Innovations in Computational Intelligence:

Trends and Technologies

Dr. Anurag Tiwari
Anurag Tiwari is an Associate Professor and Head of the Department of Computer Science
and Engineering at BBD National Institute of Technology and Management, Lucknow. A Sen-
ior IEEE Member, he has over 11 years of academic experience and has led the department to
NBA accreditation in 2023. He holds a Ph.D. in Computer Science, multiple M.Tech degrees,
and has authored several research publications. With multiple patents and consultancy pro-
jects to his credit, his research interests include machine learning and optimization techniques.

Dr. Manuj Darbari

Manuj Darbari, Senior Member IEEE (USA) and Chartered Engineer (India), is a visionary
leader in Computer Science and Engineering, specializing in AI, Cloud Computing, and IoT.
With a track record of pioneering research, patents, and numerous PhD supervisions, he is
recognized for driving innovation and excellence. Currently a Professor at BBDITM Lucknow,
Dr. Darbari has made significant contributions to cutting-edge technologies and has served as
a convener and reviewer for leading global conferences. His editorial expertise ensures that
the conference proceedings reflect the highest standards and capture the forefront of emerging
research trends.
Innovations in Computational Intelligence:
Trends and Technologies

Proceedings of the International Conference on Advances in

Emerging Trends in Computer Applications (ICAETC-2023)
December 21–22, 2023, Lucknow, India

Edited by
Anurag Tiwari
Manuj Darbari

Boca Raton London New York

CRC Press is an imprint of the

Taylor & Francis Group, an informa business
First edition published 2024
by CRC Press
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

and by CRC Press

2385 NW Executive Center Drive, Suite 320, Boca Raton FL 33431

© 2024 selection and editorial matter, Anurag Tiwari and Manuj Darbari; individual chapters, the
contributors

CRC Press is an imprint of Informa UK Limited

The right of Anurag Tiwari and Manuj Darbari to be identified as the authors of the editorial
material, and of the authors for their individual chapters, has been asserted in accordance with
sections 77 and 78 of the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form
or by any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without permission in
writing from the publishers.

For permission to photocopy or use material electronically from this work, access
www.copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive,
Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please contact
[email protected]

Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

ISBN: 9781032999012 (pbk)

ISBN: 9781003606635 (ebk)

DOI: 10.1201/9781003606635

Typeset in Sabon LT Std

by HBK Digital
 Contents

List of figures xiv

List of tables xxii

Chapter 1 A systematic review on recent controllers for an industrial wastewater

treatment plant 1
K. V. Reshma, A. Vimala Juliet, and G. Glan Devadhas

Chapter 2 Green SaaS adoption in cloud computing: An investigation into environment

sustainability13
Bhanu Priya, Bhupal Arya, K. M. Pandey, and Prashant Verma

Chapter 3 Anomaly detection in wireless sensor network 24

Mayank Shukla, Sneha Yadav, Abhay Pratap Singh, Fizza Rizvi, and Surya Vikram Singh

Chapter 4 Revolutionizing object recognition beyond CNN and YOLO with deep learning
breakthrough31
G. S. S. S. S. V. Krishna Mohan, Mahammad Firose Shaik, G. Usandra Babu,
R. G. V Prasanna, Pinagadi Venkateswara Rao, and Inakoti Ramesh Raja

Chapter 5 Investigating efficacy of emerging ICTs for the Indian construction sector 40
Sanjay Pandey and G. T. Thampi

Chapter 6 AgriTech: Empowering agriculture through integrated technology 50

Pankaj Kunekar, Anushka Popalghat, Onkar Borude, Vishal Gavali, Dnyanesh Gholap,
and Sarvesh Hadole

Chapter 7 Online and offline learning in universities: A student’s dilemma 59

Ruchi Agarwaland Rahul Pradhan

Chapter 8 Strengthening UAV network security: Advanced authentication-based intrusion

detection systems 65
Altaf C, Shanila Mahreen,C Atheeq, and Shaista Khanam

Chapter 9 Comparative analysis of wireless routing protocols for security threat—A review 74
Bhupal Arya, Amrita Kumari, and Jogender Kumar

Chapter 10 Cascadability analysis of free space optical wireless data centers 81

Anand Kumar Dixit, Meenakshi Srivastava, and Rajiv Srivastava
vi Contents

Chapter 11 Wireless network with machine learning and deep learning techniques 88
Bhupal Arya, Amrita Kumari, and Jogendra Kumar

Chapter 12 Assessing technology innovation of cellular health 100

M. Vasumathi Devi, Y. Muvvana, Y. Sai Likhitha, M. Gowri Bhavani, and M. Vishnu Priya

Chapter 13 Reform dentistry: A web-based platform for revolutionizing dentistry 107

Vineet Agrawal, Nikita Verma, Aman Yadav, and Khyati Maurya

Chapter 14 Approach to recognize fake news 114

Arya Srivastava, Palak Chaubey, Manasvi Sonkar, Rudrendra Bahadur Sing, and Vinayak

Chapter 15 Cybersecurity measures for safeguarding medical image data 121

Zeenath, K. Durga Devi, andJohn W. Carey M.

Chapter 16 Detection of phishing attacks using machine learning  129

Siddharth, Rajeev Srivastava, Harsh Raj, Shourya Dwivedi, Shourya Dwivedi,
Rohit Singh, and Nitish Chaurasiya

Chapter 17 A comprehensive exploration of aircraft detection in synthetic and satellite images 135
M. Madhu Bala, G. Ramesh Chandra, G. Pavan Teja, M. Sai Sudha, Sk. Sabeeha Kouser,
and M. Ajay Kumar

Chapter 18 Mitigating man-in-the-middle attack in UAV network using authentication

mechanism based on chaotic maps  144
C. Atheeq, Layak Ali,C. Altaf , and Aleem Mohammed

Chapter 19 A comprehensive multi-modal sentiment analysis approach for social media content
integration153
Sheela S., Suresh Balakrishnan T., Uttham Sing K., Sujitha E., Soundarrajan R., and Veena M.

Chapter 20 A proposed deep learning framework for internet-of-medical things 159

Faiyaz ahamad and Syed Hauider Abbas

Chapter 21 Analysis of novel routing protocol to minimize delay in VANET: A comprehensive

simulation for VANET  165
Kamlesh Kumar and Bobby Sharma

Chapter 22 Securing the secrets of 5G: Mitigating eavesdropping threats and enhancing network
integrity175
Mamidisetti Sai

Chapter 23 A comprehensive crop prediction model leveraging optimization algorithms and

machine learning classifiers 183
Neha Chauhan, Dinesh Sharma, and Deepika Dhaneja

Chapter 24 Malaria detection using convolutional neural networks: A deep learning approach 192
Navin Kumar Agrawal
 Contents vii

Chapter 25 An optimal emergency vehicle path finding mechanism using routing algorithms 197
Hina Naaz, Roshan Jahan, and Faiyaz Ahamad

Chapter 26 Smart city solutions: Enhancing infrastructure with LoRa multi-hop networks 203
Faraz Ahmad, Saleha Mariyam, and Faiyaz Ahamad

Chapter 27 Enhancing online customer engagement through strategic optimization:

Customized page ranking and analytical insights via web mining techniques 210
Ambareen Jameel, Mohd Usman Khan, and Faiyaz Ahamad

Chapter 28 Towards unbreakable cloud security: Symmetric key cryptography and

blockchain synergy218
Khushabu Agrawal, Gunjan Verma, Priya, Divya Sharma, and Nidhi Pruthi

Chapter 29 Neuralink’s approach to spinal cord injuries and vision impairment

through pioneering brain–machine interfaces 226
Manoj M., Banupriya N., Gayathri V., Girinath D., and Shibi Kannan A.

Chapter 30 Honeypots and honeynets: investigating attack vectors 233

Pawandeep Kaur and Harpreet Kaur

Chapter 31 A review on deep learning frameworks for pose estimation during exercise for
rehabilitation240
Banupriya N., Haris Dominic Savier, Jeeva V., and Uma R.

Chapter 32 SwiftScan SpeedPro: OpenCV-powered vehicle speed monitoring 246

Vishal Kumar Sinha, Ayush Bharti, and Selvin Paul Peter J.

Chapter 33 Hawkeye-intelligent surveillance system 255

Raj Kalantri, Umang Thakur, and Selvin Paul Peter J.

Chapter 34 Sentimental analysis for amazon product review 263

Sumathi S., Mohanapriya A., Mohammed Aashif M J., Vidhur S., and Vijayalakshmi K.

Chapter 35 Crop recommender system using machine learning approach 271

Sujatha V., Lavanya N., Karunasri V., SaiSindhu G., and Madhavi R.

Chapter 36 Cloud malware detection using heuristic techniques 277

Lakshmi Triveni D., Alapati Naresh, Sree Harshitha A., Koteswari. K., and
Bhavana Lakshmi P.

Chapter 37 Guarding your home: A seamless Iot-driven security system with high security 285
Murala Divyasree, Alapati Naresh, Tammisetty Jyothi, Perumalla Sahithi, and Vinukonda
Ankitha

Chapter 38 The challenges of enforcing cybercrime laws in the age of E-governance:

A literature review 292
Manuj Darbari, Naseem Ahmed, and Abhishek Kumar Singh
viii Contents

Chapter 39 A exhaustive review on advancements and challenges in low power wireless sensor
networks297
Manuj Darbari, Naresh Chandrab, Diwakar Yagyasenc,
Anurag Tiwarid, and Sandeep Kumar Mishrae

Chapter 40 Automated personality evaluation of children based on emotional speech 302

Alapati Naresh, Basireddy Yasaswi, Golla Sri Lakshmi, Konakandla Meghana,
and Allamudi Yamini

Chapter 41 Redefining human activity recognition with LSTM and GRU networks 308
Vasumathi Devi M., Sai Sreeja S., Anusha P., Mounika Bindu B., and Sirisha M.

Chapter 42 Cybersaferoutes: Protecting IoT networks from hybrid attack with security
measures315
Ganga Bhavani P, Lakshmi Tirapathamma S, Reshma SK, Swetha B, and Himaja N

Chapter 43 Harmful content classification in social media using gated recurrent units and
bidirectional encoder representations from transformer 321
Sujatha V., Tejaswi Y., Pravalika V., Pavani P., and Ch Sravani

Chapter 44 Detection of missprounciation using deep learning 328

Naga Vardhani B, Naga Nandini K, Nandhakeerthi M, and Rupanjali M

Chapter 45 IOT-enabled cloud solutions for reliable health monitoring 335

Lakshmi Sravani G., Lalitha Chandrika N., Lavanya S., Vijaya Lakshmi S., and
Sri Lakshmi Prasanna S.

Chapter 46 Detection of melanoma using DenseNet-based adaptive weighted loss function 341
Rishi Agrawal

Chapter 47 Advanced reconfigured solar arrays to attenuate shading loss under partial shading
conditions: experimental validation 346
Vijay Laxmi Mishra, Yogesh K. Chauhan, and Kripa S. Verma

Chapter 48 Understanding the relationship of quantum mechanics and machine learning 352
Sushamana Sharma

Chapter 49 Employability prediction: A machine learning model for computer

science students 359
Abhishek Bhattacherjee and Vinay Kukreja

Chapter 50 A comprehensive review of innovations in stray cattle detection 365

Abdullah Nizami, Rudrendra Bahadur Singh, Haris Javed, Mohd Anas Nadeem, and
Alok Kumar

Chapter 51 Enhancing emotional insight: NLP powered sentiment analysis 373

Jyoti Gaur, Stuti Singh, and Shelja Sharma
 Contents ix

Chapter 52 Decision-making using argumentation mining framework on English text 379

Rudrendra Bahadur Singh, Shobhit Sinha, Ankita Singh, Alok Kumar Thakur, Akhil Chaurasiya,
Nishtha Maurya, and Arpita Yadav

Chapter 53 Potato leaf disease classification and prediction using CNN 386
UdayPratap Singh, Sandeep Kumar Mishra, ShubhamVishwakarma, Suraj Kumar,
and Vinayak Pandey

Chapter 54 Exploring the impact of integrated technological solutions on student engagement

administrative efficiency 393
Anurag Tiwari, Ayush Maurya, Arushi Baranwal, Amritesh Gupta, and Anurag Pathak

Chapter 55 Early stage lung cancer detection using RESNET 399

Anjali Srivastava, Vinayak, Aarohi Rai, and Ainy Khan

Chapter 56 Cryptographic image concealment with neural networks 406

Chinnala Balakrishna,Shaik Saidhbi, Balajee Maram, Mallikharjuna Rao K., B. Santhosh
Kumar, and Sasibhushana Rao Pappu

Chapter 57 An efficient CNN-based approach for automated animal intrusion detection 412
Ramamani Tripathy, S V Achuta Rao, Maheswari. P, Mallikharjuna Rao K,
B. Santhosh Kumar, and Balajee Maram

Chapter 58 Next-generation smart cities: Integrating blockchain, IoT, deep learning, and
Dingo optimization 419
K. Sankar, Ramamani Tripathy, S. Nandhini, Aleem Ali, B. Santhosh Kumar,
and Balajee Maram

Chapter 59 Intelligent diagnosis of diabetic retinopathy: Leveraging machine

and deep learning 425
Balajee Maram, S. Arun Joe Babulo, B. Manivannan, B. Santhosh Kumar, T. Daniya, and
Sasibhushana Rao Pappu

Chapter 60 Digital dwelling hub: Revolutionizing real estate with AR and smart home
integration431
Tanya Omar, Rajeev Srivastava, Tanya Singh, Vaibhav Srivastava, Suraj Chhetri, and
Shaurya Kumar

Chapter 61 Mobilenet-v3: A comprehensive survey of object detection algorithms using CNN 438
Rasheeq Zehra, Omkar Sharma, Vinayak, and Rudrendra Bahadur Singh

Chapter 62 Rainstorm prediction system 445

Muskan Gupta, Ankit Khare, Kirti Pandey, Pritismita Thakur, and Muskan Gupta

Chapter 63 Advancements in assistive technologies for web accessibility: A systematic review 451
Ayushi, Padmini Mishra, Anusha Ansari, and Drishti Maurya

Chapter 64 Predicting bitcoin prices: A machine learning approach for accurate forecasting 457
Rishabh Jain, Shekhar Srivastava, and Prakhar Shukla
x Contents

Chapter 65 A comprehensive review on multi disease prediction web app: QuadraDiag 464
Khushi Saxenaa, Keerti Saxenab, Vishesh Guptac, Swatid, and Kamlesh Kumare

Chapter 66 A comprehensive review on tomato plant disease detection using

convolutional neural network  473
Rudra Pratap Singh, Sandeep Kumar Mishra, Samriddhi Chaurasia, Sameer Thakur,
and Sachin Yadav

Chapter 67 Building a decentralized non-fungible token marketplace: Leveraging blockchain

technology for secure digital asset exchange 480
Vaibhav Bansal, Shivam Arora, Sonia Deshmukh, and Ankit Saini

Chapter 68 CollegeGo: leveraging alumni data for improved college placements 487
Avneesh Agrahari, Subha Mishra, Rudrendra Bahadur Singh, Kumar Ronit,
and Pahun Ratna

Chapter 69 An in-depth review of AI/ML techniques in emotion-preserving multilingual video

translation for practical cross-cultural communication 492
Diwakar Yagyasen, Akshit Kumar Tiwari, Abhishek Kevin Gomes, Aditya Yadav, and Harsh
Anand Gupta

Chapter 70 Data analysis for revenue optimization in the hospitality sector 499
Pankaj Kunekar, Parul Patle, and Anagha Gajaralwar

Chapter 71 Technological solutions for E-waste management and disposal  506

Aditya Pratap Singh and Akshat Chaturvedi

Chapter 72 Comparative study of various transformation techniques in image fusion 513

Kunal Kishor Jha, Sumit Kumar Jha, Hari Om Shanker Mishra, Amit Dhawan,
Manish Tiwari

Chapter 73 Blockchain transaction processing: challenges and resolutions 521

Kamal Kant, Sarvesh Pandey, and Udai Shanker

Chapter 74 Fake social media profile detection using machine learning 527
Anurag Shukla, Shreya Chaurasia, Tanushri Asthana, Tej Narayan Prajapati, and
Vivek Kushwaha

Chapter 75 Prompted LLMs as chatbot modules for long open-domain conversation 534
Yash Kumar Singh, Deepak Pant, Syed Mohd Askari, and Shivam Kumar Pandey

Chapter 76 Smart locking system using machine learning technique 542

Sadhana Singh, Jhanak Verma, Aditya Gupta, and Priyanka Sharma

Chapter 77 Mind Matters: Psychological analysis using CBT Chatbot 547

Preeti Maurya, Gyanendra Kumar, Aditya Singh Yadav, Kumari Anushka Gupta,
and Adarsh Tiwari
 Contents xi

Chapter 78 Optimizing patient outcomes through integrated health analysis and management
strategies  554
Anurag Shukla, Sambhav Pathak, Rahul, Rishikant Yadav, and Raj Gupta

Chapter 79 Data leakage detection 562

Neha Singh, Sandeep Kumar Mishra, Rahul Kumar, and Sandeep Vishwakarma

Chapter 80. Vision-based lip-reading system using deep learning  569

Abhishek Pratap SinghAmit Kumar Sachan, Akram Ansari, and Abhishek Dubey

Chapter 81 Reform dentistry: A web-based platform for revolutionizing dentistry 575

Vineet Agrawal, Nikita Verma, Aman Yadav, and Khyati Maurya

Chapter 82 AI genius: Integrated intelligence platform 582

Azhar Ahmad, Aneesh Sahu, Amir Ali, Anjali Maurya, and Saurabh kumar Jain

Chapter 83 Agriculture: Soil analysis for suitable crop and fertilizer prediction 589
Shreya Singh, Saurabh Kumar Jain, and Stuti Rastogi

Chapter 84 Comprehensive examination of system surveillance method employing keystrokes

logging technology 597
Bajrangi Vishkarma, Kamlesh Kumar, Aman Verma, Atul Singh, and Abhishek Singh

Chapter 85 Health prediction application: A comprehensive literature review 604

Tushar Agrawal, Prince Kumar Singh, Aditya Kumar, and Mohit Chowdhary

Chapter 86 E-portal for case management and hearing 610

Anushka Shukla, Anurag Shukla, Ashish Kumar Yadav, and Anushka Srivastava

Chapter 87 Enhancing electoral processes: The role of “Smart Voting Systems” 616
Madhulika Sharma, Anuj Singh, and Gulshan Varshney

Chapter 88 Advancements in rainfall prediction systems: A comprehensive review of

techniques and models 622
Adarsh Vishwakarma, Ankit Khare, Amit Verma, and Aditya Pandey

Chapter 89 FindServ 628

Mayank Srivastava, Pratibha Dixit, Adarsh Verma, Sadaf Ali

Chapter 90 IoT devices security against common threats 635

Mayank Srivastava and Prabudh

Chapter 91 Predict-a-disease: Survey 642

Mayank Srivastava, Aditya Singh Chauhan, and Abhinav Bhardwaj

Chapter 92 Securing MANETs against black hole attacks: A novel approach using secure
knowledge algorithm 648
Mohammed Abdul Bari, Arshad Ahmad Khan Mohammad, Arif Mohammad, and Abdul
xii Contents

Chapter 93 AI-powered navigation of the digital frontier: Analyzing modern trends in

digital marketing and their influence on consumer purchasing patterns with
special reference to pharma industry 655
Abhay Singh Rajawat, Ved Prakash, and Shailendra Singh Chauhan

Chapter 94 Fake video detection 665

Mahendra Singh, Saurabh Jain, Laraib Khan, and Iaswarchandra

Chapter 95 Olympic odyssey: Tracing 120 years of athletes sports, and global influence 673
Shreya Verma, Shreya Mishra, Prakriti Singh, Sneha Seth, and Sanjay Kumar Gupta

Chapter 96 Crop prediction system using machine learning 680

Neha Chauhan, Devansh Tomar, Gaurav Singh, Gaurav Mishra, and Anand Mishra

Chapter 97 A mobile application to assist the women in gestation period 686

V. lavanya, Madhu Hasitha Manne, Gnana Sri Manasa Veeramachaneni, Aditya Ram Manam,
and Sajid khan Mohammed

Chapter 98 I-tourism planner 692

Aayush Kumar, Himanshu Ojha, Pratham Singla, and Amit Kumar Jaiswal

Chapter 99 Blockchain based E-vault system for legal records 697

Shekhar Srivastava, Amritanshu, Maitri Jain, and Rahul Maurya

Chapter 100 Anomaly detection in wireless sensor network 704

Mayank Shukla, Sneha Yadav, Abhay Pratap Singh, Fizza Rizvi, and Surya Vikram Singh

Chapter 101 Enhancing fault tolerance in distributed systems through machine learning
techniques711
Mohd Haroon, Manish Madhav Tripathi, Jameel Ahmad, Shish Ahmad, and Mohd Husain

Chapter 102 Predicting breast cancer risk: A machine learning approach 717
Navin kumar Agrawal

Chapter 103 A multi-objective optimization algorithm for fog computing with trust and energy
awareness  723
Ruchi Agarwal, Saurabh Singhal, and Ashish Sharma

Chapter 104 Pediatric brain tumor detection in MRI: A machine learning perspective 729
Sandeep Kumar Mishra and Sheeba Praveen

Chapter 105 Enhancing trust in the social internet of things (SIoT): A comprehensive review 736
Gyanendra Kumar and Anuradha Misra
 Contents xiii

Chapter 106 Fostering plant health: Utilizing YOLOv7 framework to identify leaf diseases 743
Rajya Lakshmi R, Sasikala V, Divya Likhitha D, Rishita Reddy Ch, and VenkataSai Vidyasri
Abhishainy A

Chapter 107 A review on-malignancy detection in lung nodules using CNN 751
Krish Yadav, Vinayak, Pratibha Yadav, Lakshmi Pandey, Himanshu Sharma, and Smita Yadav

Chapter 108 Next-generation parking system using IoT 756

K. Geetha, Gude Ganesh Satya Kumar, Raghu Ram C., and A. G. Sreedevi

Chapter 109 A review on smart health diagnosis 760

Kishan Kumar Rauniyar, Vijay Yadav, Padmini Mishra, Sanskar Singh, and
Ashish Kumar Yadav

Chapter 110 Android app threat analysis techniques: a review 765

Ujjwal Shukla, Gyanendra Kumar, Syed Haidar Abbas, and Vaibhav Paliwal

Chapter 111 AI Powered Legal Documentation Assistant 770

Yogesh Shekhawat, Utkarsh Tiwari, Syed Hasan Mehdi, and Himanshu Vaishy

Chapter 112 Ideas to boost fitness activities and assist in keeping fit 777
Neha Chauhan, Vasu Gupta, Harshita Bajpai, and Kim Gupta

Chapter 113 Using existing CCTV network for crowd management, crime prevention and work
monitoring: A novel approach for object detection 783
Manvi Agrawal
 List of figures

Figure 1.1 Automatic control flow of aeration tank DO content by the improved IPI control
algorithm6
Figure 1.2 shows the process scheme of the WWTP to which MPC is applied 7
Figure 1.3 Workflow of the proposed deep learning-driven forecasting procedure 9
Figure 1.4 Benchmark Simulation Model 1(BSM 1) 10
Figure 2.1 Green IT components 15
Figure 2.2 Greener cloud placement in cloud computing technology 15
Figure 2.3 Green cloud service model of software as a service 19
Figure 4.1 YOLO architecture 34
Figure 4.2 Feature extracting process by RCNN methodology 35
Figure 4.3 Block diagram for text to speech conversion 36
Figure 4.4 Detection of object by camera module 37
Figure 4.5 List of detected objects 38
Figure 5.1 Types of construction companies contributed to the study 43
Figure 5.2 Cadre-based classification of respondents 43
Figure 5.3 Harnessing ICTs for construction industry tasks 45
Figure 5.4 Perceived benefits of using ICTs for construction companies 46
Figure 5.5 Hindering factors for ICT adaptation in the Indian construction sector 47
Figure 5.6 Overall performance of construction companies after ICT adaptation 48
Figure 6.1 Soil testing lab locations 52
Figure 6.2 Heatmap for crop recommendation 52
Figure 6.3 Crop recommendation 52
Figure 6.4 Plant disease prediction 52
Figure 6.5 Blockchain-based marketplace 52
Figure 6.6 Crop price prediction 53
Figure 6.7 Real-time weather monitoring 53
Figure 6.8 Gamified plant growth 54
Figure 6.9 Confusion matrix 54
Figure 6.10 Accuracy chart 54
Figure 6.11 Crop recommendation flow diagram 54
Figure 6.12 Visualizing plant growth using image processing 55
Figure 7.1 Subject distribution according to the ages 60
Figure 7.2 Subject distribution according to the gender 60
Figure 7.3 Subject distribution according to education qualification 60
Figure 7.4 Feedback– Distribution of Agreement and disagreement on above issues 62
Figure 8.1 Comparison of proposed method with existing with respect to PDR 71
Figure 8.2 Comparison of proposed method with existing with respect to delay 72
Figure 8.3 Comparison of proposed method with existing with respect to overhead 72
Figure 9.1  76
Figure 10.1 Communication between the ToRs using LoS and NLoS links 82
Figure 10.2 LoS communication between the cascaded racks (BUS topology) 82
Figure 10.3 LoS communication between the racks (16 Racks topology) 83
Figure 10.4 LoS communication between the racks is 3 meters (a) FOV=70° (b) FOV=10° 84
 List of figures xv

Figure 10.5 LoS communication between the racks is 3√2 meters (a) FOV=70° (b) FOV=10° 84
Figure 10.6 LoS communication between the racks is 6 meters (a) FOV=70° (b) FOV=10° 85
Figure 10.7 LoS communication between the racks is 6√2 meters (a) FOV=70° (b) FOV=10° 85
Figure 10.8 LoS communication between the racks is 9 meters (a) FOV=70° (b) FOV=10° 85
Figure 10.9 LoS communication between the racks is 9√2 meters (a) FOV=70° (b) FOV=10° 86
Figure 10.10 Probability of Error for various distances between the racks 86
Figure 11.1 Deep learning in wireless network 90
Figure 11.2 ML using ANN and DNN 91
Figure 11.3  91
Figure 11.4  93
Figure 11.5  94
Figure 11.6  96
Figure 11.7  97
Figure 12.1 Conceptual framework.  102
Figure 12.2 Methodology schema 104
Figure 12.3 Types of admission.  105
Figure 12.4 City Code Patient Count 105
Figure 13.1 Block diagram 110
Figure 15.1 Comparison of proposed method with existing with respect to PDR 126
Figure 15.2 Comparison of proposed method with existing with respect to Delay 126
Figure 15.3 Comparison of proposed method with existing with respect to Overhead 126
Figure 15.4 Comparison of proposed method with existing with respect to Throughput 127
Figure 17.1 A Sample DOTA image of airplane category with OBB representation
(a) v1.0 (b) v1.5 and (c) v2.0waveform of BLDC motor 136
Figure 17.2 A Sample synthetic Image 137
Figure 17.3 Aircraft detection framework using YOLO 137
Figure 17.4 Architecture of Detectron2 137
Figure 17.5 FR-CNN architecture in aircraft detection  138
Figure 18.1 Comparison of proposed method with existing with respect to PDR 150
Figure 18.2 Comparison of proposed method with existing with respect to Delay 150
Figure 18.3 Comparison of proposed method with existing with respect to Overhead 151
Figure 19.1 Text data based sentiment analysis 155
Figure 19.2 Image-based sentiment analysis 155
Figure 19.3 Live video-based sentiment analysis 156
Figure 19.4 Audio-based sentiment analysis 156
Figure 19.5 Emotion detected using text 157
Figure 19.6 Emotion detected using Image 157
Figure 19.7 Emotion detected using Image 157
Figure 19.8 Emotion detected using Audio 157
Figure 20.1 The proposed model diagram 161
Figure 20.2 Comparison with existing frameworks 163
Figure 21.1 A Differential Value of VANET nodes 171
Figure 21.2 Decision of node movements 171
Figure 21.3 Simple AODV setup 172
Figure 21.4 Basic AODV 25 nodes 172
Figure 21.5 Starting simulation of basic AODV 25 nodes 172
Figure 21.6 AODV 20 nodes 172
Figure 21.7 Starting simulation of AODV 20 nodes 173
Figure 21.8 AODV 25 nodes 173
Figure 21.9 Starting simulation of AODV 25 nodes 173
Figure 21.10 AODV 30 nodes 173
xvi List of figures

Figure 21.11 Starting simulation of AODV 30 nodes 173

Figure 21.12 Bar chart during delay occurrence on different nodes 173
Figure 22.1 Eavesdropping attack 176
Figure 22.2 MOSPR working process 178
Figure 24.1 Classification of labelled image 194
Figure 24.2 Loss function for validation data 195
Figure 24.3 Accuracy on validation data 195
Figure 24.4 AUC-ROC curve 195
Figure 25.1 Grid map graph 199
Figure 25.2 Performance comparison graph 201
Figure 26.1 Layers of proposed framework for smart city 204
Figure 26.2 Proposed framework for integration of LoRa technology for cross platform
interoperability207
Figure 26.3 Significant improvement in data transmission efficiency, after integration of LoRa
technology in urban infrastructure 208
Figure 26.4 Gateway statistics facilitating inter sensor communication 208
Figure 26.5 Packet transmission statistics 208
Figure 27.1 Web data extraction techniques 211
Figure 27.2 https://www.shopify.com/in/aboutpage  213
Figure 27.3 Combined data 214
Figure 27.4 List of top countries for marketing Web Log Hit 214
Figure 27.5 List of top countries for Web Log Hit 216
Figure 27.6 Number of HITS 216
Figure 28.1 Classification of cryptography algorithm in cloud computing 220
Figure 28.2 The process of symmetric-based key algorithm 220
Figure 28.3 Represents the overall process of Proposed Cryptography algorithm using
blockchain methodology in cloud computing 221
Figure 29.1 Representation of BMI 227
Figure 29.2 Elon Musk’s Neuralink 228
Figure 29.3 BCI chip installation 229
Figure 29.4 Visual prosthesis representation 229
Figure 29.5 Prefrontal Cortex 230
Figure 29.6 Representation of Mental Synthesis Theory 231
Figure 30.1 Honeypot architecture 235
Figure 30.2 Classification of honeypot 235
Figure 31.1 The general GCN diagram from the input data to the desired output 241
Figure 31.2 Branches of classification in body 244
Figure 31.3 The dataset images for the pose estimation 244
Figure 31.4 The classification and the correction branch diagram for the image model 244
Figure 32.1 System flow 248
Figure 32.2 Framework architecture 250
Figure 32.3 Project model-object tracking 251
Figure 32.4 Speed calculation 251
Figure 32.5 Saves vehicle data 252
Figure 32.6 Use case diagram 252
Figure 32.7 Timing table 252
Figure 32.8 Stage by stage measurements 253
Figure 32.9 Stage-by-stage ROC curve 253
Figure 33.1 System flow 256
Figure 33.2 System architecture 258
Figure 33.3 Framework architecture 258
 List of figures xvii

