MIMO-VLC channels [30]. For 16×16 and 36×36 massive MIMO VLC channels, the BER performance of the AP-MSER technique was found to be better than that of existing precoding methods such as Jacobi preconditioner (JP) and Tikhonov regularisation (TR). It was

36

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

discovered that channel capacity and BER performance using Pyramid receiver (PR) and hemispheric receiver (HR) outperform spatially separated receivers, whereas proposed PR outperforms line-blocked receivers when the LED separation is large. Power imbalances of 1dB and 3dB improve BER for SM and SMP, respectively, but have no effect on BER when using repetition coding (RC). A prism array receiver with multiple receiving elements (REs) has been investigated to improve BER performance [31]. The full rank matrix is achievable at every receiver position for demultiplexing of MIMO-VLC signals, according to simulation results, as long as the number of REs is greater than the number of LEDs. Because the problem of LOS link interruption causes channel degradation, a software defined adaptive MIMO VLC has been proposed in which both modulation and MIMO techniques are dynamically adapted to changing channel conditions [32]. An obstacle with a diameter of 4.5cm was placed in the middle of the transmission line with a dynamic obstruction range of 135 cm in the experiment, and it was discovered that adaptive MIMO modes switch to spatial multiplexing SM-256, SM-64, and spatial diversity SD-256 adaptively with BER less than 10-3. An orthogonal circulant matrix transform (OCT) precoding and singular value decomposition (SVD)-based loading methods have been proposed to improve BER performance for a 2×2 MIMO-OFDM VLC system [33]. At a data rate of 1.5 Gbps, the BER is reduced from 1.7×10-2 to 4.1×10-3 and 4.7×10-4 using the two optimization schemes described above. However, in the afore mentioned research works, BER performance analysis was performed using only LOS signals, rather than NLOS signals. LOS and NLOS signals have been considered in some research works for investigating BER performance in indoor MIMO-VLC systems. N. A. Mohammed et al. [34] studied the effect of subcarrier modulation factor (SCMF), wall reflection coefficient (ρ) and number of time slots (L) on BER performance using LOS and NLOS signals. They demonstrated that increasing SCMF by 25% (i.e., to achieve 25 mW/LED) under the same conditions (L = 4, ρ = 0.8, and 1-2 Mbps) results in a remarkable BER performance of 10-9. Because of the well-conditioned channel matrix, aperture-based receivers outperform conventional non-imaging receivers with LOS and LOS plus first reflection signal [35]. The BER performance of a 4×4 MIMO-OFDM VLC system using LOS and diffused signals was used to compare rectangular and linear LED arrangements [36]. The rectangular LEDs arrangement outperforms the linear arrangement, and a data rate of 1.2 Gbps was achieved with the former configuration at a BER of 3.8×10-3. It has been observed that as receiver separation increases, higher order reflection signals exhibit more

BER Performance Comparison in a Multipath MIMO-VLC System …

37

changes in BER performance than LOS and low order reflection signals. The BER analysis, however, is based on two considerations in all of these works: First, the LEDs are mounted on the ceiling; second, the LEDs’ radiation pattern is Lambertian. These two conditions may put traditional models to the test. This is due to the fact that not all commercially available LEDs are Lambertian [37-41]. A self-powered VLC receiver that consists of a sphericalsolar-cell module and an earphone has been developed [42]. The application of such system is useful in museum, amusement parks etc. where LEDs are placed in the show window. Hence the ceiling-mounted LEDs may not fit in such types of application. Given these constraints, this chapter proposes a generalised indoor VLC model for BER analysis. Three considerations are provided for the generalised indoor MIMO-VLC model: First, the radiation pattern of the LED source may differ; second, both ceiling-mounted and wall-mounted LED scenarios are included; and third, LOS and LOS plus first reflection signals (L-R1) have been used to simulate BER. Furthermore, the channel gain distribution for LOS and first reflection signals for ceiling-mounted and wall-mounted LEDs is shown.

3.2. Indoor Geometrical Multipath Model Figure 23 depicts a multipath indoor geometrical model of a 5m×5m×3m room. Path r1 in the diagram corresponds to the LOS channel, while paths r2r3 correspond to the first reflection channel. The LOS signal’s impulse response is given as. 1 𝑟12

𝐴𝑟 𝐼(𝜑0 ) cos(𝜑0 )

0 ≤ Ѱ0 ≤ 𝐹𝑂𝑉

ℎ(0) (𝑡; 𝑆, 𝑅) = cos(Ѱ ) 𝛿 (𝑡 − 𝑟1 ) , 0

(3.1)

𝑐

{

0,

Ѱ0 ≥ 𝐹𝑂𝑉

where 𝐼(𝜑0 ) is the radiant intensity of the LED in the direction of the emission angle 𝜑0 w.r.t normal axis,Ѱ0 is the incidence angle at the receiver, r1 is the distance between source and receiver, Ar is the active area of the photodetector, c is the speed of light and 𝛿(t) is the Dirac delta function.

38

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 23. Indoor multipath MIMO-VLC geometrical model.

The radiation pattern of a reflected signal is determined by the properties of the reflection surface’s bidirectional reflectance distribution function (BDRF), as well as the geometrical position of the LED source and reflection surfaces. The LED radiation pattern also has an effect. For the purposes of analysis, we assume that the reflected radiation has a Lambertian pattern, which is true in the majority of cases. As a result, the impulse response to the first reflection signal is written as 1 𝜋𝑟22 𝑟32

ℎ(1) (𝑡; 𝑆, 𝜀𝑙 , 𝑅) =

𝐴𝜀𝑙 𝜌𝐴𝑟 𝐼(𝜑1 )

𝑐𝑜𝑠(𝜑1)𝑐𝑜𝑠(𝛼)𝑐𝑜𝑠(𝛽) 𝑐𝑜𝑠(Ѱ1 )𝛿 (𝑡 − {

0,

𝑟2 +𝑟3 𝑐

)

(3.2) 0 ≤ Ѱ1 ≤ 𝐹𝑂𝑉 Ѱ1 ≥ 𝐹𝑂𝑉

where 𝜀𝑙 is the lth grid of wall whose area is 𝐴𝜀𝑙 , r2 is the distance from source to lth grid of wall, 𝜑1 is the radiation angle of LED, Ѱ1 is the incidence angle at the receiver for first reflection signal, 𝜌 is the reflection coefficient of wall, 𝛼 is the angle of incidence at the wall, 𝛽 is the irradiance angle at the lth grid of wall w.r.t normal axis. The NR×NT channel matrix H with NR PDs and NT LEDs is written as

BER Performance Comparison in a Multipath MIMO-VLC System …

ℎ11 ℎ21 𝐻= ⋮ ℎ [ 𝑁𝑅1

ℎ12 ℎ22 ⋮ ℎ𝑁𝑅2

⋯ … ℎ𝑖𝑗 ⋯

ℎ1𝑁𝑇 ℎ2𝑁𝑇 ⋮ ℎ𝑁𝑅𝑁𝑇 ]

39

(3.3)

where hij is the channel gain between the ith photodetector and the jth LED. The expression for channel gain is same as (3.1) and (3.2) for LOS and first reflection signals respectively except that the Dirac delta function is ignored in those equations. The power received at ith PD from jth LED is given by 𝑃𝑟,𝑖,𝑗 (𝑡) = 𝑃𝑖,𝑗 (𝑡)⨂ℎ𝑖,𝑗 (𝑡)+⁡𝜎𝑖2 (𝑡) = 𝑃𝑖,𝑗 (𝑡)⨂[⁡ℎ𝐿𝑂𝑆,𝑖,𝑗 (𝑡) + ℎ𝑟𝑒𝑓,𝑖,𝑗 (𝑡)] +⁡𝜎𝑖2 (𝑡)⁡⁡⁡

(3.4)

where 𝑃𝑖,𝑗 (𝑡) is the transmitted optical power from jth LED to ith PD, ℎ𝐿𝑂𝑆,𝑖,𝑗 (𝑡) and ℎ𝑟𝑒𝑓,𝑖,𝑗 (𝑡) are LOS and reflected impulse response between ith PD and jth LED respectively, 𝜎𝑖2 (𝑡)⁡is the noise power at the ith PD which is the sum of 2 2 shot noise and thermal noise 𝜎 2 =𝜎𝑠ℎ𝑜𝑡 =+𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 , whose variance is expressed as 2 𝜎𝑠ℎ𝑜𝑡 =2𝑞𝛾𝑃𝑟 𝐵 +2𝑞𝐼𝑏𝑔 𝐼2𝐵⁡⁡⁡⁡ 2 𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 =(

8𝜋𝐾𝑇 𝐺

𝜂𝐴𝑟 𝐼2 𝐵2 +

(3.5)

16𝜋2 𝐾𝑇𝛤 𝑔𝑚

𝜂 2 𝐴2𝑟 𝐼3 𝐵3 )

(3.6)

where Ibg is background noise current, q represents electronic charge, Bequivalent noise bandwidth, I2 & I3- noise bandwidth factors, K-Boltzmann constant, T-absolute temperature, G-open loop voltage gain, 𝜂-fixed capacitance of PD per unit area, Γ-FET channel noise factor, and gmtransconductance. The received signal at ith PD is given as 𝑁

𝑇 𝑦𝑖 (𝑡) = ∑𝑗=1 𝑥𝑗 (𝑡)⨂ℎ𝑖,𝑗 (𝑡) + 𝑛𝑖 (𝑡)

(3.7)

where 𝑥(𝑡) is transmitted signal vector i.e., 𝑥(𝑡)=[𝑥1 (𝑡), 𝑥2 (𝑡)……𝑥𝑁𝑇 (𝑡)]T, 𝑛𝑖 (𝑡) is real valued additive white Gaussian noise at the ith PD with zero mean and a variance⁡𝜎 2 .

40

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The SNR at ith PD is given by 2

𝑆𝑁𝑅 =

(𝛾𝑃𝑟,𝑖 ) 𝜎𝑖2

(3.8)

where 𝛾 is the responsivity of the photodetector.

3.3. Different Radiation Patterns of LED Depending on the material and manufacturing technique, LED radiation patterns are classified as symmetrical or non-symmetrical. The light emitted spontaneously by an LED chip is the sum of three terms: refracted light from the encapsulating lens, internally reflected light inside the lens, and light reflected by the reflecting cup. As a result, the radiation pattern is altered by three structural factors: the roughness of the chip faces, the geometry and roughness of the reflective cup, and the geometry of the encapsulating lens. Because of the differences in the aforementioned factors, each LED has a distinct intensity pattern. As a result, choosing LED when designing an indoor VLC system would be advantageous. However, in this chapter, two symmetrical radiation patterns, Lambertian and Batwing, and one nonsymmetrical radiation pattern, Elliptical, have been considered for the BER performance analysis of the indoor multipath MIMO-VLC system.

3.2.1. Symmetrical Radiation Pattern Radiation patterns are typically described using spatial vectors in a spherical coordinate system with a fixed space measurement point. The luminous intensity in a symmetrical radiation pattern is symmetric with respect to the normal axis. The Lambertian pattern is characterised by bell-shaped curves that are most intense along the normal direction. The pattern’s radiant intensity is a cosine function with a semi-half power angle ϕ1/2 (𝜑) =

𝑚+1 2𝜋

𝑐𝑜𝑠 𝑚 (𝜑)

(3.9)

BER Performance Comparison in a Multipath MIMO-VLC System …

41

Figure 24. Radiation pattern of Lambertian, Batwing and Elliptical pattern (a) Cartesian plot (b) Polar plot.

where 𝜑 is the spherical polar angle w.r.t normal axis, m is the order of Lambertian pattern which is a function of ϕ1/2 given as 𝑚 = −ln⁡(2)⁄ln⁡(𝑐𝑜𝑠ϕ1/2 ). The Cartesian and polar plot of Lambertian radiation pattern as a function of 𝜑 is shown in Figure 24 (dash curve). The radiant intensity 𝐼(𝜑) can be substituted in Eq. (3.1) and (3.2) to get impulse responses for the Lambertian radiation pattern. As shown in Figure 24, the Batwing pattern has two symmetrical side lobes on either side (dash dot curve). The intensity increases with viewing angle up to a certain point, then drops sharply. As a result, the intensity is not uniform in all directions, and it has a directive nature. The intensity distribution of this symmetrical pattern is given by [41] |𝜑|−𝑔2𝑖 2

𝐼(𝜑) = ∑𝑖 𝑔1𝑖 𝑒𝑥𝑝 ⌊−𝑙𝑛2 (

𝑔3𝑖

) ⌋

(3.10)

where g11 = 0.76, g21 = 0º, g31=29º, g12 = 1.10, g22 = 45º and g32 = 21º. Substituting Eq. (3.10) in Eq. (3.1) and (3.2) for the term I(𝜑), the impulse responses of LOS and first reflection signal can be obtained for Batwing radiation pattern.

3.3.2. Non-Symmetrical Radiation Pattern The luminous intensity in a non-symmetrical radiation pattern is not symmetric with respect to the normal axis. There are several types of nonsymmetrical LED radiation patterns; however, the Elliptical radiation pattern

42

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

was used for the current analysis. The Elliptical radiation pattern, unlike the Lambertian and Batwing patterns, is not rotationally symmetric along two perpendicular azimuthal directions. The Elliptical radiation pattern has maximum intensity along the normal axis, as shown in Figure 24 (solid curve), but its radiation does not have broad angular characteristics. The Elliptical radiation pattern is described as [41] 𝑐𝑜𝑠2 ∅ 2 3𝑖 )

𝐼(𝜑, ∅) = ∑𝑖 𝑔1𝑖 𝑒𝑥𝑝 ⌊−𝑙𝑛2(|𝜑| − 𝑔2𝑖 )2 ((𝑔

𝑠𝑖𝑛2 ∅ 2 )⌋ 4𝑖 )

+ (𝑔

(3.11)

where the terms g11 = 0.13, g21 = 45º, g31 = g41 = 18º, g12 = 1, g22 = 0, g32 = 38º, g42 = 22º and ∅ is the azimuthal angle along the plane which is parallel to the radiating surface of the LED. Similar to previous cases, the impulse responses of LOS and first reflection of this radiation pattern can be found by substituting Eq. (3.11) in Eq. (3.1) and (3.2).

3.4. Results and Discussion In this chapter, we have considered two scenarios: ceiling-mounted LEDs and wall-mounted LEDs. In each case, impulse response, channel gain, and BER were simulated for a typical 5m×5m×3m room. The impulse response was calculated using the Barry method, with 1225 grids on the ceiling, floor, and walls, respectively. Based on impulse response and BER, the performance of various LED radiation patterns was compared.

3.4.1. Ceiling-Mounted Layout Case Figure 25 depicts the indoor MIMO-VLC system used for BER simulation in the ceiling-mounted layout. The four LEDs are installed on the ceiling at the coordinates shown in Table 3. Four receivers are arranged in a 20cm×20 cm pattern on the receiving plane, which is 0.85 m above the floor. These four LEDs and four PDs form a 4×4 MIMO system for which the BER performances using three radiation patterns with LOS and LOS plus reflection signals have been investigated. In our work, we considered two scenarios: in the first, the four receivers are placed in the centre of the receiving plane, and in the second, the receivers are kept at the corner of the receiving plane. The

BER Performance Comparison in a Multipath MIMO-VLC System …

43

values of the various parameters used in our simulation have been given in Table 3. Table 3. Parameters and their values Parameters Size Room Reflection coefficients of ceiling, floor, wall Number of grids on the ceiling, floor, wall Coordinates of 4 LEDs in Source ceiling-mounted layout Coordinates of 4 LEDs in wall-mounted layout Barry model Channel model Height of receiver plane Receiver above the floor Photodetector active area Coordinates of 4 PDs at centre- Scenario 1 Coordinates of 4 PDs at corner- Scenario 2 Half-angle FOV Time resolution Δt Open-loop voltage gain G Noise Equivalent noise bandwidth B Background noise current Ibg Electronic charge q Noise bandwidth factors I2, I3 Receiver area Ar Responsitivity of Photodiode R Boltzman’s constant K Absolute temperature T FET channel noise factor Γ Feedback resistor of FET based transimpedance amplifier Rf Transconductance gm Total Input capacitance ⁡𝜂

Value 5 × 5 × 3 m3 0.8 35×35, 35×35, 35×15 (1.25, 1.25, 3), (3.75, 1.25, 3), (1.25, 3.75, 3), (3.75, 3.75, 3) (0, 1.25, 1.80), (0, 1.25, 2.05), (0, 1.75, 2.05), (0, 1.75, 1.80) 0.85 m 1 Cm2 (2.4, 2.4, 0.85), (2.6, 2.4, 0.85), (2.4, 2.6, 0.85), (2.4, 2.6, 0.85) (3.75, 1.25, 0.85), (3.95, 1.25, 0.85), (3.75, 1.45, 0.85), (3.95, 1.45, 0.85) 60º 0.5 ns 10 100 MHz 5100⁡𝜇𝐴 1.6×10-19 C 0.562, 0.0868 1 cm2 0.5 A/W 1.38×10-23 J/k 298 K 1.5 1.4 KΩ

30 mS pF/cm2

44

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 25. VLC-MIMO system considered for BER simulation in ceiling-mounted LED layout.

3.4.1.1. Impulse Response due to LOS and NLOS Signals The received optical signal intensity is the product of the impulse response h(t) and the transmitted optical signal intensity. This section investigated the variation of LOS/NLOS impulse response with different radiation beam sources. The impulse response for the SISO system was computed in the first step using all three distinct radiation patterns where the LED is fixed at the centre of the ceiling and the position of the receiver is changed from centre (scenario 1) to corner of the room (scenario 2).

Figure 26. Impulse response for LOS signal at centre of room.

BER Performance Comparison in a Multipath MIMO-VLC System …

45

Figure 27. Impulse response for LOS signal at corner of room.

Figures 26 and 27 show the impulse responses to three radiation patterns at the central and corner positions. The amplitude of the impulse response due to the Elliptical pattern is greater than the other two radiation patterns, as shown in Figure 26. Because of the narrow angular characteristics of the Elliptical radiation pattern, more power is concentrated in the normal direction. At the central position, the Lambertian pattern with the broadest angular characteristics contributes the least. As shown in Figure 27, the amplitude of impulse response achieved using the Batwing pattern is greater than that achieved using the Elliptical and Lambertian patterns for the corner position. This is due to Batwing having maximum radiation intensity at a specific angle (in our case 450) from the normal axis, resulting in a strong impulse response at the corner side of the room.

Figure 28. Impulse response at the centre of room for first reflection signal.

46

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 29. Impulse response at the corner of room for first reflection signal.

Figures 28 and 29 show the impulse response for the first reflection signal at the room’s centre and corner, respectively. For both the central and corner positions, the amplitude of the impulse response due to the Elliptical pattern is the smallest, while that due to the Batwing radiation pattern is the greatest. The radiation is greatest along the side lobes at a certain angle from the normal axis in the Batwing pattern, so the greatest radiation is towards the wall surface and corner of the room. As a result, the reflected intensity using the Batwing pattern at the centre and corners is greater than the intensity achieved using the Lambertian pattern. Because elliptical irradiance is confined to the normal direction, the central and corner positions have the least reflected intensity. Because the Lambertian pattern has more angular characteristics than the Elliptical pattern, the amplitude of the impulse response obtained is greater in the corner position.

3.4.1.2. Channel Gain Distribution on the Floor Since received power at PD is dependent on channel gain, the distribution of channel gain within the room has been investigated using LOS/NLOS signals. Figures 30(a) and (b) show the distribution of channel gain on the floor for the LOS and first reflection signals, respectively. By excluding the LOS component of light rays, the reflection channel gain was simulated. According to Figure 30(a), the maximum LOS channel gain is received in the centre of the floor (vertically beneath the transmitters) and gradually decreases towards the corner and wall-side of the room. Figure 30(b) shows that the intensity of the first reflection is more uniformly distributed across the floor than the intensity of the LOS signal.

BER Performance Comparison in a Multipath MIMO-VLC System …

47

Figure 30. Channel gain distribution for ceiling-mounted layout a) LOS signal b) First reflection signal.

3.4.1.3. BER Performance In this section, we present the BER results for our 4×4 MIMO VLC system with LOS and LOS plus NLOS signals using various radiation patterns. A random NRZ-OOK bit stream with a bit rate of 64 Mb/s is generated and converted into a vector representing a time domain waveform. These data streams have been divided into four equal parallel streams, each of which is transmitted after convolution with the LED impulse responses. After propagation through the channel, the transmitted light signal reaches the receiver via the LOS or NLOS path.

48

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 31. BER performance for LOS and L-R1 signals using all three radiation patterns at the centre position of the room.

The simulated results of BER vs SNR for LOS and L-R1 signals using all three radiation patterns at the central position are shown in Figure 31. It can be seen that for LOS signals, Elliptical radiation patterns with narrow angular characteristics outperform Batwing and Lambertian radiation patterns with broad angular characteristics. The magnitude of the impulse response of the LOS signal using the three radiation patterns is of the order of 10-5, as shown in Figure 31. This is equivalent to -100dB in SNR at the receiver, which is why the LOS signal in Figure 31 has a 100dB offset. It can also be seen that the Batwing pattern outperforms the elliptical and Lambertian patterns in the case of L-R1 signals. Because the radiation from the Batwing profile is directed more towards the wall, there is more reflected power at the centre. As a result, it provides better BER curves than others. Because the magnitude of the impulse response of the Elliptical pattern is greater than that of the Lambertian pattern in the centre of the room, the BER curves for the L-R1 signal using the Elliptical pattern are better than those using the Lambertian pattern. Figure 32 depicts the BER performance at the room’s corner. The BER curves for all three radiation patterns merge together for the LOS signal, as can be seen. This contradicts the impulse response at the corner position for the LOS signal, as shown in Figure 27. The reason for this is that for the impulse response computation (Figure 27), the LED was placed in the centre position (2.5, 2.5, 3), whereas for the BER simulation, the LED coordinates

BER Performance Comparison in a Multipath MIMO-VLC System …

49

were (1.25, 1.25, 3), (3.75, 1.25, 3), (1.25, 3.75, 3), and (1.25, 3.75, 3). (3.75, 3.75, 3). As a result, the radiation intensities for the three patterns are nearly equal at the corner position, resulting in identical BER performances. For LR1 signals, the Elliptical pattern has the least reflected power due to its narrow angular radiation characteristics, causing more BER, whereas the Batwing pattern has more reflected power, resulting in better BER performance. Figures 31 and 32 show that the combined impulse response of LOS and first reflection is greatest for Batwing, then Elliptical, and finally Lambertian. The results show that if communication is dependent on LOS plus diffused signal in a ceiling-mounted layout, the Batwing radiation profile is a better choice than the Lambertian, which has been the focus of most research work. Table 4 depicts the overall comparison of both scenarios using different radiation patterns for LOS and L-R1 signals in terms of required SNR to achieve BER of 10-5.

Figure 32. BER performance for LOS and L-R1 signals using all three radiation patterns at the corner position of the room.

Table 4 shows that for each radiation pattern, the difference in required SNR between scenarios 1 and 2 increases by a greater margin for LOS signals than for L-R1 signals.

50

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Table 4. Comparison of SNR at BER of 10-5 between scenario 1 and 2 for ceiling-mounted case

Scenario 1 (Rx at centre) Scenario 2 (Rx at corner)

LOS L-R1 LOS L-R1

Lambertian 118 dB 158 dB 126 dB 162 dB

Batwing 118 dB 154 dB 126 dB 156 dB

Elliptical 116 dB 156 dB 126 dB 166 dB

3.4.2. Wall-Mounted Layout Case The performance of the impulse response, channel gain, and BER for the wallmounted scenario has been discussed in this section. Figure 33 depicts a wallmounted scenario in which LEDs are mounted symmetrically with respect to the vertical central line of the wall and receivers are placed in the centre of the floor.

Figure 33. VLC-MIMO system considered for BER simulation in wall-mounted LED layout.

3.4.2.1. Impulse Responses in Wall-Mounted Layout The impulse response in the wall-mounted layout case was simulated in the same way as the ceiling-mounted layout with a single LED on the wall. Figures 34 (a) and (b) show the impulse responses using LOS and first reflection signals at the centre of the room, respectively. As demonstrated for

BER Performance Comparison in a Multipath MIMO-VLC System …

51

the LOS signal, the magnitude of the impulse response for the Batwing is more closely followed by the Lambertian and then the Elliptical radiation patterns.

Figure 34. Impulse response for wall-mounted layout using a) LOS signal b) First reflection signal.

The impulse response in the case of the first reflection signal is more closely followed by Lambertian and then Batwing. As illustrated in Figure 34 (b), reflection from the opposite wall is uniformly distributed, and thus an Elliptical pattern with its radiation beam confined along the normal axis results in maximum reflection power at the receiver.

52

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

3.4.2.2. Channel Gain Distribution on the Floor in Wall-Mounted Layout The channel gain distribution for LOS and first reflection signals for the wallmounted layout case is shown in Figures 35 (a) and (b). Figure 35 (a) shows that the maximum intensity of the LOS signal is distributed beneath the wall where the LEDs are mounted and gradually decreases towards the opposite wall side. The intensity of the first reflection signal has a more uniform distribution than the LOS signal, as shown in Figure 35 (b). The maximum channel gain is at the upper corner side of adjacent walls (in relation to the wall on which the LEDs are mounted), because maximum reflection should occur only from these corner sides.

Figure 35. Channel gain for wall-mounted layout a) LOS signal b) First reflection signal.

3.4.2.3. BER Performance in Wall-Mounted Layout In the case of wall-mounted LEDs, the BER performance is investigated at the centre of the room. The BER performance for a LOS signal at the central position is shown in Figure 36 (a). The Batwing radiation profile has the lowest BER, followed by the Lambertian and Elliptical profiles. The reason for this is that the received LOS channel gain decreases towards the centre of the room, where the receivers are located. As a result, a radiation profile with a wide angular beam will result in better BER performance. The Batwing profile radiation has maximum directivity at 450 angles from its axis, resulting in maximum light intensity at the centre of the floor and the centre of all three walls. The Lambertian has a wider radiation beam than the elliptical and thus provides less BER. Figure 36 (b) depicts BER performance using the L-R1 signal in the centre of the room. Elliptical profiles have the lowest BER, followed by batwing

BER Performance Comparison in a Multipath MIMO-VLC System …

53

profiles and finally Lambertian profiles. Since the first reflection channel gain from the opposite wall is greater than the centre area of the adjacent walls. As a result, the Elliptical profile has the lowest BER because LEDs with high directionality transmit more power in the direction of its axis, causing more light intensity reflection from the opposite wall. Because less reflection occurs from the opposite wall, the Batwing and Lambertian profiles have higher BER than the elliptical. Figures 36 and 31 show that the BER performance at the central position for wall-mounted layout results has a higher bit error rate than for ceiling-mounted layout. The BER results in the wall-mounted layout case show that the Batwing radiation source is a better choice for LOS signal, whereas the Elliptical is appropriate if communication is based on LOS plus diffused signal. Table 5 compares various radiation patterns for LOS and LR1 signals in terms of required SNR to achieve BER of 10-5.

Figure 36. BER performance for wall-mounted scenario (a) LOS signal (b) L-R1 signal.

Table 5. Comparison of SNR at BER of 10-5 for wall-mounted case

LOS L-R1

Lambertian 171 dB 192 dB

Batwing 168 dB 200 dB

Elliptical 172 dB 188 dB

Table 5 shows that the difference in required SNR between individual radiation patterns is greater for the L-R1 signal than for the LOS signal. Though the results may differ if the LEDs are mounted on all four walls, this is not a practical option. For MIMO system implementation, all four transmitters should be placed close to each other.

54

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

3.5. Conclusion Different radiation patterns are compared in this chapter based on the BER performance of 4×4 indoor MIMO-VLC systems under various scenarios. When LEDs are mounted on the ceiling, the first case examines BER performance at the centre and corner of the room using LOS and LOS plus first order reflection signals. Because of its narrow radiation beam, the BER curve for LOS signal using elliptical radiation pattern has been found to be better than the Batwing and Lambertian when the receiver is placed in the centre position. Because of its wide radiation beam, the BER curve of the Batwing radiation pattern outperforms the Lambertian and elliptical radiation patterns for L-R1 signals. At the corner position, the BER for the LOS signal is nearly the same with all three radiation patterns, whereas the Batwing radiation pattern provides better BER performance for the L-R1 signal. The results show that the required SNR in scenario 1 using Lambertian, Batwing, and Elliptical radiation patterns for LOS signal is 118dB, 118dB, and 116dB, respectively, whereas it is 158dB, 154dB, and 156dB for L-R1 signal. For scenario 2, the required SNR for the same BER is 126dB for all radiation patterns for LOS signal and 162dB, 156dB, and 166dB for Lambertain, Batwing, and Elliptical patterns for L-R1 signal, respectively. As a result, for ceiling-mounted LEDs, a narrow beam radiation pattern is preferable for LOS signals, whereas a wide beam radiation pattern is preferable for LOS plus reflected signals. In the second case, when LEDs are mounted on the wall, the BER analysis shows that for LOS signals, a wide beam radiation pattern is preferable, whereas for L-R1 signals, a narrow beam radiation pattern is preferable. However, in this chapter, it is assumed that the VLC acts as an access point for data transmission, which is not technically correct. In practice, the VLC system requires a backbone network because it cannot serve as an information source directly. In Chapter 4, we investigated at an integrated power line communication-visible light communication system for evaluating BER performance.

Chapter 4

BER Investigation for an OFDM-Based Hybrid PLC-VLC System 4.1. Introduction Without a doubt, fibre optic networks that span cities and continents have aided in meeting broadband subscriber targets. However, due to cost and time constraints, as well as the risk of maintenance and breakage, laying cables in the last mile may not be feasible. In this context, power line communication (PLC) technology is a promising and strong alternative due to its numerous advantages such as cost-effectiveness and the ability to carry both power and data. PLC technology is already in use in home automation, smart grid, and telemetry. However, there are issues due to additive and multiplicative noises in the PLC system. Background noise and impulsive noise are examples of additive noise. The fading in the received signal strength is caused by multiplicative noise. Furthermore, reflections at impedance discontinuities cause echoes of the transmitted signal, making the power line channel a multipath environment. The transmitted signal travels to the receiver via the direct and reflected paths, resulting in intersymbol interference. Nonetheless, it has been discovered that using the discrete Fourier transform (DFT) algorithm, orthogonal frequency division multiplexing (OFDM) can mitigate these issues by distributing the effect of impulsive noise over multiple symbols. The long symbol duration time in the OFDM system aids in avoiding ISI caused by the PLC channel’s multipath effect. However, the inability of the power line to broadcast data over a large area necessitates the investigation of some integrated technology. In 2003, Komine et al. proposed integrating power line communication with visible light communication (VLC), and this integrated system is known as the hybrid PLC-VLC system [43].

56

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Visible light communication has piqued the interest of both academia and industry due to its unregulated license-free bandwidth and the dual benefit of illumination and communication. LEDs with long life, robustness, small size, and low power consumption give VLC a greater role in unlocking a wide range of opportunities and applications. However, the limited modulation bandwidth of LEDs limits VLC system high data rate transmission. This limitation is overcome by MIMO technology, which uses spatial multiplexing to increase data rate. [44]. In the VLC system, similar to the PLC system, the transmitted signal reaches the receiver via the Line-of-sight (LOS) and Non-line-of-sight (NLOS) paths. The strength of the NLOS signal is lower than that of the LOS signal due to multiple reflections from walls, floor, ceiling, table, and so on. However, NLOS signals cannot be ignored because multiple reflections cause ISI, which is a barrier to high bit rate systems. Because the mobile receiver is at desk-top height and can move around the room, the NLOS signal must be considered alongside the LOS signal in the MIMO-VLC system. The bit error rate (BER) as a function of SNR is used to assess the quality of LOS or diffused received signals. The BER performance of a Rayleigh fading PLC channel with heavy, moderate, and weak impulsive noise situations corresponding to different inter-arrival time and time duration has been investigated [45-46]. The results show that for all three situations, the bit error rate decreases for lower SNR with comparable performance, but there is no improvement for higher SNR. The results also show that impulsive noise is more harmful than background noise. A multi-hop (N) PLC system with decode-and-forward (DF) relays has been studied to combat distance-dependent attenuation [47]. For fixed transmission power (PT), it has been demonstrated that the multi-hop (N) PLC system outperforms the conventional direct transmission PLC system. For different values of attenuation parameters, the average end-to-end BER improves as PT or N increases. The BER performance of the OFDM system has been theoretically investigated for broadband PLC, and it has been discovered that only heavily distributed impulsive noise would interfere with the OFDM system, and the adverse effect of multipath is more severe than the effect of impulsive noise [48]. They also determined that a large number of carriers with optimal guard intervals are required to overcome the effect of multipath. To improve the BER performance of the MIMO-VLC system, various optimization schemes such as joint precoding, orthogonal circulant matrix transform (OCT) precoding, and SVD-based adaptive power loading of data subcarriers have been proposed [49]. BER was determined in all of those works for either the PLC system or the VLC system separately, but not

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

57

for the integrated PLC-VLC system. Through the VLC system, a user in one room (or building/office) cannot communicate with a user in another room (or building/office). This limitation can be overcome by using a Decode and Forward (DF) module in an integrated PLC-VLC system. In a hybrid PLCVLC system, the DF module bidirectionally exchanges information between PLC and VLC by demodulating and decoding signals from PLC to VLC or vice versa. Because the PLC system suffers from frequency-selective Rayleigh fading, the overall BER performance of the hybrid PLC-MIMO-VLC system differs from that of the PLC or MIMO-VLC when considered separately. Several authors investigated the performance of hybrid PLC-VLC system. W. Ding et al. [50] examined potential application of a hybrid broadband PLC and VLC system with OFDM modulation and achieved 48 Mbps data rate within an 8 MHz bandwidth using a single LED. Ma H et al. [51] proposed a hybrid PLC-VLC system that uses the spatial-optical (SO) OFDM technique to improve the signal-to-interference-plus-noise ratio (SINR) at the user. The Welch method was used to estimate the magnitude of the transfer function of the cascaded PLC-VLC channel [52]. Xu Ma et al. [53] proposed a multiservice transmission scheme for an integrated PLC-VLC system that differs in the frequency domain, time domain, and bit division multiplexing (BDM). The approach is based on the direct re-transmission concept, which does not require any changes to the original power line network layout. Under the same conditions, they discovered that the BDM scheme outperforms conventional TDM and FDM schemes. The BER performance of a hybrid PLC-VLC system in an inter-building scenario was investigated using phase-shift keying (PSK) mapping in the PLC and colour shift keying (CSK) in the VLC system [54]. They calculated BER performance by varying the magnitude of PSK symbols and discovered that 0.133 is the best value. The optimal values of impulse noise clipping/nulling threshold and DC bias for integrated PLC-VLC systems based on DCO-OFDM and ACO-OFDM were determined, and it was discovered that the probability of occurrence of impulsive noise degrades BER performance more than impulsive noise power [55-56]. BER performance of hybrid PLC-VLC system has been investigated in all of these works under restricted scenarios, such as ignoring fading effect in PLC, considering fixed spacing between LEDs, fixed room dimension, LOS signal only, and so on. As previously stated, the PLC system suffers from frequency-selective Rayleigh fading and thus cannot be excluded from the BER analysis. The channel gain between a given PD and two closely spaced LEDs is nearly the same, resulting in an ill-conditioned channel matrix. So, before implementing hybrid PLC-VLC system, the optimal spacing between LEDs must be

58

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

determined. Due to various room configurations, such as the transmitter not being placed at a sufficient height or the receiver’s mobility as required, LEDs and PDs may not be in each other’s FOV in some scenarios. As a result, the diffused signal cannot be ignored. We have reported the BER performance of an integrated PLC-MIMOVLC system in this chapter with the following considerations: 1. For PLC and VLC, the Rayleigh fading and Barry channel models are used. 2. MIMO-VLC signals with line-of-sight (LOS) and LOS plus first reflection (L-R1). 3. Configurations A and B, with room sizes of 5m×5m×3m (standard room size) and 7.5m×7.5m×3.5m (conference hall size), respectively. 4. In each configuration A and B, the LED separations are 0.2m, 0.6m, and 1m, respectively, for scenarios 1 and 2. Based on the foregoing, the following contributions are included in this chapter: 

 

For configurations A and B, we investigate the BER performance of a hybrid PLC-MIMO-VLC (HPMV) system using OFDM modulation. For different spatial separations of LEDs, we choose LOS and LOS plus first reflection (L-R1) signals for BER performance. The frequency response of the MIMO-VLC system was simulated in order to validate the BER performance for configurations A and B.

4.2. The Architecture of HPMV System Figure 37 depicts the block diagram of the HPMV system. The data is transferred from the PLC modem to the power line. Under PLC modem, pilot symbols are inserted, IFFT, cyclic prefix addition, and so on, and data is transformed into an OFDM modulated signal. While propagating from the transmitting to receiving ends, the signal is corrupted by various types of PLC noises such as background noise, narrowband noise, and impulsive noise. Before entering the VLC channel, the OFDM signal is demodulated.

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

59

Figure 37. Block diagram of proposed HPMV system.

The signal is converted into an NT parallel stream (in our case, NT = 4) in the MIMO-VLC channel, where NT is the number of LEDs. After mapping each data, it is converted into an OFDM signal using an OFDM encoder. Because LEDs can only transmit real and positive signals, OFDM encoders incorporate Hermitian symmetry. Clipping and adding DC bias make this true signal positive. Finally, the optical signal is sent via LEDs. The received optical signal is converted into an electrical signal by a photodetector on the receiver side, and then the reverse operation of VLC transmitter is done, as shown in the diagram. Finally, the decoded electrical signals are de-multiplexed using the zero-forcing (ZF) technique. The OFDM modulation (and demodulation) is performed in each PLC and VLC channel. The flow chart of process involved in PLC-MIMO-VLC transmitter is described in Figure 38.

60

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 38. Flow chart of PLC-MIMO-VLC transmitter system.

4.3. The Frequency Response of the Integrated PLC-VLC System PLC channels are random and frequency selective due to the multipath phenomenon. Because of several branches and impedance mismatches that cause multiple reflections, the signal in a PLC channel arrives at the receiving end via both direct and multipath. PLC systems are divided into two types: narrowband PLC and broadband PLC. Narrowband PLC operates at frequencies ranging from 3 kHz to 500 kHz and has a data rate of hundreds of kbps with a range of several kilometres, whereas broadband PLC operates at

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

61

higher frequencies (1.8-250 MHz), higher data rates (100 Mbps), and is used in communications with shorter ranges. The broadband PLC system has been considered in this chapter for the transmission of OFDM signals. The following mathematical expressions were used in our simulation for BER and frequency response of the integrated PLC-VLC system: The input/output model for a Rayleigh fading PLC channel is given as [45]. 𝑦 = ℎ̂𝑐 𝑥 + 𝑛̂

(4.1)

where 𝑥 is channel input, 𝑦 is channel output, 𝑛̂ is the PLC noise and ℎ̂𝑐 is complex channel gain whose envelope ℎ̂ is Rayleigh distributed with PDF given by 2

̂ ̂ ℎ −ℎ 𝑓(ℎ̂) = 𝜎̂2 exp⁡(2𝜎̂2 )

(4.2)

where 𝜎̂ 2 is the variance. In the frequency domain, the model is described as channel frequency response which is given as [50] 𝐻𝑃𝐿𝐶 (𝑓) = ∑𝐿𝑖=1 𝑔𝑖 ∙ 𝑒 −(𝛼𝑜+𝛼1𝑓

𝜒 )𝜏 𝑣 𝑖 𝑝

∙ 𝑒 −𝑗2𝜋𝜏𝑖 ⁡

(4.3)

where 𝑔𝑖 is the weighting factor of ith path, 𝛼𝑜 and 𝛼1 are attenuation factors, 𝜒 is the exponent of the attenuation factor (~1), 𝑣𝑝 is the group velocity of the signal in the power link. The power of received signal 𝑦 in (1) depends on the transmitted power and attenuation coefficients 𝛼𝑜 and 𝛼1 of PLC channel. The power line channel is plagued by a variety of noises. Background noise, periodic impulse noise synchronous with the mains, periodic impulse noise asynchronous with the mains, and aperiodic impulse noise are all present in the PLC system. The thermal noise is the background noise, and it is Gaussian and coloured. For coloured background noise, the power spectral density (PSD) 𝑆𝑏 (𝑓)⁡is modelled as a first order exponential. 𝑆𝑏 (𝑓) = 𝑎 + 𝑏|𝑓|𝐶

(4.4)

where a, b and c are positive parameters. In our simulation, the worst case has been considered with a = -145, b = 53.23, c = -0.337.