Figure 33.4 ResNet architecture 259

Figure 33.5 LSTM working 260
Figure 33.6 Beem search algorithm 260
Figure 33.7 Architecture 260
Figure 33.8 Sequence diagram 261
Figure 33.9 Class diagram 261
Figure 33.10 Upload video 261
Figure 33.11 Live captions generated 261
Figure 33.12 Comparison with other captioning models 261
Figure 34.1 General architecture 264
Figure 34.2 Implementation of dataset 266
Figure 34.3 Visualization of dataset 266
Figure 34.4 Visualization using word cloud 267
Figure 34.5 Tabulation of model 269
Figure 34.6 Visualization for F1-score 269
Figure 35.1 Architecture 272
Figure 35.2 Methodology 274
Figure 35.3 Accuracy comparison 275
Figure 35.4 Dataset distribution (area) 275
Figure 35.5 Dataset distribution (Production) 276
Figure 35.6 Heat map of the dataset 276
Figure 36.1 Generation of opcode sequences 280
Figure 36.2 Opcode sequence instructions 280
Figure 36.4 Bar graph representing opcode versus normalized frequency in benign and
malware281
Figure 36.3 Benign and malware images 281
Figure 36.5 Accuracy of CNN vs KNN, NB 283
Figure 37.1 Flowchart 289
Figure 37.2 Motion detection 290
Figure 37.3 Authentication 290
Figure 37.4 Authentication request message 290
Figure 37.5 Captcha generation 290
Figure 39.1 WSN deployed in forest area 299
Figure 40.1 Workflow of the model 304
Figure 40.2 CNN architecture 305
Figure 40.3 Output 306
Figure 41.1 LSTM and GRU model architecture 311
Figure 41.2 Applications of 52 312
Figure 41.3 Accuracy 312
Figure 41.4 Precision 312
Figure 41.5 Recall 313
Figure 41.6 F1-Score 313
Figure 42.1 Architecture diagram 318
Figure 42.2 Throughput 319
Figure 42.3 Packet delivery ratio 319
Figure 42.4 Energy delay 319
Figure 42.5 Control overhead 319
Figure 42.6 Loss 319
Figure 43.1 Hybrid GRU-BERT sentinel 324
Figure 43.2 Accuracy 325
Figure 43.3 Precision 325
xviii List of figures

Figure 43.4 Recall 325

Figure 43.5 F1-Score 326
Figure 43.6 Loss 326
Figure 44.1 CNN and BiLSTM architecture model 331
Figure 44.2 Accuracy 332
Figure 44.3 Loss 332
Figure 44.4 Precision 333
Figure 44.5 Recall 333
Figure 44.6 F1-Score 333
Figure 21.1 Architecture of reliable health monitoring 337
Figure 21.2 AES structure 338
Figure 21.3 Accuracy data 338
Figure 21.4 Precision data 338
Figure 21.5 Recall data 339
Figure 21.6 Loss data 339
Figure 21.7 F-score data 339
Figure 46.1 Samples and statistical distribution of ISIC 2019 Dataset  342
Figure 46.2 Melanoma detecting Uging DenseNet-201 344
Figure 46.3 Feature concatenation process of DenseNet  344
Figure 46.4 Proposed model performance graph for training and validation 344
Figure 47.1 Electrical circuit of a solar cell 347
Figure 47.2 Solar module connections 347
Figure 47.3 Experimental setup 348
Figure 47.4 Shading conditions 348
Figure 47.5 Output curves (a) TCT (b) PCR 348
Figure 47.6 Global power plot 349
Figure 47.7 Shading loss plot 349
Figure 47.8 Execution ratio plot 349
Figure 47.9 Fill factor plot 350
Figure 47.10 Experimental curves 350
Figure 48.1 Showing labeled data for supervised ML 354
Figure 49.1 Proposed methodology 361
Figure 51.1 Role of NLP 374
Figure 51.2 Domains of sentimental analysis 375
Figure 52.1 Data flow architecture 380
Figure 53.1 Healthy leaf/Early blight/Late blight 386
Figure 53.2 Process flow diagram 387
Figure 54.1 Process flow chart 394
Figure 55.1 AlexNet architecture 400
Figure 55.2 Problem identification using AlexNet 401
Figure 55.3 VGG16 model architecture 401
Figure 55.4 RESNET50 model architecture 402
Figure 56.1 Alice’s neural network setup  408
Figure 57.1 Animal Intrusion Detection Approaches  414
Figure 58.1 Blockchain’s Role in Smart Cities  420
Figure 61.1 Mobilenetv3 architecture 439
Figure 62.1 Diagram for rainstorm prediction system 446
Figure 64.1  459
Figure 65.1  465
Figure 67.1 System architecture 482
Figure 67.2 Working diagram 483
 List of figures xix

Figure 67.3 Image categories minted 484

Figure 67.4 Trends in transaction 484
Figure 68.1 Proposed approach for application 488
Figure 69.1 Methodological Process Flow Chart 494
Figure 69.2 Metric analysis 494
Figure 70.1 Data Model (Star Schema) 502
Figure 70.2 Final dashboard created 502
Figure 70.3 Revenue tooltip for trend by week 502
Figure 72.1 DWT based image fusion 515
Figure 72.2 DCT based image fusion 515
Figure 72.3 NSCT based image fusion 516
Figure 72.4 Curvelet transform flow graph 516
Figure 72.5 (a) Picture 1, (b) Picture 2, (c) DCT based Fusion, (d) DWT based
Fusion, (e) NSCT based fusion, (f) Curvelet based fusion 517
Figure 72.6 (a) CT Picture, (b) MRI Picture, (c) DCT based fusion, (d) DWT based fusion,
(e) NSCT based fusion, (f) Curvelet based fusion 517
Figure 76.1 Different face templates 544
Figure 76.2 Different types of matching 544
Figure 76.3 Facial Recognition system working 544
Figure 76.4 Fingerprint Recognition 544
Figure 76.5 Processing of data 544
Figure 76.6 Different types of fingerprint postures 545
Figure 76.7 Fingerprint matching 545
Figure 77.1 Architecture of the application 550
Figure 78.1 MERN stack development 557
Figure 79.1 System architecture block diagram 564
Figure 80.1 Block layout for the lip-reader system  570
Figure 80.2 Architecture of ResNet50 572
Figure 81.1 Block diagram 578
Figure 82.1 583
Figure 82.2 583
Figure 84.1 System flow diagram 602
Figure 85.1 Data driven diabetes prediction 607
Figure 85.1  612
Figure 85.2  612
Figure 88.1 Correlation matrix of major Parameters 623
Figure 88.2 Accuracy and time limitations of ML models in cloudburst prediction 624
Figure 88.3 Efficiency of ML algorithms for Cloudburst Predicitons 626
Figure 89.1 Basic working flowchart of proposed model 629
Figure 89.2 Training phase of face Identification 630
Figure 89.3 Testing phase of face Identification 631
Figure 89.4 KNN sample case image 631
Figure 89.5 PCA sample case image 632
Figure 89.6 Sample image 633
Figure 90.1 Techniques for remote security management 637
Figure 90.2 Security Server Diagram 639
Figure 90.3 Changing Default Settings 640
Figure 90.4 Downloading Firewall 640
Figure 90.5 Activating Firewall 640
Figure 91.1 Accuracy result of decision tree classifier 644
Figure 91.2 Accuracy results of Random Forest Classifier 645
xx List of figures

Figure 91.3 Numerical Distribution of Dataset 646

Figure 91.4 Part of the Training Dataset with CSV Format 646
Figure 91.5 Models Comparison 646
Figure 92.1 Demonstrates the correlation between the packet delivery ratio and the number of
malicious nodes throughout a simulation duration of 100 seconds 652
Figure 92.2 Illustrates the correlation between the packet delivery ratio and the number of
malicious nodes, employing a simulation duration of 100 seconds 652
Figure 94.1 Working procedure of fake video detection 666
Figure 94.2 Data flow diagram of CNN Based Model 667
Figure 94.3  667
Figure 95.1 Most popular sports in the Olympics 675
Figure 95.2 Gender distribution 675
Figure 95.3 Number of participants in each season 675
Figure 95.4 Distribution of medals by season 676
Figure 95.5 Age distribution of Olympic athletes 676
Figure 95.6 Medal victories by age for gold, silver, and bronze medals 676
Figure 95.7 Age distribution of Olympic athletes 676
Figure 95.8 Most popular sports for Indian athletes 677
Figure 95.9 Gender distribution of Indian athletes 677
Figure 95.10 Age distribution of Indian athletes 677
Figure 95.11 Medal distribution for India 678
Figure 96.1 Methodology of crop prediction 682
Figure 96.2 Process flow model 682
Figure 96.3 Accuracy comparison of 3.2 models implications 684
Figure 97.1 Working model 687
Figure 97.2 No of calories for each trimester 688
Figure 97.3 Customized diet chart using random method 688
Figure 97.4 Architecture of the application 689
Figure 97.5 Differences between nearest-neighbor 689
Figure 97.6 Represents the nearby hospitals 690
Figure 97.7 Nearby hospitals 691
Figure 98.1 Popular tourism platforms 693
Figure 98.2 Architecture of the platform 693
Figure 99.1  698
Figure 101.1 Distributed system architecture 712
Figure 101.2 Reliability of fault in the distributed system 713
Figure 102.1 Performance evaluation using ROC curve 720
Figure 102.2 Performance evaluation using confusion matrix 720
Figure 102.3 Learning curve using logistic model 720
Figure 103.1 Flow diagram of trust-based system 725
Figure 103.2 Correlation between Fog Node Trust and Credibility 726
Figure 103.3 Correlation Between F-Rate and Number of Incoming Tasks 726
Figure 103.4 Energy Consumption Comparison Chart 726
Figure 103.5 Resource Optimization Graph Theta 727
Figure 104.1 Pediatric brain tumor detection and classification studies different model and their
accuracy733
Figure 104.2 Flow diagram of F1 score (adapted from toward data science) 734
Figure 105.1 Proposed flow chart 740
Figure 106.1 Workflow of object detection 745
Figure 106.2 Leaf image examples for object detection 747
Figure 106.3 Accuracy 748
 List of figures xxi

Figure. 106.4 Loss 748

Figure 106.5 Precision 748
Figure 106.6 Recall 749
Figure 106.7 F1 Score 749
Figure 108.1  757
Figure 108.2  757
Figure 109.1 Data flow diagram 761
Figure 111.1 Project architecture 771
Figure 111.2 Workflow flowchart 773
Figure 112.1 Methodological process flow chart 779
Figure 113.1  784
 List of tables

Table 1.1 Comparison of different optimal control strategies of the mean EQ and EC in storm
weather condition 4
Table 1.2 Statistical results of the proposed model’s performance evaluation (R2 and MSE) for
different stages of modeling (training, validation, testing and all data sets) 5
Table 1.3 The results of three different models 7
Table 1.4 Model Performance on the training, validation and test sets (R2 values) 7
Table 2.1 State-of-the-art of green cloud 18
Table 2.2 Green cloud SAAS performance metric 21
Table 5.1 Respondent’s opinion ranking of construction tasks performed by ICTs 45
Table 5.2 Ranking of the benefits as per respondent’s opinion 46
Table 5.3 Ranking of hindering factors for ICT adaptation  47
Table 6.1 Algorithm and accuracy 56
Table 6.2 Comparison with previous work 56
Table 7.1 Participants distribution on the basis of Age 61
Table 7.2 Participants distribution on the basis of Gender 61
Table 7.3 Participant’s distribution based on their educational background 62
Table 7.4 Responses in favor of online mode of learning 62
Table 8.1 Simulation Parameters 71
Table 10.1 Simulation parameters  83
Table 12.1 Mechanisms of NPT of mHealth in medicine 103
Table 12.2 Research questions and methodology 103
Table 15.1 Simulation requirements 125
Table 17.1 Results of algorithms used 140
Table 20.1 Execution evaluation of the proposed plan of action  162
Table 20.2 Execution comparison  162
Table 21.1 VANET Comparative Tables 172
Table 26.1 Technical layers in an urban infrastructure 204
Table 27.1 Experts’ contributions in web mining 212
Table 28.1 Comparative analysis of cryptographic approaches in cloud security 223
Table 30.1 Various solutions of honeypot 237
Table 31.1 Evaluation of results according to the datasets. 244
Table 41.1 Comparison with other methods 312
Table 46.1 Comparison with other state-of-the-art methods 344
Table 47.1 Quantitative Simulation Results 349
Table 49.1 Students’ employability features 361
Table 49.2 Performance Metrics for ML Algorithms 363
Table 51.1 Dataset popularly used  375
Table 53.1 Number of images with diseases 387
Table 53.2 Model performance comparison 389
Table 56.1 Comparing cryptographic image concealment with neural networks to traditional
steganographic techniques 410
Table 56.2 Comparison of PSNR, SSI, and BER 410
Table 57.1 Results comparison 416
 List of tables xxiii

Table 59.1 Comparative Analysis 429

Table 61.1 Comparative study of different papers 440
Table 62.1 Critical analysis comparison 447
Table 66.1 Accuracy table 477
Table 69.1 Critical analysis comparison 495
Table 69.2 Parametric analysis comparison 496
Table 70.1 Key metrics list top of form 501
Table 70.2 Insights from dashboard 503
Table 72.1 Performance metrics for transform-based algorithms’ multi-focus pictures 517
Table 72.2 Performance metrics for transform-based approaches’ multi-model images 518
Table 73.1 Comparison of traditional and blockchain 523
Table 73.2 Comparative analysis of consensus algorithms based on characteristics 524
Table 89.1 Result of model based on features 633
Table 92.1 Parameters of the simulation environment 652
Table 93.1 Research methodology table 659
Table 93.2 Hypothesis test summary 660
Table 93.3 Model summary 660
Table 93.4 Statistical significance of the model 660
Table 93.5 Estimated model coefficient 661
Table 96.1 Comparative Study 683
Table 104.1  732
Table 105.1 Overview of related works 739
Table 107.1  754
Table 112.1 Critical analysis comparison 779
1 A systematic review on recent
controllers for an industrial wastewater
treatment plant
K. V. Reshma1,a, A. Vimala Juliet1,b, and G. Glan Devadhas2,c
1
Department of Electronics and Instrumentation Engineering, SRM Institute of Science and
Technology, Kattankulathur Campus, Chennai, India
2
Department Electronics and Instrumentation Engineering, Vimal Jyothi Engineering College,
Kannur, Kerala, India

Abstract: A wastewater treatment plant (WWTP), also referred to as a sewage treatment plant or a water
reclamation facility, is a facility designed to treat wastewater and sewage from residential, commercial, and
industrial sources. Wastewater treatment (WWT) facilities are essential for maintaining the well-being of our
communities by safeguarding water supplies, maintaining public health, and minimizing environmental pol-
lution. The specific design and processes employed by a WWTP can change based on variables like the size of
the facility, the characteristics of the wastewater, and the required effluent quality standards. This paper pre-
sents the comprehensive technological review of PID, FLC, PLC, and MPC controllers for an industrial WWT
plant with advantages and disadvantages of the same and results are shown. In addition, this overview includes
Artificial intelligence is revolutionizing the field of water treatment by enhancing model prediction and control
strategies.

Keywords: Deep learning model (DNN), fuzzy logic controller (FLC), machine learning (ML), model predic-
tive controller (MPC), programmable logic controller (PLC), proportional integral derivative controllers (PID),
wastewater treatment plants (WWTPs)

1. Introduction (Li et al., 2021; Fu et al., 2023). Because of

the importance of addressing sustainability
WWTPs are amenities dedicated to treating issues in WWTPs, new technologies and pro-
incoming waters and returning them to their cedures have been created often.
natural cycle, where the pollution reduction In this light, control mechanisms have
process of urban residual waters is carried been developed and implemented at WWTPs
out. The primary objective is to minimize the with the aim of keeping pollution levels
concentration of total suspended particles, within the limits set by the rules. However,
dissolved organic matter, phosphorus, and WWTPs come in a wide variety of designs, AQ1: As
per style,
nitrogen-derived components in the treated each optimized for a different set of needs. the refer-
waters such that they are safe for the aquatic As a result, a significant amount of time and ence style
should be
life and surrounding ecosystems. The cost effort has been spent over many years (Han et num-
of running a wastewater treatment (WWT) al., 2019; Iratni et al., 2019; Newhart et al., bered.
Please
plant continues to be a major concern, hence 2019) on the development of sophisticated check and
efforts are being made to improve WWT sys- controllers for WWT facilities. The conven- review for
tems while keeping operational expenses low tional PID controllers, however, impossible to clarity.

[email protected]; [email protected]; [email protected]

a
CHECKED AND CONFIRMED

DOI: 10.1201/9781003606635-1
2 Innovations in Computational Intelligence: Trends and Technologies

extensively employed for complicated process lot of data and some serious pre-processing
for treating wastewater (Subari et al., 2022) to work well. It’s crucial to remember that
because to the occurrence of nonlinearities AI can’t replace mechanical models. There-
and turbulences, as well the contemplation fore, the goal of Using Machine Learning for
of particular optimization targets for unfa- Wastewater Treatment is to make better deci-
miliar systems. The reality is that there is a sions depending on projected performance
dearth of AI-based optimization approaches for existing facilities (Sundui et al., 2021;
that work well in a complex, uncertain, and Nourani et al., 2021).
nonlinear setting. Therefore, figuring out how Because of their superior feature extrac-
to acquire intelligent optimum controllers for tion, self-learning, and generalization abilities,
complicated nonlinear systems have been deep learning algorithms have found various
seen as a crucial step in the development of fantastic Implementations in recent years
the area of advanced control. across a wide range of areas. In DL, RNNs
The utilization of single-objective opti- with recurrent links and hidden layers are
mization approaches has generally applied recommended for analyzing difficulties with
in the field of wastewater management of a a historical context, while LSTM is intended
considerable duration, with the aim of opti- to improve the system’s ability to deal with
mizing the effectiveness of the WWT process. dependencies over the long run. However, the
It’s tough to pin down the precise weighting duration of time for training is a major factor
elements, however. As a result, several dif- in DL. Recent advances in genetic based con-
ferent kinds of multi-objective optimization trol algorithms for biological WWT facilities
algorithms, such MOPSO and MODE (Han have allowed for significant improvements
et al., 2022) have been created. Traditional in effluent quality and reductions in operat-
multi-objective optimization algorithms ing costs. This type of controller is primarily
offer numerous benefits in WWT methods, intended to enhance the WWTP’s efficiency
but they also have drawbacks, such as weak by adjusting the set points of the lower-level
global search ability and high levels of unpre- Supervisors, such as PID controllers, in addi-
dictability in the ideal position section. tion to offers insights into the use of genetic
Intelligent technologies based on machine algorithms in the development of advanced
learning (ML) are now being used in the water control systems for sewage treatment facili-
industry to solve pressing problems worldwide. ties. It’s also possible to cut down on process-
Model capabilities for the dynamic aspects of a ing time by employing PSO, DE, or ACO in
WWTP operation have also been improved via the right situation.
the use of a hybrid method that blends several
ML techniques. It has been shown that artifi-
cial neural networks (ANNs) can learn and 2. Literature Review
construct mathematical models of extremely To conserve energy and cut down on emis-
non-linear interactions, making them useful as sions, researchers (Li et al., 2021) proposed
predicting tools in a variety of contexts. the development of an innovative CLSTMA
Artificial Neural Network Ant Lion for observing effluent nature within wide-
Optimization (ANN-ALO), (PSO-ANN) ranging paper industry treatment facility
(Mahadeva et al., 2022) is proposed to using continuous convolution neural network
enhance the effectiveness of conventional (CNN), long short-term memory (LSTM),
multi-objective optimization techniques on a and attention mechanism (AM). For each
global scale and prevent the algorithm from of the study’s output variables, the authors
settling to the point of reaching a regional employed standard performance measure-
maximum. However, AI models do need a ments such as Pearson correlation coefficient
 A systematic review on recent controllers for an industrial wastewater treatment plant 3

(R), the Root Mean Squared Error (RMSE), effectiveness and durability of WWT pro-
and the Mean Absolute Error (MAE). The cesses. They also emphasize the need for fur-
findings demonstrated that the hybrid deep ther research and development in this field to
learning model achieved the highest levels of overcome the existing challenges and realize
accuracy, precision, and recall compared to the full potential of control Techniques for the
the other models. The authors argue that their treatment of sewage.
model may be used to lessen the ecological The degradation of proteins and amino
toll of paper manufacturing by increasing the acids in organic waste is the principal source
efficacy of WWT systems in the paper sector. of ammonia in wastewater, Subari et al. (2022)
The RME of the CLSTMA model for the set out to compile a comprehensive overview
effluent Chemical Oxygen Demand (CODeff) of the utilization of biological treatment
was lowered by 23.3–31.55%, the MAE was techniques has emerged as a viable alterna-
condensed by 38.89–74.50%, and the R of tive strategy for the removal of ammonia in
the CLSTMA model was enhanced by 8.29– water and sewage purification facilities. This
11.86% when compared to different mod- article discusses the present worldwide pollu-
els (CNN, LSTM, and CLSTM models). The tion condition of ammonia, the sources about
RMSE of the CLSTMA model decreased by ammonia impurity, and the usual restrictions
10.26% and 9.92% for the effluent suspended on regulation of ammonia concentration pos-
solids (SSeff), the MAE decreased by 5.37% sible dangers, recorded instances, and more.
and 3.44%, and the R of the CLSTMA model In addition, AI is being used into water
rose by 15.13% and 37.21%. Results from purification model prediction and control
computer simulations show that the suggested systems. The ANN is a talented computer
CLSTMA model might be very useful for keep- mechanism that can improve both the WTP
ing an eye on the WWT system in the paper sec- and WWTP in general. Due to the nonlinear-
tor so that cleaner products can be made. ity of their dynamic behavior, variability in
PID control, as well as more modern meth- parameter values over time, constraints on
ods like MPS, FLC and neural network con- manipulated variables, interdependencies
trol, are all discussed in detail by (Iratni et between manipulated and controlled vari-
al., 2019). The authors also explore how to ables, presence of unaccounted Interruptions,
effectively regulate the WWT process by inte- delays in input and measurement processes,
grating control technology with optimization and other factors, a standalone conventional
methods and real-time monitoring systems. PID controller is inadequate for effectively
The complexity and variety of wastewater tuning parameters or achieving optimal
composition, the absence of dependable sen- control outcomes in a WTP or WWTP. It is
sors, and the high cost of installing advanced necessary to know the mathematical connec-
control technologies are all highlighted in this tion between the system’s input and output
article as problems that control systems must in order to design an appropriate controller
overcome in WWT operations. Non-linear and also compares the outcomes of several
controllers like the NMPC controller and the specific analysis involving the optimization
neural-fuzzy controller can be viable alterna- of PID controllers. Hybridization with other
tives, but they demand a high level of knowl- tuning methods, like Ziegler–Nichols (ZN) or
edge due to the complexity of their algorithms, Cohen-Coon (CC), a Fuzzy control system, or
which can lead to increased computing needs. a metaheuristic approach, like Particle Swarm
In conclusion, the authors suggest that the inte- Optimization (PSO), Gravitational Search
gration of control technologies with advanced Algorithm (GSA), or Grey Wolf Optimization
monitoring and optimization techniques can (GWO), can greatly increase the effectiveness
also lead to significant improvements in the of a PID controller.
 4 Innovations in Computational Intelligence: Trends and Technologies

Using a WWT process as an example, Han compares appropriate control options for use
et al. (2022) describe the results of a research during storms. MIOC’s optimization perfor-
into the development of a multi-objective mance may be confirmed by its best-in-class
integrated optimum control method. The mean EQ (7167) and its lowest-in-class mean
authors offer a control system that simultane- EC(3889). With a mean IAE of just (0.041),
ously maximizes the plant’s efficiency and the MIOC clearly outperforms conventional,
efficacy of the WWT process, therefore reduc- incrementally optimized control approaches.
ing both operational costs and environmen- To predict the operational effectiveness of
tal effect. Existing multi-objective optimum a desalination facility, Mahadeva et al. (2022)
control approaches, however, tend to analyze offer a model of PSO-ANN. The purpose
the procedure of maximizing efficiency and of this research is in order to improve the
the control method separately, which may dependability of projections and enhance the
lead to the set-point fluctuating wildly and benefits of the Reverse Osmosis (RO)-based
being difficult to follow, and therefore dete- desalination technology plant. Modelling and
riorate the control and practical performance simulation were accomplished with the help
of WWTPs. In order to address this issue, of MATLAB 2019a’s neural network tool-
this research suggests a Multi-Objective Inte- box. The literature implies that changes to
grated Optimal Control (MIOC) approach the modelling parameters of soft computing
for the Sludge Processing Facility. To accom- models may lead to significant gains. In order
plish synchronized optimization and control to determine what those ideal values are, this
action, a Multi-Objective model predictive study provides a comprehensive list of mod-
control framework is designed. elling parameters and an in-depth systematic
Both the model of controlled variables and examination of them. Table 1.2 displays the
the model of operational indices utilized in summary statistics for the suggested mod-
MIOC were constructed using AFNNs. In eling efficiency assessments (R2 and MSE)
BSM1, EQ and EC are used to assess WWTP’s throughout training, validation, testing, and
efficiency. Control performance and optimiza- all datasets. Table 1.2 shows details.
tion performance studies are examined using Model 2 had the greatest training results
simulation in all climates. Example: Table 1.1 (R2 = 99.8%, MSE = 0.000) while Model 1 had

Table 1.1. Comparison of different optimal control strategies of the mean EQ and EC in storm
weather condition
Optimal NH4 Ntot SS COD BOD EQ kg EC Percentages IAE
control mg L–1 mg L–1 mg L–1 mg L–1 mg L–1 poll units kW h % mg
methods L-1
MIOC 2.87 17.27 13.55 46.89 2,74 7167 3889 12.00 0.041
DMOPSO 3.07 17.48 13,02 47.79 2,78 7512 4089 7,42 0.101
(Han, Liu,
2021)
MOO 2.93 17.91 13.56 48.13 2.83 7583 4169 5.61 0.120
(Sweelapple
et al., 2014)
MOOC 2.98 17.39 13.86 48.64 2.71 7236 4410 0.15 0.054
AQ2: (Han et al.,
Please 2018)
provide
source for P1D 3.06 17.52 13.79 49.52 3.18 7970 4417 0 0.108
all tables. Source:

AUTHOR'S SELF GENERATED

 A systematic review on recent controllers for an industrial wastewater treatment plant 5

Table 1.2. Statistical results of the proposed model’s performance evaluation (R2 and MSE) for
different stages of modeling (training, validation, testing and all data sets)
Model Performance evaluations of the PSO-
ANN models by different stages
Training (75%) validation (20%) Testing (05%) All (100%)
R2(%) MSE R2(%) MSE R2(%) MSE R2(%) MSE
Model - 1 99.0 0.005 98.7 0.014 99.6 0.003 98.9 0.007
Model - 2 99.8 ~0.000 98.5 0.016 94.7 0.057 99.0 0.006
Model - 3 99.3 0.004 98.9 0.011 98.6 0.015 99.1 0.006
Model - 4 99.1 0.005 98.6 0.014 96.9 0.024 98.8 0.008
Source:
Author's self generated
the ultimate testing outcomes (R2 = 99.6%, quality is enhanced by 5.6% in Scheme 2 and
MSE = 0.006); Model 3 had the ultimate by 20.3% in Scheme 3 as compared to Scheme
outcomes across the board. Overall, this arti- 1. As a consequence, scheme 3 has shown to
cle gives important insights into the use of be an effective method for treating wastewa-
advanced machine learning techniques, such ter. A neural network model-based regulat-
as PSO-ANN, for enhancing the efficiency of ing strategy might be used in the future to
water treatment desalination facilities. The enhance effluent quality while simultaneously
optimized model presented in this study can decreasing electrical energy usage.
help plant operators make informed decisions Ning et al. (2022) propose the PLC with
and optimize the plants performance, leading incremental proportion and incremental
to improved efficiency and cost savings. (IPI) control algorithm to preserve the con-
Saravana Kumar et al. (2021), begin by servation of the aquatic ecosystem and pro-
the difficulties connected with managing the mote supportable growth by automating the
quality of effluent from WWT plants, particu- adjustment of the dissolved oxygen level in
larly in light of increasingly stringent regula- the aeration tank during the WWT process.
tory standards. They introduce the concept of A BP-NN was utilized to improve method of
fuzzy logic control, which is a type of control regulation. Automatic DO content manage-
system that uses linguistic variables to model ment within the aeration tank presented in
complex and uncertain systems. The sug- Figure 1.1 through the PLC’s enhanced IPI
gested study provides two control techniques control algorithm. A PLC-based automated
(ammonia control and total nitrogen man- control system was implemented to miti-
agement) to enhance their performance of gate the astronomical increases in operating
treating wastewater, and the authors propose expenses and energy consumption caused
using a fuzzy logic controller to achieve these by the lack of automation. Under the guid-
goals. The simulation work was done Within ance of the enhanced IPI procedure, the total
the framework of the Benchmark Simulation nitrogen content of wastewater treated in
Model No.1 (BSM1), and this study’s offers the aeration tank was minimized, and its
a supervisory FLC method enhance the efflu- variability was reduced to a minimum. The
ent quality of a WWT facility and provides authors also highlight the challenges faced
numerical evidence of the effectiveness of this by WWT plant, including fluctuating influ-
control strategy. It’s noteworthy that the efflu- ent characteristics and variations in flow rate,
ent quality is maintained despite the higher which can lead to inefficiencies in treatment
AE and PE use in this system. The effluent and increased energy consumption and cost
 6 Innovations in Computational Intelligence: Trends and Technologies

savings. Overall, the article provides a useful recycle loop. Even still, MPC may struggle to
example of how PLC-based control Systems deal with input disturbances and frequently
may be utilized to maximize productivity of needs many model coefficients to describe a
municipal WWT plant. response. The investigation found the MPC
Development and evaluation of a neuro- system was able to effectively control the
fuzzy-based Model Predictive Controller treatment process in real-time, reducing
(MPC) approaches that can estimate the key energy consumption and maintaining high
process factors and provide the appropriate levels of treatment efficiency. This paper con-
level of aeration to overcome the factors such cludes that the use of machine learning-based
as process uncertainty, seasonal variation, MPC systems able to offer significant benefits
and nonlinearity was discussed by Bernardelli to WWT plants, improving both their envi-
et al. (2020) in order to attain a streamlined ronmental and economic performance.
and cost-effective functioning. This article In order to forecast the dose of coagu-
describes the process of data collection and lant, Fang et al. (2022) employed the features
pre-processing, which involved the use of parameters such as pH, turbidity, electrical
sensors to monitor the different factors of conductivity, and flow velocity that character-
the treatment plant, such as flow rate, dis- ize influent water quality. Authors note that
solved oxygen levels. The data was then used traditional dosing algorithms for WWT plants
to train the machine learning models, which are often based on empirical formulas or trial
were integrated into the MPC system. and error, which can be inefficient and may
More testing using a weekly switch the result in suboptimal treatment outcomes.
between the two controllers (EW and CC) has In contrast, the proposed intelligent dosing
proven that the EW controller is more reli- algorithm model uses methods in the arena of
able at improving sewage quality while using artificial intelligence, such as machine learning
less energy. Especially by reducing air flow at neural networks to optimize dosing decisions
periods of low load times, but also by pre- based on real-time data from the treatment
venting oxidation during peak loads, as long plant and to suggest a sophisticated Light-
as the nutritional ceiling is never reached. GBM-based dosage prediction method, which
Figure 1.2 shows the process plan of a uses these data to estimate the dosage of coag-
WWTP that uses MPC. The dotted lines rep- ulant. The LightGBM algorithm’s findings for
resent the two control loops: the DO con- predictive power demonstrate its dominance.
trol loop controls the airflow (Ua), and the Overfitting and dataset compatibility are two
internal recycle (Qr) is controlled by the drawbacks of the LightGBM algorithm.