62

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The narrowband PSD 𝑆𝑛 (𝑓) follows a parametric Gaussian function given as 𝑆𝑛 (𝑓) =

∑𝑁 𝐾=1 𝐴𝑘

∙𝑒

−

(𝑓−𝑓𝑜,𝑘 )2 2∙𝐵𝑘

(4.5)

where 𝐴𝑘 is the amplitude of the kth narrowband signal with centre frequency 𝑓𝑜,𝑘 and bandwidth 𝐵𝑘 . At a given instant, N such narrowband interferences are present. Finally, a generalized background noise is generated by adding color background noise and narrowband noise and its PSD is given by 𝐶

𝑆(𝑓) = 𝑆𝑏 (𝑓) + 𝑆𝑛 (𝑓) = 𝑎 + 𝑏|𝑓| +

∑𝑁 𝐾=1 𝐴𝑘

∙𝑒

−

(𝑓−𝑓𝑜,𝑘 )2 2∙𝐵𝑘

⁡⁡

(4.6)

Periodic impulse noise synchronous with mains is made up of impulses with frequencies that are multiples of the main electrical network’s frequency. Laptops, rectifiers, LCD monitors, and other electronic devices are common sources of noise. The sum of damped sinusoids model is used to generate timedomain noise samples for this noise. 𝑖=𝑁 −1

𝑝(𝑡) = 𝑢(𝑡) − 𝑢(𝑡 − 𝑇𝑑 )} ∙ ∑𝑖=0 𝑑

Â𝑖 𝑒 −𝛽𝑖|𝑡| 𝑒 −𝑗2𝜋𝑓𝑖 𝑡 ⁡⁡⁡

(4.7)

where p(t) is the time domain impulse noise, u(t) is the unit step function, Nd is the number of damped sinusoids in an impulse, Â is the amplitude of impulse, Td is the duration of the impulse, β is the damping factors and f is the pseudo frequency of sinusoids. In the simulation, NI such impulses are generated for every mains cycle spread by an inter-arrival time tint. The fourth type of noise, such as periodic impulse noise asynchronous with the mains, is independent of the electrical network and is frequently generated by networkconnected switch-mode power supplies (SMPSs). Aperiodic impulse noise is caused by isolated activities such as on/off and plug/unplug, which are less frequent but more destructive. The noise is destructive because its amplitude can reach 50V with a short inter-arrival time.

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

63

Figure 39. Frequency response of PLC channel.

The frequency response of the PLC channel is depicted in Figure 39. Deep notches can be seen, indicating the frequency selective fading effect.

Figure 40. Indoor geometrical model of HPMV system.

The geometrical model of an integrated PLC with a 4×4 MIMO-VLC system is shown in Figure 40. The four LEDs are linked by power line cables in a room measuring 5m×5m×3m (configuration A) and 7.5m×7.5m×3.5m (configuration B). As shown in the figure, the signal from an Ethernet or other data source is coupled to the power line via a PLC modem. The receiver plane is 0.85m above the floor, on which four photodetectors of MIMO system are mounted.

64

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The d1 and d2-d3 in the figure represent LOS and LOS plus first reflection path, respectively. When considering an LED’s Lambertian radiation pattern, its radiant intensity⁡𝐼(𝜑)⁡ is defined by the semi-half power angle 𝜃1⁄2 𝐼(𝜑) =

𝑚+1 2𝜋

𝑐𝑜𝑠 𝑚 (𝜑)

(4.8)

where 𝜑 is the radiation angle w.r.t normal axis, 𝑚 is the order of Lambertian pattern which is a function of 𝜃1⁄2 given by 𝑚 = ln⁡(2)⁄ln⁡(cos⁡(𝜃1⁄2)). The impulse response due to LOS signal is given as 1

ℎ(0) (𝑡; 𝑆, 𝑅) =

𝑑12

𝐴𝑟 𝐼(𝜑0 ) 𝑐𝑜𝑠(𝜑0)

𝑐𝑜𝑠(Ѱ0 )𝛿(𝑡 − 0, (

𝑑1 𝑐

),

0 ≤ Ѱ0 ≤ Ѱ𝑐 Ѱ 0 ≥ Ѱ𝑐 )

(4.9)

where 𝐼(𝜑0 ) is the radiant intensity of the LED in the direction of the emission angle 𝜑0 w.r.t normal axis,Ѱ0 is the incidence angle at the receiver, Ѱc is field of view of the receiver, d1 is the distance between source and receiver, Ar is the active area of the photodetector, c is the speed of light and 𝛿(𝑡) is the Dirac function. The impulse response for the first reflection signal can be written as ℎ(1) (𝑡; 𝑆, 𝜀𝑙 , 𝑅) = 1

𝜋𝑑22 𝑑32

𝐴𝜀𝑙 𝜌𝐴𝑟 𝐼(𝜑1 ) 𝑐𝑜𝑠(𝜑1 ) 𝑐𝑜𝑠(𝛼)

𝑐𝑜𝑠(𝛽)𝑐𝑜𝑠(Ѱ1)𝛿(𝑡 − (

0,

𝑑2 +𝑑3 𝑐

),

0 ≤ Ѱ1 ≤ Ѱ𝑐

(4.10)

Ѱ1 ≥ Ѱ𝑐 )

where 𝜀𝑙 is the lth grid of wall having area 𝐴𝜀𝑙 , d2 is the distance from source to lth grid of wall, d3 is the distance from lth grid of wall to the receiver, 𝜑1 is the radiation angle, Ѱ1 is the incidence angle at the receiver for first reflection signal, 𝜌 is the reflection coefficient of wall, 𝛼 and 𝛽 are the angle of incidence and angle of irradiance at the lth grid w.r.t normal axis of wall respectively. The power received at ith PD from jth LED is given by 𝑃𝑅,𝑖𝑗 (𝑡) = 𝑃𝑇,𝑖𝑗 (𝑡)⨂ℎ𝑖𝑗 (𝑡) + ⁡ 𝜎𝑖2 (𝑡) = 𝑃𝑇,𝑖𝑗 (𝑡)⨂[⁡ℎ𝐿𝑂𝑆,𝑖𝑗 (𝑡) + ℎ𝑟𝑒𝑓,𝑖𝑗 (𝑡)] + ⁡ 𝜎𝑖2 (𝑡)

(4.11)

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

65

where ⨂ represents convolution, 𝑃𝑇,𝑖𝑗 (𝑡) is the transmitted optical power from jth LED to ith PD,⁡ℎ𝐿𝑂𝑆,𝑖𝑗 (𝑡) and ℎ𝑟𝑒𝑓,𝑖𝑗 (𝑡) are LOS and reflected impulse response between ith PD and jth LED respectively, 𝜎𝑖2 (𝑡)⁡is the noise power at the ith PD which is the sum of shot noise and thermal noise 2 2 𝜎 2 =𝜎𝑠ℎ𝑜𝑡 +𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 , whose variance is expressed as 2 𝜎𝑠ℎ𝑜𝑡 = 2𝑞𝛾𝑃𝑟 𝐵 + 2𝑞𝐼𝑏𝑔 𝐼2 𝐵 8𝜋𝐾𝑇

2 𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 =(

𝐺

𝜂𝐴𝑟 𝐼2 𝐵2 +

(4.12) 16𝜋2 𝐾𝑇𝛤 𝑔𝑚

𝜂 2 𝐴2𝑟 𝐼3 𝐵3 )

(4.13)

where 𝐼𝑏𝑔 is background noise current, q represents electronic charge, Bequivalent noise bandwidth, 𝐼2 & 𝐼3 - noise bandwidth factors, T-absolute temperature, K-Boltzmann constant, 𝜂-fixed capacitance of PD per unit area, G-open loop voltage gain, and 𝑔𝑚 - transconductance, Γ-FET channel noise factor. VLC channels are considered frequency selective due to their low pass characteristics and multipath dispersion. Multipath dispersion occurs when the signal’s bandwidth exceeds 20MHz. The total frequency response is represented as 𝑁

𝑇 𝐻𝑉𝐿𝐶 (𝑓) = ∑𝑖=1 ℎ𝐿𝑂𝑆,𝑖 exp(−𝑗2𝜋𝑓∇𝜏𝐿𝑂𝑆,𝑖 ) + 𝑇 ∑𝑁 𝑖=1 ℎ𝑑𝑖𝑓𝑓,𝑖

exp⁡(−𝑗2𝜋𝑓∇𝜏𝑑𝑖𝑓𝑓,𝑖 ) 1+𝑗(𝑓⁄𝑓𝑜 )

(4.14)

where ℎ𝐿𝑂𝑆,𝑖 and ℎ𝑑𝑖𝑓𝑓,𝑖 are the channel gains of LOS and diffused signal respectively from ith LED, ∇𝜏𝐿𝑂𝑆,𝑖 and ∇𝜏𝑑𝑖𝑓𝑓,𝑖 are the corresponding delay. The frequency response of the hybrid PLC-MIMO-VLC system is the product of the frequency responses of the PLC and the MIMO-VLC 𝐻ℎ𝑦𝑏. (𝑓) = 𝐻𝑃𝐿𝐶 (𝑓) ∙ 𝐻𝑉𝐿𝐶 (𝑓)

(4.15)

4.4. Results and Discussion In this chapter, we investigated the BER performance of an HPMV system using OFDM signals while accounting for LOS and LOS plus first reflection signals in two configurations A and B. The room size in configuration A is

66

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

5m×5m×3m, while it is 7.5m×7.5m×3.5m in configuration B. Because the BER performance of an HPMV system is affected by room dimensions, Table 6. The parameters of HPMV and their values Parameters Value PLC channel (6 path) model- Rayleigh fading Path No. 1 2 3 4 5 Path length (m) 78 126 191 256 306 Weighting factor gi -39× 156× -34× 715× -122× 10-4 10-4 10-3 10-4 10-3 ao, a1 -2.1×10-3, 8.1×10-10 Vp (m/s) 1.5×108 m/s Bandwidth 10 MHz No of impulses in one mains cycle NI 10 Inter-arrival time of impulse tint 2×10-6 s Sampling period of Rayleigh channel 1×10-3 s No of subcarrier 128 Cyclic prefix length 32 No of pilot symbols 4 No of OFDM frames 104 VLC Channel model - Barry model Room size (L×W×H) Configuration A 5m×5m×3m Configuration B 7.5m×7.5m×3.5m Reflection coefficients of walls and ceiling 0.8 Number of grids 7×L, 7×W, 7×H Position of 4 LEDs in Scenario 1 (0.01,0.01,3),(-0.01,0.01,3), (-0.01,-0.01,3),(0.01,- 0.01,3) Position of 4 LEDs in scenario 2 (0.03,0.03,3),(-0.03,0.03,3), (-0.03,-0.03,3),(0.03,- 0.03,3) Position of 4 LEDs in scenario 3 (0.05,0.05,3),(-0.05,0.05,3), (-0.05,-0.05,3),(0.05,- 0.05,3) Receiver Coordinate of 4 PDs (0.05,0.05,0.85),(-0.05,0.05,0.85), (-0.05,-0.05,0.85),(0.05,- 0.05,0.85) Photodetector active area 1Cm2 Responsitivity of Photodiode R 0.5 A/W Time resolution Δt 0.5ns Open-loop voltage gain G 10 Equivalent noise bandwidth B 100 MHz Feedback resistor of FET based transimpedance amplifier R f 1.4KΩ Transconductance gm 14mS Total input capacitance 2PF Background noise current Ibg 5100⁡𝜇𝐴 Noise bandwidth factors I2, I3 0.562, 0.0868 FET channel noise factor Γ 1.5 Absolute temperature T 298K

6 330 76× 10-3

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

67

transmitter and receiver positions, it is necessary to investigate it in various configurations and scenarios. The simulated frequency response results for MIMO-VLC systems provided side by side justify the BER results for both configurations when all three scenarios are considered. Table 6 shows the values of various parameters of PLC and VLC channels.

4.4.1. BER Performance of Integrated System for Configuration A The BER performance of the HPMV system and the frequency response of the MIMO-VLC system for a normal room size (5m×5m×3m) have been simulated in configuration A. The BER and frequency response have been demonstrated for three scenarios with spatial separations of 0.2m, 0.6m, and 1m between LEDs.

Figure 41. BER performance of HPMV system for configuration A a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m).

68

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figures 41(a), (b), and (c) show the BER performance of the HPMVOFDM system for scenarios 1, 2, and 3, respectively. Figure 41(a) shows that the LOS signal has less BER than the L-R1 signal with very little margin. At BER of 10-5, the LOS signal has a gain margin of approximately 1dB to 2dB in SNR when compared to the L-R1 signal. To better understand the nature of the curve, the inset in Figure 41 shows the BER plot for the range of SNR from 0dB to 140dB. The LOS channel gain for our considered room size is of the order of 10-4, resulting in an electrical path loss of about -80dB on the receiver side. As a result, there is an offset of 80dB in the BER curves compared to the received Es/No. In scenario-2, the separation between LEDs was increased from 0.2m to 0.6m to investigate its effect on the BER performance of the HPMV-OFDM system. Figure 41 (b) shows that as the distance between the LEDs increases, so does the difference in the BER curve between the LOS and L-R1 signals. When comparing Figures 41(a) and 41(b), it is discovered that increasing the separation between LEDs reduces the bit error rate for both LOS and L-R1 signals. The comparison shows that increasing the separation from 0.2m to 0.6m with a gain margin of approximately 20dB reduces the required SNR at BER of 10-5 from 135dB to 115dB for the LOS signal. When comparing scenario-2 to scenario-1, the BER for the L-R1 signal improved by nearly 8 dB (difference between 137dB and 129dB). The BER curve for scenario-3 is depicted in Figure 41 (c) (LED separation of 1m). It can be seen that the required SNR at BER of 10-5 using LOS and L-R1 signals is nearly the same for scenarios 3 and 2. It reveals that only optimal LED separation reduces BER for HPMV systems. It should also be noted that the BER for the LOS signal decreases by a significant margin in SNR, whereas the BER for the L-R1 signal decreases by a smaller margin. Figure 42 depicts the frequency responses of 4×4 MIMO-VLC systems using the L-R1 (LOS plus first reflection) signal in all three scenarios considered. These responses back up the previously mentioned BER performance results for each scenario. Figure 42(a) shows that there are deep notches in the frequency response for LED separations of 0.2m. The results show that when LED separation is low, deep fading is high. It also demonstrates that the attenuation is high when compared to Figures 42(b) and (c), which correspond to scenarios 2 and 3, respectively. The depth of the notches decreases as the distance between the LEDs increases. Due to the random nature of multipath components, the power received at the receiver may be zero (resulting in signal power less than the noise threshold) due to destructive interference or greater

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

69

than the transmitted signal power due to constructive interference. As a result of the random nature of multipath components in the channel, the receiver receives a range of signal power. This is also demonstrated by the fact that BER and the probability of deep fade are both inversely proportional to SNR, implying that BER is directly proportional to deep fade in the wireless communication channel. The frequency responses of figures 42(b) and (c) show that as the separation between LEDs increases, fading and attenuation decrease, which reasonably justifies the BER curves of Figures 41(b) and (c). When the distance between LEDs is small (0.2m), the multiple transmitters behave almost like a single transmitter. This event reduces transmit diversity by reducing the number of links between the transmitter and receiver. Overall, it can be seen that as the distance between LEDs in an HPMV system increases, frequency selectivity and deep fading decrease, resulting in an increase in coherence bandwidth and, consequently, BER improves.

Figure 42. Frequency response of MIMO-VLC system for configuration A a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m).

70

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The frequency response of the HPMV system for configuration A is depicted in Figure 43. When compared to Figure 42, the nature of the HPMV curve almost exactly matches that of the PLC frequency response, but the Magnitude of deep fading and overall attenuation has increased significantly. In the PLC system, the magnitude of notches ranges from -80dB to -50dB, whereas in the HPMV system, it ranges from -240dB to -200dB. The maximum deep fade in HPMV is -240dB versus -80dB, a remarkably large difference. This is why the BER of the HPMV system is higher than the BER of the PLC system.

Figure 43. Frequency response of HPMV system for configuration A.

4.4.2. BER Performance of Integrated System for Configuration B In this section, the same three scenarios are used to evaluate the BER performance of conference room hall sizes (7.5m×7.5m×3.5m). When comparing Figure 44 to Figure 41, it is clear that the BER for all three scenarios increases for LOS and L-R1 signal. The BER for LOS and L-R1 signals is higher in configuration B than in configuration A by a gain margin of 12dB and 20dB in SNR, respectively, when scenario-1 of both configurations is considered. Because the room size in configuration B is larger than the size in configuration A, the distance between source and receiver for LOS and reflected signals is greater, resulting in more BER. The bit error rate for LOS and L-R1 signals in scenario-2 of configuration B is 10dB and 15dB higher than in scenario-2 of configuration A. Similarly, in scenario-3 of both configurations, the difference in LOS and L-R1 signals is 7dB and 12dB, respectively.

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

71

However, unlike configuration A, where there is no difference in BER performance between scenarios 2 and 3, the BER in scenario 3 of configuration B is better than scenario 2. This means that as the room dimension increases, so does the required optimal separation between LEDs. As a result, the optimal LED separation should be determined by the room size for the HPMV system. Table 7 depicts the overall comparison of both configurations under three scenarios in terms of required SNR to achieve BER of 10-5. In comparison, the difference in BER between configurations A and B using both LOS and L-R1 signals decreases as the distance between LEDs increases.

Figure 44. BER performance of HPMV system for configuration B a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m).

72

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Table 7. Comparison of required SNR at BER of 10-5 between configuration A and B

Configuration A Configuration B

LOS L-R1 LOS L-R1

Scenario-1 135 dB 137 dB 148 dB 155 dB

Scenario-2 115 dB 129 dB 125 dB 145 dB

Scenario-3 115 dB 128 dB 115 dB 140 dB

Figure 45 depicts the frequency responses of configuration B in three different scenarios. In comparison to configuration A, the results show that the frequency selective fading has been shifted to a higher frequency. The comparisons of Figure 42 and 45 show that frequency selectivity, rather than attenuation of frequency response, dominates BER performance as room size increases.

Figure 45. Frequency response of MIMO-VLC system for configuration B a) scenario-1 (0.2m) b) scenario-2 (0.6m) c) scenario-3 (1m).

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

73

4.5. Experimental Demonstration of Integrated PLC-VLC We demonstrated an integrated PLC-VLC system using four LEDs and one LDR. Figure 46 depicts a schematic diagram of the experimental setup. Figure 47 shows the hardware configuration of our PLC-VLC system. Both the transmitter and receiver of a PLC system are comprised of an Arduino and a modem for signal transmission and reception. MATLAB Simulink generates the signal, which is sent to Arduino pin 9 of the transmitter. The signal is coupled to the modem at pin 9 and converted into an FSK modulated signal. The FSK signal is transmitted over the PLC line and then received by the modem at the PLC end. The FSK signal is demodulated by the modem and then sent to Arduino. The signal is delivered from Arduino pin 9 to the LED circuit, where the electrical signal is converted into an optical signal and finally transmitted for indoor communication. The transmitted light signal is captured and converted into an electrical signal by the LDR circuit. DSO displays the received electrical signal. Figure 48 depicts the Simulink model of the generated and received signals from pins 9 and 10. Figure 49 depicts the PLC transmitter and receiver. The hardware circuits of transmitter and receiver are shown in Figures 50 and 51. Figure 52 shows the experimental results obtained from the output of a digital storage oscilloscope (DSO). Figure 52 (a) shows the Simulink generated square wave signal that is fed to pin 9. Figure 52 (b) depicts the FSK modulated signal from the modem output that is connected to the 230V main power line. The FSK signal is transmitted through the PLC channel and then demodulated to obtain the recovered signal as displayed in Figure 52 (c). Figure 52 (d) shows the received signal via LDR.

Figure 46. Schematic diagram of experimental set up.

74

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 47. The experimental set up of PLC-VLC system.

Figure 48. Generated signal on pin no. 9 and received signal on pin no. 10 by interfacing Arduino hardware with MATLAB Simulink.

Figure 49. Transmitter and receiver of PLC.

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

Figure 50. Hardware of VLC transmitter circuit using four LEDs.

Figure 51. Hardware of VLC receiver circuit using single LDR.

75

76

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 52. (a) The generated signal taken from output of pin 9 of Arduino (b) FSK modulated signal from output of PLC modem (c) FSK demodulated signal at PLC receiver (d) Received signal from LDR output.

BER Investigation for an OFDM-Based Hybrid PLC-VLC System

77

4.6. Conclusion The BER performance of an HPMV system using OFDM modulation technique has been investigated in this chapter for configurations A and B using LOS as well as LOS plus first reflection signal. The simulation was run for three different scenarios in both configurations, with LED separations of 0.2m, 0.6m, and 1m, respectively. The frequency response of the MIMO-VLC system supports the BER results in all three scenarios. For both configurations A and B, the results show that the BER decreases as the separation between the LEDs increases. There is an optimal separation between individual LEDs beyond which BER does not decrease, and this optimal separation is determined by the room’s dimensions. The decrease in BER with increasing LED separation is significant for the LOS signal but less so for the L-R1 signal. Overall, the study shows that while OFDM can mitigate the negative effects of multipath and selective fading, if optimal separation between LEDs is not chosen, fading can cause more BER at the receiver side. In the BER performance of an HPMV system, frequency selectivity is more detrimental than attenuation factor. As a result, the optimal separation between LEDs should be chosen based on the room size to avoid fading and multipath effects in an HPMV system using OFDM technology. Nonetheless, the OFDM degrades the spectral efficiency. In the next chapter we investigate the BER performance using a more spectral efficient multiple access technique i.e., NOMA.

Chapter 5

Downlink Multipath Multi-User NOMA-VLC System Performance Analysis 5.1. Introduction As the demand for data-driven applications grows, visible light communication (VLC) has emerged as a promising supplement to the traditional radio frequency (RF) communication system. The VLC system has the advantage of both illumination and communication. It also has numerous advantages over RF systems, such as unregulated license-free spectrum, low cost, less power requirement, long life of light emitting diode (LED), and so on [57-59]. However, the limited modulation bandwidth of LED limits the VLC system’s achievable capacity [60]. Conventional multiple access technologies such as frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA) have been used in this context [61]. Wavelength-Division Multiple-Access (WDMA) has been used to support a larger number of users while maintaining high data transmission rates. Each user is assigned a unique wavelength [6263]. Nevertheless, the above-mentioned techniques cannot achieve sufficient resource reuse. Because of its frequency selective channel and ability to reject intersymbol interference, orthogonal frequency division multiple access (OFDMA) allows resources to be reused at the subcarrier level and has been widely investigated in the VLC system [64]. Nonetheless, orthogonal multiple access techniques degrade spectral efficiency and bit error rate performance. Non-orthogonal multiple access (NOMA) has recently received a lot of attention as a promising multiple access scheme for 5th generation (5G) wireless networks [65-66]. NOMA employs superposition coding on the transmit side to facilitate simultaneous signal transmission to multiple users from a single source. Signals intended for different users are multiplexed by assigning inverse power levels based on channel gains. According to Z. Ding et al. [67], the NOMA performs significantly better in high signal-to-noise ratio (SNR) conditions. As a result, an indoor VLC system with a high SNR due to the short distance between the LEDs and the photo detector (PD) can

80

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

use the NOMA in the downlink VLC system. However, because the received signal is a superposition of multiple users’ signals, SIC techniques cannot completely eliminate interference, resulting in error propagation. M. Obeed et al. [68] proposed and evaluated a new scheme that combines the benefits of cooperative non-orthogonal multiple-access (Co-NOMA), hybrid VLC/RF, and energy harvesting techniques to address such issues. A. Alqahtani et al. [69] conducted a performance analysis for a hybrid indoor-outdoor NOMA user, where both users can be served via outdoor-to-indoor and outdoor-tooutdoor channels. These research articles, however, did not investigate the bit error rate (BER), which is a metric for evaluating wireless communication systems. Several pioneering works on the BER performance analysis for indoor VLC systems have been completed. V. Dixit et al. [70] proposed a dynamic FOV strategy for a two-user MIMO-NOMA-VLC system with 2×2 users. They examined different FOV values in the BER analysis, including 600,400,300, and 200, and discovered improvements in performance. They also demonstrated that a 2×2 NOMA-MIMO-VLC system outperforms a SISONOMA-VLC system. This type of dynamic FOV strategy can be useful in LIFi, airports, and railway stations, among other places. B. Lin et al. [71] proposed a single carrier-based NOMA scheme for the VLC system and discovered an optimum power allocation ratio (PAR) of 0.29 for best BER performance. Chen et al. [72] proposed a downlink VLC system with a flexible-rate SIC-free NOMA technique. They discovered that for high power allocation range (PAR), the near user can achieve nearly the same BER performance using the proposed constellation portioning coding (CPC) method. H. Li et al. [73] proposed a NOMA hierarchical pre-distorted layer asymmetrically clipped optical OFDM (HPD-LACO-OFDM) scheme. They demonstrated experimentally that the proposed method outperforms dc-biased optical OFDM (DCO-OFDM) and asymmetrically clipped optical OFDM (ACO-OFDM) in terms of BER performance for the same signal power. X. Liu et al. [74] obtained the closed form expression of BER and discovered that as modulation order increases, the difference in BER performance among users decreases at the expense of higher power consumption. H. Marshoud et al. [75] proposed a novel gain ratio power allocation (GPRA) scheme that outperforms the conventional static power allocation method in terms of BER performance. They also demonstrated that using the GPRA method to tune both the semi-half power angle of the LED and the FOV of the receivers provides a higher sum rate than using a fixed semi-angle of the LED and a fixed FOV of the receiver. V. Dixit et al. [76] studied the BER performance

Downlink Multipath Multi-User NOMA-VLC System Performance …

81

of a two-user downlink OOK and L-PPM modulated NOMA-VLC system with perfect and imperfect CSI. They simulated BER for various power allocation coefficient, FOV, and receiver position values. They discovered that the error probability for L-PPM (L>2) modulation is lower than that of OOK modulation, and that the BER performance of near and far users improves as the FOV decreases. They also discovered that channel estimation error has a negative impact on higher order L-PPM modulation in the NOMA-VLC system. Similarly, Prakriti Saxena et al. [77] developed a mathematical expression for BER and sum rate for two randomly positioned users and investigated the probability of error with different FOV and power allocation coefficient values. Nonetheless, the NOMA-based BER performance has been investigated in all of the preceding works using only line-of-sight (LOS) signals. H. Hesham et al. [78] proposed a hybrid technique that combines NOMA, asymmetrically-clipped optical (ACO), and filter bank multicarrier (FBMC) to reduce the number of unserved (blocked) users. They demonstrated that the power allocation coefficient is 0.2 to achieve a threshold BER of 10-3. The authors also proposed positioning, clustering, and resource allocation algorithms and discovered that the NOMA-FBMC technique outperforms the NOMA-OFDM scheme by 1.8 times. V. Dixit et al. [79] used a first order reflection signal to investigate the BER performance of a NOMA-VLC system for different values of FOV, PD responsiveness, wall reflection coefficient, and receiver location (blocked LOS). They discovered that when the reflection coefficient of the wall and the responsiveness of the PD are reduced, the BER decreases. Furthermore, as the FOV of the receivers increases, so does the BER. The investigation was conducted using a single-user system with either LOS or first reflection signal, but in practice, there are multi-users who receive LOS as well as reflected signal. There are some cases where the BER performance varies with different signal components. There may be various cases in a practical indoor VLC system, such as LOS blockage, tilted receiver, and so on, where the light signal is not captured via the LOS path. Furthermore, the LOS component may be comparable or even smaller than the NLOS component in cases where the horizontal separation between LED and PD is large enough that not all incident light rays are received by the receivers’ limited FOV. Besides, when the LED is in the centre and the PD is on either the centre or the wall side of the room, the received power from the second reflection is greater than the received power from the first reflection. As a result, it is necessary to investigate the BER performance of a multipath multiuser NOMA-VLC system under various scenarios.

82

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

In this chapter, we have reported BER performance of downlink NOMAVLC system with the following considerations: i. Barry channel models for VLC system. ii. Line-of-sight (LOS) and LOS plus first reflection (L-R1) signals for NOMA-VLC system. iii. Scenario with blocked LOS. iv. Two-user and three-user system. v. The semi-angles of the LED are 200, 300, 400, 500, 600 & 700. Based on the above considerations, following are the main contributions of this chapter. 

  

The BER performance of NOMA-VLC has been investigated with single LED for two-user & three-user system by changing the semiangle of the LED using LOS and L-R1 signals. The sum rate vs. offset has been examined for different number of users using LOS and blocked LOS signal. The optimal power allocation factor of far user has been investigated for two-user system. Also, the outage probability has been simulated for blocked signal.

The proposed research can be useful in following ways:  

  

The research work can be useful in enabling technology for massive connectivity and coverage enhancement in 5G. Moreover, if LED with proper semi-angle LED is chosen for indoor NOMA-VLC system then users with poor channel conditions can utilize higher transmit power due to severe intercluster interference. The investigation on choosing power allocation factor can maximize the sum rate of all the users. The NLOS signal based NOMA-VLC system can be useful in situation with blocked path between transmitter and receiver. The proposed research work can lead to spatial reuse of frequency resources in better way.

Downlink Multipath Multi-User NOMA-VLC System Performance …

83

5.2. NOMA-VLC Multipath Downlink Model The model of an indoor multipath NOMA-VLC system has been described in this section in a room size of 5m×5m×3m, as shown in Figure 53. For the sake of simplicity and generality, a single LED located in the centre of the ceiling and k users were considered. In this chapter, simulations of two-user (k = 2) and three-user (k = 3) systems are presented. Table 8 shows the coordinates of LED and users-1, user-2, and user-3. Because the users are 0.85m above the ground, the vertical distance between the LED and the receiving plane is 2.15m. Each user’s photodetector is oriented vertically upward. The light signal from the source travels to the receiver via the LOS and reflected paths. In our case, we considered first order reflection signal as well as LOS path signal. The LOS signals emitted at angle 𝜑𝑜 from the LED with respect to the normal axis (unit normal vector (𝑛 ̂))) reach the PD via paths d1 and d2 for 𝑠 user-1 and user-k, respectively. At user-1 and user-k, the LOS signal is incident at angles 𝜓𝑜,1 and 𝜓𝑜,𝑘 . The NLOS signal radiated by angle 𝜑1 is incident at the lth grid of the wall at angle 𝛼 and then reflected by angle 𝛽 with respect to the normal axis of the wall (unit normal vector (𝑛 ̂)). The reflected 𝑡 light ray is incident at angle 𝜓1,1 and 𝜓1,𝑘 on the user-1 and user-k w.r.t to normal axis of receiver (unit normal vector 𝑛 ̂). 𝑟 The LOS along with first reflection (L-R1) signal for user-1 and user-k is shown by the paths d3-d4 and d3-d5 respectively. Since, the LED is assumed at the center of ceiling so maximum coverage radius of the light beam of the LED is L = 2.5m. The LED is assumed to have Lambertian radiation, which is a cosine function, as shown below: 𝐼(𝜑) =

𝑚+1 2𝜋

𝑐𝑜𝑠 𝑚 (𝜑)

(5.1)

Where 𝜑 is the irradiance angle with respect to the normal axis, and m is the Lambertian order, which is a function of the LED’s semi-half angle φ1/2 and is given as m = −ln⁡(2)⁄ln⁡(cosφ1/2 ). The receiving side wireless optical channel gain is inversely proportional to the square of the distance between the optical source and the PD and directly proportional to the PD’s effective collection area. On the receiving side, the PD is represented by an active area 𝐴𝑟 that collects the light signal. For the kth user, the LOS and non-line-of-sight (first reflection) channel gains are denoted by ℎ𝑘𝑜 and ℎ1𝑘 , which are expressed as

84

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 53. Indoor downlink multipath multi-user NOMA-VLC model.

1

ℎ𝑘𝑜

𝑑22

𝑔(𝜓𝑜,𝑘 )𝐴𝑟 𝐼(𝜑𝑜,𝑘 )

= {𝑐𝑜𝑠 𝑚 (𝜑

𝑜,𝑘 ) cos(𝜓𝑜,𝑘 ) ,

0,

0 < 𝜓𝑜,𝑘 < 𝜓𝐹𝑂𝑉 𝜓𝐿,𝑘 ≥ 0

(5.2)

1 𝑔(𝜓1,𝑘 )𝜇𝐴𝑙 𝐴𝑟 𝐼(𝜑1,𝑘 ) 𝜋𝑑32 𝑑52 1 ℎ𝑘 = 𝑐𝑜𝑠(𝜑1,𝑘 ) 𝑐𝑜𝑠(𝛼) 𝑐𝑜𝑠(𝛽) 𝑐𝑜𝑠(𝜓1,𝑘 ), 0 < 𝜓1,𝑘 < 𝜓𝐹𝑂𝑉 0, 𝜓1,𝑘 ≥ 𝜓𝐹𝑂𝑉 { (5.3) where 𝐴𝑟 is the active area of the PD, 𝐴𝑙 is the small reflective area of lth grid, 𝜇 is the reflection coefficient of wall, 𝛼 and 𝛽 are the incidence and reflected angles from the grid of the wall, 𝜓𝐹𝑂𝑉 is the field of view of the concentrator. Light rays that are incident within the receiver’s field of view will be captured. The factor 𝑔(𝜓) is the gain of optical concentrator given by

Downlink Multipath Multi-User NOMA-VLC System Performance … 𝑛2

⁡𝑔(𝜓) = {𝑠𝑖𝑛2 (𝜓𝐹𝑂𝑉 ) 0,

, 0 ≤ 𝜓 ≤ 𝜓𝐹𝑂𝑉

85

(5.4)

𝜓 ≥ 𝜓𝐹𝑂𝑉

where 𝑛 is the refractive index of concentrator. Here 𝜓 can be either 𝜓𝑜,𝑘 or 𝜓1,𝑘 .

5.3. NOMA-VLC System Principle In NOMA, the bipolar message signal for the different users are superposed in the power domain and a DC bias is added at the LED as given by [80]: 𝑥 = ∑𝐾 𝑖=1 𝛼𝑖 √𝑃𝑒𝑙𝑒𝑐 𝑆𝑖 + 𝐼𝐷𝐶

(5.5)

where 𝑃𝑒𝑙𝑒𝑐 is the total electrical power of all the message signals, 𝐼𝐷𝐶 is the DC bias added to the LED before the signal transmission, 𝑆𝑖 and 𝛼𝑖 are the modulated message signal and power allocation coefficient for the ith user (i = 1, 2,…k….K). Two criteria must be met in NOMA: more signal power should be allocated to the user despite poorer channel quality, and power allocation coefficients must satisfy the total power constraint. This implies that 2 𝛼1 ≥ ·∙∙ 𝛼𝑘 ≥∙∙∙≥ 𝛼𝐾 and ∑𝐾 𝑖=1 𝛼𝑖 = 1 are true. The received signal is given to th the k user by [80]: 𝑘−1 ∑𝐾 𝑦𝑘 = √𝑃𝑒𝑙𝑒𝑐 ℎ𝑘 (∑ 𝑘 𝑠𝑘 + ⏟ 𝑖=𝑘+1 𝛼𝑖 𝑠𝑖 ) + 𝑛𝑘 ⏟𝑖=1 𝛼𝑗 𝑠𝑗 + 𝛼⏟ SIC

signal

(5.6)

interference

In the above equation, the first term is SIC, the second term is the signal intended for the kth user, the third term is interference, and the fourth term (𝑛𝑘 ) is noise. SIC is performed at the kth user to remove the message signal for other users with poorer channel conditions, i.e., SIC term. The message signal for users with higher channel gains than the kth user is treated as noise, i.e., an 2 interference term. The noise has a zero mean and variance ⁡𝜎𝑛𝑜𝑖𝑠𝑒 , which includes shot noise and thermal noise, as shown below:

86

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai 2 2 2 ⁡𝜎𝑛𝑜𝑖𝑠𝑒 = 𝜎𝑠ℎ𝑜𝑡 + 𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 = (2𝑞𝐼𝑏𝑔 𝐼2 𝐵 + 2𝑞𝑅𝑃𝑟 𝐵ℎ) + 8𝜋𝐾𝑇

(

𝐺

𝜂𝐴𝑟 𝐼2 𝐵2 +

16𝜋2 𝐾𝑇Γ 2 2 𝜂 𝐴𝑟 𝐼3 𝐵3 ) 𝑔𝑚

(5.7)

where q is electronic charge, R-responsitivity of the receiver, B-equivalent noise bandwidth which is equal to the modulation bandwidth, 𝐼𝑏𝑔 -background noise, 𝑃𝑟 -transmitted optical power, h-channel gain, K-the Boltzmann constant, T-absolute temperature, G-open loop voltage, 𝜂-input capacitance of PD, G-open loop voltage gain, Γ-FET channel noise factor, 𝑔𝑚 -FET transconductance. The achievable data rate per bandwidth for the kth user after optical to electrical conversion is given as [80] 1 2

𝑅𝑘 = {

𝑙𝑜𝑔2 (1 + 1

(ℎ𝑘 𝛼𝑘 )2 1 𝐾 ∑𝑖=𝑘+1(ℎ𝑘 𝛼𝑖 )2 + 𝜌

𝑙𝑜𝑔2 (1 + 𝜌(ℎ𝑘 𝛼𝑘 )2 ), 2 where 𝜌 =

𝑃𝑒𝑙𝑒𝑐 𝑁0 𝐵

) , 𝑘 − 1, … . , 𝐾 − 1 (5.8) 𝑘=𝐾

represents the transmit SNR and scaling factor of ½ is due to

the Hermitian symmetry. Figure 54 depicts the schematic block diagram of the downlink NOMAVLC system. It is assumed that user-1,... user-k..., user-K are sorted in ascending order based on their channels, i.e., h1 ≤ h2 ≤ ··· ≤ hK. Users are served by the transmitter at the same time/code/frequency, but with different power levels P1, …Pk…, PK, according to the NOMA concept [25]. Based on the channel state information (CSI), the modulated bipolar message signals S1, S2 … Sk...SK are superposed in the power domain so that the message to the user with the weaker channel condition is allocated more transmission power, ensuring that this user can detect its message directly by treating the other user’s information as noise. Then, to obtain a unipolar signal for intensity modulation and direct detection (IM/DD), a fixed DC bias is added [25]. Before being captured by the photodetector, the signal is finally transmitted through the LED and AWGN is added to it in the indoor multipath channel. The received optical signal is converted into an electrical signal on the receiver side by a photodetector. Finally, after the DC term is removed, the SIC is performed at the kth user, so that the user with the stronger channel condition must first detect the message for its partner, then subtract this message from its observation, and finally decode its own information.

Downlink Multipath Multi-User NOMA-VLC System Performance …

87

Figure 54. Downlink k-user NOMA-VLC system schematic block diagram.

5.4. Results and Discussion The channel gain of each user determines the SIC-based decoding order between the multi-users of the LED. The channel gain can be stated more simply as 1

ℎ ∝ 𝑑(𝑚+3)𝑠𝑖𝑛2 (𝐹𝑂𝑉)⁡

(5.9)

The above Eqn. shows that the channel decoding order is determined by three factors: the Euclidean distance between the LED and the kth user, the FOV of the receiver, and the semi-angle of the LED. All simulations were carried out using the MATLAB (2019 version) software. Table 8 lists the parameters and their corresponding values used in simulation. Figure 55 shows the flow chart of BER and sum rate simulation using LOS and first reflection signal.