AQ3:
Please
provide
source Figure 1.1. Automatic control flow of aeration tank DO content by the improved IPI control
for all algorithm.
figures.
Source: Author's own resource
 A systematic review on recent controllers for an industrial wastewater treatment plant 7

Figure 1.2. shows the process scheme of the WWTP to which MPC is applied.
Source:
AUTHOR'S SELF GENERATED

Table 1.3 provides an evaluation rubric capability is shown by these results. Overall,
for predicting outcomes using various this paper highlights the potential benefits of
approaches. Evaluating the efficacy of three using artificial intelligence methods that may
models using experimental data. The experi- be used to boost the performance of waste-
mental findings demonstrate that the R2 water remediation systems.
value reaches a maximum of 0.9957 for the Matheri et al. (2021) describe a study that
dataset model created by LightGBM, while was done by a group of researchers from dif-
the RMSE and MAE values are reduced to a ferent institutions in South Africa. They used
minimum of 0.3049 and 0.1474 correspond- data from a WWT plant to train and test
ingly. LightGBM’s advantage in predictive ANN. The models were built using a Multi-
layer Perception model and were utilized for
Table 1.3. The results of three different models forecasting potential issues with WWT sys-
tems, such as chemical composition, inherent
Model R2 RMSE MAE MAPE flow rate, and better safety factor. The result
MLR 0.9952 0.7504 0.4458 6.29% demonstrates that ANNs are an effective tool
DT 0.9748 0.4949 0.2978 4.78% for developing integrated wastewater systems
LightGBM 0.9957 0.3049 0.1474 1.34% and simulating the designs of WWTPs. For
the purpose of the real-time prediction of
Source:
nutrient removal efficiency, a neural network
based on the LSTM network (a descendant
Table 1.4. Model Performance on the training, of the RNN) is also suggested. However,
validation and test sets (R2 values) ANN is computationally intensive, difficult
Training Validation Test set to explain, and data-intensive. WWT process
set set issues were predicted using the artificial intel-
TSSe_RF 0.934 0.862 0.920 ligence (AI)-based ANN prediction model,
developed on the popular MATLAB platform
TSSe_ 0.935 0.892 0.920
DNN
and enhanced with machine learning (deep
learning). The research also stresses the need
PO4_RF 0.905 0.870 0.886
of taking into account numerous removal
PO4_DNN 0.904 0.908 0.872 input factors while creating such models to
Source: guarantee precise forecasts.
8 Innovations in Computational Intelligence: Trends and Technologies

Wang et al. (2021) present a unique ML- and require more training data to learn effec-
based method that aims to enhance effluent tively. In addition, GMM is highly sensitive
quality control in WWTPs by showing how to outliers. The best models for each algo-
practical factors and discharge parameters are rithm under test were determined by running
related. The author collected a large dataset multiple tests with different combinations of
of process data, including influent and efflu- their main parameters and selecting the ones
ent water quality parameters, operational lowest errors according to metrics taken into
conditions and environmental factors. This account, including MSE, RMSE, MAE, and
dataset was used to train several ML methods MAPE. With least RMSE, MSE, MAE, and
such as ANNs, decision trees, random forests, MAPE, the Transformer approach obtained
and linear regression. The framework is com- the highest Performance for predicting all of
prised of Random Forest (RF) models, Deep the critical variables.
neural network (DNN) models, and Variable When it comes to forecasting TSS and
Importance Measure (VIM) models and the NO3, LSTM methods fared the best, while
partial Dependence Plot (PDP) analyzers, and ANN and GMM algorithms produced the
it uses a unique method to explain temporal best results for NH4 and BOD5, respectively.
delays between processes. Thirty-two opera- Concerning the predictability of outcomes, it
tional variables and two effluent parameters— is lastly worth noting that the Transformer
Total Suspended Solids in Effluent (TSSe) and algorithm necessitates more time and effort
Phosphate in Effluent (PO4e)—are analyzed to train and make predictions. In accord-
in order to create RF models, which are then ance with the selected metrics, both ANN
verified using DNN models as references and and LSTM provided accurate predictions
demonstrated to be appropriate for VIM and while simultaneously using less computing
PDP in terms of clarifying the specific effects resources and a shorter amount of time to
of these two operational variables on TSSe train and provide predictions.
and PO4e. The model performance of DNN Classification models were shown by
and RF on training, validation and test data Wodecka et al. (2022) to be a useful tool
set as shown in Table 1.4. In conclusion, both for predicting changes in effluent character-
the RF and DNN models accurately forecast istics at the intake of WWT facilities, when
all data sets and demonstrated outstanding these values are directly proportional to the
Performance in generalization for TSSe and flow rate of incoming wastewater. In the
PO4e. considered proposal, the levels of quantified
Salles et al. (2022) suggested using machine wastewater characteristics parameters were
learning algorithms to forecast important divided into lower (lower-than-average indi-
variables (monitoring, control, and optimi- cator values), average (normal and maximum
zation of industrial processes) in the WWT values), and higher (increasing values). The
method. The value of the TSS, BOD5, NH4, results of the computations showed that the
and NO3 variables were approximations values of the chosen wastewater quality indi-
using the Gaussian Mixture Models (GMM), cators (BOD5, COD, TN, TP) can be detected
LSTM, ANFIS, Transformer and ANN algo- with adequate accuracy using the calculated
rithm. Greater computational resources and ensemble models blending support vector
longer preparation time are needed for the machines and boosted trees. The statistical
transformer algorithm. ANFIS has a limi- parameters (minimum, average, maximum,
tation of the type, quantity and position of and standard deviations) of seasonal fluc-
membership functions, the curse of dimen- tuation in the amount and quality of BOD5,
sionality, and the fact that LSTM models COD, TN, and TP were identified, with a
are more complex than conventional RNNs split between winter and autumn/spring. For
 A systematic review on recent controllers for an industrial wastewater treatment plant 9

the purpose of maximize the effectiveness of included. The results of this study will help
WWT, this research finds that machine learn- professionals manage WWTPs based on data.
ing approaches may be utilized to forecast the The authors concluded that deep WWTP
quality of incoming wastewater. The findings learning-based models have the potential to
of this investigation could be applied to cre- be useful tools for forecasting key features
ate a real-time monitoring system for WWT of WWT plant and advise future research to
plants, which allow operators to improve take into account sensor networks that can
the treatment process based on the predicted provide data sets with greater dimensions
quality of incoming wastewater. and frequency.
Cheng et al. (2020) utilized training Figure 1.4 depicts the core of the work
and testing data collected over time from a of Do et al. (2021), which is the incorpora-
WWT facility in Norway. Standard LSTM tion of genetic algorithms (GAs) into the
and GRU, exponentially smoothed LSTM, Benchmark Simulation Model1 (BSM1) for
adaptive LSTM, and smoothed LSTM are the purpose of designing a higher-level con-
only some of the six deep learning models trol system for WWTPs. This research article
built using RNN-based soft sensors. Models proposes a unique Genetic Optimization for
built on GRUs are more efficient than LSTM Biological Wastewater Treatment Plants with
models because they converge more quickly. the objectives of increasing discharge qual-
For all important aspects, the LSTM soft- ity while decreasing operational expenses.
sensor performs best overall, succeeded by The suggested controller is meant to optimize
the exponentially-smoothed GRU and the the WWTP by adjusting the set points of the
LSTM. The dynamic models underperformed lower-level controllers, such as the PID con-
the other models in terms of their ability to trollers. This research provides the outcomes
predict the future. All that is needed are the of a simulation research was carried out to
data itself, as the offered models produce test the performance of proposed control-
good forecasting results. Figure 1.3 is a sche- ler. In this work, a GA-based hierarchical
matic illustration of the forecasting system controller was presented, with the bottom
that we propose. Using RNN-based soft sen- tier controller regulating S0,5 by manipula-
sors, this research constructed RNN-based tion of Kla,5, and the upper-level controller
models to predict WWTP essential param- adjusting S0,5 per SNH,4. In the higher-level
eters such influent flow, temperature, BOD, control, GA is utilized to figure out KSO,5
chloride, and energy consumption. To fur- and BSO.5 in order to achieve three prede-
ther reprocess unprocessed Data before deep termined goals in a variety of climates and
learning, the exponential smoothing filter is atmospheric circumstances.

Figure 1.3. Workflow of the proposed deep learning-driven forecasting procedure.

Source:

AUTHOR'S SELF GENERATED

10 Innovations in Computational Intelligence: Trends and Technologies

A tank 5 has recently implemented the for designing a suitable treatment system
desired control. For each of the three weather to ensure the safety, efficacy, and quality of
scenarios investigated, the simulated out- the treated wastewater. Here, the limitations
comes will be compared to those obtained faced by the existing research are described as
by using simply the lowest controller, which follows. PID, FLC, PLC, and MPC controllers
is an ILC paired with a PI regulator for the are all commonly used in WWT processes;
wastewater influent data. As a result, the However, the PID controller’s inefficient
suggested strategy might be used to various parameter coordination and substantial time
contexts to further decrease the OCI and delay make it difficult to provide precise con-
EQI. Overall, the research sheds light on trol, especially on complex systems or sys-
how genetic algorithms might be included tems that demand high performance.
into the planning of advanced controls for Moreover, FLC controller has disadvan-
WWT facilities. There may be far-reaching tages of lower speed and longer run time and
consequences for the environment and pub- lack of real time responses. This PLC-based
lic health if the suggested controller is not control system is difficult to adapt or replace,
implemented to enhance the performance of requires extensive training and is always diffi-
WWTPs. Furthermore, time spent might be cult to identify errors. In addition, MPC may
cut down by using PSO, DE, or ACO in effi- struggle to deal with input disturbances and
cient ways. frequently needs many model coefficients to
describe a response.
Since traditional mechanistic models suffer
3. Discussion from a number of drawbacks, scientists have
This paper comprises the discussion about turned to Machine Learning (ML) techniques
the waste water and their various waste to simulate WWTP processes. Therefore, the
water treatment methods which provides existing research utilized LightGBM, RF,
different types of techniques from different ANN, ANFIS, LSTM, and GMM. However,
methods. WWT processes have advanced fast ANN requires lots of computational power
with scientific breakthroughs in recent dec- hard to explain and it requires lots of data,
ades. The treatment effectiveness with respect over fitting and data set compatibility are two
to their benefits, uses, and limitations have drawbacks of the LightGBM algorithm. RF
been thoroughly addressed. Understanding model is not able to deal with unbalanced
the characteristics of wastewater are essential and missing data, also its runtimes are fast,

Figure 1.4. Benchmark Simulation Model 1(BSM 1).

Source:

AUTHOR'S SELF GENERATED

 A systematic review on recent controllers for an industrial wastewater treatment plant 11

greater computational resources and longer Acknowledgement

preparation time are needed for the trans-
former algorithm. ANFIS has a limitation of The authors gratefully acknowledge the stu-
the type, quantity and position of member- dents, staff, and authority of Electronics and
ship functions, the curse of dimensionality Instrumentation department for their coop-
and the fact that LSTM models are complex eration in the research.
than conventional RNNs and require more
training data to learn effectively. In addition, References
GMM is highly sensitive to outliers. Moreo- [1] Bernardelli, A., Marsili-Libelli, S., Manzini,
ver, DNN is utilized to investigate the Indus- A., Stancari, S., Tardini, G., Montanari, D., ...
trial Wastewater and Venier, S. (2020). Real-time model pre-
Treatment however, DNN algorithm are dictive control of a wastewater treatment
not able to perform with smaller data and plant based on machine learning. Water Sci-
critical interpretability. ence and Technology, 81(11), 2391–2400.
[2] Cheng, T., Harrou, F., Kadri, F., Sun, Y., and
Leiknes, T. (2020). Forecasting of wastewa-
4. Conclusion ter treatment plant key features using deep
learning-based models: A case study. Ieee
The findings and analyses presented in this Access, 8, 184475–184485.
study underscore the paramount importance [3] Do, H. T., Van Bach, N., Van Nguyen, L.,
of WWT. Failing to address wastewater can Tran, H. T., and Nguyen, M. T. (2021). A
have dire consequences for both human well- design of higher-level control based genetic
being and environmental factors. algorithms for wastewater treatment plants.
Looking ahead, the significance of water Engineering Science and Technology, an
treatment is set to escalate both globally and International Journal, 24(4), 872–878.
particularly in India. The gradual deteriora- [4] Fang, X., Zhai, Z., Zang, J., and Zhu, Y.
tion of groundwater quality, exacerbated by (2022, April). An intelligent dosing algo-
rithm model for wastewater treatment plant.
industrial waste and pollution, has led to the
In Journal of Physics: Conference Series (Vol.
emergence of hazardous diseases. To combat 2224, No. 1, p. 012027). IOP Publishing.
these challenges, WWT methods will play an [5] Fu, X., Zheng, Q., Jiang, G., Roy, K., Huang,
increasingly pivotal role. These techniques L., Liu, C., ... and Wang, Z. (2023). Water
aim not only to minimize water wastage but quality prediction of copper-molybdenum
also to ensure that an ample supply is con- mining-beneficiation wastewater based on
served for future generations. the PSO-SVR model. Frontiers of Environ-
In our perspective, researchers should mental Science and Engineering, 17(8), 98.
give top priority to the creation of WWT [6] Han, H., Liu, Z., Hou, Y., and Qiao, J.
systems that excel in efficiency, energy con- (2019). Data-driven multiobjective pre-
servation, fouling prevention, cost-effective- dictive control for wastewater treatment
ness, and compactness. The survey results process. IEEE Transactions on Industrial
strongly indicate the potential for employ- Informatics, 16(4), 2767–2775.
[7] Han, H. G., Chen, C., Sun, H. Y., and Qiao, J.
ing hybrid methodologies, which could
F. (2022). Multi-objective integrated optimal
yield enhanced precision and superior out- control for a wastewater treatment process.
comes for Industrial Wastewater Treatment Control Engineering Practice, 128, 105296.
facilities. By focusing on these aspects, the [8] Iratni, A., and Chang, N. B. (2019). Advances
field of WWT can usher in transformative in control technologies for wastewater treat-
advancements that address environmen- ment processes: status, challenges, and per-
tal concerns and industrial demands more spectives. IEEE/CAA journal of automatica
effectively. sinica, 6(2), 337–363.
12 Innovations in Computational Intelligence: Trends and Technologies

[9] Li, X., Yi, X., Liu, Z., Liu, H., Chen, T., Niu, using machine learning models. In 2022
G., ... and Ying, G. (2021). Application of International Joint Conference on Neural
novel hybrid deep leaning model for cleaner Networks (IJCNN) (pp. 1–9). IEEE.
production in a paper industrial wastewater [16] Saravana Kumar, S., and Latha, K. (2021).
treatment system. Journal of Cleaner Pro- A supervisory fuzzy logic control scheme
duction, 294, 126343. to improve effluent quality of a wastewater
[10] Mahadeva, R., Kumar, M., Patole, S. P., and treatment plant. Water Science and Technol-
Manik, G. (2022). An optimized PSO-ANN ogy, 84(10-11), 3415–3424.
model for improved prediction of water [17] Subari, F., Harisson, H. F., Kasmuri, N. H.,
treatment desalination plant performance. Abdullah, Z., and Hanipah, S. H. (2022).
Water Supply, 22(3), 2874–2882. An overview of the biological ammonia
[11] Matheri, A. N., Ntuli, F., Ngila, J. C., Seodi- treatment, model prediction, and control
geng, T., and Zvinowanda, C. (2021). Per- strategies in water and wastewater treat-
formance prediction of trace metals and cod ment plant. Malaysian Journal of Chemical
in wastewater treatment using artificial neu- Engineering and Technology (MJCET), 5(1),
ral network. Computers and Chemical Engi- 8–28.
neering, 149, 107308. [18] Sundui, B., Ramirez Calderon, O. A., Abdel-
[12] Newhart, K. B., Holloway, R. W., Hering, dayem, O. M., Lázaro-Gil, J., Rene, E. R.,
A. S., and Cath, T. Y. (2019). Data-driven and Sambuu, U. (2021). Applications of
performance analyses of wastewater treat- machine learning algorithms for biological
ment plants: A review. Water research, 157, wastewater treatment: updates and perspec-
498–513. tives. Clean Technologies and Environmen-
[13] Ning, S., and Hong, S. (2022). Programma- tal Policy, 23, 127–143.
ble logic controller-based automatic control [19] Wang, D., Thunéll, S., Lindberg, U., Jiang,
for municipal wastewater treatment plant L., Trygg, J., Tysklind, M., and Souihi, N.
optimization. Water Practice and Technol- (2021). A machine learning framework to
ogy, 17(1), 378–384. improve effluent quality control in waste-
[14] Nourani, V., Asghari, P., and Sharghi, E. water treatment plants. Science of the total
(2021). Artificial intelligence-based ensem- environment, 784, 147138.
ble modeling of wastewater treatment plant [20] Wodecka, B., Drewnowski, J., Białek, A.,
using jittered data. Journal of Cleaner Pro- Łazuka, E., and Szulżyk-Cieplak, J. (2022).
duction, 291, 125772. Prediction of wastewater quality at a waste-
[15] Salles, R., Mendes, J., Araújo, R., Melo, C., water treatment plant inlet using a system
and Moura, P. (2022, July). Prediction of based on machine learning methods. Pro-
key variables in wastewater treatment plants cesses, 10(1), 85.
2 Green SaaS adoption in cloud
computing: An investigation into
environment sustainability
Bhanu Priyaa, Bhupal Aryab, K. M. Pandeyc, and Prashant Vermad
Department of Computer Science and Engineering, Roorkee Institute of Technology, Roorkee, India

Abstract: The advent of cloud computing has advanced tremendously in the realm of information technology,
resulting in a transformative influence on the IT sector. Cloud computing provides the provision and delivery of
services, software, and computing over the Internet. Due to the utilization of resources such as the virtualiza-
tion of servers, networks, and storage, as well as workload distribution through load-balancing processes, it
consumes a lot of power and energy. In this paper, we have defined the “Green IT component” which refers to
the utilization of energy-saving and environmentally friendly techniques within data centers and their compo-
nents as processing, storage, and network elements. This paper gives the concept of a “greener cloud” and its
implications for sustainability. We have examined various research papers discussing the challenges and oppor-
tunities associated with green cloud computing, with the application of green software-as-a-service models such
as desktop as a service, integration as a service, prognostic, and health management as a service.. Green cloud
technology ensures high performance and productivity through an energy-efficient approach, enabling busi-
nesses while also promoting a sustainable future for the planet. In future work, we will propose a new architec-
ture for a greener cloud for the utilization of different software-as-a-service models of delivery.

Keywords: Sustainability, energy efficiency, resource optimization, carbon footprint, green IT, energy-aware
design

1. Introduction until 2025, indicating a rising demand for

data center deployment in the country in the
The rapid growth of digital services has
foreseeable future. It was predicted that the
resulted in an extensive surge in worldwide
volume of the data center trade would dou-
internet traffic, growing 20 times since 2010
ble from 400 megawatts to 1008 megawatts
and accompanied by a doubling of internet
users. As a result, there is a pressing necessity by 2023. Hence the operation of servers in
to prioritize energy efficiency measures to alle- data centers requires a significant amount of
viate the considerable power consumption of energy, resulting in a corresponding demand
data centers. Google and Microsoft have set for power for resource components such
targets for 2030 to procure and utilize zero- as CPU, hard disc, and memory elements.
carbon electricity around the clock in every Moreover, this energy consumption generates
grid where their operations are present. A ris- excess heat that must be efficiently cooled
ing number of organizations are also striving and managed.
to achieve continuous access to carbon-free According to the International Energy
energy 24/7. The data center market in India Agency CO2 emission report 2022, the largest
is projected to experience a steady growth sectorial rise in emissions during 2022 was
rate of 5% compound annual growth rate) observed in the electricity and heat generation

[email protected]; [email protected]; [email protected]; [email protected]

a

DOI: 10.1201/9781003606635-2
14 Innovations in Computational Intelligence: Trends and Technologies

sector, with a 1.8% increase or 261 Mt. application businesses can eliminate the need
Notably cloud infrastructure is designed to for on-premises hardware and infrastructure,
minimize expenses and maximize resource resulting in lower energy consumption and
utilization, making it more environmentally reduced carbon footprint. PaaS platforms
friendly compared to traditional data centers frequently provide frameworks and tools
(IEA, 2022). In terms of service delivery mod- that encourage effective coding techniques,
els, cloud computing offers infrastructure as optimal resource utilization, and scalability,
a service (IaaS), platform as a service (PaaS), which results in energy savings and decreased
and software as a service (SaaS). Through the waste.
internet, virtualized computer resources are All spheres of life are starting to adopt
made available by IaaS, giving users control cloud computing technologies, particularly in
and management over their infrastructure. the area of data storage. Increasingly, Con-
PaaS provides a framework for creating, sumers are okay with endpoint devices having
testing, and deploying applications with- less storage since they prefer using the cloud.
out requiring the management of the foun- While this has been happening, some com-
dational infrastructure. SaaS eliminates the mercial cloud development firms, like Google,
requirement for local installation and main- Amazon, and IBM Cloud, have offered a vari-
tenance by delivering software programs via ety of services, including infrastructure and
the internet (Gulati, no date). Green cloud services for scientific computing.
solutions offer a means to reduce operational
expenses, minimize greenhouse gas emissions, 2.2. Green IT
and conserve energy resources, thereby fos-
tering a positive impact on the environment, The Environmental Protection Agency initi-
all while ensuring the maintenance of high- ated the Energy Star program in the United
quality service levels (QoS) (Raghavendran et States back in 1992. The program specifi-
al., 2017). cally focuses on enhancing energy efficiency
in computer devices and related technologies.
The term “green computing” gained popular-
2. Background ity subsequent to the inception of the Energy
Star program (Brown, Webber and Koomey,
2.1. Cloud computing 2002)a voluntary labeling program designed
A fundamental transformation in the utili- to identify and promote energy-efficient prod-
zation and management of hardware and ucts, in order to reduce carbon dioxide emis-
software resources are used and managed is sions. Since then, the EPA, now in partnership
brought about by cloud computing. Robust with the US Department of Energy (DOE.
data replication and failover mechanisms Around the same period, TCO Devel-
ensure high reliability and mitigate the risks opment, an organization based in Sweden,
of data loss and downtime. In the context introduced the TCO certification program
of the energy-efficient cloud model, IaaS (Joumaa and Kadry, 2012). This initiative
allows businesses to optimize their resource was aimed at promoting and certifying a
utilization such as provisioning, mapping, wide range of IT equipment include net-
adaptation, discovery, brokering, estimation, work accessories, computer systems, video
and modeling. It emphasizes the benefits of adapters, monitors, keyboards, and even
resource management in IaaS, including scal- mobile phones with a primary focus on ful-
ability, quality of service, cost-effectiveness, filling distinct environmental and sustainabil-
and simplified interfaces (Mary, 2013). SaaS ity standards. The use of energy-saving and
refers to cloud-based software applications environmentally friendly techniques in data
that are accessed through the internet. In SaaS centers, with a focus on the CPU, memory,
 Green SaaS adoption in cloud computing: An investigation into environment sustainability 15

computing. The middle layer, which includes

“cloud service” and “green IT,” likely repre-
sents the components and technologies that
enable environmentally sustainable cloud
services.

3.1. Virtual machine management

When we talk about virtual machine (VM)
AQ1: management, to handle cloud services,
Please
provide VM migration is crucial. In the practice of
source VM migration (Jain and Inderveer, 2016),
for all the Figure 2.1. Green IT components.
figures
both live and non-live patterns are utilized
Source: to ­ transfer complete virtual servers across
Author's own resource
­physical machines within a data center.
Through the migration process, our objec-
storage, and network components, is referred tive is to balance the VM load, enhance power
to as a “green IT component,” as shown in efficiency, and streamline system maintenance
Figure 2.1. and fault tolerance capability.
Green IT in cloud computers uses tech-
nology like solid-state drives, which use
less power than conventional spinning hard 3.2. Workload consolidation
disc drives, to increase storage efficiency. By With the growing demand for computational
removing superfluous data and optimizing power, to achieving energy and power effi-
data location, reduplication, compression, ciency involves consolidating active VMs to
and data tiring techniques further mini- reduce the server and transitioning dormant
mize storage energy use. green IT focuses servers into sleep or shut down mode consid-
on power-saving networking hardware and ering lower power states.VM consolidation
protocols that preserve high-speed and low- methods can be categorized into two vari-
latency connections while using less energy. ants, namely static and dynamic, based on

3. Greener cloud: challenges and

opportunities
The term “greener cloud” encompasses the
environmentally friendly nature of cloud com-
puting and its efficient service delivery. Green
cloud serves as a strategic plan to reduce
operational expenses by leveraging virtual-
ization, multi-tenancy for resource optimiza-
tion, load balancing, and maintaining quality
of service (QoS) and manageability.
In the context of cloud, “green cloud
placement” mentions to the strategic place-
ment of cloud services to optimize energy effi-
ciency and minimize environmental impact. Figure 2.2. Greener cloud placement in cloud
Figure 2.1. depicts a visual representation of computing technology.
the different layers involved in green cloud Source: Author's own resource
 16 Innovations in Computational Intelligence: Trends and Technologies

the initial condition of the data centers they resources, storage, network usage, data trans-
start with. These techniques aim to minimize fer, load balancing, and additional services.
relocation overhead and maximize consolida- By considering factors such as instance usage,
tion by incorporating it into their modeling data storage, network traffic, and support
process (Arshad et al., 2022). options, the calculator provides an estimation
of cloud service costs.
3.3. Efficient task scheduling By adopting these practices, cloud service
providers (CSPs) contribute to a more sus-
In cloud data centers, an effective task sched- tainable IT industry and offer environmen-
uler is responsible for efficiently utilizing the tally responsible services to their customers.
resources available in the data center for Greener cloud projects offer businesses a
task execution. The efficiency of a scheduling variety of chances to cut costs, support the
algorithm is typically evaluated based on two environment, innovate, and stand out from
key factors: first is makespan which measures the competition while promoting a greener
how long it takes to finish every work on the future as:
cloud, and second is energy consumption,
which concentrates on using energy as effi- 3.6. Environment sustainability
ciently as possible when carrying out activi-
ties (Zambuk et al., 2021)such as security, the In order to reduce the negative effects on the
efficient allocation of resources, which in turn environment, cloud computing sustainabil-
results in the waste of resources. Researchers ity, especially in SaaS offerings, entails uti-
have explored a number of approaches over lizing resource-efficient and environmentally
the past decade to overcome these challenges. friendly practices such as server virtualiza-
The main objective of this research is to tion, resource sharing, and dynamic scaling
explore the task scheduling of cloud comput- also implementing a system to monitor and
ing using multi-objective hybrid Ant Colony report on energy consumption.
Optimization (ACO).
3.7. Innovation technological
3.4. Power and energy calculator advancements
Calculating carbon emissions in data centers The pursuit of energy-efficient cloud com-
involves considering various factors such as puting drives innovation in the development
AQ2:
Please
PUE as it quantifies the ratio of the overall of new technologies, software, and hardware
define energy consumed by a data center compo- solutions. This fosters technological advance-
PUE. nent to the energy consumed exclusively by ments, such as improved virtualization tech-
the IT infrastructure within that facility. A niques, energy-efficient data center designs,
PUE: PUE value of 1.0 represents ideal efficiency, and intelligent workload management systems.
Power indicating that total power utilization by the
Usage data center components is utilized by the IT 3.8. Government incentives and
Effectiveness
hardware and infrastructure. regulations
Governments worldwide are recognizing the
3.5. Cost calculator importance of energy efficiency and environ-
A data center cost calculator estimates mental sustainability. They may offer incen-
expenses including infrastructure, energy, tives, grants, or tax benefits to organizations
maintenance, cooling, personnel, and miscel- adopting green cloud practices. Compliance
laneous costs. Cloud services cost calculator with green regulations can also open doors to
involves estimating expenses for compute new markets and partnerships.
 Green SaaS adoption in cloud computing: An investigation into environment sustainability 17

3.9. Collaboration and knowledge field, present patterns, and upcoming research

sharing issues. Katal (Katal, Dahiya and Choudhury,
2022) explain energy efficiency in containers
Green cloud initiatives promote collabora- and present a solution to reduce the power
tion among industry stakeholders, including consumption in data centers. Instead of
CSPs, researchers, and policymakers. Sharing solely showcasing new possibilities in green
best practices, research findings, and experi- cloud computing, this paper emphasizes
ences can accelerate the adoption of sustain- the importance of long-term technological
able cloud technologies and foster collective advancement and directs attention toward
efforts toward a greener future. new technologies applicable at the operating
system, software and virtualization and also
4. Study of the state-of-the-art application levels.
CSPs offer various IT infrastructure and
techniques and research services over the internet, governed by service
Zambuk (Zambuk et al., 2021)such as secu- level agreements (SLAs) between CSPs and
rity, the efficient allocation of resources, customers/clients. Gebreslassie et. al (Gebre-
which in turn results in the waste of resources. slassie and Sharma, no date) aims to address
Researchers have explored a number of metering and billing system as client-side for
approaches over the past decade to overcome cloud services, empowering customers to
these challenges. The main objective of this verify measured and billed services, including
research is to explore the task scheduling of quality attributes like greenness.
cloud computing using multi-objective hybrid Park (Park, Han and Lee, 2023) investi-
Ant Colony Optimization (ACO explores gates the impact of cloud computing on users’
proposes a multi-objective hybrid approach energy efficiency using a novel industry-level
called ant colony optimization with bacterial measure. Findings show that IT cloud ser-
foraging (ACOBF). ACOBF achieves this by vices give the efficient result on energy con-
classifying requests based on protocol sen- sumption in the cloud, particularly after
sitivity, scheduling tasks considering dead- 2006. Chaurasia (Chaurasia et al., 2023)
lines, and creating a VM cluster to minimize proposes strategies for achieving green cloud
energy consumption. This research contrib- computing through efficient replica manage-
utes to improving the efficiency of internet ment, virtualization, resource scheduling, and
services by enhancing cloud computing task power scaling. It explores the utilization of
scheduling. sustainable resources, such as thermal-heat
Stoyanova (Stoyanova et al., 2020) explain recovery and free cooling systems, to reduce
in the context of IoT-based investigations, the environmental impact of high energy
consumers’ privacy rights and forensics suc- consumption.
cess must be balanced. It investigates the IoT- Public health issues and epidemics, such
based cloud security challenges in legal and as obesity, diabetes, cardiovascular illnesses,
private terms. The paper offers a safeguard cancer, osteoporosis, and dental diseases,
of privacy during data extraction and ensures pose significant challenges to human well-
the integrity of evidence through dispersed being. Priya (Priya et al., 2020) provide a
blockchain-based solutions and also explores promising approach for intelligent disease
the forensics-as-a-service to reduce the data diagnosis by machine learning and artificial
and use of forensics intelligence techniques. intelligence. The proposed paper introduces a AQ3:
Please
Patil (Patil and Patil, 2019) presents a thor- novel hybrid algorithm and prediction model define
ough examination of green cloud computing, that combines the features of KNN and CNN KNN and
covering its attributes, prior successes in the algorithms. CNN.