88

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Table 8. Parameters and their values Parameters

Value

Room

Size

5m× 5m × 3m

Reflection coefficient of walls

𝜇𝑁𝑜𝑟𝑡ℎ = 𝜇𝑆𝑜𝑢𝑡ℎ = 𝜇𝐸𝑎𝑠𝑡 = 𝜇𝑊𝑒𝑠𝑡

0.7

Number of grids on the ceiling Number of grids on floor Number of grids on wall Coordinate of LED

35×35 35×35 35×15 (0m, 0m, 3m)

Semi-angle at half power, φ1/2

20°, 30°, 40°, 50°, 60°, 70°

Source

Channel model

Barry channel model

Receiver

Height of receiver plane above the floor

0.85m

Active area (AR)

1Cm2

Coordinate of user-3

(0.5m, 0.5m, 0.85m)

Coordinate of user-2

(-1m, -1m, 0.85m)

Coordinate of user-1

(-1.5m, -1.5m, 0.85m)

Half-angle FOV

60°

Noise

Ambient light noise Spectral irradiance of ambient light noise Pbn Receiver preamplifier noise

5.8uw/cm2.nm

Electronic charge q

1.6×10-19

Noise bandwidth factors I2, I3

0.562, 0.0868

Background noise 𝐼𝑏𝑔

5100μA

Receiver area Ar

10-6 m2

Responsitivity of Photodiode R

0.5 A/W

Boltzman’s constant K

1.38×10-23 J/k

Absolute temperature T

300 K

FET channel noise factor Γ

0.82

Feedback resistor of FET based transimpedance amplifier Rf Transconductance gm

1.4 KΩ

Total Input capacitance η

2 pF

14 mS

Downlink Multipath Multi-User NOMA-VLC System Performance …

89

Figure 55. Flow chart for BER and sum rate simulation using LOS and L-R1 signal.

90

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

5.4.1. BER Performance Analysis In practice, the receiver captures not only LOS signals but also non-line-ofsight (NLOS) signals, and thus the BER performance using LOS as well as LOS along with first reflection (L-R1) signal has been analysed in this section. Figures 56a and 56b show the BER performance for a two-user and three-user system using a LOS signal versus the power spectral density of an AWGN. In the BER simulation, three different values of the semi-angle φ1/2 were considered: 300, 500, and 700. Figure 56a shows that as angle φ1/2 decreases, the error probability of users 1 and 2 increases. For example, at 70dB SNR, user-1’s BER is 7.49×10-3, 4.88×10-3, and 1.03×10-1 for φ1/2 = 700, 500, and 300, respectively. Similarly, when φ1/2 = 700, 500, and 300, the error probability is 4.89×10-2, 5.84×10-2, and 3.73×10-1, respectively. It can be seen that the best BER performance is obtained when φ1/2 equals 500 and 700 for user-1 and user2, respectively. However, the BER curves in Figure 56b show that the error probability for user-1 at φ1/2 of 300, 500, and 700, respectively, degrades to 8.31×10-3, 4.46×10-2, and 2.67×10-1. Similarly, using φ1/2 = 300, 500, and 700, the error probability for user-2 increases to 4.86×10-1, 3.02×10-1, and 2.59×10-1, respectively. Also, when φ1/2 is 300, 500, or 700, the error probability for user-3 is 4.86×10-1, 3.3×10-1, and 2.9×10-1, respectively. As a result, the optimal value of φ1/2 for user-1 is 300, while it is 700 for users 2 and 3. Thus, for a standard room size of 5m×5m×3m, there is an optimal value of semiangle of LED at which best BER best performance is achieved for multi-user NOMA-VLC system using LOS signal. However, the optimal value of the LED’s semi-angle can vary depending on the number of users, so proper investigation is required before implementing an indoor multi-user NOMAVLC system. The BER curves in Figure 56b also show that for three user system, successive interference cancellation becomes inefficient at large semiangles of the LED (φ1/2= 700). The performance of the far user is superior to that of the near user because the power allocation coefficient for the far user is greater in the NOMA system. The total transmitted power, on the other hand, remains constant, and thus the power allocation coefficient for each user decreases as the number of users increases. Furthermore, an increase in semiangle results in a larger optical attocell, which increases path loss [80]. Also, as the semi-angle of the LED increases, so does the channel correlation between multi-users (when the number of users increases from two to three). All of these factors contribute to poor successive interference cancellation for users as the semi-angle of the LED increases in tandem with the number of users. The saturation of BER graphs appears to occur abruptly when the semi-

Downlink Multipath Multi-User NOMA-VLC System Performance …

91

angle of the LED increases from 500 to 700, but this is not the case. Figure 56c depicts the gradual increase in the probability of error as the semi-angle increases from 500 with a 20 increment, i.e., 500, 520, 540, 560. It can be seen that for semi-angles greater than 540, the SIC method fails to decode the signals and there is saturation in the BER curves, implying that the BER does not decrease. Because the channel gain is of the order of 10-4, the electrical path loss at the receiver side is -80dB, resulting in an 80dB offset with respect to the received Es/No in Figure 56.

Figure 56. LOS BER performance of a NOMA-VLC system using different values of φ(1/2) for (a) a two-user system, (b) a three-user system, and (c) a gradual increase in BER with increasing LED semi-angle.

92

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 57. L-R1 BER performance of a NOMA-VLC system using various values of φ1/2 for (a) a two-user system and (b) a three-user system.

Similarly, the BER for user-2 is 4.90×10-2, 8.62×10-2, and 5.35×10-1 with φ1/2 of 700, 500, and 300, respectively. Thus, the optimal value of φ1/2 is 700, which gives the lowest bit error rate in a two-user system. Because the number of reflected light rays increases as φ1/2, the number of captured light rays through the receiver’s FOV increases as well. Nonetheless, as shown in Figure 57b, as the number of users increases to three using the L-R1 signal, the SIC method fails to decode the user’s intended signal. The reason for this is that channel correlation grows as the number of users grows. As a result, in order to accommodate more users without degrading performance, we must increase transmit power. Furthermore, when the number of users is large, a new problem known as SIC error propagation occurs. In a three-user system, for example, user-3 must perform SIC on the data of users 1 and 2. If user-1’s data is decoded incorrectly, a wrong signal is subtracted in the SIC process, resulting in incorrect decoding of user-2’s data, and this error propagates to the decoding of user-3’s data. In such a case, the concept of optical attocells, as proposed by H. Haas [81], can be used to make the channel gain uncorrelated. As a result, selecting the optimal value of the LED’s semi-angle depends not only on the number of users, but also on the reflected signals. To achieve optimal BER performance, factors such as indoor room scenarios as well as NOMAVLC system requirements, such as number of users, position of transmitters and receivers, semi-angle of the LEDs, and so on, must be considered.

Downlink Multipath Multi-User NOMA-VLC System Performance …

93

Figure 58. LOS BER vs. Power allocation coefficient (a) Two-user system (b) Three-user system (c) Two-user system at SNR of 123dB for α values ranging from 0.4 to 0.85 [79].

We have simulated the error probability Vs. power allocation coefficient (of far user) for two-user and three-user system with different values of semiangle i.e., 700, 500 & 300 using LOS signal as shown below in Figures 58a & b. The result shows that the BER performance for user-1 (far user) improves for higher values of 𝛼 (greater than 0.5). The reason behind the improved performance of far user is that as 𝛼 increases the power allocated to far user also increases. For user-2, the BER increases for region 0 < 𝛼 < 0.2 then remains constant up to α = 0.5 thereafter decreases for 0.5 < 𝛼 < 0.8 and then increases for 𝛼 > 0.8. It can be observed that when 𝛼 is less than 0.2, the BER for user-2 using φ1/2 of 300 is more than BER achieved using φ1/2 = 500 & 700. Also, the BER for 𝛼 > 0.5, the user-2 has less error probability when φ1/2 = 700 than the BER achieved using φ1/2 of 300 & 500. The result displays that the average (of both users) error probability is minimum at 𝛼 of

94

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

0.8. It means that if we want both users to fairly benefit from NOMA, the optimal power allocation coefficient is 0.8. For three-user system, it is apparent from Figure 58b that the user-1 exhibits the low bit error rate as compared to user-2 & user-3 using φ1/2 = 300, 500 and 700. Therefore, the result also justifies the outcome of Figure 56b in context that when number of users increases then SIC method fails to decode the higher order users. This is because as the number of users increases in the network then strongest user will be allocated less and less power. Figure 58c is representing 58a only with minor changes. The plot is shown for varying 𝛼 (0.4 to 0.85) at SNR of 123dB. In this plot, we have done simulation for additional semi-angle e.g., 600 to show comparison with previous research work [79]. From the plot it can be seen that when 𝛼 is increasing, the BER for user-2 (near user) is increasing and for user-1 (far user), the BER is decreasing. The simulated BER graph for far user and near user using 600 semi-angle value matches with the results demonstrated by V. Dixit et al. [79].

5.4.2. Sum Rate Analysis Figure 59 depicts the simulated sum rate vs. normalised offset (l/L) curves for various LED semi-angle values using a LOS signal. The normalised offset of user-1 with respect to user-2 is defined as l/L, where L is the maximum coverage range within the room. The sum rate vs. normalised offset relationship has been simulated for various LED semi-angle values (200, 300, 400, 500, 600, and 700). The result shows that as user-1 moves closer to the edge of the room, the sum rate decreases. The decrease in sum rate is not clearly visible in Figure 59 (curves appear straight horizontal line), so it has been shown in the inset diagram with only three curves corresponding to 500, 600, and 700. The result shows that as the semi-angle increases, the sum rate increases and reaches its maximum value before decreasing as the semi-angle increases further. As shown in Figure 60, the average value of each curve for different offset values has been calculated and plotted against semi-angles. It can be observed that the peak average sum rate of 278.84 Mbps is achieved at φ1/2 of 50°. Thus φ1/2 = 50° is the optimal value of semi-angle of the LED for two user system at which maximum sum rate is achieved in the room whose dimension is 5m×5m×3m. The optimal value of φ1/2 can be different for different number of users (discussed previously). Consequently, it can be noted that the semi-angle of the LED can’t be increased arbitrarily to any value in order to improve BER.

Downlink Multipath Multi-User NOMA-VLC System Performance …

95

Figure 59. Sum rate vs. normalized offset using LOS signal.

Because reflected signals were considered in the previous discussion for BER performance analysis, the sum rate vs. offset has been simulated using the first reflection signal. Figure 61 depicts the sum rate versus offset for a blocked LOS signal (first order reflection signal). The result shows the opposite trend as shown in Figure 59, namely that the sum rate increases as the offset increases. This is for the reason that user-1 receives more optical power as it moves closer to the edge of the wall, because the wall acts as a light source (reflected light) for the user. The result of Figure 62 shows that the average sum rate increases as the LED’s semi-angle increases. Because a greater number of light rays are incident on the walls when the LED’s semiangle is increased, the number of reflected light rays captured through the FOV of the receiver is also increased. Furthermore, the simulation curves show that the sum rate is nearly zero for φ1/2 = 200 and 300. Because the emitted light rays from the LED are confined in a very narrow beam, no light rays are incident on the surface of the walls, leaving users in the dark. As can be seen, the multipath reflected signal as well as the semi-angle of the LED play an important role in the multi-user NOMA-VLC system.

96

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 60. Avg. sum rate vs. semi-angle of LED.

Figure 61. Sum rate vs. normalized offset using blocked-LOS signal.

Downlink Multipath Multi-User NOMA-VLC System Performance …

97

Figure 62. Avg. sum rate vs. semi-angle for blocked LOS.

Figure 63. Sum rate Vs. SNR (dB) for two-user & three-user system using different values of φ1/2.

98

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 63 depicts the sum rate versus SNR for two and three-user systems with different values of φ1/2. The results show that as the semi-angle of the LED increases from 300 to 700, so does the sum rate of the two-user and threesystem. Each user’s PD receives enough optical power by increasing the LED’s semi-angle. It is also observed that the achieved sum rate for a threeuser system is greater than the sum rate for a two-user system. Because NOMA provides greater capacity than orthogonal multiple access, the sum rate increases as the number of users increases (OMA). However, this only works if the interference levels are manageable and the strongest user is given a respectable amount of power (although strongest user is allocated least power). It has been discovered that as the number of users in a NOMA network increases, the network’s total capacity initially increases, then drops and saturates. The drop off point can be thought of as the maximum number of users that can be supported in the network without causing performance degradation. However, in order to accommodate a greater number of users without degrading performance, transmitted power must also be increased. For better comparison and clarity, the sum rate simulation for two-user and threeuser systems has been shown separately as well as in a common figure.

5.4.3. Outage Performance Analysis The essence of shadowing is incomplete without a discussion of outage probability, so the outage performance analysis is presented in this section. Figure 64 depicts the outage probability for a blocked LOS scenario versus the far user’s target rate. When the target rate of the far user increases, the outage probability increases with a decrease in the semi-angle of the LED. For semiangle values of 300, the outage probability (for target rates greater than 3bps/Hz) reaches 100% for both users, i.e., the coverage area becomes 0%. This is because the signal reduction caused by shadowing is greater than the margin when the semi-angle decreases. This result backs up the results discussed in the previous section for sum rates of 200 and 300 values of semiangle. Table 9 compares the pros and cons, as well as the future scope of our proposed work, to previously published research works. Table 10 compares our proposed simulated results to those of other research articles.

Room Size 4m×4m×3m

5m×5m×3m

4m×4m×3m

5m×5m×3m

References

H Marshoud et al. [75]

V. Dixit et al. [76]

X. Liu et al. [74]

Proposed work

Two-user system & Three-user system

Two-user system

Two-user system

Three-user system

No of users

OOK

M-PSK, MPAM, M-QAM

OOK

Modulation technique OOK

LOS+NLOS

LOS only

LOS only

LOS/ NLOS LOS only

Semi-angle of LED. Power allocation coefficient of far user. Position of far user w.r.t to near user. SNR at receiver. Target rate of far user.

Different order of modulation

FOV

Performance analysis by changing FOV

Not found optimal α at which lowest average BER is achieved. Outage or coverage probability not discussed. Not shown the effect of increase in number of users on the performance. Not analysed the performance using multipath signal. Not discussed the effect of number of users on user’s performance. Not shown the effect of offset. Not shown the coverage probability. Not discussed the impact of multipath signal. The work can be extended to MIMONOMA-VLC system. The impact of higher order i.e., second order reflection signal can be investigated.

Shown only BER results. Not shown the impact of number of users on the performance. Not shown the effect of multipath signal.

Limitations/Future scope

Table 9. Comparison of the benefits and drawbacks of the proposed work with various research articles

Normalized offset Proposed work Obeed et al. [82]

10-1 10-2 10-3 10-4

Avg. BER

Semi-angle of the LED Modulation technique Number of users BER 10-1 10-2 10-3 10-4 OOK

OOK

Two Three Two Two Near User Far User Near User Far User Near User Far User Near User Far User 60 68 105 97 128 124 105 90 69 75 112 110 131 127 110 95 72 77 117 115 132 129 112 97 74 79 118 117 133 131 115 99 Comparative analysis of BER Vs. power allocation factor for simulated result of Figure 58a Power allocation factor for far Power allocation factor for Power allocation factor for far user Power allocation factor for far user user in proposed work (Figure far user [75] (Figure 3c) [76] [74] 58a) 0.55 0.15 0.76 0.01 0.65 0.25 0.92 0.05 0.70 0.27 0.95 0.08 0.75 0.30 0.99 0.14 Comparative analysis of Sum rate (Mbps) Vs. Normalized offset for Figure 59 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 274.15 273.94 274.02 274.01 272.83 271.72 271.1 269.5 267.4 265.8 252 254 253 252 250 245 240 230 220 190

OOK

BPSK

φ1/2 = 50°

φ1/2 = 50°

φ1/2 = 50°

φ1/2 = 60°

Required SNR (dB) [74] (Figure 9)

Comparative analysis of BER Vs. SNR for simulated result of Figure 56a Required SNR (dB) in proposed Required SNR (dB) [75] Required SNR (dB) [76] work (Figure 56a) (Figure 4) (Figure 3)

Table 10. Comparative analysis of simulated results with previous research work

Downlink Multipath Multi-User NOMA-VLC System Performance …

101

Figure 64. Outage probability of blocked LOS scenario.

5.5. Conclusion In this chapter, the BER and sum rate performance of NOMA-VLC using a single LED, two-user, and three-user systems for LOS and L-R1 signals were investigated using different LED semi-angle values. The results show that at optimal semi-angle, the BER as well as the sum rate of LOS and reflected signals perform best. However, the optimal LED semi-angle value varies depending on the number of users and the order of signal reflection. The outage probability for a blocked LOS signal displays that the outage probability for a small angle of LED reaches 100%. Overall, the results show that the semi-angle of the LED, in addition to the multipath signal, plays an important role in the multi-user NOMA-VLC system. As a result, proper research must be conducted prior to implementing a downlink multipath multiuser NOMA-VLC system. So far, we have investigated the LED effect from the perspective of room configuration. However, there is another factor that influences performance of VLC system. This factor is LEDs’ limited linear operating region, which will be discussed in the next chapter.

Chapter 6

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System Using a Union of Precoder and Companding 6.1. Introduction In the last two years, there has been an unprecedented increase in the use of high-speed indoor data communication networks for various services such as work from home (WFH), online education, internet of things (IoT), cloud storage, and so on, as well as ubiquitous internet connectivity, resulting in a massive increase in global data traffic. The current radio frequency (RF) bandwidth scheme has significant difficulty dealing with the required spectrum. The availability of vast unregulated visible spectrum necessitates a paradigm shift in the existing wireless system toward mobile internet connectivity via lighting infrastructure. With the exception of serving illumination, visible light communication (VLC) uses the luminescence of LEDs to achieve high transmission capacity, particularly in indoor wireless networks. The VLC contributes to the green aspect of next-generation 5G/6G wireless networks with energy-efficient data transmission due to the vast unlicensed spectrum, robust security, low cost, low latency, immunity to electromagnetic interference (EMI), and so on [83]. In addition to providing low-cost illumination, LED-based VLC systems have recently been used to provide extended radio services and medical applications [84, 85]. The low 3dB bandwidth (~20 MHz) of LEDs used as transmitters limits the achievable system capacity in an optical wireless system. Because of its numerous benefits, orthogonal frequency division multiplexing (OFDM) modulation has been the industry standard for increasing system capacity over the last decade. Because of its scalability with multiple input multiple output (MIMO) technologies, OFDM is expected to play a critical role in the development of 5G communication networks [86]. Because of its lower latency and higher spectrum efficiency, nonorthogonal multiple access (NOMA) is a viable approach for increasing system capacity in the upcoming fifth-generation (5G) VLC network [87].

104

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Users in NOMA, as opposed to orthogonal frequency division multiplexing access (OFDMA), have simultaneous access to both frequency and time and are distinguished by their power domains [88]. NOMA combined with the OFDM modulation format is expected to provide low latency, improved spectral efficiency, and wider cell coverage [89]. Several efforts have been made to implement OFDM schemes in order to increase the capacity of the NOMA VLC system [90, 91]. In these works, the DCO-OFDM method is used to improve the user holding range in a NOMA VLC system. The accumulation of highly coherent data symbols, on the other hand, frequently raises the peak magnitude of the output signal during the inverse fast Fourier transform (IFFT). The peak-to-average power ratio (PAPR) is used to quantify this phenomenon. Because of the nonlinear voltage-current (V-I) relationship of the LED used as a transmitter, a high PAPR OFDM signal is undesirable in an optical communication system. The main disadvantages of a high PAPR system are nonlinear distortion and a low bit error rate (BER). To compensate for nonlinear distortion, the quiescent point in the linear zone of the power amplifier must be backed off. It reduces the signal-to-noise ratio (SNR) at the receiver and the amplifier’s efficiency. Using a power amplifier with a wide dynamic range, on the other hand, raises the overall system cost. The majority of recent research contributions pertaining to the implementation of a precoder exist for RF-based NOMA systems [92-97]. In this paper, the precoder proposed by S.B. Slimane in [98] is used in conjunction with µ-law compander because of its several advantages, including low complexity, downward compatibility, and ease of symbol detection at the receiver, among others, to reduce PAPR and nonlinear characteristics in NOMA-based OFDM-VLC systems for 5G applications. The precoding matrix must be orthogonal and meet the symbol separability condition given by Eqs. (21) and (23) in [98] to maintain the symbol detectability property. The orthogonality property of the precoding matrix used in [98] has been verified while the symbol detectability condition has been maintained. The approach for PAPR reduction and non-linearity mitigation in DCO-OFDM and ADO-OFDM systems using the orthogonal multiplexing access (OMA) scheme has been successfully implemented [99, 100]. The NOMA protocol was used to multiplex two users using a single base station. The DCO-OFDM method, which is widely used in visible light communication to obtain non-negative signals, has been used for unipolar signal transmission [101]. It is widely acknowledged, however, that DCOOFDM does not provide the full LED dimming range for illumination and energy savings. To keep the effective LED brightness of fast O-OFDM signals

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

105

in sync with the industry-standard pulse-width modulation (PWM) light dimming technique for reliable message transmission, approaches such as reverse polarity O-OFDM (RPO-OFDM) and variable M-QAM OFDM have been discussed in [102,103]. The manuscript’s main focus, however, is on non-linearity reduction and improving BER performance. The following are the proposed approach’s major contributions:  



A conventional NOMA DCO-OFDM system has a PAPR of 13.3 dB. The proposed scheme significantly reduces the PAPR to 4.3dB. For a reference BER of 10-4, the proposed hybrid scheme has an EVM of only 12.3 dB, compared to 14 dB, 15 dB, and 19dB for simple NOMA DCO-OFDM, precoded NOMA DCO-OFDM, and combined NOMA DCOOFDM, respectively. To reduce non-linearity in a system, the system’s input back-off (IBO) power must be increased. The increased IBO power reduces the system’s power efficiency. The proposed NOMA DCOOFDM system has the lowest EVM (%) for an IBO of 0-3dB when compared to the precoded and combined NOMA DCO-OFDM systems.

6.2. Proposed NOMA DCO-OFDM Model This study considers the multiplexing of two users on a single base station. In practice, the NOMA application considers two users to reduce both detection delay and computational complexity at the receiver [103]. Figure 65 depicts the proposed NOMA O-OFDM system implementation. The modulated data of users 1 (U1) and 2 (U2) are individually preprocessed in the transmitter section of the proposed hybrid NOMA scheme using a precoding matrix (PM), as shown in the figure. The superposition coding (SC) technique is used to transmit both users’ precoded information at the same time, and power allocation (PA) is performed based on the channel state information (CSI) of U1 and U2 [104]. To obtain a real-valued timedomain signal, the Hermitian symmetry (HS) operation is performed before the inverse fast Fourier transform (IFFT). To generate a non-negative real signal, direct current (DC) bias clipping is applied to the IFFT output. Before passing the time domain signal through the OSRAM SFH 4230, a high power LED used for message broadcast, the µ-law compander is applied. Through the process of photodetection, the photodiode (PD) used as the receiver converts the optical signal to an electrical signal. After passing

106

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

through the inverse µ-law compander, the received electrical signal is deprecoded. The principle of successive interference cancellation (SIC) is applied to the composite signal to separate the information of individual users. Finally, the demodulator recovers the transmitted data from U1 and U2.

Figure 65. Transceiver of the proposed NOMA DCO-OFDM system.

6.3. PAPR Computation Let X1 and X2 denote the modulated symbols of user-1 (U1) and user-2 (U2). The modulated data are pre-processed in the frequency domain with a predetermined matrix PM of order R×Q, as defined by [97]. 𝑝0,0 𝑃𝑀 = [ ⋮ 𝑝𝑅−1,0

⋯ 𝑝0,𝑁−1 ⋱ ⋮ ] ⋯ 𝑝𝑅−1,𝑄−1

(6.1)

where R is N+NP, Q is the number of base-band modulated data sub-carriers, NP are nulls carriers used with 0≤ NP < N. The precoded output signals are given as 𝑌1 = 𝑃𝑀 𝑋1

(6.2)

𝑌2 = 𝑃𝑀 𝑋2

(6.3)

The unit element of precoder matrix is defined as [97] 𝑚

𝑃𝑚,𝑘 = 𝑝𝑚,0 𝑒 −2𝜋 𝑁

(6.4)

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

107

where pm,0 is given by (−1)𝑚 √𝑁

𝜋𝑚 sin (2𝑁 ), 𝑝

(−1)𝑚

𝑝𝑚,0 =

√𝑁 (−1)𝑚

{

√𝑁

,

sin (𝜋(𝑚−𝑁) ), 2𝑁𝑝

0 ≤ 𝑚 < 𝑁𝑝 𝑁 ⁡⁡⁡⁡⁡⁡⁡⁡⁡ 𝑝 ≤ 𝑚 < 𝑁 𝑁𝑝 ≤ 𝑚 < 𝑁

(6.5)

where m ∈ {0, 1, 2, ...R−1} and k ∈ {0, 1, 2, .....N −1}. Assuming that channel state information of U1 is superior to U2, a fixed power NOMA approach is employed without losing generality. The precoded NOMA signal S[k]PNOMA obtained as a result of superposition coding is expressed as [105] 𝑆[𝑘]𝑃𝑁𝑂𝑀𝐴 = √𝛼𝑃𝑡𝑜𝑡 𝑌1 + √(1 − 𝛼)𝑃𝑡𝑜𝑡 𝑌2

(6.6)

where 𝛼 is the power allocation factor and 𝑃𝑡𝑜𝑡 is the total power of Y1 and Y2. The Hermitian symmetry (HS) yields an output vector that is double the length of original input sequence [98] 𝑆[𝑘] = [𝑆0 ⋯ 𝑆𝑁 ⋯ 𝑆2𝑁−𝑘 ⋯ 𝑆2𝑁−1 ]𝑇

(6.7)

𝑆2𝑁−𝑘 = 𝑆𝑘∗

(6.8)

and

𝑆2𝑁−𝑘 = 𝑆𝑘∗ is the complex conjugate of the vector 𝑆𝑘 . The inverse fast Fourier transform (IFFT) followed by DC bias clipping yields a real-valued time-domain precoded NOMA-DCO signal, which is shown in [98] 𝑌(𝑛) =

1 √2𝑁

(2𝑅𝑒 ∑𝑁−1 𝑖=0 𝑆𝑖 𝑒

𝑗2𝜋𝑖𝑛 𝑁

)

(6.9)

The µ-law companding is applied to the time-domain precoded NOMA signal. The recommended transmitted signal is given as a result of the compander output 𝑆(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑 =

𝑣𝑙𝑜𝑔(

1+𝜇|𝑌(𝑛)| )𝑠𝑔𝑛(𝑌(𝑛)) 𝑣

log⁡(1+𝜇)

(6.10)

108

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

where µ denotes the compression control factor, which ranges from 0-255, V denotes the maximum amplitude, and sgn denotes the signum function. The PAPR of the proposed transmitted time domain signal is defined as the maximum power divided by the mean power. It is calculated mathematically as PAPR(dB)= 10⁡𝑙𝑜𝑔10

|𝑆(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑,𝑚𝑎𝑥 |

2

2

𝐸|𝑆(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑 |

(6.11)

2

2

where |𝑆(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑,𝑚𝑎𝑥 | denotes maximum power and 𝐸|𝑆(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑 | represents the mean power of the transmitted signal.

6.4. Proposed Receiver The base station (BS) of the NOMA network receives the signal transmitted by each mobile user. Under the assumption that the transmitter and receiver are reciprocal, the received signal is expressed as 𝑅(𝑛) = ∑𝐿𝑘=1 ℎ𝑘 √𝛼𝑃𝑡𝑜𝑡 𝑆𝑘 (𝑛) + 𝑛

(6.12)

L defines the overall number of users, hk relates to the kth user’s channel coefficient, Ptot indicates total common usage power, n is the normally distributed AWGN with 0 mean and standard deviation σ: 𝑛~𝒩(0, 𝜎 2 ) As illustrated in Figure 65, the inverse µ-law is applied to the received user signal, followed by the FFT operation and signal de-precoding. The signal obtained from the inverse µ-law operation is written as 𝑣

𝑅(𝑛)𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑 (1+𝜇)

𝑅 ′ (𝑛) = 𝜇 𝑒𝑥𝑝 (|

𝑣

| − 1)

(6.13)

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

109

6.4.1. Theoretical Assessment of the NOMA DCO-OFDM System Proposed The high PAPR in the NOMA DCO-OFDM system causes clipping distortion, which reduces the system’s net throughput. The proposed scheme employs a precoder in conjunction with a µ-law compander to improve transmission performance. A predetermined matrix is used in the precoding matrix (PM) approach to pre-process each user’s modulated data in the frequency domain. The modulated data of each user is multiplied by a preset matrix PM in the precoding matrix (PM) approach, as shown in Eqs. (6.2) and (6.3). PM preprocesses frequency domain data by generating a cyclic shifted copy of the subcarrier waveforms. The variation in sub-carrier shape reduces the degree of auto-correlation among sub-carriers. As a result of the combination of a low number of similar phase sub-carriers, the system’s PAPR decreases. The addition of a µ-law compander to a precoded signal increases its average power. The increase in average power reduces the system’s overall PAPR. The use of a precoder reduces clipping distortion, which leads to an improvement in the system’s BER. The uniform distribution of SNR among data carriers reduces the BER. In a conventional OFDM system, the SNR of the ith subcarrier is calculated as [105]. |𝑋(𝑖)|2

𝑆𝑁𝑅⁡(𝑑𝐵) = 10𝑙𝑜𝑔10 𝛿2 +ℎ−1 𝜓2 𝑖

𝑖

(6.14)

𝑖

where |𝑋(𝑖)|2 is the power of ith signal, 𝛿𝑖2 represents clipping distortion power, and ℎ𝑖−1 𝜓𝑖2 is the channel noise power. The SNR of ith precoded signal is given by 𝑆𝑁𝑅⁡(𝑑𝐵) = 10𝑙𝑜𝑔10

|𝑋(𝑖)|2

(6.15)

1 ∑𝑁−1(𝛿 2 +ℎ𝑖−1 𝜓𝑖2 ) 𝑁−1 𝑛=1 𝑖

The use of precoder in presence of nulls modifies the ℎ𝑖−1𝜓𝑖2 ) of the precoded NOMA DCO-OFDM signals. If

1 𝑁−1 1 𝑁−1

2 ∑𝑁−1 𝑛=1 (𝛿𝑖 + 2 ∑𝑁−1 𝑛=1 (𝛿𝑖 +

ℎ𝑖−1𝜓𝑖2 ) > ⁡ 𝛿𝑖2 + ℎ𝑖−1𝜓𝑖2, the precoder lowers the SNR and BER of the system 1 2 −1 2 2 −1 2 ∑𝑁−1 degrades. However, for ith carrier if 𝑛=1 (𝛿𝑖 + ℎ𝑖 𝜓𝑖 ) < ⁡ 𝛿𝑖 + ℎ𝑖 𝜓𝑖 , 𝑁−1

the precoder enhances the SNR to improve BER of the system. So, using a precoder with overhead (10% nulls) redistributes the SNR among different subcarriers. As a result, each data carrier achieves an average value. In the

.

110

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

presence of PM, the overall improvement is obtained if the average SNR is sufficient for the 16-QAM demodulator.

6.5. Results and Discussion The simulation results presented in this section are used to evaluate the performance of the proposed NOMA DCO-OFDM scheme. The proposed scheme’s performance is evaluated using the complementary cumulative distribution function (CCDF), error vector magnitude (EVM), and BER. Before IFFT, the modulated data is pre-processed using a predetermined matrix in this work. The output of IFFT is passed through a µ-law compander. The complete set of simulation parameters are presented in Table 11. Table 11. Simulation parameters (NOMA DCO-OFDM) Data sub-carriers

64

IFFT lengths

32

Null carriers

3

DC Bias

5.1dB

Cyclic prefix

25%

The CCDF is a statistical measure used to explain the PAPR probability distribution. A CCDF comparison of NOMA DCO, precoded NOMA DCO, companded NOMA DCO, DCO-OFDM with precoder and compander, and the proposed NOMA system is shown in Figure 66. The conventional NOMA DCO -OFDM CCDF is the same as that obtained in [106]. The PAPR of the proposed hybrid NOMA DCO VLC system is nearly 4.3dB, compared to 13.3dB for the conventional NOMA. The pre-processing of modulated symbols with a predetermined matrix followed by the µ-law algorithm raises the average power of the signal, resulting in a low PAPR. The error vector magnitude (EVM) is defined as the root mean square (r.m.s) of the error vector, a metric that quantifies non-linearity. A high EVM indicates that the system has a high BER. Figure 67 depicts a BER comparison for various NOMA-OFDM systems. The results show that for a reference BER of 10-4, the proposed system has the least non-linearity of LED.

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

111

Figure 66. PAPR comparison of different NOMA DCO-OFDM systems.

Figure 67. BER comparison of different NOMA DCO-OFDM systems.

As shown in Figure 67, combining a precoder and a compander in a conventional DCO-OFDM system results in a very low PAPR by a very small margin. However, using NOMA leads to increased capacity by supporting more users. As a result, the proposed NOMA-based O-OFDM system appears to be an excellent choice for high burst rate transmission to support multiple users. Figure 68 depicts an evaluation of EVM performance for NOMA systems. For data transmission, the OSRAM SFH 4230 LED was used. The sixth order curve fitting approach, as described in previous works, matches the V-I

112

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

response curve of the LED [98, 99]. In the presence of an LED, the EVM is computed for five different bias voltages ranging from 1.6 V to 1.7 V. The EVM performance of various NOMA DCO schemes at 1.6 V with input backoff power ranging from 0 to 8dB is shown in Figure 68 (a).

(b)

(a)

(d)

(c)

(e)

Figure 68. Mean EVM comparison for different LED Bias Voltages (a) LED Bias = 1.6 V, (b) LED Bias = 1.625 V, (c) LED Bias = 1.675 V and (e) LED Bias = 1.7 V.

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

113

The data in Table 12 summarises each scheme’s performance. Because of its nonlinearity, the -NOMA has the highest EVM for all back-off values. In the range of 0-5dB back-off power, the proposed scheme outperforms traditional NOMA, NOMA, and Precoded NOMA. The precoded NOMA, on the other hand, matches the performance of the proposed schemes at high back-off power levels of more than 6dB. The performance of various schemes at 1.65 V is depicted in Figure 68 (c), and the results are summarised in Table 13. The EVM of the proposed scheme drops to 32.49% at 0 dB back-off, as shown in the table, representing a 13% improvement over the same back-off at 1.6 V. In comparison to 1.6 V, the NOMA DCO scheme employing the union of precoding and compander provides a more than 10% improvement for higher back-off values of more than 5dB. Table 12. EVM comparison at LED bias = 1.6 V Back-off power (dB) 0 1 2 3 4 5 6 7 8

µ-NOMA 61.30 56.86 53.03 49.69 46.92 44.98 43.69 42.56 41.91

NOMA 56.50 53.35 50.50 48.62 45.90 44.36 43.29 42.56 41.91

EVM (%) Precoded NOMA 48.73 45.99 43.66 41.63 39.96 38.81 38.65 37.88 37.00

Proposed NOMA 45.60 43.75 42.07 40.61 39.36 38.45 38.05 37.38 37

Table 13. EVM comparison at LED bias = 1.65 V Back-off power (dB) 0 1 2 3 4 5 6 7 8

µ-NOMA 44.56 41.30 38.48 36.03 34.00 32.60 31.67 30.86 30.40

NOMA 40.47 30.28 36.31 34.59 33.12 32.06 31.33 30.76 30.40

EVM (%) Precoded NOMA 35.16 33.22 31.55 30.17 29.02 28.24 27.73 27.28 27.30

Proposed NOMA 32.49 31.31 32.24 29.30 28.51 27.93 27.73 27.28 27.30

114

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

As shown in Table 14, the combination of precoder and compander has the lowest EVM of 21.70% at 0dB at 1.7 V, compared to 30.51% and 26.98% for µ-NOMA and traditional NOMA, respectively. Table 14. EVM comparison at LED bias = 1.7 V Back-off power (dB)

µ-NOMA

NOMA

0 1 2 3 4 5 6 7 8

44.56 41.30 38.48 36.03 34.00 32.60 31.67 30.86 30.40

40.47 30.28 36.31 34.59 33.12 32.06 31.33 30.76 30.40

EVM (%) Precoded NOMA 35.16 33.22 31.55 30.17 29.02 28.24 27.73 27.28 27.30

Proposed NOMA 32.49 31.31 32.24 29.30 28.51 27.93 27.73 27.28 27.30

The proposed system reduces non-linearity for 0-5dB input back-off power when compared to conventional NOMA and Companded NOMA systems, resulting in higher power efficiency.

Figure 69. SIC performance comparison for 3 user systems.

Successful interference cancellation (SIC) is used at the receiver in a NOMA-based system to decode each user’s intended signal. Figure 69 depicts a SIC receiver’s performance as the number of users increases. The graph

Mitigation of LED Nonlinearity in a NOMA-OFDM VLC System …

115

compares the performance of two and three user systems in terms of BER. According to the results, User-1 of the U2 system has a lower BER than User1 of the U3, and User-2 of the U2 system has a lower BER than User-2 of the U3. The channel correlation among users grows as the number of users grows. As a result, the BER for fixed power allocation increases as the number of users increases. As a result, in order to accommodate a greater number of users while maintaining performance, the transmitter power must be increased.

6.6. Conclusion One of the potential multiple access techniques for next-generation 5G wireless systems is power domain NOMA. A hybrid approach combining a precoder and a compander is proposed in this chapter for PAPR reduction and BER improvement in a NOMA-based DC-biased optical OFDM system. In a conventional NOMA DCO-OFDM system, the proposed method supports low PAPR and improved non-linearity. The proposed hybrid NOMA DCO VLC system has a PAPR of around 4.3dB, compared to 13.3dB for the conventional NOMA. In comparison to the OMA-VLC system, the use of NOMA in optical VLC systems results in improved capacity regions. As a result of the improvement in non-linearity as well as the low PAPR, the proposed system is a good choice for high-speed 5G indoor mobile data network. In the next chapter, we will discuss the novel approach for enhancing the capacity of 5G system by using integrated VLC-RF system.

Chapter 7

SWIPT Integrated VLC/RF System Performance Evaluation Using the Hybrid-OMA-Cooperative-NOMA Scheme 7.1. Introduction By 2023, the number of internet users is expected to reach 5.3 billion, accounting for 66% of the global population. Machine-to-Machine (M2M) connections will be the fastest-growing device and connection category, growing nearly 2.4-fold (19% compound annual growth rate) to 14.7 billion connections by 2023 [107]. As a result, the already constrained RF spectrum does not appear to address the bottleneck of meeting ever-increasing demands for connectivity, reliability, and data rates in 5G networks and beyond. In this context, the visible light communication (VLC) system has been proposed as an alternative solution for RF networks. According to one survey, the majority of people spend 80% of their time indoors, making VLC, with its dual advantage of illumination and communication, a more promising option. LEDs add value to the VLC system because of their long life, energy efficiency, low cost, and ease of availability. Despite its numerous advantages, the VLC system does have some drawbacks, one of which is uneven qualityof-service (QoS) distribution [108]. Because light intensity is inversely proportional to the square of the distance between the LED and the users, the VLC channel degrades significantly with distance. Also, the conic area of the LED light beam is small and easily blocked by obstacles [109]. As a result, in a practical indoor scenario, there is a significant difference in performance between the near and far user. The receiver’s field-of-view (FOV) and the halfpower semi-angle of the LEDs are two other factors that influence the received signal power of a line-of-sight (LOS) signal. The large FOV or semi-angle of the LEDs increases the likelihood of LOS connectivity between transmitter and receiver, but it also reduces the received power [110]. However, reducing the FOV of the receiver or the semi-angle of the LEDs has the opposite effect. We proposed and thoroughly tested a hybrid orthogonal multiple access (OMA)-cooperative Non-Orthogonal Multiple Access (NOMA) scheme

118

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

for the SWIPT-based integrated RF/VLC system, motivated by the aforementioned VLC design challenges. NOMA has recently gained popularity as a promising multiple access scheme for 5th generation (5G) wireless networks [111]. NOMA uses superposition coding on the transmitter side to enable simultaneous signal transmission to multiple users from a single source. Signals for different users are multiplexed by assigning inverse power levels based on channel gains. The NOMA performs significantly better in high signal-to-noise ratio (SNR) conditions, as demonstrated by Z. Ding and colleagues [112]. Because of the close proximity of the LEDs to the photodetector (PD), an indoor VLC system with a high SNR can employ the NOMA in the downlink VLC system. However, the NOMA scheme cannot address the issue of blocked LOS paths or expand the coverage area for distant users. To address the aforementioned issue, the cooperative NOMA (C-NOMA) has been proposed [113]. The near user decodes the far user’s data while performing successive interference cancellation so that the near user can relay the information to the far user to assist it. As a result, even if the LED does not send a LOS signal, the far user can still receive data from the close user. As a result, the C-NOMA can improve the overall coverage probability of the far user. C-NOMA, however, has not been implemented in the VLC network due to physical constraints. For example, an obstacle between the near and far users could prevent information from being transmitted to the far user via the light signal. T. Rakia et al. [114] proposed integrating VLC and RF systems to solve this problem in the C-NOMA system. The idea is to convert the light signal at the nearest user into an RF signal and then send the signal to the far user via an RF link [115]. Nonetheless, as we progress toward a more advanced communication system, device power consumption has become an important consideration. Battery drain in wireless networks with a large number of IoT sensors may cause the sensor to become inactive. To address this issue, simple RF circuits and RF energy harvesting technology must be implemented. The two primary RF energy harvesting techniques are time switching (TS) and power splitting (PS). In the time switching protocol, energy is harvested in the first fraction of the time slot, and the harvested power is then used for signal transmission in the next fraction of the time slot. The power splitting protocol works on the principle of splitting the received signal power in order to harvest energy and decode information at the same time. As a result, this method is referred to as Simultaneous Wireless Information and Power Transfer (SWIPT).