KNN: K-Nearest Neighbors

CNN: Convolutional Neural Network

 18 Innovations in Computational Intelligence: Trends and Technologies

Azambuja (Azambuja et al., 2013) provides valuable insights to researchers

addressed the issue of inefficient construction working on energy consumption, highlight-
management, where delayed feedback leads ing the efficiency of the GreenCloud simu-
to resource wastage. They proposed a solu- lator compared to other simulators’ system
tion by utilizing cloud-based tools to partici- (Atiewi and Yussof, 2014).
pate with suppliers and the genuine demand Wang (Wang, 2021) analyse the fusion
at the construction site, thereby reducing the technology as the system’s subject, object,
accumulation of large inventories and mini- and means of evaluating green investments.
mizing material wastage on-site. The study demonstrates the significant impact
In greener cloud methodology the author of cloud computing and information fusion
evaluates three popular simulators: Cloud- technology on the green investment evalu-
Sim, CloudAnalyst, and GreenCloud, which ation system. Ahmad et. al (Ahmad et al.,
all measure energy consumption based on 2015) explain the use of virtualization and
their respective capabilities. This survey VM migration in modern cloud data centers

Table 2.1. State-of-the-art of green cloud

Author/Year Approach Technique Performance metric
AQ4: Zambuk et. al ACOBF (ant colony Classification of Convergence, solution
Please
(Zambuk et al.,
provide optimization with protocol sensitivity, diversity
2021)
citation bacterial foraging) scheduling, vm cluster
for Table
Stoyanova et. al
2.1. and Privacy-preserving data Theoretical past and Data reduction and
source(Stoyanova et al., extraction, current models in digital forensics intelligence
info 2020) forensics
Patil et. al (Patil Green cloud computing Green cloud computing, Developmental stage and
and Patil, 2019) and its characteristics cutting-edge solutions implementation barriers
Katal et. al (Katal, Problem-solving Dynamic, power Clear measures at each
Dahiya and approaches to reduce modeling at OS level level for efficient power
Choudhury, 2022) power consumption and application level
Gebreslassie
AUTHOR'S et. al Introducing a client-side
SELF GENERATED DPS-Yemane (simulated Performance, service
(Gebreslassie and metering and billing model) for client-side outage, updates,
Sharma, no date) system SLAs violation
Park et. al (Park, Environmental impact of Novel industry-level Estimated energy cost
Han and Lee, cloud computing measure savings for users in the
2023) us economy
Chaurasia et. al Emphasized for network Greedy-based Thermal-heat recovery
(Chaurasia et al., traffic management scheduling, load and free cooling systems
2023) balancing
Priya et. al(Priya et Machine learning and Leveraging cloud Uncovering hidden
al., 2020) artificial intelligence storage for high-speed patterns in extensive
prediction analysis medical data
Azambuja et. al Utilizing cloud Cloud-based technology Material wastage on-
(Azambuja et al., computing in the integration of suppliers site, accumulation of
2013) construction industry and demand large inventories
Atiewi et. Empirical qualitative, CloudSim, cloud analyst, Analysis of green
al(Atiewi and quantitative analysis, and Greencloud attributes
Yussof, 2014) data integration,
Source:
 Green SaaS adoption in cloud computing: An investigation into environment sustainability 19

to efficiently manage resources and reduce

computational cost and energy consumption.
Yan (Yan et al., 2022) represents a detailed
design and implementation of a deep rein-
forcement learning approach to address the
real-time job allocation problem in cloud
environments. The primary objective is to
optimize the consumption of energy while
ensuring a high QoS for users.

5. Green cloud software-as-a- Figure 2.3. Green cloud service model of

software as a service.
service: A provider’s perspective
Source:
AUTHOR'S SELF GENERATED
Cloud computing refers to the storage,
administration, and data processing carried DAAS helps to unravel many worries such
out using an internet-connected network of as data security, low computational cost, and
distant computers as an alternative to a local efficient use of storage and technology, and
server or personal computer. The green cloud to build a green environment (Charul Jag-
software-as-a-service (SaaS) model which is tap et al., 2020)we have integrated Cloud
shown in Figure 2.2 illustrates environmen- Computing with Desktop Virtualization. In
tally sustainable practices and energy-efficient our research and project, we have deployed
strategies within cloud computing. a cloud using openstack along with Desk-
top Virtualization. This paper aims for the
5.1. Information as a service security and cost reduction for institutional
As a service, information (IAS) is a cloud ser- purpose. I.INTRODUCTION As IT sectors
vice model that provides data to its users. The and Institutions are expanding rapidly day
information is given in an enterprise-friendly by day, need for computer systems are also
and user-friendly style. IAS Green technol- increased. new machines are being purchased.
ogy focuses on giving perspectives based on The IT companies can easily afford buying
analyzed and processed data, as well as infor- new machines as per their requirements. But
mation based on knowledge or a trustworthy what about the institutions and educational
source (Taryana, Fajar and Utama, 2018). sectors? Hence an effective way should be
Validation and search of zip codes and determined for the usage. N-Computing i.e.
addresses, payment processing, and data vali- Desktop Virtualization with Cloud provides
dation and completion services are all exam- us with the solution. It helps to resolve many
ples of Information as a Service. concerns such as Cost reduction, Data Security
and Efficient use of storage and technology.
Cloud Computing helps to provide maxi-
5.2. Desktop as a service mum storage with low computational cost.
Desktop as a service (DAAS) providers man- Hence the Institutions find this method more
age tightly closed desktop services that help affordable and promising. which reduces the
businesses to set up a whole far-off computer total hardware with software.Enabling tech-
surroundings for their employees. Using cloud nology of cloud computing, virtualization
desktop virtualization, we are providing cus- provides a realistic solution for resource con-
tomized software from any device in secure solidation and simplifying the management.
storage with flexible time and also providing On the other hand it helps an organization
anywhere in the world. to build a green computing environment with
20 Innovations in Computational Intelligence: Trends and Technologies

the increasing popularity of cloud computing real-time guarding of data, and restoration
with virtualization technology. By using soft- of vulnerabilities.
ware as a service(SAAS.
5.6. Security as a service
5.3. Database as a service SaaS model of cloud computing faces many
Database as a service (DBaaS) model is a troubles when we build the cloud infrastruc-
cloud computing managed service that allows ture because cloud providers generally are
users to line up, operate, manage, and scale not concerned about the security and pri-
with some style of access to information while vacy of the tenant or customers. Hence the
not the necessity to set up the physical hard- tenant uses specific protection gear together
ware, putting in code, or configuring it for with anti-virus and host-based intrusion
performance and heading for by themselves detection structures to stabilize their digital
(Al Shehri, 2013). Green DBaaS in enterprise machines additionally providers afford flex-
manager is applied all the way through 4 ibility to tenants to determine how a great
options which are: VM based, joint cluster, deal ­manage they want to have over their
collective installation, and shared database. very own VM.
payments are per-usage and software propri- SaaS provider model based on the base-
etors can get entry to their utility statistics line security framework (Varadharajan and
according to their need. Tupakula, 2014) to secure its own infra-
structure and make sure that malicious ten-
ants aren’t attacking the cloud infrastructure
5.4. Integration as a service or maybe the web website hosting malicious
Integration as a service (INaaS) is a machine software.
of equipment and applied sciences that con-
nects a number of applications, systems, 5.7. Prognostics and health
repositories, and IT environments for the management as a service
real-time trade of facts and processes. Many
industries prefer the SAAS for most consum- Prognostics and Health Management (PHM)
ers because they do not want more machines. service of cloud computing represents a great
So the center of attention is to invoke the chance to notice coming failures by predicting
functionalities and develop the legacy appli- future behavior. There are many challenges to
cation in heterogeneous cloud services such predicting the integrating behavior such as
as web service (Kumar and Bari, 2017). complex data of industries, large volumes of
data stored in different locations, high com-
putational resources, and continuous moni-
5.5. Compliance as a service toring of logistics and infrastructure. This
Compliance as a service (CaaS) is a cloud management service provides PHM solutions
SLA whose principal intent is to regulate and to have the ability of fault detection, fault iso-
the guard consumers’ privatizes and furnish lation, and estimation of remaining useful life
security by imposing greener attributes such (Terrissa et al., 2016).
as confidentiality, integrity, availability, and Table 2.2 Provides a list of critical consid-
accountability (CIAA) (Yimam and Fernan- erations for developing and overseeing envi-
dez, 2016). ronmentally sustainable SaaS and provides
CaaS offers partitioning of duties, annual some comparison and recognition of the ser-
risk evaluation, management of differ- vice’s environmental impact, especially with
ent applications, facts and figures detec- regard to carbon emissions and the ecosys-
tion, policy creation and enforcement, tem’s overall footprint.
 Green SaaS adoption in cloud computing: An investigation into environment sustainability 21

Table 2.2. Green cloud SAAS performance metric

Green SaaS Green Software as a service affecting factors
Service Energy efficiency Optimize resource Carbon footprint Environmental impact
IAS High Variable Moderate Depends on data usage
DAAS High High Low Reduced e-waste
DBAAS High High Low Reduced hardware
INAAS High High Low Reduced resource usage
CAAS High High Low Streamline processes
PHM High High Low Efficiently maintained
Source:
Author'sOwn Generated

6. Conclusion and future directions the environmental impact of SaaS solutions.

This paper concludes that the green cloud
Green cloud computing has emerged as an SaaS model aligns with the principles of sus-
important area of research and development tainability by promoting energy efficiency,
in recent years. Green data centers provide a resource optimization, and carbon footprint
wide array of advantages, such as decreased reduction in services.
ongoing operational expenses, reduced space
demands, lower carbon footprints, mini-
mized water usage, decreased waste gen- Acknowledgment
eration, diminished electricity consumption,
and a strong commitment to sustainable and This paper and the research behind it would
renewable data center methodologies. Vari- not have been possible without the excep-
ous strategies and techniques have been pro- tional support of my colleague, Bhupal Arya.
posed and implemented to achieve greener I am also grateful for the insightful comments
clouds, including virtualization, workload offered by the anonymous peer reviewers at
consolidation and task scheduling. this conference. The generosity and exper-
By optimizing resource utilization and tise of one and all have improved this study
improving energy efficiency, green IT can in innumerable ways and saved me from
measure and reduce the carbon footprint many errors; those that inevitably remain are
of data centers and contribute to the over- entirely my own responsibility.
all sustainability of the IT industry as it can
help lower operational costs for businesses References
and enable more scalable and flexible cloud AQ5: As
services. This paper discusses the various [1] Ahmad RW, Gani A, Hamid SHA, Shiraz M, per style,
references
parameters of green cloud placement in cloud Xia F, Han J. A survey on virtual machine and its
services and surveyed various technique and migration and server consolidation frame- citations
works for cloud data centers. J Netw Com- should be
measure their performance metric used in num-
put Appl. 2015;52:11-25. doi:10.1016/j.
cloud environment. jnca.2015.02.002
bered.
Please
While significant progress has been made [2] Arshad U, Zia MA, Siddique K, et al. Utiliz- check and
in green cloud computing, there are still sev- ing power consumption and SLA violations confirm
eral avenues for future research and develop- using dynamic VM consolidation in cloud
for
clarity.
ment such as developing energy-aware SaaS data centers. Renewable Sustainable Energy
architectures and design patterns, and devel- Rev. 2022;167:112782. doi:10.1016/j.
oping regular metrics and tools to evaluate rser.2022.112782
CHECKED AND CONFIRMED
22 Innovations in Computational Intelligence: Trends and Technologies

[3] Atiewi S, Yussof S. Comparison between Cluster Comput. Published online 2022.
cloud sim and green cloud in measuring doi:10.1007/s10586-022-03713-0
energy consumption in a cloud environ- [14] Kumar M, Bari P. Integration of
ment. In: Proceedings - 3rd International cloud based services. Published online
Conference on Advanced Computer Sci- 2017;2(4):101-105.
ence Applications and Technologies, ACSAT [15] Mary MA. Survey on resource management
2014. IEEE; 2014:9-14. doi:10.1109/ technique in cloud computing. Published
ACSAT.2014.9 online 2013;2(12):1723-1731.
[4] Azambuja M, Camargo M, Filho LM, et al. [16] Park J, Han K, Lee B. Green cloud? An empir-
Enabling lean supply with a cloud comput- ical analysis of cloud computing and energy
ing platform - An exploratory case study. In: efficiency. Manag Sci. 2023;69(3):1639-
21st Annual Conference of the International 1664. doi:10.1287/mnsc.2022.4442
Group for Lean Construction 2013, IGLC [17] Patil A, Patil DR. An analysis report on
2013. 2013:815-824. green cloud computing current trends and
[5] Brown R, Webber C, Koomey JG. Status future research challenges. SSRN Electron
and future directions of the ENERGY STAR J. Published online 2019. doi:10.2139/
program. Energy. 2002;27(5):505-520. ssrn.3355151
doi:10.1016/S0360-5442(02)00004-X [18] Priya L, Nallakannu G, Abirami C, Kumar
[6] Jagtap C, Rathore K, Patel P, et al. Desk- SV, Pavithra A. A novel intelligent diagno-
top virtualization with cloud computing. sis and disease prediction algorithm in green
Int J Eng Res Technol. 2020;9(1):255-257. cloud using machine learning approach. J
doi:10.17577/ijertv9is010157 Green Eng. 2020;10(7):3421-3433.
[7] Chaurasia N, Chauhan R, Agarwal H, [19] Raghavendran CV, Guhan P, Balasubrama-
Kumar N, Jaiswal RK. Shifting from cloud nian P, Sathyakumar M. A study on cloud
computing to green cloud and edge comput- computing services. Int J Eng Res Technol.
ing. SSRN Electron J. Published online 2023. 2017;4(34):1-7.
doi:10.2139/ssrn.4387781 [20] Al Shehri W. Cloud database database as a ser-
[8] Gebreslassie Y, Sharma DP. DPS-Yemane- vice. Int J Database Manag Syst. 2013;5(2):1-
Shareme CSMM model for client-side SLA 12. doi:10.5121/ijdms.2013.5201
of green cloud services measuring and [21] Stoyanova M, Nikoloudakis Y, Panagiotakis
monitoring. S, Pallis E, Markakis EK. A survey on the
[9] Gulati R. A systematic review on the various internet of things (IoT) forensics: challenges,
approaches used for achieving energy con- approaches, and open issues. IEEE Com-
sumption in cloud. Published online. mun Surv Tutor. 2020;22(2):1191-1221.
[10] IEA. CO2 emissions in 2022. Glob doi:10.1109/COMST.2019.2962586
Energy. 2022;62(10):20-21. Available at: [22] Taryana U, Fajar AN, Utama DN. Informa-
https://www.iea.org/news/global-co2-emis- tion as a service on cloud computing tech-
sions-rebounded-to-their-highest-level-in- nology: a review. In: 2018 International
history-in-2021. Seminar on Research of Information Tech-
[11] Jain N, Inderveer K. Energy-aware virtual nology and Intelligent Systems, ISRITI
machine migration for cloud computing - a 2018. IEEE; 2018:39-42. doi:10.1109/
firefly optimization approach. J Grid Com- ISRITI.2018.8864468
put. Published online 2016. doi:10.1007/ [23] Terrissa LS, Sahnoun M, Zghal M, et
s10723-016-9364-0 al. A new approach of PHM as a ser-
[12] Joumaa C, Kadry S. Green IT: Case studies. vice in cloud computing. In: Colloquium
Energy Procedia. 2012;16(PART B):1052- in Information Science and Technology,
1058. doi:10.1016/j.egypro.2012.01.168 CIST. IEEE; 2016:610-614. doi:10.1109/
[13] Katal A, Dahiya S, Choudhury T. Energy CIST.2016.7804958
efficiency in cloud computing data cent- [24] Varadharajan V, Tupakula U. Security as a
ers: a survey on software technologies. service model for cloud environment. IEEE
 Green SaaS adoption in cloud computing: An investigation into environment sustainability 23

Trans Netw Serv Manag. 2014;11(1):60-75. [27] Yimam D, Fernandez EB. A survey of com-
doi:10.1109/TNSM.2014.041614.120394 pliance issues in cloud computing. J Inter-
[25] Wang P. Application of cloud computing net Serv Appl. 2016;7(1). doi:10.1186/
and information fusion technology in green s13174-016-0046-8
investment evaluation system. J Sensors. [28] Zambuk FU, Anwar A, Zangeneh F, et al.
2021. doi:10.1155/2021/2292267 Efficient task scheduling in cloud comput-
[26] Yan J, Zhang S, Yu W, Yuan Y. Energy-aware ing using multi-objective hybrid ant colony
systems for real-time job scheduling in cloud optimization algorithm for energy efficiency.
data centers: a deep reinforcement learning Int J Adv Comput Sci Appl. 2021;12(3):450-
approach. Comput Electr Eng. 2022;99:107688. 456. doi:10.14569/IJACSA.2021.0120353
doi:10.1016/j.compeleceng.2022.107688
3 Anomaly detection in wireless sensor
network
Mayank Shuklaa, Sneha Yadavb, Abhay Pratap Singhc, Fizza Rizvid,
and Surya Vikram Singhe
Department Computer Science and Engineering, Babu Banarasi Das Institute of Technology and
Management, Lucknow, Uttar Pradesh, India

Abstract: With the recent emergence of the notion of the “Internet of Things,” wireless sensor networks (WSNs),
or WLANs, are critical and indispensable platforms for the future. They are employed in the military, health-
care, industry, and habitat for the purposes of tracking, monitoring, and regulating numerous applications.
However, anomalies that arise for a variety of causes, including node failures, reading mistakes, odd events, and
malicious attacks, have an impact on the quality of data that sensor nodes collect. Thus, before sensor data is
used to inform choices, anomaly detection is a crucial step in ensuring its quality.
We outline the difficulties in detecting anomalies in WSNs and the conditions that must be met in order to
create models that are both successful and efficient. Next, we examine the most recent developments in data
anomaly detection research in WSNs and group existing detection techniques into five major classes according
to the detection techniques used to create these techniques. Variations of the most advanced models in each
class are discussed, and their drawbacks are emphasized to give insight into possible future research directions.
In addition, the methods under assessment are contrasted and assessed according to how well they fulfill the
specified criteria. In conclusion, the overall constraints of existing methodologies are referenced and potential
avenues for future research are proposed and deliberated.

Keywords: IoT, anomaly detection, wireless sensor network, machine learning, neural network

1. Introduction reliability of WSNs. It empowers these net-

works to identify unusual patterns, irregulari-
Wireless sensor networks (WSNs) have ties, and deviations from expected behavior,
emerged as a cornerstone of modern infor- which can stem from various sources – envi-
mation and communication technologies, ronmental changes, equipment malfunctions,
enabling the collection of data from remote or malicious intrusions.
and often challenging environments. From
monitoring environmental conditions and
industrial processes to facilitating health-
2. Methodology
care and smart city applications, WSNs have
become pervasive. However, their ubiquity
2.1. Define research objectives
also exposes them to a wide range of vulner- The computer science project aims to address
abilities and threats. This is where the critical the problem of anomaly detection in WSNs
field of Anomaly Detection in WSNs comes by designing and implementing efficient algo-
into play. rithms that can accurately identify abnor-
Anomaly detection is a fundamental com- mal events while considering the limitations
ponent of ensuring the integrity, security, and of resource-constrained sensor nodes. The

a
[email protected]; [email protected]; [email protected];dRizvif952@
gmail.com; [email protected]
DOI: 10.1201/9781003606635-3
 Anomaly detection in wireless sensor network 25

specific objectives and approaches in this pro- These applications may include early detec-
ject include: tion of environmental incidents, intrusion
detection in security applications, equipment
2.2. Algorithm selection and fault monitoring in industrial settings, and
more.
development
The project involves a comprehensive review
of existing anomaly detection algorithms, 3. Thematic Overview
both from the domains of machine learn- Anomaly detection in WSNs is a project topic
ing and statistical analysis. After evaluating that involves the identification of unusual
these methods, the project aims to develop a patterns or behaviors within the data col-
customized anomaly detection algorithm tai- lected by sensors in a wireless network. The
lored to the unique characteristics of WSNs. overarching goal is to develop techniques and
algorithms that can effectively distinguish
2.3. Resource efficiency between normal and anomalous activities,
helping to enhance the overall security, reli-
Given the resource constraints of sensor ability, and efficiency of the wireless sensor
nodes, the project focuses on creating an algo- network.
rithm that minimizes the computational and Here’s a thematic overview of such a
memory requirements. This ensures that the system:
anomaly detection process can be executed on
sensor nodes with limited resources without
compromising the network’s overall efficiency. 3.1. Sensor data characteristics
Analyzing the characteristics of sensor data,
2.4. Real-time detection such as signal strength, frequency, and tem-
poral patterns, to identify anomalies that
An important aspect of the project is to deviate from expected norms.
achieve real-time anomaly detection, allow-
ing the network to respond promptly to
abnormal events. This involves optimizing 3.2. Machine learning models
the algorithm for quick decision-making and Utilizing machine learning algorithms to
minimizing communication overhead, as well detect anomalies in user behavior based on
as energy consumption. historical data. This may involve supervised,
unsupervised, or semi-supervised learning
2.5. Evaluation and validation approaches.
The project involves extensive experimen-
tation using both synthetic and real-world 3.3. Pattern recognition
datasets. It aims to assess the performance Identifying patterns of normal user behavior
of the developed algorithm in terms of accu- and flagging deviations from these patterns as
racy, false-positive rate, detection time, and potential anomalies. This can involve statisti-
resource utilization. The algorithm will be cal methods or advanced pattern recognition
compared against existing methods to dem- techniques.
onstrate its advantages.
3.4. User profiling
2.6. Practical applications Creating user profiles based on their typical
The research project also explores practical interactions with the wireless sensor network.
applications of anomaly detection in WSNs. Anomalies are then detected when a user’s
26 Innovations in Computational Intelligence: Trends and Technologies

behavior deviates significantly from their 3.11. Integration with access control

established profile. systems
Integrating anomaly detection with access
3.5. Context-aware anomaly
control mechanisms to dynamically adjust
detection user privileges based on detected anomalies,
Considering contextual information, such as enhancing network security.
the physical location of users, time of day,
or specific network conditions, enhances the 3.12. User feedback and interaction
accuracy of anomaly detection and reduces
false positives. Incorporating user feedback and interaction
data to improve the accuracy of anomaly
detection models and reduce false positives
3.6. Energy-efficient anomaly by considering user-specific preferences and
detection habits.
Developing anomaly detection methods that are
energy-efficient, considering the resource con- 3.13. Scalability and robustness
straints of sensor nodes in wireless networks.
Designing anomaly detection systems that
are scalable to large wireless sensor networks
3.7. Real-time detection and response and robust against various types of anomalies
Implementing real-time anomaly detec- and environmental changes.
tion mechanisms to promptly identify and
respond to abnormal user behavior, minimiz- 3.14. Data fusion techniques
ing potential security threats or disruptions.
Employing data fusion techniques to com-
bine information from multiple sensors and
3.8. Collaborative anomaly detection sources for a more comprehensive under-
Enabling collaboration between sensor nodes standing of user behavior and improved
to collectively analyze user behavior pat- anomaly detection.
terns and detect anomalies more accurately
through information sharing. 3.15. Threat intelligence integration
Integrating threat intelligence feeds to
3.9. Adaptive models
enhance anomaly detection by incorporating
Creating adaptive anomaly detection mod- information about known threats and attack
els that can continuously learn and evolve patterns.
based on changing user behavior and network
conditions.
4. Critical Analysis
3.10. Security and privacy
4.1. Accuracy and false positives
considerations
Addressing the security and privacy impli-
4.1.1. Strength
cations of user anomaly detection, ensuring Many anomaly detection algorithms can
that sensitive information is protected and achieve high accuracy in identifying unu-
that the detection process itself is resistant to sual patterns. However, achieving a balance
malicious attacks. between sensitivity and specificity is crucial
 Anomaly detection in wireless sensor network 27

to minimize false positives, as these can lead between efficiency and accuracy is essential,
to unnecessary alarms and resource wastage. especially in applications where energy con-
servation is critical.
4.1.2. Weaknesses
4.4. Security considerations
Achieving high accuracy may come at the
cost of increased false positives, especially in 4.4.1. Strength
dynamic and noisy environments. Striking the
Anomaly detection is a fundamental com-
right balance is challenging, and the trade-off
ponent of security systems in wireless sensor
between sensitivity and specificity needs care-
networks. It helps identify malicious activities
ful consideration.
and potential threats.

4.2. Adaptability to dynamic 4.4.2. Weaknesses

environments
Anomaly detection itself can be vulner-
4.2.1. Strength able to evasion and adversarial attacks.
Effective anomaly detection systems should Adversaries may manipulate sensor data
adapt to dynamic changes in the wireless sen- to evade detection, highlighting the need
sor network environment. Machine learning for robust security measures and continu-
models, particularly those based on unsuper- ous improvement of anomaly detection
vised learning, can learn, and adapt to new techniques.
patterns over time.
4.5. Scalability
4.2.2. Weaknesses 4.5.1. Strength
Adapting to rapid changes can be challeng- Scalability is essential for large-scale wire-
ing for some algorithms. Sudden variations less sensor networks. Efficient algorithms and
in sensor data due to environmental changes distributed approaches can enable effective
or network conditions may lead to delayed anomaly detection in networks with numer-
or inaccurate anomaly detection. Ensuring ous sensor nodes.
real-time adaptability remains a significant
challenge. 4.5.2. Weaknesses
As the network size increases, the complex-
4.3. Resource efficiency ity of anomaly detection also grows. Ensuring
4.3.1. Strength scalability without compromising accuracy
and response time requires careful system
Energy-efficient anomaly detection is crucial
design.
for wireless sensor networks with resource-
constrained nodes. Certain lightweight algo-
rithms and edge computing approaches can 4.6. Context awareness
minimize energy consumption. 4.6.1. Strength
Context-aware anomaly detection enhances
4.3.2. Weaknesses the accuracy of identifying abnormal behav-
Some resource-efficient algorithms may sac- ior by considering environmental factors and
rifice detection accuracy. Striking a balance user context.
28 Innovations in Computational Intelligence: Trends and Technologies

4.6.2. Weaknesses 4.9.2. Weaknesses

Implementing context awareness can be com- In highly dynamic environments, distinguish-
plex and may introduce additional computa- ing between anomalous behavior and normal
tional overhead. The challenge lies in defining
relevant context parameters and integrating 4.10. Gaps, inconsistencies, and
them into the anomaly detection process controversies
effectively.
Variations (noise) remain challenging. Potential areas for future research and devel-
Robustness to noise is crucial for avoiding opment in anomaly detection for WSNs:
false positives and maintaining the reliability
of the detection system. 4.10.1. Deep learning and AI
techniques
4.7. Privacy concerns: To improve the precision and effectiveness of
4.7.1. Strength anomaly identification in WSNs, deep learn-
ing models—such as deep neural networks
Privacy-preserving anomaly detection meth- (DNNs), convolutional neural networks
ods aim to protect sensitive information while (CNNs), recurrent neural networks (RNNs),
still identifying abnormal behavior. or hybrid architectures—are included. use
of pre-trained models and transfer learning
4.7.2. Weaknesses to enhance anomaly detection performance,
Balancing privacy and accuracy is a delicate particularly in situations with a shortage of
task. Some privacy-preserving techniques labeled training data.
may result in a loss of detection efficacy, espe- Investigate the deployment of anomaly
cially when limited information is available detection algorithms directly on edge devices
for analysis. within the WSN, enabling real-time process-
ing and reducing the need for transmitting
raw sensor data to a central server. Develop
4.8. Human-computer interaction lightweight anomaly detection models opti-
4.8.1. Strength mized for edge devices with constrained com-
putational resources and energy.
Involving user feedback and interaction can
enhance the accuracy of anomaly detection
models. 4.10.2. Federated learning and
collaborative anomaly detection
4.8.2. Weaknesses Research on federated learning approaches
where local WSN nodes collaborate to collec-
Depending solely on user feedback may intro-
tively train a global anomaly detection model
duce biases and subjective judgments. Strik-
without sharing raw data.
ing a balance between automated detection
Centrally, ensuring privacy and efficiency.
and user input is crucial for a reliable system.
Study how to aggregate and utilize insights from
multiple WSNs for improved anomaly detection
4.9. Robustness to noise in large-scale, distributed IoT environments.
4.9.1. Strength
Anomaly detection algorithms often include 4.10.3. Robustness and security
mechanisms to handle noise and outliers in Investigation of adversarial attacks and tech-
sensor data. niques to improve the robustness of anomaly
 Anomaly detection in wireless sensor network 29

detection models against malicious adver- integrity and confidentiality of data

saries attempting to deceive the system. transmitted across the network.
Development of anomaly detection meth-
­ 3. Real-time responsiveness:
ods that can detect anomalies resulting from • Synthesis: Anomaly detection systems
cyber-­physical attacks and security breaches that operate in real-time enable swift
in WSNs. identification and response to abnor-
mal events, minimizing the impact
4.10.4. Human-in-the-loop anomaly of disruptions and ensuring timely
detection interventions.
• Implications: In scenarios where rapid
Study the integration of human expertise and response is crucial, such as emergency
feedback into the anomaly detection process, management or critical infrastructure
leveraging human knowledge to enhance monitoring, real-time anomaly detec-
the interpretation and validation of detected tion enhances the effectiveness of the
anomalies. Design interactive and interpret- wireless sensor network in providing
able anomaly detection systems that allow actionable insights.
domain experts to understand, validate, and
refine anomaly detection results.
6. Conclusion
5. Synthesis and implication Concluding a computer science project on
“Anomaly Detection in WSN” should effec-
5.1. Patterns, trends, and tively summarize the key findings, outcomes,
relationships and the significance of your research. Here’s a
sample conclusion for such a project:
1. Improved network reliability: In conclusion, the research and implementa-
• Synthesis: In wireless sensor networks, tion of anomaly detection in WSN have yielded
anomaly detection helps to maintain several important insights and contributions.
overall network reliability by spotting This project aimed to address the critical issue
and addressing anomalous activity of detecting abnormal behavior or events in
that could jeopardize data integrity or WSNs, which is vital for ensuring the reliability
interfere with communication. and security of these networks in various appli-
• Implications: Enhanced network cations, including environmental monitoring,
reliability leads to improved trust- surveillance, and industrial automation.
worthiness in data collection and dis- Key findings and contributions of this pro-
semination, critical for applications ject are as follows:
such as environmental monitoring,
healthcare, and industrial automation. 1. Algorithm evaluation: Through an in-
2. Enhanced security: depth analysis of existing anomaly
• Synthesis: Anomaly detection acts as detection algorithms, we assessed their
a frontline defense against malicious performance in the context of WSNs. Our
activities and security breaches in project involved the evaluation of various
wireless sensor networks, helping to machine learning and statistical methods,
identify and respond to unauthorized such as Isolation Forest, One-Class SVM,
access or tampering. and K-means clustering, to identify their
• Implications: By bolstering network suitability for WSN anomaly detection.
security, anomaly detection systems 2. Customized solution: We developed a
safeguard sensitive information and customized anomaly detection algorithm
prevent potential threats, ensuring the tailored to the specific characteristics
30 Innovations in Computational Intelligence: Trends and Technologies

of WSNs. This algorithm considers the technology. Here are some potential areas for
constraints of limited resources, wire- future research and development in anomaly
less communication, and distributed data detection for WSNs:
gathering in WSNs, making it a practical
choice for real-world deployments.
7.1. Deep learning and AI techniques
3. Performance evaluation: Extensive exper-
imentation was conducted on both syn- To improve the precision and effectiveness of
thetic and real-world datasets. Our results anomaly identification in WSNs, deep learn-
demonstrated that the proposed algo- ing models—such as DNNs, CNNs, RNNs,
rithm outperformed existing methods in or hybrid architectures—are included. use
terms of accuracy, false-positive rate, and of pre-trained models and transfer learning
resource efficiency. It exhibited a high to enhance anomaly detection performance,
detection rate while minimizing energy particularly in situations with a shortage of
consumption and communication over- labeled training data.
head, which is crucial in resource-con- Investigate the deployment of anomaly
strained WSNs. detection algorithms directly on edge devices
within the WSN, enabling real-time process-
In conclusion, the successful development ing and reducing the need for transmitting
and evaluation of an efficient anomaly detec- raw sensor data to a central server. Develop
tion algorithm for Wireless Sensor Networks lightweight anomaly detection models opti-
have the potential to enhance the reliability mized for edge devices with constrained com-
and security of WSNs in various applications. putational resources and energy.
This project contributes to the ongoing efforts
to make WSNs more resilient and practical for
real-world deployments. As technology contin- References
ues to evolve, so will the challenges and solu- [1] Akyildiz I. Wireless sensor networks: A sur-
tions in the field of anomaly detection in WSNs. vey. Comput Netw. 2002;38(4):393–422.
[2] Barreca A, Curto R, Malavasi G, and
7. Future Work Rolando D. Energy retrofitting for the
modern heritage enhancement in weak real
Future work in anomaly detection for WSNs estate markets: The Olivetti housing stock
is likely to focus on addressing emerging in Ivrea. Sustainability. 2022;14(6):3507.
challenges and leveraging advancements in doi:10.3390/su14063507.
4 Revolutionizing object recognition
beyond CNN and YOLO with deep
learning breakthrough
G. S. S. S. S. V. Krishna Mohan1,a, Mahammad Firose Shaik2,b,
G. Usandra Babu3,c, R. G. V Prasanna1,d, Pinagadi Venkateswara
Rao4,e, and Inakoti Ramesh Raja5,f
1
Department of Electronics and Communication Engineering, Aditya Institute of Technology and
management, Tekkali, India
2
Department of Electronics and Instrumentation Engineering V R Siddhartha Engineering College
Vijayawada, India
3
Department of Electronics and Communication Engineering, Aditya Engineering College,
Surampalem, India
4
Department of AI and ML, School of Engineering, Malla Reddy University, Hyderabad, Telangana,
India
5
Department of Electronics and Communication Engineering Aditya College of Engineering and
Technology, Surampalem, India

Abstract: In the world of computer vision and object detection, there’s a strong connection between these areas.
Object recognition is all about recognizing specific things in pictures and videos. But this paper goes a step fur-
ther than the usual methods of finding objects. It tries to understand images in more detail, like how our eyes
do. The work starts by looking at deep learning and well- known object detection systems like CNN, R-CNN,
and YOLO. These systems can typically find only a few objects in a picture, and they work best at distance of
5–6 meters. However, our new model is much better at this task and has an interesting feature it can even tell
you the names of the objects it sees using Google Translate. This is especially helpful for people with vision
problems because it helps them understand what’s around them better. In summary, the research combines
computer vision, deep learning, and real-time object recognition to enhance visual perception and offer valuable
assistance to individuals with visual impairments.

Keywords: Deep learning, CNN, YOLO, RCNN, visual impairs, computer vision

1. Introduction over 220 million people throughout the

world would experience visual impairment,
The human eye, a vital component of our visual presenting substantial daily challenges [2].
system, is not only integral to perceiving the While the traditional white cane remains a
world but also susceptible to external factors, widely used tool for the visually impaired [3],
including potential contact with pathogens. its limitations and safety concerns propel this
Eye infections and other visual impairments work towards a clear and compelling objec-
may result from some viruses that enter the tive: Developing an affordable, yet incred-
body via the eyes [1]. By 2019, the World ibly useful device that would enable the blind
Health Organization (WHO) predicted that and visually impaired to move about their

a
[email protected]; [email protected]; [email protected]; drgvprasanna@
gmail.com; [email protected]; [email protected]
DOI: 10.1201/9781003606635-4
32 Innovations in Computational Intelligence: Trends and Technologies

environment more quickly, easily, confidently, systems that perform optimally up to 5–6
and with a distinct sense of mobility. Visual meters, our innovative model excels in rec-
impairment extends beyond mere eyesight ognizing various objects, even articulat-
issues, stemming from various causes like ing their names using Google Translate.
trouble with eyesight, cataract development, This distinctive feature proves particularly
issues in the rear side of the eye, or problems advantageous for individuals facing vision
with the optic nerve [4]. People facing visual challenges, enhancing their comprehen-
impairment encounter numerous challenges sion of the surrounding environment. By
in daily life, from navigating roads safely to integrating computer vision, deep learn-
discerning traffic light signals or avoiding ing, and real-time object recognition, our
obstacles. Pursuing education often involves work strives to advance visual perception.
learning Braille, a system of raised dots felt Ultimately, our goal is to foster inclusivity
with the fingers. Basic activities, like visiting in technology, offering valuable support to
the doctor, become challenging as individuals individuals with visual impairments for a
rely on touch and hearing to navigate their more profound understanding of the world
world, introducing an element of motion around them.
awareness.
In the contemporary landscape, artificial
intelligence (AI) stands out as a transforma- 2. Literature Review
tive force, shaping our future with its smart Cheng et al. proposed a comprehensive
capabilities [5]. Advancements in AI, fueled by approach to enhance visual localization for
abundant data and powerful computers, find individuals with visual impairments [6].
applications in diverse fields like retail and Their system comprises multiple components,
delivery services. AI facilitates efficient inven- including a deep descriptor network, 2D–3D
tory management in retail, making shopping geometric validation, and online sequence
more convenient by tracking stock and offer- comparison. In this workflow, a Dual Descrip-
ing personalized product recommendations tor network is integrated with RGB, Infrared,
based on preferences. In the delivery realm, and depth images, enabling the generation
AI plans optimal routes to transport items of robust, context-rich descriptors and local
seamlessly, introducing a dynamic element of features.
motion. Researchers at Springer-Verilog presented
AI’s transformative influence extends to a LiDAR-based technique for predicting
education, enabling personalized support by distances and effectively measuring obsta-
helping teachers understand student progress. cles [7]. They utilize an EfficientDet-LiteV4
In security, AI aids law enforcement in iden- lightweight deep learning model for obsta-
tifying suspects through image processing in cle detection and calculate anticipated dis-
surveillance videos and monitors buildings, tances using LiDAR-generated depth maps.
alerting authorities to any unusual activities. To test their approach, they implemented it
Even in healthcare, when doctors conduct on a Raspberry Pi4 platform integrated with
thorough eye examinations using a contact LiDAR. The results of their experiments dem-
fundus examination tool, they capture not onstrate high accuracy in both obstacle detec-
only static images but also dynamic insights tion and distance estimation.
into motion, aiding in the diagnosis and treat- Shen et al. proposed a two-stage aggregate
ment of various eye conditions.” learning model called VRTMM, designed for
Our research explores a smarter approach interpreting remote sensing images [8]. This
to understanding images beyond merely model combines a variational autoencoder as
spotting objects. In contrast to traditional well as reinforcement learning. Convolutional
 Revolutionizing object recognition beyond CNN and YOLO with deep learning breakthrough 33

Neural Networks (CNNs) are first tuned A helpful system has been developed for
using a variational autoencoder. In the later individuals with partial vision using a Rasp-
stage, a transformer-based model that com- berry Pi 3 and optical character recognition
bines semantic and geographical data gener- (OCR) technology, as discussed in [13]. This
ates textual descriptions. system assists people with limited vision by
Rahman et al. introduced an Internet of converting text from documents into spo-
Things (IoT)-enabled system for automated ken words. However, one drawback is that
object recognition in assisting individuals the complexity increases as a computer is
with visual impairments [9]. Their system required to process the images.
facilitates the recognition of obstacles and
enable audio feedback to the user when nec-
essary. It includes four laser sensors employed 3. Technologies
for object recognition and currency denomi- The contemporary atmosphere has increased
nation identification in real-time scenarios, the risk of inhaling disorders, namely asthma,
both indoors and outdoors. To accomplish chronic obstructive pulmonary disease
these objectives, the system makes use of Ten- (COPD), and also few other disorders. That
sorFlow Lite, MobileNet, and a Single-Shot risk has increased due to increase in air pol-
Detector (SSD). lutants like PM2.5, PM10, etc., a respirator
A tool that reads text from images and can be utilized as an immediate countermeas-
turns it into spoken words is created using ure on an individual level safety measure as
a method called Optical Character Recogni- bringing down pollution levels require much
tion (OCR). In this method [10], a specific longer time than the severity of the problem
algorithm called MSER is used to make sure is allowing.
it’s accurate. The setup involves connecting a
Raspberry Pi to a camera for taking pictures,
and then the MSER algorithm is used to pull 3.1. Deep learning
out the words. On average, this system gets it The neural network in the brain is repli-
right about 85–87% of the time. cated by a specific type of computer program
An advanced audio assistance system is known as deep learning [14]. It is frequently
designed to aid blind individuals by reading referred to as ‘deep learning’ because it
letters or text from documents and identify- employs deep neural networks, which are a
ing objects in their surroundings. This sys- subset of machine learning. Connected lay-
tem, as described in [11], uses technology to ers are utilized in the construction process of
detect both text in documents and obstacles deep learning systems.
nearby. Once detected, the system produces a The surface layer is addressed as the input
voice-based audio output, providing valuable layer, while the base layer is addressed as the
information to help individuals with visual output layer. “Hidden layers” refers to the
impairments navigate their environment more intermediary phases. The term ‘deep’ refers to
effectively. A special tool has been created for neural networks that connect neurons across
people who can’t see well [12]. This tool uses more than two layers. Each hidden layer com-
smart technology to help them read books. prises neurons with interconnections between
It works with a small computer called Rasp- them.
berry Pi 2 and is programmed using Python. The input signal undergoes processing
There’s a monitor to check how the program by the neuron before being sent to the layer
is doing, and it’s especially helpful for visually above it. The weight, bias, and activation
impaired students. However, it can read only function that are applied dictate the inten-
one language. sity of the signal that is sent into the neuron
 34 Innovations in Computational Intelligence: Trends and Technologies

which lies in the layer below. This very net- Regression analysis, which yields the
work consumes large volumes of input data, greatest likelihood photos, is how YOLO
processing them through several layers and formulates the object identification process.
learning more intricate characteristics of the YOLO utilizes convolutional neural net-
data at each layer. works (CNN) to swiftly diagnose entities. As
Neural networks can be grouped into the name indicates, the approach recognizes
shallow and deep types. In a shallow neu- objects using one forward propagation over
ral network, there is a solitary hidden layer a neural network. This indicates that the pre-
connecting both the input and output layers. diction procedure will be performed through-
Deep learning networks, on the other hand, out the whole image with a single algorithm
include multiple layers, such as the 22 layers run. CNN anticipates numerous class prob-
in the Google LeNet model for recognizing abilities and bounding boxes in a single go.
images. Nowadays, deep learning is utilized The YOLO algorithm comes in a wide
in various applications, ranging from TVs variety, with Tiny YOLO and YOLOv5 being
and autonomous vehicles to cell phones and a couple of the more well-known versions.
Google’s fraudulent activity tracking. The YOLO algorithm functions by employ-
ing three distinct strategies:
3.2. YOLO (You Only Look Once)
• Residual blocks
algorithm • Bounding box regression
You Only Look Once (YOLO) is a cutting- • Intersection Over Union (IOU)
edge problem-solving object identification
technique that was originally published in the 3.3. Region-based convolutional
esteemed academic publication “You Only neural network (RCNN)
Look Once: Unified, Real-Time Object identi-
fication” in 2015. The authors of this work are The multifaceted field of computer vision
Joseph Redmon, Santosh Divvala, Ross Gir- surged in popularity with the advent of
shick, and Ali Farhadi, as shown in Figure 4.1. Convolutional Neural Networks (CNNs),
This real-time technique can be employed to particularly in the context of self-driving
identify various objects in images. automobiles taking the forefront. Within

AQ1:
Please
provide
source
for all the Figure 4.1. YOLO architecture.
figures
Source:
AUTHOR'S SELF GENERATED
 Revolutionizing object recognition beyond CNN and YOLO with deep learning breakthrough 35

computer vision, object detection plays a piv- Sophisticated neural networks often employ
otal role. This technique simplifies tasks such COCO sets, a kind of rigorous, better visual
as pose estimation, vehicle identification, sur- dataset. As an illustration, COCO has been
veillance, and more. used extensively to evaluate and ­compare real-
In contrast to classification algorithms, time object identification skills. The formats
object identification techniques use a bound- of the COCO dataset are quickly interpreted
ing box to encircle the object of interest by sophisticated neural network frameworks.
inside an image. Predicting the exact number The various attributes of COCO are
of bounding boxes in an object identification
scenario that accurately represent the numer- • Segmentation of objects with comprehen-
ous places of interest may be challenging. The sive prototype annotations
variable length of the output layer, reflect- • Contextualizing apprehension
ing the changing number of items of interest, • Super pixel stuff segmentation
makes it impractical to address this issue by • More than 200’000 images of the total
simply adding a fully connected layer on top 330’000 are labelled
of a conventional convolutional network. • 1.5 Mio object prototype
To overcome these challenges, a solution • 80 object assorts, the “COCO classes”,
involves identifying specific regions of interest which comprises “things” for which indi-
and applying a CNN algorithm to determine vidual instances may be promptly labelled
whether an object is located in the specified (human-being, automobile, any kind of
region. The initial approach of creating a furniture etc.,)
large number of areas could lead to computa- Let’s utilize the image from the previous
tional challenges. As a result, algorithms like section as input for an object detection pro-
R-CNN, YOLO, and others have been devel- cess, which encompasses two tasks [8]: object
oped to swiftly identify instances of interest. localization and image classification. Initially,
object localization draws bounding boxes
4. Proposed Method around one or more objects in an image, as
illustrated in Figure 4.2. Subsequently, image
The following arguments make the YOLO classification predicts the type or class of an
algorithm crucial: object by comparing it with a predefined
dataset and assigning a class label to the
1. Speed: This technique is able to predict
image. Consequently, the YOLO algorithm
objects in real time, which speeds up
is employed for both object localization and
detection.
image classification.
2. High accuracy: With minute background
error, the YOLO forecast approach yields • The advantages of this approach include:
decisive results. • Facilitating the correct identification of
3. Learning capability: Due to the algo- objects through speech.
rithm’s superior learning capabilities, it is
capable of distinguishing object represen-
tations and use them for detecting

4.1. Data set

The term COCO, which stands for Common
Objects in Context, refers to the original goal Figure 4.2. Feature extracting process by RCNN
of the picture collection, which was to improve methodology.
visual understanding. Computer vision Source:
AUTHOR'S SELF GENERATED
36 Innovations in Computational Intelligence: Trends and Technologies

• Contributing to decision-making processes.

• Mitigating human risk.
• Eliminating the need for batteries
implanted within the body.

4.2. Software packages used

• NumPy: A versatile package designed for
array handling is known as NumPy. It
provides a multidimensional array object
with exceptional speed and features for
efficient collaboration with these arrays.
Figure 4.3. Block diagram for text to speech
• OpenCV: An open-source computer
conversion.
vision library, OpenCV, enables machines
Source:
to identify faces and other objects. AUTHOR'S SELF GENERATED
• pyttsx3: A Python library for text-to-
with image processing methods such
speech conversion, pyttsx3 is compatible
as noise removal, followed by feature
with both Python 2 and 3 and operates
extraction to identify lines, regions, and
offline, distinguishing it from interactive
potentially emphasize certain textures.
libraries.
The comprehension and extraction of
• PyWin32: For utilizing the Win32 appli-
objects in an image or video are achiev-
cation programming interface (API) func-
able through computer vision technology
tionality with Python on Windows, the
known as object detection. Specifically,
PyWin32 library offers valuable extensions.
object detection employs RCNN technol-
• OpenCV- In addition to the OpenCV
ogy to create bounding boxes around the
module found in the open-source library,
identified objects.
the system requires OpenCV contrib
• Object Tracking: Using a camera as well as
to effectively implement SURF feature
blob detection, a moving object is tracked
descriptions.
over time. Blob detection involves search-
• Imutuls: Basic image processing opera-
ing for areas in a digital image that exhibit
tions such as boundary detection, sort-
distinct characteristics from the surround-
ing contours, skeletonization, translation,
ing areas, such as differences in brightness
rotation, and scaling are made easier
or color. In a broader sense, a blob is a
using OpenCV. Python 2.7 and Python 3
distinct section of an image. In the con-
are supported.
text of object recognition, these identified
• winsound: The winsound module provides
zones may signify the presence of objects
access to primary sound-playing com-
or portions of objects in the visual domain.
ponents offered by Windows systems. It
The combination of YOLO with RCNN is
includes various constants and functions,
a widely adopted strategy for blob detec-
such as the ability to beep the PC’s speaker.
tion. Because it can produce forecasts for
the whole image in a single algorithm run,
4.3. Text to speech conversion YOLO is particularly helpful in this regard.
Primarily, an image or video can be provided • Deep Neural Network (DNN): Deep Neu-
as input, as illustrated in Figure 4.3. ral Networks simulate the neuronal net-
work in the brain. Deep neural networks
• Object Detection: The process of detecting are a part of the “deep learning” disci-
objects in any image typically commences pline of machine learning. Deep learning
 Revolutionizing object recognition beyond CNN and YOLO with deep learning breakthrough 37

algorithms consist of interconnected lay- given time, the camera will identify and label
ers. The hidden layers are found in the 5–6 objects, as illustrated in Figure 4.4.
front-end layer, which is addressed as the These steps offer an overview of the key
input layer, and the back-end layer, that procedures involved in running your Python
which is addressed as the output layer. program, starting from importing libraries
The term ‘deep’ indicates a network that to the continuous video processing loop, and
connects neurons across more than two concluding with the program’s termination.
layers. Each Hidden layer comprises neu- Keep in mind that executing the program
rons that are activated during the process. requires running it within a suitable Python
The network consumes large volumes development environment or IDE.
of input data, processing them through
several layers, and progressively learns 1. Import necessary libraries (cv2, imutils,
intricate characteristics of the data at numpy, pyttsx3).
each layer. This iterative learning process 2. Define a text-to-speech (TTS) function.
occurs at each stage of the layer, allowing • Initialize the TTS engine.
the network to understand progressively • Adjust the TTS engine’s properties.
complex features of the input data. • Use TTS to speak labels.
• Trained Data: In this work, a pre-defined 3. Load the YOLO model and class names.
dataset has been imported, consisting of a 4. Open a camera feed for video capture.
network trained on more than 50 images. 5. Start a loop to continuously process
This trained network excels in classify- frames from the camera.
ing photos into various object categories, • Resize the frame for processing.
encompassing a wide range, from animals to • Detect objects in the frame using
everyday objects like keyboards, mice, and YOLO.
various stationery items. The pre-trained • Filter and display the detected objects.
model utilized in this context is COCO • Use TTS to announce the labels of
(Common Objects in Context). Leverag- detected objects.
ing the OpenCV-contrib-python package, • Continue processing frames until the
the recognized objects are compared to a user presses ‘q’ to quit.
pre-trained database. The identified images 6. Release the video capture when done.
are then labeled through a tagging process,
translating the visual information into
text. Subsequently, the spelled-out items
are announced using the winsound pack-
age, and the text is converted into speech
through the pyttsx3 package.

5. Results and Discussion

The pre-trained database, named ‘coco,’ is
sourced from the internet and includes pre-
trained objects. Alternatively, another data-
base can be chosen from the internet.
Upon running the code, the camera mod-
ule will open using the OpenCV package, Figure 4.4. Detection of object by camera
detecting objects in front of the camera. The module.
detected objects will be labeled, and at any Source:
AUTHOR'S SELF GENERATED
38 Innovations in Computational Intelligence: Trends and Technologies

After detecting objects, the identified References

objects will be spelled out using the winsound
[1] Pablo Argüeso. 2022. Human ocular
package. The winsound module provides
mucins: The endowed guardians of sight,
access to the fundamental sound-playing Advanced Drug Delivery Reviews. Volume
capabilities offered by Windows platforms, 180. 114074. https://doi.org/10.1016/j.
including various functions and constants. addr.2021.114074.
Additionally, the detected objects can be [2] Boyce P. Light. 2022. Lighting and
logged into a separate file, as depicted in human health. Lighting Research and
­Figure 4.5. Technology. 54(2): 101–144. doi:
10.1177/14771535211010267.
[3] V. Adusumilli, M. F. Shaik, N. Kolavennu, L.
B. M. T. Adepu, P. A. V., and I. R. Raja. 2023.
Reading Aid and Translator with Raspberry
Pi for Blind people. 9th International Con-
ference on Advanced Computing and Com-
munication Systems (ICACCS. Coimbatore,
India, pp. 327–331).
[4] Bi, Siguo, Cong Wang, Jilong Zhang, Wutao
Huang, Bochun Wu, Yi Gong, and Wei Ni.
2022. A Survey on Artificial Intelligence
Aided Internet-of-Things Technologies in
Emerging Smart Libraries. Sensors 22(8):
2991. https://doi.org/10.3390/ s22082991.
[5] S. M. Zekavat, et al. 2022. Deep Learning of the
Retina Enables Phenome- and Genome-Wide
Figure 4.5. List of detected objects.
Analyses of the Microvasculature, Circula-
Source: tion 145:134–150. https://doi.org/10.1161/
AUTHOR'S SELF GENERATED
CIRCULATIONAHA.121.057709.
6. Conclusion [6] Ruiqi Cheng, Weijian Hu, Hao Chen,
Yicheng Fang, Kaiwei Wang, Zhijie Xu,
In conclusion, YOLO, a speedy and versatile and Jian Bai. 2021. Hierarchical visual
object detection method, efficiently identi- localization for visually impaired people
fies a range of objects in diverse scenarios. using multimodal images. Expert Systems
It transforms images into text and spoken with Applications, Volume 165. https://doi.
words, aiding individuals with visual impair- org/10.1016/j.eswa.2020.113743.
ments in comprehending visual content. The [7] Kuriakose, B., Shrestha, R., Sandnes, F.
study underscores YOLO’s rapid object rec- E. 2022. LiDAR- Based Obstacle Detec-
ognition and its role in enhancing accessibil- tion and Distance Estimation in Naviga-
ity and understanding of visual information, tion Assistance for Visually Impaired. In:
particularly for those with vision challenges. Antona, M. and Stephanidis, C. (eds), Uni-
By amalgamating various technologies, the versal Access in Human- Computer Inter-
action. User and Context Diversity. HCII
research strives to boost the speed of object
2022. Lecture Notes in Computer Science,
identification and improve the overall com-
vol 13309. Springer, Cham. https://doi.
prehension of visual content. Essentially, org/10.1007/978-3-031-05039-8_35.
YOLO emerges as a crucial player in this [8] Xiangqing Shen, Bing Liu, Yong Zhou, Jiaqi
endeavor, showcasing its effectiveness beyond Zhao, and Mingming Liu. 2020. Remote
conventional methods and contributing sig- sensing image captioning via Variational
nificantly to a more inclusive approach to Autoencoder and Reinforcement Learn-
interpreting the visual world. ing. Knowledge-Based Systems. Volume
 Revolutionizing object recognition beyond CNN and YOLO with deep learning breakthrough 39

203. 105920, https://doi.org/10.1016/j. Technologies in Electronics, Computing and

knosys.2020.105920. Communication (ICETECC): 1–6.
[9] Md. Atikur Rahman and Muhammad Sheikh [12] Harum, N. B., Zakaria, N. A., Emran, N.
Sadi. 2021. IoT Enabled Automated Object A., Ayop, Z., and Anawar, S. 2019. Smart
Recognition for the Visually Impaired. Com- Book Reader for Visual Impairment Person
puter Methods and Programs in Biomedicine using IoT Device. International Journal of
Update, Volume 1, https://doi.org/10.1016/j. Advanced Computer Science and Applica-
cmpbup.2021.100015. tions. 10(2): 251–259.
[10] Ganesh, A. 2019. OCR based Image Pro- [13] Vinaya Phutak, Richa Kamble, Sharmila
cessing with Audio Output for Visually Gore, Minal Alave, and R. R. Kulkarni. 2019.
Challenged People. International Journal Text to Speech Conversion using Raspberry –
for Research in Applied Science and Engi- PI. International Journal of Innovative Science
neering Technology. 7. 599–604. 10.22214/ and Research Technology. 4(2): 291–293.
ijraset.2019.3104. [14] Alzubaidi, L., Zhang, J., Humaidi, A. J. et al.
[11] Sarwar, S., Turab, M., Channa, D., Chan- 2021. Review of deep learning: concepts, CNN
dio, A., Sohu, M. U., and Kumar, V. 2022. architectures, challenges, applications, future
Advanced Audio Aid for Blind People. 2022 directions. J Big Data 8, 53 (2021). https://
International Conference on Emerging doi.org/10.1186/s40537-021-00444-8.
5 Investigating efficacy of emerging ICTs
for the Indian construction sector
Sanjay Pandeya and G. T. Thampib
Thadomal Shahani Engineering College Bandra(w), Mumbai, India

Abstract: The construction sector, being one of the least digitized and major contributors in developed or
developing economies faces lots of challenges. We see an opportunity for reorienting the sector with cutting-
edge information and communication technologies (ICTs) integration. The construction sector challenges
namely poor productivity, lower efficiencies, safety, and sustainability can be addressed by ICT adaptation. The
research objective is to investigate the efficacy of utilizing ICTs as a technological intervention for the Indian
construction sector. The methodology adopted in this research study encompasses a hybrid approach, combin-
ing qualitative interviews of industry professionals and quantitative analytics of data gathered from question-
naire survey of construction projects. The findings of the questionnaire survey highlight that we can harness
emerging digital technology in various construction tasks to overcome challenges of the sector, and also identify
numerous potential benefits of utilizing ICTs. Additionally, the study also explores the challenges and barriers
faced by stakeholders in the adoption and implementation of cutting-edge ICTs within construction companies.