SWIPT Integrated VLC/RF System Performance Evaluation …

119

We know that NOMA serves multiple users on the same frequency at the same time. However, [116] discovered that once the number of users exceeds a certain threshold, the sum rate begins to fall. The maximum number of users that can be supported in the network without degrading performance is referred to as the drop-off point. However, in order to accommodate more users without degrading performance, transmitted power must be increased as well. As a result, because transmitted power is limited, the number of users in the NOMA system cannot be increased beyond a certain limit. As a result, a multiple access scheme must be implemented in order to support more users. As a result, to accommodate a larger number of users, a hybrid combination of NOMA and any other OMA system can be used. OFDM has recently been proposed as a multicarrier system by a number of authors. OFDM, on the other hand, cannot be considered an effective solution for optical wireless communication because it does not meet the requirement of positive and realvalued signals [117]. The bandwidth efficiency is cut in half even when Hermitian symmetry is used to obtain the true signal. Furthermore, the cyclic prefix increases transmission overhead, lowering throughput, whereas a high peak-to-average power ratio (PAPR) necessitates the use of highly efficient linear amplifiers. As a result, we propose in this chapter a hybrid of C-NOMA and Time Division Multiple Access (TDMA) for a SWIPT-based integrated RF/VLC system.

7.2. Related Work B. Lin et al. [118] have experimentally demonstrated a non-orthogonal multiple access (NOMA) scheme for optical MIMO visible light communications (VLC) with single carrier transmission and frequency domain successive interference cancellation, which offers a low peak to average power ratio, a good balance between throughput and fairness, and a higher system capacity for a larger number of users. With efficient channel equalisation, MIMO demultiplexing and inter-user interference mitigation are realised. In the case of four users, the optimal power allocation ratio is around 0.29. Chen et al. [119] used numerical simulations to investigate the performance of an indoor 2×2 MIMO-NOMA-based multi-user VLC system. The obtained results show that using NOMA with the proposed gain difference power allocation (NGDPA) method can significantly improve the achievable sum rate of the 2×2 MIMO-VLC system. In the 2×2 MIMO-VLC system with three users, NOMA with NGDPA achieves a sum rate improvement of up to

120

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

29.1% when compared to NOMA with the gain ratio power allocation method. J. Shi et al. [120] proposed and experimentally demonstrated the first OQAMOFDM-based multi-user VLC MIMO-NOMA. With the optimised power ratio, their scheme outperforms the conventional MIMO scheme, achieving an aggregate capacity of 3.2 Gbit/s. Assaidah et al. [121] investigated the BER performance versus SNR variation for each level of LED neighbour interference in a MIMO VLC network serving six user equipment (UEs). Based on NHS-OFDM and NOMA schemes, the MIMO configurations used were 2x3 and 3×2. They discovered that the required power ratio for a successful MIMO 2×3 NOMA is 1:2:4, whereas the MIMO 3×2 configuration requires a power ratio of 1:2. To achieve BER below the FEC limit, the minimum SNR for each MIMO configuration should be greater than 20dB, with a maximum interference level of 3% for MIMO 3×2 and 1% for MIMO 2×3. When the SNR is 30dB, the maximum interference level allowed is 13% for MIMO-3×2 and 10% for MIMO-2×3, respectively, to keep the BER below FEC limit. However, all of the preceding work has focused on NOMA, and as previously stated, the NOMA scheme cannot increase the coverage area of distant users on its own. To address this issue, cooperative non-orthogonal multiple access (CNOMA) has been proposed, which uses a dedicated relay to increase NOMA users’ capacity and coverage [122]. According to the findings, successive relaying techniques outperform traditional half-duplex (HD) schemes. Furthermore, in [123], the authors proposed a number of cooperative schemes, including decode-and-forward (DF) relay, also known as fixed relaying (FR), selective DF with coordinated direct and relay transmission (SDF-CDRT), and incremental-selective DF (ISDF) relaying. Z. Ding et al. [124] proposed and investigated user pairing in the C-NOMA-VLC system to reduce system complexity. In terms of performance, they demonstrated that their proposed system outperforms the C-NOMA-based system. Nonetheless, none of these works address simultaneous wireless information and power transfer (SWIPT) to overcome practical energy consumption constraints. Bariah et al. investigated the error rate performance of NOMA-based relay networks with energy harvesting in a C-NOMA study presented in [125]. According to the results, the optimal power splitting factor is between 0.3 and 0.5. They investigated the effect of path loss on energy harvesting by varying the distance of the relay from the base station. They demonstrated that the power allocation coefficient and the design of the electric circuit have an effect on NOMA performance. In addition, [126] proposed a novel wireless energy harvesting DF relaying protocol for underlay cognitive networks that allows

SWIPT Integrated VLC/RF System Performance Evaluation …

121

secondary users to harvest energy from primary users. To mitigate interference in multi-cell VLCs, M. Obeed et al. [127] propose the use of C-NOMA, energy harvesting, and hybrid VLC/RF techniques. They discovered a significant improvement in C-NOMA performance due to the fact that each weak user can choose between the hybrid RF/VLC and direct VLC links, whereas in the NOMA scheme, each weak user can only be served via the direct VLC link. An RF link connects each weak user to one strong user, and the VLC link serves the strong user directly. As a result, system performance is influenced by user pairing, link selection, and message power. In present chapter we have proposed combining NOMA and TDMA to serve a larger number of users. A few research articles on this topic are discussed below. B. Li et al. [128] proposed a reconstructed hybrid RHO-OFDM that uses simultaneous transmission of multiple Optical-OFDM signals to improve spectral efficiency over ACO-OFDM. They discovered that, when compared to traditional HACO-OFDM, RHO-OFDM can eliminate error propagation in NOMA-VLC systems while retaining high spectral and power efficiency. T. Uday et al. [129] proposed a DCO-OFDM NOMA hybrid scheme for multiple access channels and broadcast channels in an indoor VLC system. They demonstrated that their proposed system provides less BER for SNR values greater than 40dB. H. Ren et al. [130] proposed a new decoding method known as EAC to improve the performance of the OFDM-NOMA VLC system. The BER performance was investigated using the EAC decoding method, with the modulation formats for each user being 4-QAM, 16-QAM, and 64-QAM, respectively. The impact of power allocation coefficients and peak clipping on BER performance was also investigated. When high order M-QAM (M > 4) signals are applied to the two users, the simulation results show that the proposed EAC method outperforms the SIC method in terms of BER performance. A. Abdul et al. [131] created two novel filter bank-based multicarrier techniques for visible light communication. To generate orthogonal multicarrier signals, they proposed a new signal design that combines pulse shaping with square root raised cosine filtering and frequency translation. They proposed a filter bank modulation technique for multipleinput-multiple-output visible light communication based on this. They also create another filter bank modulation method that uses spatial indexing to improve data rate and spectral efficiency. However, none of the preceding works jointly investigate a hybrid combination with SWIPT in an integrated VLC/RF system.

122

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

7.3. Contributions We addressed all of the issues raised in this chapter by implementing a hybrid combination of C-NOMA, TDMA, and SWIPT for an integrated VLC/RF system. With various user combinations, we compared the performance of single carrier NOMA (SC-NOMA), TDMA, and the proposed HOCN system. The results show that the proposed HOCN, TDMA, and SC-NOMA outperform each other across a wide range of input parameters, indicating that each is significant in its own right.

7.4. Structure The rest of the chapter is organised as follows. Section 7.5 describes our proposed model’s indoor room system. The simulation results and discussion are presented in Section 7.6. The chapter concludes with Section 7.7.

7.5. Model of the System 7.5.1. Proposed System Model The following section describes our proposed HOCN model, which is based on a VLC/RF system with SWIPT. As shown in Figure 70, we consider a typical VLC indoor downlink scenario with one LED installed on the ceiling of a Lx × Ly × Lz (length, width, height) room and four users. The strong users (U1 and U2) who have a Line of Sight (LOS) link with the LED are located within the Ro coverage area’s horizontal radius. The weak users (U3 and U4) are surrounded by a circle with a horizontal radius Rv (Ro HOCN (N-F) > SC-NOMA HOCN (N-F) > SC-NOMA > HOCN (N-N, F-F) > TDMA SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F) > TDMA

100dB120dB 120dB140dB

SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F) > TDMA SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F) > TDMA

60dB-80dB

Remark The order of sum rate using different schemes remains constant for all values of semi-angle. SC-NOMA overtakes HOCN (N-N, F-F) & TDMA at semi-angle of 350 SC-NOMA overtakes HOCN (N-N, F-F) & TDMA at a semi-angle of 200. Also, SCNOMA overtakes HOCN (N-F) at a semiangle of 250. SC-NOMA overtakes all other schemes at a semi-angle of 150. SC-NOMA outperforms all other schemes for all values of semi-angle.

The results of Figure 71 indicate that TDMA is a better option for lower SNR ranges, whereas our proposed HOCN outperforms all other schemes for medium SNR ranges. The SC-NOMA provides the worst sum rate performance in the lower range of SNR values because overloading all users onto the same carrier causes interference issues. This issue can be alleviated by increasing transmitted power to accommodate a larger number of users. This logic is justified because it can be seen that the SC-NOMA outperforms other schemes over a wider range of SNR values. The proposed HOCN with N-F user pair performs better in the 80dB-100dB SNR range but not in the lower and higher SNR ranges. This is due to the fact that N-F user pairing performs best when the channel conditions between the two users are distinct. In either the lower or higher SNR range, the channel condition between two users is indistinguishable. The Figure 72 explains why our proposed HOCN system performs better than NOMA:

128

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 71. Sum rate Vs. semi-angle of the LED for (a) SNR = 40dB (b) SNR = 60dB (c) SNR = 80dB (d) SNR = 100dB (e) SNR = 120dB (f) SNR = 140dB.

Figure 72. Concept of TDMA, NOMA, NOMA+TDMA.

SWIPT Integrated VLC/RF System Performance Evaluation …

129

Consider a time slot with a duration of T ms and a capacity of four users. The TDMA will divide the T ms into four T/4 ms slots, one for each user. The SC-NOMA will assign T ms slots to all users. As a result, performing successive interference cancellations (SIC) will be difficult, as will processing delay. The HOCN system, which is a hybrid of TDMA and NOMA, will divide the T ms into two T/2 ms slots and assign two NOMA users to each slot. So, the HOCN system can serve all users with less complexity. Aside from the SNR value, the sum rate is also affected by the LED’s semi-angle range. Though for higher SNR values, the SC-NOMA scheme provides more sum rate because the NOMA scheme would allocate all of the power to the strong user and zero power to the blocked one, improving the sum rate but with poor fairness (because all of the power would be assigned to the strongest channel). Fairness is lacking because the blocked user suffers (there is no user cooperation). Due to the inclusion of C-NOMA in our proposed HOCN, the blocked user will be served by the relay user, resulting in a trade-off between sum rate and fairness for the higher SNR range. This is due to the fact that the blocked user can be served by the RF link in the proposed HOCN system with integrated VLC/RF, and the proposed scheme would split the power between the strong and weak users, improving fairness. It can also be seen that user pairing affects the sum rate at various SNR levels. As the number of users increases, it is critical to identify the best user pairing strategy for making the best use of power and bandwidth resources. For various SNR values, Figure 73 depicts the sum rate vs power allocation coefficient (of far users). Figures 73a and b show that when the SNR is less than 60dB, the sum rate achieved by SC-NOMA is lower than that of the HOCN system, confirming the results of Figures 71a and b. Furthermore, when the power allocation coefficient is between 0 and 0.5 in the HOCN system, N-F user pairing achieves a higher sum rate than N-N, F-F user pairing. In the range of 0.5 to 1, however, the N-N, F-F user pairing achieves a higher sum rate than the N-F user pairing. Furthermore, when the power allocation coefficient is 0.5, the sum rate obtained by using both user pairing strategies is the same because power is distributed equally among strong and weak users. Figures 73c-f show that the sum rate achieved by SC-NOMA is greater than that of the HOCN system for a given range. The range of SNR over which SC-NOMA outperforms the HOCN system expands as the SNR value increases. Furthermore, in HOCN, the N-F user pairing achieves a higher sum rate than the N-N, F-F user pairing (0,1). Table 17 summarizes the findings of Figure 73.

130

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 73. Sum rate Vs. power allocation factor for (a) SNR = 40 dB (b) SNR = 60dB (c) SNR = 80dB (d) SNR = 100dB (e) SNR = 120dB (f) SNR = 140dB.

SWIPT Integrated VLC/RF System Performance Evaluation …

131

Thus, Table 17 shows that when 𝛼 ∈ (0, k); k SC-NOMA; α ϵ (0, 0.5) HOCN (N-F) > HOCN (N-N, F-F) > SC-NOMA; α ϵ (0, 0.65) HOCN (N-F) > HOCN (N-N, F-F) > SC-NOMA; α ϵ (0, 0.65) SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F); α ϵ (0, 0.75) SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F); α ϵ (0, 0.8) SC-NOMA > HOCN (N-F) > HOCN (N-N, F-F); α ϵ (0, 0.9)

Order of sum rate HOCN (N-N, F-F) > HOCN (N-F) > SC-NOMA; α ϵ (0.5, 1) HOCN (N-N, F-F) > HOCN (N-F) > SC-NOMA; α ϵ (0.65, 1) HOCN (N-N, F-F) > HOCN (N-F) > SC-NOMA; α ϵ (0.65, 1) HOCN (N-F) > HOCN (N-N, F-F) > SC-NOMA; α ϵ (0.7, 1) HOCN (N-F) > HOCN (N-N, F-F) > SC-NOMA; α ϵ (0.8, 1) HOCN (N-F) > HOCN (N-N, F-F) > SC-NOMA; α ϵ (0.9, 1)

So far, we’ve looked at how the semi-angle of the LED affects coverage area and uniformity illuminance. To cover a larger area, the semi-angle must be wide. The coverage area of an LED can still be changed by adjusting the receiver plane. The field of view angle ψFOV of the receiver must also be considered in terms of the communication function related to connectivity [135]. The field of view ψFOV of the receiver must also be chosen, especially if an energy harvesting scheme is included. There will be two possibilities: ψFOV < 𝜑1/2 and ψFOV > 𝜑1/2 . It has been discovered that decreasing ψFOV allows for more received power and a higher data rate, resulting in improved performance. As a result, we consider sum rate performance using all of the schemes in two different scenarios in this section. In the first scenario, the ψFOV of near and far users is assumed to be 600 and 400, respectively. In the second scenario, the ψFOV of near and far users is assumed to be 400 and 600,

132

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

respectively. Figure 74a shows two distinct scenarios, scenarios 1 and 2. Figures 74b and 74c depict the sum rate versus semi-angle for the first and second scenarios, respectively. Figures 74b-c show that the sum rate in the first scenario is HOCN (N-N, F-F) > TDMA > HOCN (N-F) > SC-NOMA, whereas the sum rate in the second scenario is TDMA > HOCN (N-F) > SC-NOMA. According to Figure 74d, scenario 2’s TDMA system achieves the highest sum rate, followed by scenario 1’s HOCN (N-N, F-F) system. This result is consistent with Figure 71a. It is also worth noting that in scenario 2, the HOCN (N-F) system outperforms its counterpart in scenario 1. For SCNOMA, Scenario 1 is preferable to Scenario 2. Table 18 summarises the entire discussion. Thus, our findings indicate that the FOV combination is critical for user pairing in both a HOCN and a C-NOMA system in order to ensure high-quality communication. (a)

(b)

Figure 74. (Continued)

SWIPT Integrated VLC/RF System Performance Evaluation …

133

(c)

(d)

Figure 74. (a) FOV of scenario 1 & scenario 2. (b) Sum rate vs. semi-angle of LED for scenario 1 (c) sum rate vs. semi-angle for scenario 2 (d) Comparison of scenario 1 & scenario 2 in same figure.

Table 18. Summary for selecting the best option from scenarios 1 and 2 Scheme TDMA HOCN (N-N, F-F) HOCN (N-F) SC-NOMA

Better option from scenario 1 & scenario 2 Scenario 2 Scenario 1 Scenario 2 Scenario 1

134

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 75. Achievable rate for far user and total sum rate for (a) Transmit power = 40dB (b) Transmit power = 60dB (c) Transmit power = 80dB (d) Transmit power = 120dB.

We simulated the sum rate vs power harvesting efficiency for various transmit power values to investigate the effect of harvested energy on sum rate performance, as shown in Figure 75. The power allocation factor in this simulation is 0.8 for far users and 0.2 for near users. The target rate has been set at 1bps/Hz for both distant and close users. The results show that as the power harvesting efficiency of the circuit improves, so do the far user rate and sum rate. Furthermore, as the LED’s transmitted power increases, so does the amount of harvested power, and thus the far user’s achievable rate as well as the sum rate. At 80 percent power harvesting efficiency, for example, when the transmitter power is 40dB, both the far user rate and the sum rate become equal. The required efficiency for power harvesting, however, decreases as transmitted power increases; for example, at 120dB transmitted power, the far user rate and sum rate become equal at 50% efficiency.

SWIPT Integrated VLC/RF System Performance Evaluation …

135

Figure 76. Power harvested vs semi-angle of the LED.

The effect of the LED’s semi-angle and the FOV of the receiver on the harvested power is then discussed. Figure 76 shows the power harvested vs. LED semi-angle for various FOV values. As the FOV of the nearby user increases, the harvested power decreases. However, increasing the semi-angle of the LED results in more harvested power. Because of the increased FOV, the signal at the receiver is attenuated, resulting in less harvested power. The increased semi-angle of the LED results in more optical signal at the receiver and thus more harvested power. When the semi-angle is less than 200, no light rays reach the receiver, resulting in zero harvested power. As a result, the harvested power is influenced by the LED source’s semi-angle as well as the receiver’s FOV.

7.7. Conclusion To reduce SIC complexity, we proposed a hybrid combination of NOMA and TDMA for a SWIPT-based integrated RF/VLC network. The HOCN system’s sum rate performance (different user pairing) was simulated and compared to SC-NOMA and TDMA systems. Using all of the aforementioned schemes, we also investigated the effect of the power allocation coefficient, LED semiangle, and various FOV combinations on the sum rate and harvested power. Furthermore, the sum rate performance has been evaluated in order to determine the relay user’s power harvesting efficiency. The findings indicate

136

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

a trade-off between the sum rate and fairness. Though SC-NOMA has a higher sum rate than our proposed system in higher SNR ranges, our proposed system is a better option in terms of fairness because it considers users who are not in the coverage area. Furthermore, we demonstrated two different scenarios with different FOV combinations and showed that different schemes perform better in different scenarios. Finally, the effect of semi-angle and field of view on harvested power was investigated. The findings indicate a trade-off between the sum rate and fairness. Though SC-NOMA has a higher sum rate than our proposed system in higher SNR ranges, our proposed system is a better option in terms of fairness because it considers users who are not in the coverage area. However, achieving high data rate transmission using a MIMO-VLC system does not completely solve the problem unless and until the receiver is capable of receiving the transmitted data via the VLC channel. A wide FOV and high gain receiver system is required to collect the maximum light rays. In the next Chapter, we proposed a novel lens that is small in size, has a wide field of view, and has a high gain.