Keywords: Construction sector, cutting-edge ICTs, efficacy, productivity, efficiency, safety, cost-effectiveness

1. Introduction possible with the conventional approach.

With this fresh insight into the projects,
Recently, significant growth has been seen in stakeholders can make better decisions more
the construction sector due to urbanization quickly and intelligently which results reduc-
and infrastructure development. However, tion in time, cost, and risks, thus improving
the sector is facing challenges with efficiency, the performance of the overall project [1–3].
cost-effectiveness, and safety. Driven by rapid In this paper, the effectiveness and impact of
urbanization, expanding infrastructure, and emerging ICTs on the Indian construction
growing demand for innovative construction sector will be thoroughly examined. In the
technologies and solutions, the Indian con- conventional approach of construction, tasks
struction sector needs to undergo a dynamic are very slow and time-consuming as they are
transformation. So, by keeping those chal- labor and paper-based resulting in slippage
lenges in mind, the inclusion of emerging of schedule and budget overrun which ulti-
ICTs such as drones, IoT, 5G, Cloud, AI/ mately force projects to fall into the catego-
ML, etc. has become an important aspect of ries of failed or challenged projects. On the
transforming the conventional construction other hand, the ICT-enabled smart approach
approach into a smarter approach. Emerg- of construction where tasks are carried out
ing ICTs can change the way construction at a reasonably faster speed with optimum
projects are designed, monitored, and exe- scale as they are digitally enabled, results
cuted. For instance, ICTs have the capability in the completion of tasks on schedule and
of fetching real-time data that can provide budget which in turn enhances the productiv-
insight into projects that were previously not ity and efficiency of the projects. Therefore,

[email protected]; [email protected]
a

DOI: 10.1201/9781003606635-5
 Investigating efficacy of emerging ICTs for the Indian construction sector 41

ICT adaptation in different construction extract meaningful insight from it, which ulti-
tasks proliferates the probability of success- mately will help in better decision-making.
ful completion of construction projects. In Doing so all stakeholders can benefit from it
this research work, we have identified various and coordination and communication will be
construction tasks where ICTs can be utilized, in a very efficient and effective manner. This
numerous advantages of using ICTs in con- sharing of information will curtail the time
struction companies, and different hindering required for data gathering, processing, and
factors coming in the way of ICT adaptation decision-making [8, 9]. Researchers have also
by the construction sector [4-6]. So, to carry highlighted that the utilization of wireless
out our research work we have conducted a networking omits the need for fellow work-
series of interviews with industry profession- ers, not to be present physically at the site,
als and recorded their valuable opinions and at the same moment. Web-based companies
suggestions. We have also prepared a ques- can get an edge in their operations [10]. Few
tionnaire, distributed it among the industry researchers have highlighted virtual modeling
expert, obtained their inputs, and lastly ana- for building design. It includes all relevant
lyzed the data obtained. The utilization of design information needed for every con-
emerging ICTs in Indian construction compa- struction phase during the project life cycle.
nies is a technological paradigm shift that has Any modification in the models gets updated,
the potential not only to change the way the and any design-related information may get
industry is working but also to enhance pro- extracted from this database, automatically
ductivity and efficiency, resulting in a higher [11, 12]. Some researchers are also talking AQ1:
Please
success rate of the projects. about the formation of a virtual team for col- provide
laborative purposes to improve communica- reference
tion, coordination and sharing of information citation
for refer-
2. Literature Review of ICT among stakeholders, teams, and enterprises. ences 13,
Adaptation in the Construction The ideation and creation of virtual teams for 14, 20,
21.
construction companies have been advised
Companies by the researchers [15]. ICTs may support
After the review process of the literature, we the virtual team for routine operations, but
have found that Some research papers high- unfortunately, construction enterprises are
lighted that the industry is using some basic much behind in adopting them.
ICTs like MS Office, AutoCAD, Wi-Fi, etc. [6, The present scenario of ICTs utilized in
7] for construction tasks. Many authors con- different construction company activities is
cluded that the utilization of ICTs can aid in discussed in this section. We have studied sev-
enhancing the present situation of construction eral conference papers, journal papers, blogs,
companies, apart from the fact that construc- and articles on different websites, discussing
tion companies are digitally divided, they need old and current trends of the different ICTs
to embrace ICTs to maximize the productivity for construction tasks or applications. It has
and efficiency of the construction enterprises. been noticed that many construction activities
Transfer of data and information among harnessed BIM, AutoCAD, mobile/web-based
the stakeholders are very slow and limited tools and technologies, virtual conferencing,
during and after the construction work as and information sharing systems, etc. [16,
many professionals are not interested in shar- 17]. Few articles, white papers, and bloggers
ing information and keeping it in their silos. have mentioned adaptation of emerging ICTs
So, with the advent of these newer commu- viz. drone, IoT, 5G, Cloud, Blockchain, and
nication technologies in the construction sec- AI/ML, etc. may be good options for different
tor, companies can get this precious data and construction activities/tasks. Deployment of
42 Innovations in Computational Intelligence: Trends and Technologies

various ICTs for construction activities is in 3. Methodology

the early stage and good attention is needed
by the researcher to grow the present situa- 3.1. Research objectives
tion of construction companies [18, 19]. It has
been noticed that many construction activi- Before carrying out research work, objectives
ties are using AutoCAD, mobile/web-based were finalized. The prime objective of the
applications, BIM, information systems, research is to estimate the efficacy of emerg-
video conferencing, etc. [16, 17] for their day- ing ICTs for the Indian construction sec-
to-day activities to enhance their productivity tor. Quantifying the overall performance of
[22]. According to many research, increasing construction companies planning to harness
productivity and building efficiencies may be ICTs, is another goal of the research project.
possible with the ICT adaptation. Planning,
designing, and implementation of a project 3.2. Research questionnaire
need effective knowledge sharing among var-
Based on the evaluation of the present sce-
ious stakeholders. It can be obtained through
nario, the research questionnaire was formu-
proper adaptation of ICT by the construction
lated for the Indian construction companies.
companies [23, 24]. Construction industries,
whether they are small, medium, or large • What is the impact of harnessing emerg-
can use emerging ICTs to enhance efficiency ing ICTs for construction sector tasks?
and productivity by minimizing dependabil- • What are the different benefits of utilizing
ity. The utilization of ICTs by construction ICTs in construction firms?
companies has various advantages, including • What are various hindering factors for
enhanced productivity, cost reduction, time ICT adaptation in the Indian construction
savings, information sharing, waste manage- sector?
ment, and many more [25]. Construction site
employees with limited technical proficiency,
fear of virus attacks, and incompetence when 3.3. Data collection and analysis
handling computing devices have limited their To collect industry data, questionnaires were
ability to use them properly [26, 27]. Also, designed and distributed. During the early
professionals think that ICT embracement is phase of the survey, we created a Google
a positive initiative to enhance the efficiency form and floated the questionnaire electroni-
and productivity of construction companies. cally among the industry professionals with
Construction firms have limited success the humble request to participate in research
due to the low penetration of emerging ICT work, but unfortunately received negligi-
solutions. We did an exhaustive literature ble responses. Therefore, we decided to go
survey where we noticed many hurdles to the offline survey and approach industry profes-
embracement of ICTs in construction compa- sionals personally. Many professionals have
nies [28, 29]. ICT adoption is confined by the not shown their interest in participating in
complex structure of the construction sector, the survey but a few interested professionals
low productivity, professionals with relatively seriously contributed by providing genuine
low work experience, schedule overruns, and responses to the questionnaire and also will-
budget overruns. A good no of researchers ingly participated in the interview process to
has done their research to assess the embrace- provide important insight about the activities
ment and utilization of ICT for construction which helped me for a meaningful contribu-
work and tried to earmark various reasons tion to the existing body of knowledge. Here
that are acting as a hurdle for construction data were obtained from primary sources of
works [30–32]. information including questionnaires and
 Investigating efficacy of emerging ICTs for the Indian construction sector 43

interviews. In 223 questionnaires were dis- middle management and 3rd cadre is lower
tributed for data collection and we obtained management. The participation of the 1st
118 responses. cadre is 37% whereas 41% of respondents
A Likert scale of 3-points has been used for are playing their role under the 2nd cadre and
analysis of data, where the scale ranges from belonging in the middle management catego-
1 to 3. The scale of “Agree” was 3, “Disagree” ries. The lower management category share
was scored 2, and “Neutral” was scored 1, is 22%.
The statistical tool used for data analysis and
visual representation of this research study
was Microsoft Excel. 4. Results and Discussion
After the questionnaire survey of the con-
3.4. Types of construction companies struction companies, it has been noticed
that emerging ICTs viz. drone, IoT, and 5G
that contributed to the study
technology and cloud, etc. can be a power-
Figure 5.1 shows the types of construction ful digital solution for the construction sector
enterprises that participated in the research to improve productivity and efficiency with
study. The contribution of large-scale enter- the enhanced safety of the personnel. These
prises is 33% whereas 41% of enterprises are cutting-edge ICTs can be used for various
in medium-scale categories, collectively these construction tasks and it has the potential to
two categories make it nearly three-fourths change the ways construction tasks are per-
of the participating organization. Small-scale formed through conventional approaches.
and microscale enterprise shares are 15% and
11% respectively. 4.1. ICTs can be utilized to perform
different construction sector’s
3.5. Cadre-based classification tasks
(CBC) of respondents
4.1.1. Topographic mapping and land
Figure 5.2 shows the Cadre-based classifica-
tion of the respondents and this cadre depends survey
upon the type of role they performed in their Topographic mapping and land survey-
enterprises. There are three cadre of respond- ing using ICT, particularly drones have
ents that participated in the research study. become increasingly popular due to their
1st cadre is higher management, 2nd cadre is time-efficient, cost-effective and ability to fly

AQ2:
Please
provide
source
for all the Figure 5.1. Types of construction companies Figure 5.2. Cadre-based classification of
figures contributed to the study. respondents.
and tables
Source: Source:
AUTHOR'S SELF GENERATED
AUTHOR'S SELF GENERATED
44 Innovations in Computational Intelligence: Trends and Technologies

hard-to-access terrain. 82% of respondents 4.1.5. Personnel safety

are willing to adopt drones for topographic
IoT-enabled wearables can monitor health
mapping and land surveys in their enterprises
parameters like body temperature, pressure
as shown in Figure 5.3.
even ECG of workers. Construction work-
ers frequently have to climb unsteady heights
4.1.2. Equipment tracking and navigate around hazardous conditions.
Industry professionals can effectively use ICTs, particularly drones can be used to
ICTs for equipment tracking, functioning replace workers in such a dangerous situa-
status of equipment, enhancing security, etc., tion and mitigate the risk. According to 73%
which can reduce downtime, and improve of respondents, worker safety in the construc-
asset/equipment visibility and overall man- tion industry is of utmost importance since it
agement, and this is supported by 67% of can seriously harm personnel and the indus-
respondents who like to adopt ICTs for try as well, as shown in Figure 5.3.
tracking equipment in their organizations as
shown in F ­ igure 5.3. 4.1.6. Structural inspection
Emerging ICTs, particularly drones fitted
4.1.3. Real-time progress reporting with sensors, cameras, and other devices can
Harnessing ICTs can fetch real-time data from examine the fine details and stability of struc-
construction sites and equipments, allowing tures. Drones equipped with thermal sensors
all stakeholders (Clients, contractors, design- can detect temperature variations, which can
ers project managers, etc.) to keep track of the reveal hidden problems such as electrical,
project’s progress and identify any deviations water, or insulation problems. In the long run,
from the original plan of action. This enables this quality assurance facilitates the develop-
timely adjustments and helps in meeting pro- ment of client relationships. 58% of respond-
ject deadlines, thus increasing the possibili- ents support the use of ICTs for structural
ties of successful project completion and this inspection. as shown in Figure 5.3.
real-time collected data can be used by spe-
cialized software to generate visual reports. 4.1.7. Supply replenishment
As per our survey findings, 80% of respond- Counting of materials becomes easy when
ents are willing to embrace ICTs for real-time supply units are hooked up with IoT-ena-
progress reporting of the project as shown in bled tags. As soon as the system notices that
Figure 5.3. the counting of units drops below a certain
threshold level it automatically generates
4.1.4. Security surveillance order requests, thus waiting time goes down,
and the probability of completion of projects
The ICTs-driven security surveillance pro-
increases. This also reduced the cost incurred
vides an efficient solution for monitoring large
with the project. As shown in Figure 5.3, 75
areas, enhancing situational awareness, and
% of respondents are likely to adopt ICTs for
responding to security incidents. The sensors
supply replenishment purposes.
of drones can ensure whether the equipment
is in a secure enough location or not. Drone
surveillance cameras can also ensure unau- 4.1.8. Concrete curing
thorized access to the site. our survey findings Sensors deployed in concrete can transmit
reflect that 72% of professionals would like temperature, strength, and humidity data to
to adopt it for security surveillance as shown phones or tablets via a 5 G-enabled mod-
in Figure 5.3. ule, and then managers can plan curing
 Investigating efficacy of emerging ICTs for the Indian construction sector 45

Table 5.1. Respondent’s opinion ranking of

construction tasks performed by ICTs
Sr Construction Respondent’s
No. industry tasks Opinion
Performed by Percentage Ranking
ICTs (%)
1 Topographic map 82 1
and land survey
Figure 5.3. Harnessing ICTs for construction
industry tasks. 2 Equipment 67 8
tracking
Source:
3 Real-time 80 2
AUTHOR'S SELF GENERATED
progress reporting
4 Security 72 6
accordingly. As depicted in Figure 5.3, 76% surveillance
of the respondents are willing to use emerg-
5 Personnel safety 73 5
ing ICTs for concrete curing.
6 Structural 58 9
inspection
4.1.9. Waste management 7 Supply 74 4
Sensor-enabled smart bins or vehicles help in replenishment
handling waste properly, doing so penalties 8 Concrete curing 76 3
for the various contractors can be avoided. 9 Waste 69 7
Figure 5.3 shows that 69% of respondents management
are interested in adopting ICT-enabled waste
Source:
management for their enterprises AUTHOR'S SELF GENERATED

4.2. Respondent’s opinion ranking of 4.3. Perceived benefits of using ICTs

construction tasks performed by for construction companies
ICTs
The integration of ICTs in construction com-
Respondent’s opinion ranking of construc- panies brings forth a range of substantial
tion tasks performed by ICTs has been shown benefits that positively impact various aspects
in Table 5.1. As per this ranking table, 82% of the construction process. Numerous ben-
of respondents are willing to use ICTs, par- efits of using ICTs for performing different
ticularly drones for topographic mapping and tasks in the construction industry have been
land surveys of their enterprises, and in the perceived.
ranking table, this task is ranked as 1. Real-
time progress reporting is having rank 2 with
80% response. Concrete curing has 3rd rank 4.3.1. Time saving
in the rank table with 76%, supply replace- ICTs have the potential to enhance time effi-
ment is ranked 4th with 74%, personnel safety ciency in the construction industry by con-
is ranked 5th with 73% and Security Surveil- tributing in several ways whether it is through
lance is ranked 6th with 72% of responses, surveying and mapping, site inspection, mon-
waste management, and equipment tracking itoring construction progress, safety inspec-
having 69% and 67% responses respectively, tions, Concrete curing, etc. As per our finding
lastly, structural inspection has 9th rank with 88% of respondents were willing to use ICTs
58% positive responses. for time saving as shown in Figure 5.4.
46 Innovations in Computational Intelligence: Trends and Technologies

4.3.2. Budget controlling

ICTs can streamline several construction
activities like topographic mapping and land
surveys, real-time progress monitoring, etc.
which can lead to substantial cost and time
savings. Our survey reflects that 82 % of
respondents are looking to embrace digital
solutions for overall budget control of pro-
jects as depicted in Figure 5.4. Figure 5.4. Perceived benefits of using ICTs for
construction companies.
Source:
4.3.3. Enhance safety and security AUTHOR'S SELF GENERATED

Drones equipped with sensors can improve

the security of the construction site by pro-
4.4. Ranking of the benefits of using
viding 24/7 monitoring and surveillance. Any ICTs as per respondent’s opinion
attempt made by the trespasser or unauthor- Table 5.2 depicts a rank matrix of various
ized personnel can be detected immediately advantages of using ICTs for different con-
and a response to the security threats shall be struction tasks. One of the significant bene-
sent. As per Figure 5.4, survey results reflect fits is time-saving, having rank 1 in the rank
that 80% of respondents feel that ICTs may matrix is supported by 88% of respondents.
uplift the present safety and security scenario 86% of respondents support the improve-
of construction works and sites. ment of productivity which is ranked 2nd in
the matrix. Budget controlling is having 3rd
4.3.4. Reduced paperwork rank with the 82% responses. Enhancement
of safety and security has 80% support and
We have identified multiple ways in which is ranked 4th. 69% of respondents support a
ICTs can facilitate the minimization of reduction in the paper which will contribute
paperwork such as digitally enabled sur- to green construction and having 5th rank.
veys and mapping, progress tracking and
reporting, inventory management, etc. In
our survey, 69% of respondents feel that Table 5.2. Ranking of the benefits as per
ICTs can digitize the construction process respondent’s opinion
hence reducing paperwork, as shown in Sr Perceived benefits Respondent’s Opinion
Figure 5.4. This also promotes sustainable No. of adopting ICTs Percentage Ranking
construction practices by reducing carbon for construction (%)
footprint. firms
1 Time-saving 88 1
4.3.5. Improved productivity 2 Budget 82 3
controlling
Incorporating ICTs in the construction indus-
try can help to improve productivity in several 3 Enhanced safety 80 4
and security
ways such as streamlining various processes,
enhancing communication among the stake- 4 Reduced 69 5
holders, fetching real-time data, etc..86% of paperwork
respondents would like to adopt ICTs for 5 Improved 86 2
various tasks to enhance the productivity of productivity
enterprise as shown in Figure 5.4. Source:
AUTHOR'S SELF GENERATED
 Investigating efficacy of emerging ICTs for the Indian construction sector 47

4.5. Hindering factors for ICT management support is ranked 2nd with 81%
adaptation in the indian of respondent’s opinion. 79% of Respondents
construction sector are in the notion that the training of employ-
ees to use ICTs is also a serious hurdle and
As depicted in Figure 5.5, the primary barrier ranked as 3rd. Fear of job security is also a
to ICT adoption in the construction sector prime concern for the respondents with 71%,
is the higher cost of purchasing and main- ranked as 4th. Government support for ICT
taining ICTs. 85% of respondents say that integration is at 5th place with 68% and lastly,
the biggest concern is the higher cost of ICT legal and regulatory compliance have the last
adoption. 81% of respondents feel manage- position in the matrix with 64%. Apart from
ment support is 2nd biggest concern. 79% of the above hindering factors respondents also
responders rate training on ICT as a serious highlighted the limited skill of personnel with
concern. 71% of respondents also think that low expertise in managing/handling software
job security is a major hurdle for ICT adapta- solutions and digital devices, data analysis,
tion in construction companies and concerns and its interpretation.
regarding legal and regulatory compliance
are shared by 64% of respondents. Addition-
4.7. Overall performance of
ally, according to 68% of respondents, govt.
support poses a significant obstacle to ICT construction companies after
adoption in the construction industry. ICT adaptation
The utilization of emerging ICTs for construc-
4.6. Ranking of hindering factors tion companies is in the nascent stage and
for ICT adaptation as per has the potential to enhance the overall per-
respondent’s opinion formance of the enterprises. The majority of

The following ranking Table 5.3 depicts the

ranking of the different hindering factors for Table 5.3. Ranking of hindering factors for ICT
ICT adaptation. One of the biggest obstacles adaptation
in the adaptation of ICTs in Indian construc- Sr ICT hindering Respondent’s opinion
tion firms is the “higher cost of ICT adapta- No. factors Percentage Ranking
tion” in the industry which is ranked as 1, in (%)
rank Table 5.3 with 85% of responses. Top 1 Higher cost of 85 1
adaptation
2 Training of 79 3
employees to use
ICTs
3 Government 68 5
support
4 Management 81 2
support
5 Fear of job 71 4
security
6 Legal and 64 6
Figure 5.5. Hindering factors for ICT regulatory
adaptation in the Indian construction sector. compliance
Source: Source:
AUTHOR'S SELF GENERATED
AUTHOR'S SELF GENERATED
48 Innovations in Computational Intelligence: Trends and Technologies

Regulatory compliance remains a significant

hurdle, necessitating ongoing dialog between
industry stakeholders and regulatory bodies
to establish clear guidelines. This quantified
investigation reaffirms the strategic embrace-
ment of emerging ICTs for Indian construc-
tion sectors and holds immense promise
for revolutionizing traditional practices of
accomplishing numerous construction tasks.
Figure 5.6. Overall performance of construction The benefits of enhanced efficiency, safety,
companies after ICT adaptation. cost-effectiveness, and sustainability have the
Source: AUTHOR'S SELF GENERATED
potential to reshape the construction sector.
However, the study also sheds light on the
respondents agree that overall performance challenges that accompany the integration of
will be enhanced by 85%, after the adapta- emerging ICTs in Indian construction compa-
tion of cutting-edge ICTs by the construction nies. This research work is of good use for the
companies, as shown in Figure 5.6. construction sector professionals, research-
ers, and academic fraternity as ICTs reflect its
usefulness for a group of construction tasks
5. Conclusion with multiple benefits.
In the dynamic landscape of Indian construc-
tion enterprises, the integration of ICTs has References
emerged as a pivotal catalyst for construc-
tion sector transformation. The study find- [1] Dixit S, Stefan A, Musiuk A, and Singh P.
ings affirm that the utilization of ICTs in Study of enabling factors affecting the adop-
construction projects has demonstrated sig- tion of ICT in the Indian built environment
sector. Ain Shams Eng J. 2021;12:2313-2319.
nificant improvements in various critical
[2] Bello SA, Oyedele LO, and Akinade OO.
areas. The ability of ICTs to conduct precise Cloud computing in the construction indus-
site surveys, monitor progress in real-time, try: Use cases, benefits and challenges.
collect accurate data, supply replenishment, Autom Constr. 2021;122:103441.
waste management, etc., has enhanced pro- [3] Sawant RS, Ravikar A, Bagdi N, and Bel-
ject productivity, efficiency, and decision- lar V. Drone technology in the construction
making processes. Additionally, the improved industry in the Indian construction industry.
safety measures through remote inspections VIIRJ. 2020;Special Issue:2319-4979.
and early hazard detection contribute sub- [4] Mahajan G. Applications of drone technol-
stantially to the overall well-being of the ogy in construction industry: A study 2012-
workforce. The time and cost savings realized 2021. Int J Eng Adv Technol. 2021;11(1).
from streamlined processes, reduced rework, [5] Pandey S, and Thampi GT. Literature review
enhanced safety and security, and optimized on efficacy of ICT in realm of construc-
tion industry. J Emerg Technol Innov Res.
resource allocation can help to enhance the
2022;9(8).
project’s performance. The financial impli- [6] Gamil Y. Internet of things in construction
cations of ICT implementation cannot be industry revolution 4.0. Department of
understated and initial investment costs for Civil and Environmental Engineering, Uni-
ICT may be a serious concern. Skill develop- versiti Tun Hussein Malaysia, Batu Pahat,
ment is another critical aspect, requiring con- Malaysia.
certed efforts in training and upskilling the [7] Onyegiri I, Nwachukwu CC, and Onyegiri J.
workforce to maximize the potential of ICTs. Information and communication technology
 Investigating efficacy of emerging ICTs for the Indian construction sector 49

in the construction industry. Am J Sci Ind [20] Chien HJ, and Barthorpe S. The current state
Res. 2011. of information and communication technol-
[8] Information and communications technol- ogy usage by small and medium Taiwanese
ogy in construction. Designingbuildings. construction companies. J Inf Technol Con-
co.uk. Last edited March 30, 2022. Accessed str. 2014;15(5):75–85.
September 28, 2024. [21] Mesaros P, and Mandiak T. Impact of ICT on
[9] Paudyal G. Role of ICT in construction. In: performance of construction companies in
National Students Conference on Informa- Slovakia. IOP Conf Ser Mater Sci Eng. 2017.
tion Technology (NaSCoIT). Nepal; 2016. [22] Griffis FH, Hogan DB, and Li W. An analy-
[10] Orihuelaa P, Orihuelab J, and Pachec S. sis of the impact of using three-dimensional
Information and communications technol- computer models in the management of con-
ogy in construction: A proposal for pro- struction. Constr Ind Inst (CII) Austin, TX,
duction control. In: Creative Construction USA. 1995;Research Report 106–11.
Conference 2016, CCC 2016; 2017:25–28. [23] Kang Y, O’Brien WJ, Thomas S, and Chap-
[11] Sharma DK. Digital revolution 4.0: Informa- man RE. Impact of information technologies
tion and communication technology (ICT) on performance: Cross-study comparison. J
in construction industry. JK Lakshmipat Constr Eng Manag. 2008;134:852–863.
University, Jaipur; July 16, 2020. [24] Owolabi OSB, and Olufemi ODK. Effect of
[12] Hosseini MR, Chileshe N, Zuo J, and the use of ICT in the Nigerian construction
Baroudi B. Approaches of implementing ICT industry. Int J Eng Sci. 2018;7(5):71–76.
technologies within the construction indus- [25] Mu’azu DA. The role of the professional
try. Univ South Aust. 2013. builder in the Nigerian construction industry.
[13] Chinowsky PS, and Rojas E. A guide to suc- ATBU J Environ Technol. 2015;1(1):29–31.
cessful implementation: A report on con- [26] Yisa SO. An investigation into the use of
struction industry institute. Univ Texas ICT in the Nigerian construction industry.
Austin, Constr Ind Inst. 2012. Unpublished HND Project, Department of
[14] Brandon PS, Kocaturk T, and Foundation R. Building Technology, The Federal Polytech-
Virtual futures for design, construction and nic, Ado Ekiti. 2014.
procurement. Malden, Ma: Blackwell Pub; [27] Oladapo AA. The impact of ICT on pro-
2018. fessional practice in the Nigerian construc-
[15] Zhang Y, Gregory M, and Shi Y. Global engi- tion industry. Electron J Inf Syst Dev Ctries.
neering networks (Gen): Drivers, evolution, 2016;24(2):1–19.
configuration, performance and key patterns. [28] Leverage on ICT in construction industry.
J Manuf Technol Manag. 2008;19:299-314. New Strait Times. May 13, 2018.
[16] Chen C, and Messner JI. A recommended [29] Al-shammary S, and Ali AA. ICT hindering
practices system for a global virtual engi- factors applied in Jordan construction pro-
neering team. Archit Eng Des Manag. jects. Civ Eng Archit. 2017;5(3):83–88.
2014;6:207–221. [30] Hassan A, and Afrah MHK. Identification
[17] Adwan EJ, and Al-Soufi A. A review of ICT and analysis of hindering factors of ICT
technology in construction. Int J Manag Inf adoption in project management in Iraq. J
Technol. 2016;8(3/4). Univ Babylon Eng Sci. 2019;27(3).
[18] Sarshar M, and Isikdag U. A survey of ICT [31] Shah K, Soni N, and Shah Z. Scrutinizing
use in the Turkish construction industry. Eng attributes influencing role of information
Constr Archit Manag. 2015;11(4):238-247. communication technology in building con-
[19] Mutesi N. Application of ICT in the con- struction. Int Res J Eng Technol. 2020;7(3).
struction industry in Kampala. In: Advances [32] Vasista TG, and Abone A. Benefits, barri-
in Engineering and Technology: Proceedings ers and applications of information com-
of the Second International Conference on munication technology in construction
Advances in Engineering and Technology, industry: A contemporary study. Int J Eng.
Kampala. 2013. 2016;105:933–939.
6 AgriTech: Empowering agriculture
through integrated technology
Pankaj Kunekara, Anushka Popalghatb, Onkar Borudec, Vishal
Gavalid, Dnyanesh Gholape, and Sarvesh Hadolef
Department of Information Technology Vishwakarma Institute of Technology, Pune, India

Abstract: Agriculture plays a pivotal role in ensuring global food security and economic stability. In an era of
rapidly advancing technology, harnessing the power of data and innovation is vital to address the multifaceted
challenges faced by farmers. AgriTech is a comprehensive agricultural technology system designed to empower
farmers at every farming stage. Leveraging a range of technologies, including Full Stack Web Development,
Ethereum, Alan AI and Various Machine learning algorithms, this multifaceted solution offers features such as
nearby soil testing laboratory location, crop recommendations, gamified plant growth mentoring, plant disease
detection, crop price prediction, expert consultations, an integrated e-commerce marketplace, and Realtime
weather monitoring with location-based alerts. This paper provides an overview of AgriTech’s architecture,
emphasizing its technology integration, machine learning applications, and potential impact on modern farm-
ing practices, including enhanced productivity, sustainability, and resource efficiency. AgriTech represents a
pioneering approach to digital agriculture, bridging the gap between traditional farming and the digital era.

Keywords: Agriculture technology, precision farming, crop price prediction, soil testing, disease detection, crop
recommendation

1. Introduction By harnessing the power of machine learn-

ing and real-time data analysis, it furnishes
Traditionally, agriculture relied on time-
farmers with personalized insights, data-
tested practices passed down through genera-
driven recommendations, and a platform to
tions. However, the contemporary farming
connect with agricultural experts, transcend-
landscape demands a departure from conven-
tional methods. AgriTech represents a pivotal ing geographical boundaries and traditional
response to this demand, offering farmers an constraints. Moreover, its gamified approach
integrated and technologically advanced ally to plant growth mentoring and innovative
in their quest for sustainable and productive reward mechanisms promotes knowledge
farming practices. acquisition and sustainable farming practices.
AgriTech seamlessly integrates critical This paper delves into the architecture,
aspects of farming, from soil quality analysis methodologies, and potential impact of
to crop selection, growth management, dis- AgriTech, as it endeavors to bridge the gap
ease detection, price forecasting, expert con- between age-old agricultural practices and
sultations, and real-time weather monitoring. the digital age. In so doing, it unlocks a new
The genesis of AgriTech is rooted in a pro- era characterized by enhanced productivity,
found understanding of the intricate interplay resource efficiency, and sustainability in farm-
between data, technology, and agriculture. ing. AgriTech embodies not just technological

a
[email protected]; [email protected]; [email protected]; dvishal.gavali21@vit.
edu; [email protected]; [email protected]
DOI: 10.1201/9781003606635-6
 AgriTech: Empowering agriculture through integrated technology 51

advancement but also a testament to the syn- field which encompasses a wide range of tech-
ergy between human ingenuity and the relent- nologies, including precision farming, robot-
less pursuit of food security and ecological ics, artificial intelligence, and biotechnology,
harmony in an ever-changing world. which have already begun transforming the
agricultural landscape [8]. Discuss the impact
of Internet of Things on farming, including
2. Literature Review using sensors for soil, water, light, humidity,
The authors provide a comprehensive review and temperature management [9]. The paper
of recent advances in image-processing tech- proposes a Blockchain-based framework for
niques for automating leaf pests and disease data quality management in the context of
recognition. They discuss the various stages COVID-19, aiming to ensure a high level of
involved in the image-processing pipeline, data trust, security, and privacy. It explores
from image acquisition to feature extrac- the use of innovative technologies such as
tion and classification [1]. The study Intro- Blockchain, Artificial Intelligence, and Big
duces a framework for crop price forecasting Data in tracking infectious disease outbreaks
in emerging economies, emphasizing the and improving reporting accuracy [10].
importance of accurate price predictions for
farmers and policymakers. The proposed 3. Methodology
framework includes data quality assessment,
In response to the evolving agricultural land-
feature selection, and modeling, offering
scape, the AgriTech project emerges as a com-
advantages such as improved data quality
prehensive agricultural technology platform.
and feature selection techniques for enhanced
This support ranges from initial crop selec-
accuracy [2]. The system’s two-tier architec-
tion to the monitoring of crop growth and
ture and various sensors offer advantages like
culminates in facilitating marketing and sales.
real-time monitoring and portability, while
The platform leverages an amalgamation of
challenges include data quality and real-time
cutting-edge technologies and data-driven
performance [3]. This research Focuses on
methodologies to provide a holistic and User-
online price forecasting for cryptocurrencies
friendly agricultural support system.
like Bitcoin, Ethereum, and Ripple, utilizing
Frontend technologies and functionalities:
artificial intelligence techniques. The model’s
Frontend stack: The front of the AgriTech
advantages include periodic predictions and
platform is built using React, a renowned
customization options.[4].
JavaScript library for developing user inter-
The authors address crop price predic-
faces. This choice of technology ensures a
tion in developing nations, using decision tree
User-friendly and interactive experience.
regression techniques [5]. Authors Delves into
User interface: The frontend features a
crop yield and price forecasting using machine
user-friendly map-based interface, designed
learning, employing the decision tree algo-
to enhance accessibility and usability for
rithm. This research enhances crop yield pre-
farmers.
diction based on weather and price trends [6].
Features:
The authors introduce a neighborhood-
based strategy to optimize parameters and
obtain maximum yield, expanding knowl- 3.1. Soil testing labs map
edge through population exploration. The Farmers can easily locate nearby soil testing
paper presents a method to increase soil fertil- laboratories, simplifying the crucial process
ity performance by providing nutrient recom- of soil quality assessment. This feature facili-
mendations for optimal crop development [7]. tates data-driven decisions about crop selec-
The study examines the Modern agricultural tion and soil management.
 52 Innovations in Computational Intelligence: Trends and Technologies

3.3. Plant disease detection and

AQ1:
Citations pesticide recommendation
for all fig-
ures are Utilizing image recognition technology, Agr-
missing in iTech can accurately identify plant diseases
the text,
please
from images of leaves. In addition to disease
provide detection, it recommends appropriate pesti-
the cides, aiding in timely disease management
appro-
priate and prevention.
citations Figure 6.1. Soil testing lab locations.
AQ2: Source: AUTHOR'S SELF GENERATED
Please
provide
source
3.2. Personalized crop recommendations
for all the
figures
The platform employs machine learning algo-
rithms to provide personalized crop recom-
mendations. [11,12]

Figure 6.4. Plant disease prediction.

Source: AUTHOR'S SELF GENERATED

3.4. Coin-based e-commerce
marketplace
AgriTech implements a coin-based reward
system during the plant growth mentoring
process. These earned coins can be redeemed
within an integrated e-commerce market-
Figure 6.2. Heatmap for crop recommendation. place, fostering user engagement and partici-
Source:AUTHOR'S SELF GENERATED
pation.[16,17]

By analyzing soil data and real-time

weather conditions, it assists farmers in
choosing the most suitable crops for their
specific geographic locations[13-15]. This pro-
motes optimal resource utilization and crop
yield improvement.

Figure 6.5. Blockchain-based marketplace.

Source: AUTHOR'S SELF GENERATED

3.5. Plant price prediction

The platform includes a sophisticated price
prediction functionality, leveraging cutting-
Figure 6.3. Crop recommendation. edge technologies to provide farmers with
Source: valuable insights into future crop prices.
AUTHOR'S SELF GENERATED
 AgriTech: Empowering agriculture through integrated technology 53

Machine learning algorithms, specifically 3.6. Real-time weather monitoring

decision tree regression techniques, are and alerts
employed to analyze historical data and fore-
cast crop prices [18]. The platform offers real-time weather moni-
These algorithms consider factors such as toring, including temperature, humidity, and
market trends, historical price fluctuations, rainfall data. It provides farmers with essential
and relevant variables to generate accurate weather insights, aiding in informed decision-
price predictions. By harnessing the power making regarding planting and harvesting.
of machine learning and a robust database
system, AgriTech empowers farmers to make 3.7. Gamified plant growth
informed decisions regarding crop selection In the context of plant height estimation from
and pricing strategies. images, one crucial step involves color space
conversion. Typically, the input image is rep-
resented in the BGR (Blue, Green, Red) color
space. Still, for precise color-based analysis,
conversion to the HSV (Hue, Saturation,
Value) color space using OpenCV’s `cv2.cvt-
Color` is a valuable technique. In HSV, the
image is separated into three components:
Hue, capturing color information; Satura-
tion, reflecting color intensity; and Value, sig-
nifying brightness.
Ultimately, the plant’s height is estimated
Figure 6.6. Crop price prediction. by calculating the vertical extent of the larg-
Source: AUTHOR'S SELF GENERATED est contour concerning the image’s overall

Figure 6.7. Real-time weather monitoring.

Source:
AUTHOR'S SELF GENERATED
54 Innovations in Computational Intelligence: Trends and Technologies

height. This is achieved by determining the

ratio of the contour’s vertical extent to the
image’s height and subsequently scaling it
based on a constant base height, typically in
specified units such as centimeters or pixels.

3.8. Algorithm implementation

In the context of agricultural decision-mak-
ing, the project employs a robust model
known as the support vector machine (SVM)
with a radial basis function (RBF) kernel. Figure 6.10. Accuracy chart.
This machine-learning model is harnessed to
Source:
provide crop recommendations based on a
AUTHOR'S SELF GENERATED
comprehensive set of seven crucial features,
including nitrogen levels, phosphorus lev-
els, potassium levels, pH levels, temperature,
humidity, and rainfall.
SVM with the RBF kernel is particularly
valuable for this task as it excels in captur-
ing non-linear relationships between these

Figure 6.11. Crop recommendation flow

diagram.
Source: AUTHOR'S SELF GENERATED

features and crop suitability. By leveraging

this combination, the project offers farmers
Figure 6.8. Gamified plant growth. data-driven insights into selecting the most
Source: AUTHOR'S SELF GENERATED appropriate crops for their specific soil and
environmental conditions, ultimately enhanc-
ing agricultural productivity and resource
optimization.
As seen in this diagram, the first step in
the environmental assessment process is to
gather and analyze crucial data on soil and
environmental conditions. This information
is used to make well-informed choices regard-
ing land use and resource management. Once
the data is collected, a thorough data pre-pro-
cessing is carried out to guarantee the quality
of the data. Sophisticated machine learn-
ing algorithms are utilized in the analysis to
unveil hidden insights and patterns. To cre-
Figure 6.9. Confusion matrix. ate a model with enhanced accuracy for fore-
Source: casting environmental changes and assessing
AUTHOR'S SELF GENERATED
 AgriTech: Empowering agriculture through integrated technology 55

interventions, these insights are put into

practice. At the end of the day, the model’s
outcomes are used to deliver well-informed
suggestions, guiding decisionmakers in the
implementation of sustainable measures for
responsible land and resource management.
In the project, plant disease prediction
plays a crucial role in identifying and miti-
gating diseases affecting various crops. The Figure 6.12. Visualizing plant growth using
project employs the MobileNetV2 model, a image processing.
state-of-the-art deep learning architecture, for Source:
AUTHOR'S SELF GENERATED
this task. With access to a diverse dataset com-
prising 38 different plant diseases, sourced
from the Kaggle dataset repository (acces- As seen in this diagram, visualizing plant
sible at https://www.kaggle.com/vipoooool/ growth through the implementation of image
newplantdiseases-dataset), MobileNetV2 is processing comprises a holistic procedure
trained to recognize and classify these diseases comprising multiple essential components.
accurately[19]. By leveraging deep learning The process commences with the collection
and the extensive dataset, the model can rap- of images that document the various devel-
idly diagnose plant diseases from images of opmental phases of the plants. Once these
leaves, offering a swift and effective solution raw images are obtained, various enhance-
to disease management and crop protection ment techniques are applied to enhance their
in agriculture. overall quality and clarity. After the enhance-
ment, the application of image segmenta-
3.8.1. Image processing tion is used to differentiate the plants from
To estimate the height of a plant from the background and measure their growth
images, several image processing techniques patterns. Data visualization is done using
are applied. Initially, blurring is employed the segmented data to create graphical rep-
to reduce noise in the input image, enhanc- resentations of plant growth metrics, which
ing subsequent processing robustness. The offer valuable insights. Ultimately, the usage
image is then converted from the BGR color of image processing techniques is crucial in
space to the HSV color space to facilitate every step of this process. These techniques
color-based analysis. In the HSV space, a are essential for extracting valuable informa-
green color range is defined to create a mask tion from plant images and enhancing our
isolating the plant’s green parts. Morpho- comprehension of plant development and
logical operations, including opening and health.
closing, further refine the mask, eliminating
noise and connecting object parts. Contour
detection identifies the plant’s boundaries,
4. Data Collection
with the largest contour selected as the plant. The datasets sourced from GitHub reposito-
This largest contour is drawn on the origi- ries provide essential resources for agricul-
nal image to visualize the region of interest. tural research.
Finally, height estimation is calculated by The [Plant Disease Detection Dataset]
assessing the vertical extent of the largest (https://github.com/manthan89-py/PlantDis-
contour relative to the image’s height and easeDesection) features a diverse collection
scaling it based on a predefined base height of labeled plant leaf images afflicted by vari-
value. ous diseases, facilitating the development and
 56 Innovations in Computational Intelligence: Trends and Technologies

evaluation of machine learning models for impact on modern farming practices. The
automated disease detection. results from table 6.1 and table 6.2 can be sum-
The [Crop Price Prediction Dataset] -marized as follows:
(https://github.com/posi2/Crop-Price-Pre- 1. Improved crop recommendations:
diction) offers historical data on crop prices, Through the utilization of the SVM with
market trends, and relevant factors, serv- an RBF kernel, AgriTech successfully pro-
ing as a foundation for crop price forecast- vided personalized crop recommenda-
ing models that benefit both farmers and tions to farmers.
policymakers. 2. Efficient disease detection: The implemen-
Meanwhile, the [Crop Growth Stage tation of the MobileNetV2 deep learning
Dataset] (https://github.com/jahaniam/Crop- model for plant disease detection exhib-
GrowthStage) provides valuable insights into ited remarkable accuracy in identifying
the growth stages of different crops, aiding and classifying various plant diseases
precision agriculture and crop management from images of leaves. This functionality
research. offered rapid disease diagnosis, allowing
These datasets collectively empower agri- farmers to take timely measures for dis-
cultural advancements, from disease identi- ease management and crop protection.
fication to crop price prediction and growth 3. Enhanced user engagement: The coin-
monitoring. based reward system within the integrated
e-commerce marketplace effectively
5. Results and Discussion increased user engagement. Farmers
actively participated in the plant growth
The AgriTech platform has undergone rigor- mentoring process to earn coins, which
ous testing and real-world implementation, could be redeemed for agricultural prod-
AQ3: resulting in several key findings and outcomes ucts and tools.
Cita- that demonstrate its efficacy and potential 4. Real-time weather insights: AgriTech’s
tions for
both the real-time weather monitoring and alerts
tables are Table 6.1. Algorithm and accuracy provide farmers with essential weather
missing in
the text, Algorithm Accuracy Functionality data, including temperature, humid-
please ity, and rainfall. Farmers reported that
provide SVM + RBF 98.72% Crop
the Kernel Recommendation these insights significantly contributed
appro- to informed decision-making regarding
priate MobileNetV2 98.9% Disease Prediction
CITATION
citations
ADDED
planting and harvesting, resulting in more
Random Forest 97.1% Crop Price efficient resource utilization and crop
AQ4: Regression Prediction management.
Please
provide
Source:
AUTHOR'S SELF GENERATED 5. Secure and transparent transactions: The
source integration of the Ethereum blockchain
for all the
Tables Table 6.2. Comparison with previous work into AgriTech’s e-commerce marketplace
ensured secure and transparent transac-
Algorithm Previous Current tions. Users appreciated the added layer
Accuracy Accuracy
of trust and transparency in their mar-
SVM 98.2% [14] 98.72% ketplace interactions, which is especially
CNN 95.05% [20] 98.9% critical in the agricultural sector.
6. Improved communication: Agora SDK’s
Random Forest 98% [21] 98%
video conferencing capabilities facili-
Regression
tated real-time video calls and chat func-
Source: tionality, enhancing communication and
AUTHOR'S SELF GENERATED
 AgriTech: Empowering agriculture through integrated technology 57

support for farmers. This improved com- India’s Livelihoods. 2023:107. Terborgh J.
munication enabled farmers to seek expert Preservation of natural diversity. BioScience.
consultations and collaborate effectively 2009;24:715–22.
with peers. [3] Reyana A, Kautish S, Karthik PS, Al-Baltah
7. Conversational AI: The integration of IA, Jasser MB, and Mohamed AW. Acceler-
ating crop yield: multisensor data fusion and
Alan AI for voice commands enhanced
machine learning for agriculture text classifi-
the platform’s accessibility, allowing users
cation. IEEE Access. 2023;11:20795–20805.
to interact with AgriTech using natural [4] Sundaram P, and Suparna K. Crop recom-
language. This feature proved particu- mendation system using machine learning. J
larly useful for users who may have lim- Eng Sci. 2023;14(09).
ited typing abilities or prefer voice-based [5] Sharma P, Dadheech P, and Senthil ASK. AI-
interactions. enabled crop recommendation system based
on soil and weather patterns. In: Artificial
Intelligence Tools and Technologies for
6. Conclusion Smart Farming and Agriculture Practices.
In conclusion, the AgriTech project represents IGI Global; 2023:184–199.
a pioneering and comprehensive agricultural [6] Baishya DR. International Journal for
technology platform that bridges the gap Research in Applied Science and Engineer-
ing Technology. 2020;8(IV).
between traditional farming practices and
[7] Ahmed U, Lin JCW, Srivastava G, and Dje-
the digital age. AgriTech empowers farm-
nouri Y. A nutrient recommendation system
ers at every stage of the agricultural process, for soil fertilization based on evolution-
from crop selection to growth management, ary computation. Comput Electron Agric.
disease detection, and price forecasting. The 2021;189:106407.
platform’s machine learning capabilities pro- [8] Green S. Reshaping the future of agriculture
vide personalized insights and data-driven through agri-tech innovations. Int Multidis-
recommendations, fostering enhanced pro- cip J Sci Technol Bus. 2023;2(02):1–4.
ductivity, sustainability, and resource effi- [9] Rathod ML, Shivaputra A, Umadevi H,
ciency. The implementation of MobileNetV2 Nagamani K, and Periyasamy S. Cloud com-
for plant disease detection and the integration puting and networking for SmartFarm Agr-
of the Agora SDK for real-time video commu- iTech. J Nanomater. 2022.
nication further demonstrate the platform’s [10] Ezzine I, and Benhlima L. Technology against
COVID-19: a blockchain-based framework
commitment to offering holistic support to
for data quality. In: 2020 6th IEEE Con-
farmers. As digital agriculture continues to
gress on Information Science and Technol-
evolve, AgriTech serves as a testament to the ogy (CiSt). IEEE; 2021:84–89.
potential of technology to revolutionize and [11] Yu J, Yin H, Xia X, Chen T, Li J, and Huang
improve farming practices, making a signifi- Z. Self-supervised learning for recommender
AQ5:
cant contribution to the well-being of farm- systems: a survey. IEEE Trans Knowl Data
Refer- ing communities worldwide. Eng. 2023.
ences are [12] Bandara P, Weerasooriya T, Ruchirawya T,
missing
Nanayakkara W, Dimantha M, and Paba-
cita-
tions in
References sara M. Crop recommendation system. Int J
the text. [1] Ngugi LC, Abelwahab M, and Abo-Zahhad Comput Appl. 2020;975:8887.
Please
ensure
M. Recent advances in image processing [13] Iqbal M, Kumar V, and Sharma VK. Krishi
that all techniques for automated leaf pest and dis- Portal: web-based farmer help assistance. Int
references ease recognition: a review. Inf Process Agric. J Adv Sci Technol. 2007;29:4783–4786.
are cited
2021;8(1):27–51. [14] Bakthavatchalam K, Karthik B, Thiru-
and in
sequential [2] Kanitkar A, and Wadekar I. Digital Agri- vengadam V, Muthal S, Jose D, Kotecha
order. culture: Is this the Future of ‘New’. State of K, and Varadarajan V. IoT framework for
58 Innovations in Computational Intelligence: Trends and Technologies

measurement and precision agriculture: International Conference of Safe Produc-

predicting the crop using machine learning tion and Informatization (IICSPI). IEEE;
algorithms. Technologies. 2022;10(1):13. 2020:462–465.
[15] Kang M, Wang X, Wang H, Hua J, de Reffye [18] Mulla SA, and Quadri SA. Crop-yield and
P, and Wang FY. The development of AgriV- price forecasting using machine learning. Int J
erse: past, present, and future. IEEE Trans Anal Exp Modal Anal. 2020;12:1731–1737.
Syst Man Cybern Syst. 2023. [19] Kosamkar PK, Kulkarni VY, Mantri K,
[16] Fathi M, Haghi Kashani M, Jameii SM, and Rudrawar S, Salmpuria S, and Gadekar N.
Mahdipour E. Big data analytics in weather Leaf disease detection and recommenda-
forecasting: a systematic review. Arch Com- tion of pesticides using convolution neu-
put Methods Eng. 2022;29(2):1247–1275. ral network. In: 2018 Fourth International
[17] Su Z, Wang H, Wang H, and Shi X. A finan- Conference on Computing Communica-
cial data security sharing solution based tion Control and Automation (ICCUBEA).
on blockchain technology and proxy re- IEEE; 2018:1–4.
encryption technology. In: 2020 IEEE 3rd
 7 Online and offline learning in
universities: A student’s dilemma
Ruchi Agarwala and Rahul Pradhanb
Department of Computer Engineering and Applications, GLA University, Mathura, Uttar Pradesh,
India

Abstract: The pandemic, heatwave brought life to a standstill, if not for older but for young ones and students
it brings things to a standstill, presenting a significant obstacle to humanity. However, online learning offers
flexibility, interaction, and cooperation. This article aims to compare and contrast the results of student surveys
and explore the differences between live online instruction and conventional classroom instruction, as well
as the future of education in universities and colleges. The study was conducted at several universities in the
Mathura region of Uttar Pradesh, India, with 150 participants selected from those enrolled in online and offline
university courses, aged between 16 and 40 years. The students were then asked to complete a detailed question-
naire. 58.95% of students mostly agreed with the question in the survey, while 36.69% disagreed. As per the
survey analysis, online learning seems to be more effective. It is concluded that traditional classroom teaching
has advantages for students, saves them time, and improves their academic performance in the current environ-
ment of the COVID-19 epidemic. Higher education institutions must collaborate to make these initiatives effec-
tive, especially by decentralizing educational teleconferences and granting access to surrounding institutions to
ensure that all students have reliable internet connectivity.

Keywords: COVID-19, E-learning, offline learning, online learning, students, survey

1. Introduction technology will revolutionize education,

teaching, and learning Lee (2006); Romeo
E-learning designs, delivers and manages for-
(2006). Integration of educational technolo-
mal and informal learning and knowledge
gies presents learning obstacles and problems.
exchange anywhere, anytime. Some courses
The COVID-19 pandemic has stopped
are delivered fully online, while others are
hybrid. Online learning offers flexible tim- most human activities, posing a huge obsta-
ings, self-paced, more interactivity, and sup- cle to humanity. Although in this era where
port Gedera, Williams, and Wright (2013). computers and the internet reach each indi-
With e-expansion, teachers and students can vidual’s pocket, this tiny microbe is devas-
AQ1: As
construct knowledge and enhance learning tating Akbar, Rashid, et al. (2018). Even the
per style,
the refer- outside the classroom. According to Shafaati school system was closed as everyone was
ence list
et al., 2022 Shafaati, Valizade, Rahimi, and affected. These stalled face-to-face classes dis-
and its
citation Panahi (2022), 4,600,000 college students in turbed student training. Access to technology
should be the US take at least one online class. By 2014, in rural areas causes barriers to learning. The
num-
bered, this figure will rise to 18,650,000 Shafaati et research literature validates students’ online
please al. (2022). Despite learning of the popular- needs and challenges Aboagye, Yawson, and
check and
confirm
ity of e-Commerce Bell and Federman (2013), Appiah (2021); Chase et al. (2018a); Chung,
for positive results are not guaranteed Alexander Subramaniam, and Dass (2020); Lorenzo
clarity. (2001). Some researchers are unsure whether (2017). Some people think differently about

CHECKED AND CONFIRMED

a
[email protected]; [email protected]
DOI: 10.1201/9781003606635-7
 60 Innovations in Computational Intelligence: Trends and Technologies

Internet connections. In-efficient internet

facilities in universities do not affect internet
use Apuke and Iyendo (2018). Another study
highlighted the correlation between inter-
net connectivity and online learning in rural
areas Ahmed, Khan, Faisal, and Khan (2017).
In another survey, the majority of respond-
ents were ready for e-learning Muthuprasad,
Aiswarya, Aditya, and Jha (2021). In another
paper, Hossain Hossain and Rahman (2017)
and colleagues suggested in 2017 that uni- Figure 7.3. Subject distribution according to
versities provide students with internet facili- education qualification.
ties and an environment. Tarimo Tarimo and Source:
Kavishe (2017) and colleagues revealed in
2017 that 82% of children use the Internet
to go to school. A study showed negative atti-
tudes toward online learning management Learning tools or equipment are also
systems Serhan (2020). Taking into account important in this study. Although we live in
these concepts and facts. Internet connec- the age of technology, not everyone has. A
AQ2: tivity is a problem for students, faculty, and study in South Asia showed that respondents
Please institutions. had computers and internet at home and at
provide
in-text
college. In another study, the students owned
citation various smartphones. This is a possibility as
for Figure demand for phones increases and prices are
1 to 3
competitive. Research shows that students
rely on smartphones to learn and surf the
Internet. They found no correlation between
smartphone ownership and nomophobia
AQ3: in college students Gezgin (2017). Another
Please study found that mobile learning devices can
provide
improve student learning efficiency Chase et
source Figure 7.1. Subject distribution according to the
for all the ages. al. (2018b). The biggest problem is the resist-
figures ance of students to blended online learning.
Source:
AUTHOR'S SELF GENERATED In the Philippines, Surveys revealed mixed
sentiments toward internet connectivity and
device usage for online education. Lack of
reliable Internet connection is a barrier to
learning. Fabito, Trillanes, and Sarmiento
(2020). Authors noted that few pupils had
internet access, preventing them from using
the e-learning platform Casillano (2019).
Poor students lack laptops, desktops, and
Internet access, according to another sur-
vey Cleofas and Rocha (2021). They found
Figure 7.2. Subject distribution according to the mobile devices can be used for learning Jin
gender. and Junio-Sabio (2018). Another study found
Source: that owning devices plays a crucial role in
AUTHOR'S SELF GENERATED
 Online and offline learning in universities: A student’s dilemma 61

learning readiness Estira (2020). Another and Solomon (2021). Atika Qazi et al.2020
state university survey found that students are Qazi et al. (2020) 320 students participated
ready for online lessons, but renting comput- in study about learning in different modes
ers and the Internet in cafes is a burden Yra, during the COVID-19 pandemic. The study
Castillo, Bautista, Camayang, and Camayang concluded that online education is the only
(2020). This study assesses students’ internet solution and effective way Qazi et al. (2020).
connection and learning devices for online or The vast majority (70.4%) of the survey
flexible learning. The study will show school respondents said they would advise use of
administrators how students and staff will e-learning material. Online learning scores
modify teaching-learning procedures. The 4/5 for effectiveness, while offline learning
result will be used to institutionalize flexible scores 3/5 Yang et al. (2022).
learning in the future.
3. Material and Method
2. Literature Review
Favale et al. 2020 Favale, Soro, Trevisan, 3.1. Study design
Drago, and Mellia (2020), The infrastructure The online survey is used to collect data and
for the internet is under pressure as a result respondents’ responses. A questionnaire was
of this sudden spike in usage. E-learning designed to capture respondents’ opinions.
tools like Microsoft Teams and Zoom were The purpose of this study is to see the mood of
quickly adopted. Students are more aware of students at the university level about learning
online learning. E-learning, however, may be techniques adopted by their higher education
extremely challenging for teachers Favale et institutions during the pandemic covid-19 in
al. (2020) who not habitual of making pow- Braj area of Uttar Pradesh which constitutes
erpoint ppts. Agra, Mathura, and some parts of Aligarh.
Padmalini Singh et al. 2021 Singh et al.
(2021), conducted a study at India, Indo- Table 7.1. Participants distribution on the basis
nesia, and Malaysia on learning with 100 of Age
respondents. they compare online and offline
Age Frequency Percentage AQ4:
learning. their study shows traditional teach- Please
ing is more effective than online teaching 16–20 40 26 provide
21–25 46 31 in-text
Singh et al. (2021). Elyassi et al. 2021 Elyassi citation
et al. (2021), various changes in the psychic 26–30 25 16.6 for Tables
of students due to lockdown and pandemics 31–35 19 13
1 to 3.

Elyassi et al. (2021). >35 20 13.3 AQ5:

Biswas 2021, study India for learning. they Please
Total 150 100 provide
concluded that the face-to-face offline class- source
room is better than online resources can be Source: for all the
AUTHOR'S SELF GENERATED
used as a supplement. After life returns to Tables

normal in the post-COVID-19 era, we need Table 7.2. Participants distribution on the basis
to return to traditional classroom teach- of Gender
ing Dalai, Dey, Chatterjee, Chakravorti, and Gender Frequency Percentage
Bhattacharya (2014). Sheba et al. 2021 sur- Female 85 56.6
veyed 450 students for effectiveness of online
Male 65 43.3
study material. students found the mate-
rial was effective when used with anima- Total 150 100
tions, PPT, and visuals Darius, Gundabattini, Source:
AUTHOR'S SELF GENERATED
62 Innovations in Computational Intelligence: Trends and Technologies

3.2. Investigation Participants According to the data above, as much as

36.6% of respondents disagreed compared to
For this study, we have selected 150 partici-
63.3% respondents chose to agree who chose
pants according to age between 16 and 40
to agree that “During the COVID-19 pandemic,
years, sex, and educational qualification reg-
students prefer online learning”. “Online Learn-
istered at universities for online and offline
ing saves time while studying” has a tie where
courses.
50% says they agree to the statement while
50% disagree to the statement. For the third
3.3. Investigation content question, 56.6% agree to be satisfied with the
Feedback on the learning material provided, leaning outcomes in online mode. Figure 7.4
instruction techniques, usage of ICT, and indicates the Student feedback graph.
mode of teaching was collected through
questionnaires. 4. Discussion
Table 7.3. Participant’s distribution based on Covid-19 impacted education and popular-
their educational background ized online education. This study aimed to find
Education
Frequency Percentage
Qualification
Bachelor’s Degree 45 30
Master 65 43.3
PhD 19 12.6
Others 21 14 Figure 7.4. Feedback– Distribution of
Total 150 100 Agreement and disagreement on above issues.
Source: AUTHOR'S SELF GENERATED Source:
AUTHOR'S SELF GENERATED

Table 7.4. Responses in favor of online mode of learning

S.No. Feedback Agree Disagree Agree (%) Disagree (%)
1 Online learning is preferred during pandemic 95 55 63 36.66
CoVID-19
2 Online Learning saves time while studying. 75 75 50 50
3 Satisfaction with the learning effect 85 65 56.6 43.33
4 Online mode of learning is beneficial for 100 50 66.66 33.33
students
5 Students prefer online lectures to physical 105 45 70 30
lectures.
6 Online class materials were useful and accurate 96 54 64 36
7 Exams were based on material covered in the 87 63 58 42
assignment and lectures
8 Communicated clearly 90 60 60 40
9 The teacher showed an interest in helping 87 63 58 42
students learn
10 The method of evaluating my work were fair 65 85 43.33 56.6
Total 58.95 36.69
Source:
AUTHOR'S SELF GENERATED
 Online and offline learning in universities: A student’s dilemma 63

the effectiveness of new techniques adopted internet struggling province in the philip-
by various higher educational institutes and pines. Indian Journal of Science and Tech-
government guidelines during COVID-19 for nology, 12 (10), 1–4.
continued learning. The study was conducted [8] Chase, T. J., Julius, A., Chandan, J. S., Powell,
in the Indian state of Uttar Pradesh in the Brij E., Hall, C. S., Phillips, B. L., and Fernando,
B. (2018a). Mobile learning in medicine: an
region. A total of 150 participants partici-
evaluation of attitudes and behaviours of
pated and responded to questions. We con-
medical students. BMC medical education,
clude that in our study online education and 18 (1), 1–8.
techniques are effective and most students are [9] Chase, T. J., Julius, A., Chandan, J. S., Powell,
satisfied with it. E., Hall, C. S., Phillips, B. L., . . . Fernando, B.
(2018b). Mobile learning in medicine: an eval-
uation of attitudes and behaviours of medical
5. Conclusion students. BMC medical education, 18 (1), 1–8.
We can conclude based on the responses col- [10] Chung, E., Subramaniam, G., and Dass, L.
lected, upon analyzing the responses, online C. (2020). Online learning readiness among
learning could be adopted. Online learning university students in malaysia amidst covid-
19. Asian Journal of University Education,
could be more effective if devices and inter-
16 (2), 45–58.
net costs could be subsidized and reach rural
[11] Cleofas, J. V., and Rocha, I. C. N. (2021).
areas. Demographic, gadget and internet profiles
as determinants of disease and consequence
related covid-19 anxiety among filipino col-
References lege students. Education and Information
[1] Aboagye, E., Yawson, J. A., and Appiah, K. Technologies, 26 (6), 6771–6786.
N. (2021). Covid-19 and e-learning: The [12] Dalai, S., Dey, D., Chatterjee, B., Chakra-
challenges of students in tertiary institutions. vorti, S., and Bhattacharya, K. (2014). Cross-
Social Education Research, 1–8. spectrum analysis-based scheme for multiple
[2] Ahmed, S. S., Khan, E., Faisal, M., and Khan, power quality disturbance sensing device.
S. (2017). The potential and challenges of IEEE Sensors Journal, 15 (7), 3989–3997.
moocs in pakistan: a perspective of students [13] Darius, P. S. H., Gundabattini, E., and Solo-
and faculty. Asian Association of Open Uni- mon, D. G. (2021). A survey on the effective-
versities Journal, 12 (1), 94–105. ness of online teaching–learning methods for
[3] Akbar, M. A., Rashid, M. M., et al. (2018). university and college students. Journal of
Technology based learning system in inter- The Institution of Engineers (India): Series
net of things (iot) education. In 2018 7th B, 102 (6), 1325–1334.
international conference on computer and [14] Elyassi, H., et al. (2021). Economics of the
communication engineering (iccce) (pp. financial crisis: any lessons for the pandemic
192–197). downturn and beyond? Contemporary Eco-
[4] Alexander, S. (2001). E-learning develop- nomics, 15 (1), 100–121.
ments and experiences. Education+ Train- [15] Estira, K. L. A. (2020). Online distance
ing, 43 (4/5), 240–248. learning readiness of business administra-
[5] Apuke, O. D., and Iyendo, T. O. (2018). Uni- tion students in one state university in the
versity students’ usage of the internet resources philippines. Journal of Critical Reviews, 7
for research and learning: forms of access and (12), 826–832.
perceptions of utility. Heliyon, 4 (12). [16] Fabito, B. S., Trillanes, A. O., and Sarm-
[6] Bell, B. S., and Federman, J. E. (2013). iento, J. R. (2020). Barriers and challenges
E-learning in postsecondary education. The of computing students in an online learn-
future of children, 165–185. ing environment: Insights from one private
[7] Casillano, N. F. B. (2019). Challenges of university in the philippines. arXiv preprint
implementing an e-learning platform in an arXiv:2012.02121.
64 Innovations in Computational Intelligence: Trends and Technologies

[17] Favale, T., Soro, F., Trevisan, M., Drago, I., Conventional to online education during
and Mellia, M. (2020). Campus traffic and covid-19 pandemic: Do develop and under-
e-learning during covid-19 pandemic. Com- developed nations cope alike. Children and
puter networks, 176, 107290. Youth Services Review, 119, 105582.
[18] Gedera, D., Williams, P. J., and Wright, N. [26] Romeo, G. (2006). Engage, empower, enable:
(2013). An analysis of moodle in facilitating Developing a shared vision for technology in
asynchronous activities in a fully online uni- education. Engaged learning with emerging
versity course. International Journal of Sci- technologies, 149–175.
ence and Applied Information Technology [27] Serhan, D. (2020). Transitioning from face-
(IJSAIT), 2 (2), 6–10. to-face to remote learning: Students’ atti-
[19] Gezgin, D. M. (2017). Exploring the influ- tudes and perceptions of using zoom during
ence of the patterns of mobile internet use covid-19 pandemic. International Journal of
on university students’nomophobia levels. Technology in Education and Science, 4 (4),
European Journal of Education Studies. 335–342.
[20] Hossain, M. A., and Rahman, M. H. (2017). [28] Shafaati, A., Valizade, R., Rahimi, A., and
Comparative study of internet usage among Panahi, A. (2022). The role of smart urban
university students: A study of the university develop- ment in the development of the
of dhaka, bangladesh. European Scientific dilapidated areas of tabriz city: A case study
Journal, 13 (34), 134–150. of 8th district of tabriz. Geography and
[21] Jin, W., and Junio-Sabio, C. (2018). Poten- Environmental Planning, 33 (1), 113–132.
tial use of mobile devices in selected public [29] Singh, P., Sinha, R., Koay, W. L., Teoh, K.
senior high schools in the city of manila phil- B., Nayak, P., Lim, C. H., . . . Aryani, D. N.
ippines. International Journal of Learning, (2021). A comparative study on effective-
Teaching and Educational Research, 17 (4), ness of online and offline learning in higher
102–114. education. International Journal of Tourism
[22] Lee, K. T. (2006). Creating ict-enriched and Hospitality in Asia Pasific (IJTHAP), 4
learner-centred environments: Myths, gaps (3), 102– 114.
and challenges. Engaged learning with [30] Tarimo, R., and Kavishe, G. (2017). Inter-
emerging technologies, 203–223. net access and usage by secondary school
[23] Lorenzo, A. R. (2017). Comparative study students in morogoro municipality, tanza-
on the performance of bachelor of secondary nia. International Journal of Education and
education (bse) students in educational tech- Development using ICT, 13 (2).
nology using blended learning strategy and [31] Yang, S., Tan, C., Madsen, D. Ø., Xiang, H.,
traditional face-to-face instruction. Turkish Li, Y., Khan, I., and Choi, B. J. (2022). Com-
Online Journal of Educational Technology- parative analysis of routing schemes based
TOJET, 16 (3), 36–46. on machine learning. Mobile Information
[24] Muthuprasad, T., Aiswarya, S., Aditya, K. S., Systems, 2022.
and Jha, G. K. (2021). Students’ perception [32] Yra, J. F. P., Castillo, R., Bautista, R. G.,
and preference for online education in india Camayang, J. G., and Camayang, A. G. G.
during covid-19 pandemic. Social sciences (2020). Students’ online learning readiness
and humanities open, 3 (1), 100101. and internet connectivity: Bases for the cus-
[25] Qazi, A., Naseer, K., Qazi, J., AlSalman, H., tomization of qsu e-aral. American Journal
Naseem, U., Yang, S., . . . Gumaei, A. (2020). of Educational Research, 8 (11), 878–884.
8 Strengthening UAV network security:
Advanced authentication-based
intrusion detection systems
Altaf C1,a, Shanila Mahreen2,b, C Atheeq3,c, and Shaista Khanam2,d
1
Department of ECE, Lords Institute of Engineering and Technical, Hyderabad, India
2
Department of ECE, Nawab Shah Alam College of Engineering and Technical, Hyderabad, India
3
Department of CSE, GITAM University, Hyderabad, India

Abstract: This article delves into contemporary Intrusion Detection Systems (IDS) designed for Unmanned
Aerial Vehicle (UAV) networks, emphasizing authentication-based approaches. With UAVs playing pivotal roles
across sectors, securing their communications is paramount. Traditional IDS often falls short in UAV scenarios,
necessitating innovative strategies. The proposed IDS employs advanced authentication mechanisms, including
cryptography, biometrics, and dynamic key generation. Cryptographic techniques ensure secure key exchange,
while biometric authentication ties UAV activities to verified identities. Dynamic key generation, bolstered by
chaotic maps, enhances security, making it computationally challenging for attackers. Anomaly detection and
machine learning contribute to the IDS’s adaptability, distinguishing normal and malicious UAV behavior.
Simulated evaluations using tools like ns3 quantify system performance, with metrics like Packet Delivery
Ratio, Delay, and Overhead providing insights. This authentication-based IDS presents a robust solution for
safeguarding UAV network integrity, addressing unique security challenges in diverse operational scenarios.