Chapter 8

Concave-Convex Lens with High Gain and Volume Efficiency for MIMO-VLC Systems 8.1. Introduction Visible light communication (VLC) is a new technology that aims to provide lighting as well as high data rate transmission in an indoor environment. It operates on the principle of intensity modulation of light-emitting diodes (LED). The VLC system has several advantages over radio frequency communication, including immunity to interference, energy efficiency, high security, vast unlicensed bandwidth, environmental friendliness, and ease of integration with existing lighting infrastructure. While much attention has been paid in the VLC system to modulation-demodulation schemes [136], channel modelling [137], multiple-input multiple-output (MIMO) [138-139], and so on, little research has been done on the design of receiver lenses. LEDs, for example, have a limited modulation bandwidth, so the MIMO technique has been used to improve data rate using the spatial multiplexing concept [140, 141, 142]. Different types of LED layout arrangements, such as circular, square, and so on, have been compared using a genetic algorithm to increase illumination with uniform power distribution [143]. Similarly, various optimal optical attocell configuration models for improving optical received power and SNR distribution have been proposed [144-145]. Furthermore, the impact of LED tilt on the accuracy of received signal strength in visible light positioning (VLP) has been studied using the Monte-Carlo method [146]. For an indoor MIMO-VLC system to collect light from the above-mentioned layouts, the receiver’s field of view (FOV) must be large. Because increasing the size of the photodetector reduces the receiver’s bandwidth, an optical antenna with a large FOV, high gain, and small size is required. These optical antennas in VLC systems are optical concentrators that increase the receiver’s effective collection area, optimise power consumption, and allow for greater distance between transmitters and receivers.

138

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

To improve the FOV and gain of the receiver in an optical wireless system, a compound parabolic concentrator (CPC) has been proposed [147]. A dielectric totally internally reflecting concentrator (DTIRC) that can provide more concentration than a CPC has been introduced [148]. The light collection efficiency of DTIRC has been investigated using ray tracing by varying the angle of incidence. Dia wang et al. [149] proposed an indoor MIMO-VLC system using a gradient-index lens with a compound parabolic shape. They discovered that the gain is 1.72 times greater than that of a bare CPC (uniform refractive index) based receiver. The large length of CPC and DTIRC, on the other hand, encourages researchers to look for alternative lens designs because the receiver’s size should be small and compact [150]. For a MIMO-VLC system with a wide FOV, an imaging-lens with a hemispherical shape was used [151]. It was discovered that the proposed imaging system provides adequate channel gain for angles of incidence as large as 700 after investigating various receiver positions and tilted receiver scenarios in an indoor system. The hemispherical lens’s wide field of view makes it suitable for mobile use in an indoor VLC system. Unfortunately, the gain of a hemispherical lens remains constant at n2 (n-refractive index of material) over a fixed FOV of 900, necessitating the search for a more effective alternative. Recently, H. G. Hao et al. [152] proposed combining Fresnel and hemispherical lenses for wide FOV and high gain in an indoor MIMO-VLC system. Similarly, Y. Wang et al. [153] proposed a scheme to integrate CPC with hemispherical lenses and discovered that by using ray tracing in TracePro software, the radius of spot size of received light was reduced from 5 mm to 3 mm at the receiving system. Although a cascaded or integrated antenna system increases FOV and gain, it may cause errors if the optical axis of the combined antenna system is not properly aligned. Furthermore, the collective system requires more volume and may not be suitable for the VLC system’s compact receiver. An angle diversity technique has been proposed to increase the FOV of receiver antennas [154, 155]. However, the angle diversity technique causes signal crosstalk and decreases data rate. In this chapter, we propose a lens for MIMO-VLC receiver systems that has 1.22 times the volume of a hemispherical lens but has a larger entrance aperture area, a wider FOV, and a higher gain. It has been demonstrated in ray trajectories using COMSOL software that the proposed lens, due to its unique design, provides smooth convergence of light rays from the exit aperture using a single LED in two cases: the first case, the light beam is incident parallel to the axis of the lens, and the second case, the incident beam is at 450 degrees. The investigation was then expanded by using two LEDs with light beams

Concave-Convex Lens with High Gain and Volume Efficiency …

139

incident at 450 degrees from the lens axis. It has been demonstrated that, unlike hemispherical lenses, the proposed lens does not suffer from spherical aberration due to its unique design. Furthermore, it produces distinct images of individual LEDs, confirming that the concave-convex lens can support higher diversity orders in the MIMO system. The MATLAB simulation results show that it provides more received power for a 4×4 MIMO-VLC system than not only hemispherical lenses but also uniform refractive-index and gradientindex based CPC lenses. Furthermore, by varying the angle of incidence of light rays and the semi-half power-angle of the LED, the simulation of channel gain was compared using the proposed one and hemispherical lens. The outcome shows that the proposed lens outperforms the hemispherical lens.

8.2. Concave-Convex Lens Architecture Figure 77 depicts the proposed lens, as well as hemispherical, CPC, and DTIRC options. The structures of these lenses are depicted in the same figure to demonstrate the volume efficiency comparison. When compared to hemispherical and concave-convex lenses, CPC and DTIRC require more vertical length and the PD must be placed at a greater distance from the lens. Our proposed lens has a convex entrance aperture and a concave exit aperture. The proposed antenna can be modelled in the following section by judicious modification of the existing hemispherical lens, where the overall lightcollecting entrance aperture area and gain are greater than the hemispherical lens.

Figure 77. 3D structure of (a) Hemispherical lens (b) Concave-convex lens (c) CPC lens (d) DTIRC lens designed in AutoCAD.

140

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

8.2.1. The Design of Concave-Convex Lens We determined the fraction by which the concave-convex lens provides more received power than the hemispherical lens because the structure of our proposed lens has been modified in the existing lens. The cross-sectional view of a hemispherical and concave-convex lens with portions OAO’B and OAPB is shown in Figure 78. The segment APB is carved out of a circle (shown in white) with a radius of OA = OP, and when combined with the hemispherical section OAO’B, it results in a concave-convex configuration.

Figure 78. Cross-sectional view of concave-convex (OAPB) and hemispherical (OAO’B) lens.

The total power at the entrance aperture of a receiving antenna lens is [149]. P = ∫ E(x, y, z, φ)ds⁡⁡

(8.1)

where φ is the angle between the incident ray (at point U) and z-axis, E(x, y, z, φ) is the total illuminance on inlet surface and integration is performed over the entrance aperture area. The illuminance E(x, y, z, φ) for a spatially uniform radiation can be written as φ

𝐸 = ∫φ max. 𝑝𝑁𝑑𝜑⁡⁡⁡ min.

(8.2)

where p is the power of each ray and N is the number of rays per unit angle.

Concave-Convex Lens with High Gain and Volume Efficiency …

141

Thus from Eq. (8.1) & (8.2), the power received using concave-convex (𝑃𝐶 ) and hemispherical lens (𝑃𝐻 ) is given by l′ +R−√R2 −r2

PC = ∫l′

φ

(∫φ max. Npdφ) 2πRdz⁡⁡ min.

φ

𝑃𝐻 = 2𝜋𝑟 2 (∫φ max. Npdφ) min.

(8.3) (8.4)

respectively. From Eq. (8.3) & (8.4) it can be observed that the total power received using lens depends on the product of illuminance and area of entrance aperture. Assuming that the illuminance over both lenses is same, it can be concluded that ratio of power received through proposed lens and hemispherical lens will be proportional to the ratio of their respective entrance surface area i.e., PC PH

A

∝ A C ⁡⁡⁡ H

(8.5)

where AC and AH are entrance aperture area of concave-convex and hemispherical lens respectively.

Figure 79. Geometrical configuration to find the ratio of entrance aperture area of concave-convex lens to the hemispherical lens.

To get the surface area (Ac) of the entrance aperture (APB in Figure 78) of concave-convex lens, we consider a circular disc (displayed by shaded region in Figure 79) that subtends angle dθ and has arc length given by dl = Rdθ, where R is the radius of the sphere. The circumference of circular disc is c = 2πr, where r (r = R sin θ) is the radius of hemispherical lens over

142

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

which the portion APB is to be appended. The area of the circular disc is dA = circumference⁡of⁡disc × dl. To get the area of portion APB (fig. 78), the area of circular disc dA is integrated from θ = −arc sin(r/R) to θ = arc sin(r/R) or simply it is twice the area integrated from θ =0o to θ = arc sin(r/R). Hence, the area of the entrance aperture=2of∫ concave-convex lens can be written as (2πr × Rdθ) θ=arcsin ⁡ (r⁄R ) θ=00

θ=arcsin⁡(r⁄R)

AC=2 ∫ dA = ∫θ=00

dA

θ=arcsin⁡(r⁄R) =2 ∫θ=00 (circumference⁡of⁡disc × θ=arcsin⁡(r⁄R) =2 ∫θ=00 (2πr × Rdθ)

dl)

Figure 80. Radius of convex aperture of proposed lens is R = 4.77 mm corresponding to radius of hemispherical lens r = 2.5 mm as designed in AutoCAD.

On integration and simplification = 4πR2 |1 − cos⁡(arcsin(r/R))|

(8.6)

The entrance aperture area of hemispherical lens is 2πr2 . Hence, the ratio of entrance aperture area of concave-convex lens to the hemispherical lens is AC AH AC AH

= =

4πR2 |1−cos⁡(arcsin(r/R))| 2πr2 2|1−cos⁡(arcsin(r/R))| (r/R)2

(8.7) (8.8)

From Eq. (8.5) & (8.8), it is observed that the ratio of received power using the proposed lens and the hemispherical lens PC/PH is the function of ratio of radius of hemispherical lens (r) to the radius of convex aperture (R) i.e., r/R. We have considered the radius of hemispherical lens as 2.5 mm and

Concave-Convex Lens with High Gain and Volume Efficiency …

143

correspondingly the radius of convex aperture of proposed lens comes out to be 4.77 mm as designed in AutoCAD (illustrated in Figure 80). The plot of the ratio PC/PH as a function of ratio r/R is shown in the Figure 81. Thus, for the given ratio r/R = 2.5 mm/4.77 mm = 0.5242, the received power using proposed lens is 1.081 times the power received through hemispherical lens as shown in the figure. With increase in ratio (r/R), the surface area of convex aperture also expands and it augments the light collection efficiency of the proposed lens as compared to hemispherical lens. Keeping in mind that the optical lens should be of optimal size, standard values of radius can only be considered. The radius of 2.5 mm value for hemispherical lens has been considered in [151] and hence to compare the simulated results we have also taken the same value of radius for hemispherical lens and correspondingly 4.77 mm as radius of convex aperture for our proposed lens.

Figure 81. The ratio of power received using concave-convex lens and the hemispherical lens as a function of (r/R).

8.3. Results and Discussion In our study, we compared the performance of a concave-convex lens to that of a hemispherical lens, as well as bare CPC (uniform refractive-index) and graded-index-based CPC. The comparison was made using various factors such as ray tracing, received power, and channel gain.

144

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

8.3.1. Ray Trajectories Using Different Types of Lenses Because the received light signal must be focused at the photodetector using a lens, ray trajectories have been simulated in Comsol software using the Ray tracing module to compare the performance of various types of lenses. To investigate their competence in MIMO-VLC systems, two scenarios for ray trajectories were considered: in the first scenario, light rays are parallel to the z-axis, and in the second scenario, light rays are incident at 450. Figures 82 (a), (b), (c), and (d) show the light ray trajectories for the first scenario using hemispherical, concave-convex, CPC, and DTIRC lenses, respectively.

Figure 82. Ray trajectories for the scenario when rays are incident parallel to z-axis for (a) Hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC.

As a result, our proposed lens provides the best convergence for refracted light rays from the exit aperture. Because VLC works on the principle of intensity modulation and direct detection of light, the light rays are reflected from

Concave-Convex Lens with High Gain and Volume Efficiency …

145

hemispherical and CPC lenses, lowering the quality of the received signal. In the case of DTIRC, the refracted light rays from the exit aperture experience both convergence and divergence. Light ray divergence can cause interference on the receiver side of a MIMO-VLC system. Thus, when the light rays are parallel to the lens’s axis, our proposed lens performs better.

Figure 83. Ray trajectories for scenario when rays are incident at 45 0 w.r.t z-axis (a) hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC.

Because the LEDs in the MIMO-VLC system are distributed on the ceiling in various layout forms, it is necessary to compare the performance of the emitted light rays from the LEDs using different lenses at large angles of incidence. The light ray trajectories for the second scenario are shown in Figure 83 when the angle of incidence is 450 w.r.t. the z-axis. As can be seen in Figure 83 (a), the hemispherical lens has a spherical aberration problem. Spherical aberration causes many focal points, resulting in a blurred image on the imaging plane.

146

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Spherical aberration is caused primarily by three factors: lens design, glass material quality, and lens placement. Because the latter two factors are the same, we can conclude that our proposed lens has a better design that avoids the problem of spherical aberration, as shown by the ray trajectories in Figure 83 (b). Figures 83 (c) and (d) show the ray trajectories with the CPC and DTIRC lenses, respectively. The findings show that light rays refracted from the exit aperture are diverging. Furthermore, the number of rays refracted from the exit aperture is reduced. This is better explained by taking a close look at the ray trajectories inside the lens with a shorter path length, as shown in Figure 84.

Figure 84. Close view of ray trajectories inside (a) Hemispherical lens (b) Concaveconvex lens (c) CPC (d) DTIRC. greater in the case of hemispherical lenses, and thus the receiver requires more volume. The

As shown in Figure 84 (b), the transmitted light rays from the entrance aperture bend inwards, causing the refracted rays from the exit aperture to properly converge, resulting in a single focal point. However, because hemispherical lenses have a flat entrance aperture, the transmitted rays are less inward than concave-convex lenses, resulting in less convergence from the exit aperture. The number of rays that experience total internal reflection is also higher in hemispherical, CPC, and DTIRC lenses than in the proposed lens, reducing the received power at the photodetector. Thus, when light rays are incident at a large angle, our proposed lens performs better.

Concave-Convex Lens with High Gain and Volume Efficiency …

147

Figure 85. Ray trajectories using two sources whose light rays are incident at φ = 450 (a) Hemispherical lens (b) Concave-convex lens (c) CPC (d) DTIRC (e) Rotated version of Figure 85 (d) to show scattered rays from the exit aperture.

Channel correlation has an effect on the performance of a MIMO-VLC system, so it is necessary to investigate how clearly the images of different LEDs can be distinguished at the receiver side. Hence, we simulated ray trajectories with two LEDs whose light rays are incident at 450. Figures 85 (a) and (b) show that hemispherical and concave-convex lenses produce distinct

148

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

images of the two LEDs. Images produced by a hemispherical lens, on the other hand, are less separated and are located lower in comparison to the proposed lens. It means that the distance between the imaging plane (over which the array of PDs is placed) and the lens is diversity order supported by a concave-convex lens would be greater because, unlike a hemispherical lens, images would still be distinguishable as the number of LEDs increased. Figures 85 (c) and (d) depict ray trajectories with CPC and DTIRC lenses, respectively. The results show that images using these lenses are not distinguishable, and thus channel correlation would be high, potentially leading to a non-invertible channel matrix, making signal decoding in the MIMO-VLC system difficult. The front view of Figure 85 (d) appears to show refracted rays from the exit aperture bunched together, so a rotated version of the same figure is provided in Figure 85 (e) to demonstrate that rays are actually scattered.

8.3.2. Power Received Using Concave-Convex and Other Lenses To evaluate the quality of performance of our lens in a MIMO-VLC system, it is necessary to compare received power with other lenses. For this, we considered a 5m×5m×3m room with four LEDs installed on the ceiling and an array of four PDs on the receiver side, as shown in Figure 86. The receiver is positioned 0.85 m above the floor. Table 19 shows the rest parameters and their values. Figure 87 depicts the geometrical model of a concave-convex lens receiver using a spherical coordinate system. As shown in the figure, the spherical coordinate’s origin O’ is located at the centre of the lens (the coordinate system of indoor room and lens is separate). A LED is placed at point S (lsinφcosθ, lsinφsinθ, lcosφ) l m from the lens, where and represent the angle between SO’ and the positive z axis, and the angle between S’O’ and the positive x axis, respectively. The light ray emitted by the LED strikes the lens’s entrance aperture (point A) at angle α1 and is refracted at angle α2. At an angle of incidence of α3, the refracted light ray strikes the interior surface (point B) of the exit aperture. The refracted light ray (angle α4) then strikes the photodetector array at point F, which is on the receiver system’s imaging plane. Point A’s coordinate is (rcosβ, rsinβ, 0), where r and β are the radius and polar angle of the polar coordinates of the X’O’Y’ plane.

Concave-Convex Lens with High Gain and Volume Efficiency …

149

Figure 86. Geometrical model of indoor VLC system with four LEDs and array of four PDs.

Table 19. Parameters and their corresponding values for indoor VLC system Parameters Room size Reflection coefficients of wall Power of each LED Centre position of four LEDs LEDs chip interval Array of LEDs Receiver’s height from floor Radius of hemispherical lens Radius of entrance aperture of concaveconvex lens

Value 5 m×5 m×3 m 0.7 20 mW Tx1 = [-1.25,1.25,3],Tx2=[-1.25,-1.25,3], Tx3 = [1.25,-1.25,3], Tx4=[1.25,1.25,3] 0.01 m 60×60 0.85 m 2.5 mm 4.77 mm

150

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

Figure 87 depicts the geometrical model of a concave-convex lens receiver using a spherical Figure 87. Geometrical model showing ray tracing in concave-convex lens with photodetector array.

The power dP1 received by the lens at point A through infinitesimal area rdrdβ is given as dP1 =

Pt I(ϕ)rdrdβcosφ

(8.9)

𝑙2

where Pt power transmitted by LED, φ is incident angle and I(ϕ) is the radiation intensity of Lambertian pattern of LED given by I(φ) =

(m+1) 2π

cos m ϕ

(8.10)

where ϕ is angle of emission w.r.t normal optical axis of LED, m is the Lambertian order which depends on half power semi-angle ϕ1/2 of the LED given as m = −ln⁡(2)⁄ln⁡(cos⁡ϕ1/2 ). Accordingly, (8.9) is expressed as dP1 =

Pt (m+1)cosm ϕcosφ 2π𝑙2

rdrdβ

(8.11)

Concave-Convex Lens with High Gain and Volume Efficiency …

151

The power of incident light ray is partially lost due to reflection from the entrance aperture surface of the lens. The part of the light ray which is reflected is given by [151] n cosα −n cosα

R p (α1 , α2 ) = n1cosα2+n2cosα1 1

2

2

1

(8.12)

and n cosα −n cosα

R s (α1, α2 ) = n1 cosα1+n2cosα2 1

1

2

2

(8.13)

Where n1 , n2 are refractive indices of air and lens respectively, R p and R s are reflection coefficients of p-polarized and s-polarized light respectively. Therefore, the power transmission coefficient of entrance aperture of the lens for the un-polarized light is given as 1

Tair−lens (α1 , α2 ) = 1 − {R2s (α1 , α2 ) + R2p (α1, α2 )} 2

(8.14)

The power dP2 of light ray that passes through the lens is expressed by dP2 = Tair−lens (α1 , α2 )dP1 =

Pt (m+1)cosm ϕcosφ 2π𝑙2

Tair−lens (α1 , α2 )rdrdβ

(8.15)

The transmitted light ray refracts away from the point B of interior surface of exit aperture towards photodetector. The proportion of the power which is refracted out of the lens is calculated in similar manner described in Eq. (8.14) and is given by 1

Tlens−air (α3 , α4 ) = 1 − 2 {R2s (α3 , α4 ) + R2p (α3 , α4 )}

(8.16)

Thus, the power of ray after refracting photodetector is written as

=

𝑑𝑃0 = 𝑇𝑙𝑒𝑛𝑠−𝑎𝑖𝑟 (𝛼3 , 𝛼4 )𝑑𝑃2 𝑃𝑡 (𝑚+1)𝑐𝑜𝑠𝑚 𝜙𝑐𝑜𝑠𝜑 𝑇𝑎𝑖𝑟−𝑙𝑒𝑛𝑠 (𝛼1 , 𝛼2 ) 2𝜋𝑙2 𝑇𝑙𝑒𝑛𝑠−𝑎𝑖𝑟 (𝛼3 , 𝛼4 )𝑟𝑑𝑟𝑑𝛽

(8.17)

152

Ajit Kumar, Nishant Sharan and Swapan Kumar Ghorai

The channel gain which is ratio of power received at the PD and the transmitted power by LED is obtained by integration of all the rays that reach at PD 𝑇𝑐𝑔 =

∬

Po Pt

=

(m + 1)cosm ϕcosφ 2π𝑙2

Tair−lens (α1 , α2 )Tlens−sir (α3 , α4 )rdrdβ

α3