Keywords: attack, UAV, authentication, chaotic maps, network, IDS

1. Introduction offering real-time data acquisition from van-

tage points that might be inaccessible or haz-
Unmanned Aerial Vehicle (UAV) networks ardous for human operators. This capability
have emerged as transformative technologies has proven invaluable in enhancing security,
with multifaceted applications across vari- border control, and monitoring large-scale
ous industries. These aerial platforms, com- events. In agriculture, UAVs equipped with
monly known as drones, are equipped with specialized sensors contribute to precision
diverse sensors and communication systems, farming by assessing crop health, optimizing
enabling them to perform tasks that were irrigation, and facilitating crop management
once deemed challenging or impractical for [2]. This not only enhances productivity but
traditional methods [1]. The significance of also minimizes resource utilization, aligning
UAV networks lies in their ability to offer with sustainable agricultural practices.
cost-effective, efficient, and flexible solutions UAV networks play a pivotal role in dis-
in domains such as surveillance, agriculture, aster management, where their swift deploy-
disaster management, and infrastructure ment can provide rapid aerial assessments
inspection. of affected areas, aiding in search and res-
In the realm of surveillance, UAVs pro- cue missions, damage evaluation, and plan-
vide a dynamic and adaptable approach, ning efficient relief efforts. Furthermore, in

[email protected]; [email protected]; [email protected]; [email protected]

a

DOI: 10.1201/9781003606635-8
66 Innovations in Computational Intelligence: Trends and Technologies

infrastructure inspection and maintenance, sensitive information but also in ensuring the
UAVs offer a safer alternative to manual uninterrupted and reliable operation of UAVs
inspection, particularly in complex and ele- across various applications. UAVs heavily rely
vated structures [3]. They can capture high- on seamless communication for tasks such as
resolution images and data, facilitating the navigation, data acquisition, and command
early detection of potential issues and reduc- execution. Any compromise in communica-
ing the risks associated with traditional tion integrity can lead to severe consequences,
inspection methods. The versatility of UAV including loss of control, unauthorized access
networks is underscored by their adaptabil- to mission-critical data, and potential threats
ity to diverse environments and tasks. Their to public safety.
compact size, agility, and ability to operate One primary consideration in securing
in challenging terrains make them indispen- UAV communication is the protection of
sable tools in scenarios where conventional mission data. In applications like surveil-
methods may fall short. However, this tech- lance, infrastructure inspection, or disaster
nological revolution is not without its chal- response, the data collected by UAVs often
lenges, particularly concerning the security contains sensitive and confidential informa-
and integrity of the communication systems tion [6]. Unauthorized interception or tam-
within UAV networks. pering of this data not only compromises
As UAVs rely heavily on wireless com- the mission’s objectives but can also lead to
munication for control, navigation, and data privacy breaches and misuse of information.
transfer, ensuring the confidentiality, integrity, Therefore, robust security measures are essen-
and authenticity of the transmitted informa- tial to safeguard the confidentiality and integ-
tion is paramount. Unauthorized access, data rity of the data exchanged between UAVs and
tampering, and interception pose significant ground stations.
threats to the secure operation of UAV net- Moreover, securing communication is
works [4]. Therefore, implementing robust integral to maintaining control over UAV
security measures, including Intrusion Detec- operations. In scenarios where UAVs operate
tion Systems (IDS), becomes imperative to autonomously or semi-autonomously, secure
safeguard these networks against evolving communication ensures that commands from
cyber threats. ground control stations reach the UAV accu-
In the face of these challenges, the inte- rately [7]. Any interference or manipulation
gration of authentication-based IDS emerges in these commands could result in the UAV
as a promising solution. By implementing deviating from its intended path or executing
advanced authentication mechanisms, such as unintended actions, posing risks to both the
dynamic key generation with chaotic maps, UAV and its surroundings. Security breaches
UAV networks can enhance their resistance in communication can also open avenues for
against malicious attacks. This article delves malicious actors to take unauthorized control
into the intricacies of these authentication- of UAVs, leading to potential safety hazards
based approaches, exploring their potential and misuse of the technology [8].
to fortify the security posture of UAV net- Beyond the immediate operational con-
works and ensure the continued advancement cerns, the importance of secure communica-
of this transformative technology [5]. tion extends to the public’s perception and
Securing communication in Unmanned regulatory compliance. As UAVs become
Aerial Vehicle (UAV) operations is of para- more integrated into civilian airspace and
mount importance, given the critical nature daily life, ensuring their secure operation is
of the tasks these aerial platforms undertake. crucial to building public trust and meeting
The significance lies not only in protecting regulatory standards. The potential for UAVs
 Strengthening UAV network security: Advanced authentication-based intrusion detection systems 67

to be used for nefarious purposes, such as bandwidth management become paramount

spying or carrying out malicious attacks, fur- in ensuring seamless and secure UAV oper-
ther underscores the need for robust commu- ations [12]. The constraints on onboard
nication security measures. resources, including processing power and
energy, further complicate the implementa-
2. Literature Review tion of sophisticated security measures.
Securing UAV networks, therefore, is a
2.1. Discussion on the vulnerabilities multifaceted endeavor. It requires not only
addressing specific technical vulnerabilities
and challenges faced by UAV
but also considering the operational con-
networks text in which UAVs operate. The develop-
Unmanned Aerial Vehicle (UAV) networks, ment of adaptive communication protocols,
representing a technological marvel, bring encryption techniques resilient to signal
forth a spectrum of vulnerabilities and chal- interference, and real-time threat detection
lenges intrinsic to their operational character- mechanisms becomes imperative. Addition-
istics. The wireless communication paradigm, ally, collaborative efforts between industry,
a cornerstone of UAV networks, introduces regulators, and cybersecurity experts are
susceptibility to diverse cyber threats [9]. essential to establish standardized security
The reliance on radio frequency signals ren- practices that can be universally applied to
ders communication channels vulnerable to UAV networks [13,14].
interception, eavesdropping, and jamming
attempts. Given the varied applications of 2.2. Overview of traditional IDS
UAVs, spanning from surveillance to delivery and their limitations in UAV
services, the potential impact of a compro-
mised communication link is substantial. scenarios
Furthermore, the deployment of UAVs in Traditional Intrusion Detection Systems (IDS),
dynamic and diverse environments adds lay- stalwarts in safeguarding network integrity,
ers of complexity to their security landscape. encounter distinct challenges when applied
Urban areas, characterized by high signal to the dynamic and resource-constrained
interference and potential hacking threats, domain of UAV networks. While traditional
pose different challenges than remote regions IDS have proven effective in terrestrial set-
where the communication infrastructure may tings, their seamless integration into UAV
be limited. This diversity demands a nuanced operations requires a nuanced understanding
understanding of the threat landscape to of the unique characteristics defining aerial
devise adaptive security mechanisms. Cyber- missions.
physical threats, such as GPS spoofing or A fundamental limitation arises from the
manipulation of sensor data, underscore the reliance of traditional IDS on predefined pat-
need for robust security protocols to main- terns and signatures for identifying malicious
tain the integrity of UAV operations [10,11]. activities. In the fluid and dynamic airspace
Operational challenges intertwine with where UAVs operate, the ability to adapt to
technical vulnerabilities, shaping the unique emerging threats not captured by known pat-
security concerns of UAV networks. The terns becomes critical. Traditional IDS may
dynamic nature of UAV missions, often struggle to keep pace with rapidly evolving
involving rapid changes in altitude, speed, threats that do not conform to pre-established
and environmental conditions, demands signatures, necessitating the development of
communication links that can swiftly adapt. anomaly detection approaches tailored to
Issues such as signal reliability, latency, and UAV environments [15].
68 Innovations in Computational Intelligence: Trends and Technologies

The resource constraints inherent in UAVs Securing Data Transmission: Authentica-

pose another significant challenge. Traditional tion plays a pivotal role in securing the trans-
IDS often assume a stable network environ- mission of sensitive data between UAVs and
ment with ample computational resources. ground stations. By confirming the identity of
In contrast, UAVs operate with limited communicating entities, authentication miti-
onboard processing capabilities and energy gates the risk of data interception or manipu-
reservoirs. Implementing resource-intensive lation during transmission.
IDS solutions may compromise the overall Establishing Trust in UAV Operations: Trust
performance and endurance of UAVs [16]. is fundamental in UAV operations, especially
Achieving an optimal balance between the in applications like surveillance, reconnais-
computational demands of intrusion detec- sance, and critical infrastructure monitoring.
tion and the resource constraints of UAVs Authentication mechanisms build trust by ver-
becomes a paramount consideration. ifying the legitimacy of UAVs within the net-
Moreover, the decentralized and autono- work, fostering confidence in the reliability of
mous nature of UAV operations further com- data and commands exchanged.
plicates the application of traditional IDS. In Preventing Spoofing and Impersonation:
scenarios where continuous connectivity is UAVs are susceptible to various cyber threats,
not guaranteed, IDS solutions need to adapt to including spoofing and impersonation. Effec-
intermittent communication links and oper- tive authentication mechanisms act as a deter-
ate in a standalone fashion. This demands a rent against these threats, ensuring that UAVs
paradigm shift in IDS design, moving away only respond to legitimate commands and
from centralized models to distributed archi- communications from authenticated sources.
tectures that align with the decentralized
nature of UAV networks. 3.1.2. Challenges in UAV
authentication
3. Methodology While authentication is paramount, imple-
menting effective authentication in UAV
3.1. Authentication mechanisms in networks presents unique challenges due to
UAV networks the operational characteristics of these aerial
systems.
3.1.1. Importance of authentication
Limited Computational Resources: UAVs
in UAV networks often operate with constrained computa-
Authentication is a cornerstone of ensuring the tional resources, necessitating lightweight yet
security and integrity of communications within secure authentication protocols. Traditional
UAV networks. In the context of Unmanned methods designed for resource-rich environ-
Aerial Vehicles (UAVs), where these autono- ments may prove impractical, requiring the
mous systems often operate in sensitive and development of specialized solutions tailored
mission-critical environments, the significance to UAV constraints.
of robust authentication cannot be overstated. Dynamic and Ad-Hoc Network Topol-
Protection Against Unauthorized Access: ogies: UAV networks frequently exhibit
Authentication safeguards UAV networks dynamic and ad-hoc topologies, making it
from unauthorized access attempts. Ensuring challenging to establish and maintain con-
that only authenticated and authorized enti- tinuous authentication. Traditional methods
ties can participate in communication pro- relying on stable network structures encoun-
cesses is essential for preventing malicious ter difficulties in adapting to the dynamic
actors from compromising the system. nature of UAV communications.
 Strengthening UAV network security: Advanced authentication-based intrusion detection systems 69

Vulnerability to Jamming and Interference: communicating parties to enable them to

UAV communication links are susceptible encrypt and decrypt messages. In UAV net-
to jamming and electromagnetic interfer- works, where data integrity is paramount,
ence. Authentication methods must contend robust key exchange protocols are essential.
with these challenges to ensure the reliable Biometric Authentication: Explore the
exchange of authentication signals between integration of biometric authentication to
UAVs and ground stations. establish a strong connection between UAV
Harsh Environmental Conditions: UAVs activities and verified identities.
operate in diverse environmental conditions, Step 1: Fingerprint = {P1, P2,..., Pn}: Col-
including extreme temperatures and varying lect a set of n fingerprint data points.
weather patterns. Authentication hardware Step 2: Template = f(Fingerprint): Process
and protocols must be resilient to these con- the fingerprint data to create a biometric
ditions to guarantee continuous and reliable template.
performance. Step 3: Database = {Template1, Template2,
In overcoming these challenges, the ..., Templatem}: Store the template securely in
development of authentication mechanisms a database.
tailored for UAV networks is imperative. Step 4: CapturedData=f(OperatorFingerp
Solutions should strike a balance between rint): Capture the operator’s biometric data
robust security measures and adaptability during UAV operation.
to the unique operational characteristics of Step 5: Match = g(CapturedData,Database):
UAVs. Match the captured data with stored
templates.
Step 6: If match successful, authorize
3.2. Advanced authentication
access: Grant access to the UAV operator.
techniques Dynamic Key Generation with Chaotic
Cryptography and Key Exchange: Cryptogra- Maps: Explain the use of dynamic key gen-
phy and key exchange form the cornerstone eration, enhanced by chaotic maps, to bolster
of secure communication systems, especially the security of UAV communications.
in the realm of Unmanned Aerial Vehicle Step 1: InitialConditions = {IC1, IC2, ...,
(UAV) networks where sensitive informa- ICk}: Set initial conditions for the chaotic map.
tion is transmitted. The primary objective Step 2: xn+1 = f(xn, InitialConditions): Iter-
of this process is to establish a secure chan- ate the chaotic map to generate a sequence.
nel for the exchange of cryptographic keys, Step 3: Kn = g(xn): Generate a crypto-
ensuring confidentiality and integrity during graphic key from the chaotic map.
communication. Cryptography involves the Step 4: Kn+1 = h(Kn): Update the crypto-
use of mathematical algorithms and cryp- graphic key for the next iteration.
tographic techniques to encode information Step 5: KeyDistribution = {K1, K2, ..., Km}:
in such a way that only authorized entities Distribute the generated keys securely.
can decipher it. In UAV networks, crypto- Step 6: EncryptedData = E(OriginalData,
graphic mechanisms play a pivotal role in Kn): Use the key for encryption/decryption.
securing data transmitted between UAVs and Step 7: Synchronization = i(Map1, Map2):
ground stations. The emphasis is on prevent- Synchronize chaotic maps for consistency.
ing unauthorized access, data tampering, and Step 8: SecurityAnalysis = j(Data,
eavesdropping. ­KeyDistribution): Perform security analysis on
Key exchange is a critical aspect of data and key distribution.
secure communication. It involves the secure The Biometric Authentication Algorithm
sharing of cryptographic keys between in UAV networks operates by first acquiring
70 Innovations in Computational Intelligence: Trends and Technologies

and processing unique biometric data, specifi- Biometric authentication, a key element in
cally fingerprints, to create templates stored UAV network security, establishes a robust
in a secure database. During UAV opera- connection between UAV activities and veri-
tion initiation, real-time fingerprint data is fied identities. The algorithm initiates by
captured, matched against stored templates, collecting a set of n fingerprint data points
and upon successful authentication, access (Fingerprint = {P1, P2, ..., Pn}). These data are
authorization is granted to the identified then processed to create a biometric template
operator, ensuring only authorized individu- (Template = f(Fingerprint)), securely stored
als control the UAV. This robust authentica- in a database (Database = {Template1, Tem-
tion mechanism links UAV activities with plate2, ..., Templatem}). During UAV operation,
verified operator identities, enhancing overall the operator’s biometric data (Captured-
network security. Data = f(OperatorFingerprint)) is captured
On the other hand, the Dynamic Key and matched against stored templates. If
Generation with Chaotic Maps Algorithm the match is successful, access is authorized,
focuses on fortifying UAV communications ensuring that only authorized operators con-
through the dynamic generation of crypto- trol the UAV. This process enhances overall
graphic keys. Utilizing chaotic maps with network security by linking UAV activities to
unpredictable iterations based on initial verified operator identities.
conditions, the algorithm generates crypto- The Dynamic Key Generation with Cha-
graphic keys that are dynamically updated. otic Maps Algorithm fortifies UAV commu-
These keys are securely distributed across nications through the dynamic generation
the UAV network, serving as the foundation of cryptographic keys. The algorithm begins
for encryption and decryption processes. by setting initial conditions for a chaotic
Chaotic map synchronization ensures con- map (InitialConditions = {IC1, IC2, ..., ICk}).
sistency among communicating entities, and Through iterations, a sequence is generated
a comprehensive security analysis evaluates (xn+1 = f(xn, InitialConditions)), and crypto-
the algorithm’s effectiveness in safeguard- graphic keys (Kn = g(xn)) are derived. The keys
ing UAV communications. This dynamic key are dynamically updated for subsequent iter-
generation approach adds an extra layer of ations (Kn+1 = h(Kn)) and securely distributed
security, mitigating the risk of unauthorized across the UAV network (KeyDistribution =
access in UAV networks. {K1, K2, ..., Km}). Encrypted data (Encrypted-
Cryptography and key exchange play Data = E(OriginalData, Kn)) utilizes the key
a pivotal role in securing communications for encryption/decryption. Chaotic map
within Unmanned Aerial Vehicle (UAV) net- synchronization (Synchronization = i(Map1,
works, where sensitive information transmis- Map2)) ensures consistency, and security
sion demands confidentiality and integrity. In analysis (SecurityAnalysis = j(Data, KeyDis-
UAV networks, cryptographic mechanisms tribution)) evaluates the algorithm’s efficacy
utilize mathematical algorithms to encode in safeguarding UAV communications. This
information, thwarting unauthorized access, dynamic key generation approach enhances
data tampering, and eavesdropping. Key security, mitigating the risk of unauthorized
exchange, a critical component of secure com- access in UAV networks.
munication, involves sharing cryptographic
keys between parties to facilitate encryption
and decryption. This process ensures data 4. Results and Discussions
integrity, a crucial aspect in UAV networks, Network Simulator 3 (ns-3) played a piv-
where robust key exchange protocols are otal role as a powerful tool in simulat-
imperative. ing and evaluating the proposed Chaotic
 Strengthening UAV network security: Advanced authentication-based intrusion detection systems 71

Map-Based Authentication Algorithm 4.1. Packet delivery ratio

within a virtual Unmanned Aerial Vehicle
Packet Delivery Ratio (PDR) stands as a criti-
(UAV) network environment. Leveraging
cal performance metric that signifies the pro-
ns-3’s capabilities, we constructed a robust
portion of successfully transmitted packets in
framework for modeling complex network
relation to the total sent. This metric serves as
scenarios, allowing for a thorough assess-
a barometer for network reliability, where a
ment of the algorithm’s performance under
heightened PDR signifies efficient data deliv-
diverse conditions. By utilizing ns-3, we
ery. Conversely, a diminished ratio suggests
efficiently generated simulation results that
potential challenges such as packet loss or
facilitated a comprehensive examination of
disruptions in transmission. PDR thus pro-
key metrics, including packet delivery ratio,
vides a crucial insight into the overall robust-
delay, and overhead. This integration of
ness and effectiveness of the network’s data
ns-3 significantly enhanced the credibility
transmission capabilities.
of our research, providing valuable insights
The graphical representation in Figure 8.1
into the algorithm’s efficacy in ensuring
clearly illustrates the superior PDR achieved
secure communication among UAVs. More-
by the proposed CHAOS authentication sys-
over, the simulation allowed us to assess
tem, emphasizing its effectiveness in improv-
the algorithm’s resilience against poten-
ing data delivery rates. This highlights the
tial cybersecurity threats, particularly in
strength of chaotic maps in authentication,
the context of Intrusion Detection Systems
presenting a more dependable and efficient
(IDS) designed for UAV networks. Evalua-
approach when compared to existing meth-
tion parameters used in the proposed model
ods. The distinct difference in the PDR graph
is listed in Table 8.1.
accentuates the competitive advantage offered
by chaotic authentication in the domain of
secure communication within UAV networks.
Table 8.1. Simulation Parameters
Parameter Description 4.2. Delay
Mobility Models Random Waypoint Network delay, which includes propagation,
Number of UAVs 30 transmission, and processing delays, gauges
Communication 300 meters the time required for data packets to trav-
Range erse the network. The reduction of delay is
Altitude and Height 150 meters
Communication Directional Antennas
Models
Propagation Models Two-Ray Ground
Traffic Models CBR with 5 Mbps
Routing Protocols AODV
Propagation Delay 10 milliseconds
Link Quality SNR of 20 dB
AQ1: Data Packet Size 512 bytes
Please
provide MAC Protocols CSMA/CA
source Energy Models Energy Harvesting Figure 8.1. Comparison of proposed method
for all the
figures Model with existing with respect to PDR.
and table Source: Source:
AUTHOR'S SELF GENERATED AUTHOR'S SELF GENERATED
72 Innovations in Computational Intelligence: Trends and Technologies

paramount in guaranteeing swift and respon-

sive communication within the network.
Minimizing these delays enhances the overall
efficiency and responsiveness of the commu-
nication infrastructure.
The delay graph underscores the remark-
able efficiency of the proposed CHAOS
authentication system in minimizing com-
munication delays, as depicted in Figure 8.2.
This emphasizes the robustness of chaotic
maps in the authentication process, leading
Figure 8.3. Comparison of proposed method
to reduced delays and ensuring timely data with existing with respect to overhead.
transmission in UAV networks. The compari-
Source:
son with traditional methods highlights the AUTHOR'S SELF GENERATED
superior performance of CHAOS in optimiz-
ing network responsiveness. HT, and DKRP. This showcases CHAOS as
a promising solution for optimizing network
resources and enhancing overall system per-
4.3. Overhead
formance in UAV communication.
In the realm of networking, overhead sig-
nifies the supplementary data or resources
beyond the fundamental requirements of 5. Conclusion
transmission, encompassing error check- In conclusion, the contemporary landscape
ing and control information. While crucial, of Unmanned Aerial Vehicle (UAV) networks
an overabundance of overhead can result in demands sophisticated security measures,
inefficiencies, underscoring the importance of and the proposed Intrusion Detection System
a well-balanced approach to ensure optimal (IDS) significantly advances the field. By pri-
utilization of the network. oritizing authentication-based strategies and
In terms of overhead, Figure 8.3 illustrates recognizing the limitations of traditional IDS
CHAOS’s notable advantage in mitigating in UAV scenarios, the system incorporates
unnecessary burdens on the network. The use advanced mechanisms such as cryptogra-
of chaotic maps in authentication proves to phy, biometrics, and dynamic key generation.
be more streamlined, offering a significant The synergy of these elements, coupled with
reduction in overhead compared to QKD, the robustness introduced by chaotic maps,
not only ensures secure key exchange but
also poses computational challenges for
potential attackers. The adaptability of the
system, augmented by anomaly detection
and machine learning, further distinguishes
between normal and malicious UAV behav-
ior. Simulated evaluations utilizing tools like
ns3 provide tangible insights into the system’s
performance metrics, including Packet Deliv-
ery Ratio, Delay, and Overhead. This com-
prehensive approach culminates in a robust
Figure 8.2. Comparison of proposed method solution that safeguards UAV network integ-
with existing with respect to delay. rity, effectively addressing the unique security
Source: challenges prevalent in diverse operational
AUTHOR'S SELF GENERATED
 Strengthening UAV network security: Advanced authentication-based intrusion detection systems 73

scenarios. The proposed authentication- flying ad-hoc networks”, Ad Hoc Networks,

based IDS emerges as a pivotal advancement, Vol 133, 2022.
promising heightened security in the dynamic [7] C Atheeq, M Munir Ahamed Rabbani,
realm of UAV communications. “CACK—A Counter Based Authenticated
ACK to Mitigate Misbehaving Nodes from
MANETs” Recent Advances in Computer
Acknowledgement Science and Communications Vol 14, no. 3,
Dr. Atheeq and Dr. Altaf provided essential pp. 837–847, 2021.
insights and guidance, shaping the theoretical [8] Pandey, Gaurav K., et al. “Security threats
and mitigation techniques in uav commu-
framework and methodology. Dr. Shanila’s
nications: A comprehensive survey.” IEEE
expertise significantly enriched the data anal-
Access 2022.
ysis and interpretation phases. Ms. Shaista [9] Li, B. et al., “Anti-honeypot enabled optimal
played a crucial role in the implementation attack strategy for industrial cyber-physical
and experimental setup. It is noteworthy that systems”, IEEE Open Journal of the Com-
this research was independently conducted puter Society, Vol 1, pp. 250–261,2020.
without external sponsors, highlighting [10] Alshaer, N. et al., “Reliability and security
our unwavering commitment to advancing analysis of an entanglement-based QKD
knowledge and fostering innovation through protocol in a dynamic ground-to-UAV FSO
both individual and collective efforts. communications system”, IEEE Access, Vol
9, pp. 168052–168067, 2021.
[11] C Atheeq, M Munir Ahamed Rabbani,
References “Secure Data transmission in integrated
[1] Lateef, M.A. et al., “Data Aegis Using Che- internet MANETs based on effective trusted
byshev Chaotic Map-Based Key Authentica- knowledge algorithm” Indian Journal of Sci-
tion Protocol”, In Intelligent Manufacturing ence and Technology. Vol. 9 No. 47, 2016.
and Energy Sustainability: Proceedings of [12] Conrad, A. et al., “Drone-based quantum
ICIMES, pp. 187–195, 2022. key distribution (QKD)”, In Free-space laser
[2] C Atheeq, M Munir Ahamed Rabbani, communications XXXIII Vol. 11678, pp.
“Mutually Authenticated Key Agreement 177–184, 2021.
Protocol based on Chaos Theory in Integra- [13] Assaf, Tasneem, et al. “High-Rate Secret Key
tion of Internet and MANET” International Generation Using Physical Layer Security
Journal of Computer Applications in Tech- and Physical Unclonable Functions.” IEEE
nology, Vol. 56, No. 4, pp.309–318, 2017. Open Journal of the Communications Soci-
[3] Rosch-Grace, D. and Straub, J., “Analysis of ety 4, pp. 209–225, 2023.
the likelihood of quantum computing pro- [14] Tu, Yu-Ju, and Selwyn Piramuthu. “Security
liferation”, Technology in Society, Vol 68, and privacy risks in drone-based last mile
pp.101880, 2020. delivery.” European Journal of Information
[4] Gómez Hernández, J.A. et al, “Crypto-Ran- Systems, pp. 1–14, 2023.
somware: A Revision of the State of the Art, [15] Calcara, Antonio, et al. “Drones and Offen-
Advances and Challenges”, Electronics, Vol sive Advantage: An Exchange–The Authors
12, no. 21, p.4494, 2023. Reply.” Security Studies 32.3, pp. 582–588,
[5] S. Syed Abdul Syed et al., “Enhanced Chaotic 2023.
Map Based Key Agreement Mitigate Packet [16] Omolara, Abiodun Esther, Moatsum
Dropping Attack from MANETs” Interna- Alawida, and Oludare Isaac Abiodun.
tional Journal of Electronic and Communi- “Drone cybersecurity issues, solutions, trend
cation Engineering, Vol 10, no 7, 2023. insights and future perspectives: a survey.”
[6] Tsao et al., “A survey of cyber security threats Neural Computing and Applications 35.31,
and solutions for UAV communications and pp. 23063–23101, 2023.
9 Comparative analysis of wireless
routing protocols for security threat—A
review
Bhupal Arya1,a, Amrita Kumari2,b, and Jogender Kumar3,c
1
Research Scholar, Department of Computer Science and Engineering, Quantum University
Uttarakhnad, Rorrkee, India
2
Assistant Professor, Department of Computer Science and Engineering, Quantum University
Uttarakhnad, Rorrkee, India
3
Assistant Professor, Department of Computer Science, GBPIET Ghurdauri Pauri Garhwal, Pauri,
India

Abstract: There are various security issues, challenges, and threats in wireless networks. In this paper, we review
recent work in this field that primarily addresses such issues and challenges. Different researchers have taken
various approaches to justify their research based on different parameters. Methods such as enhanced packet
delivery, residual energy, and link quality, on-demand link and energy aware dynamic multipath (O-LEADM),
Rate aware congestion control (RACC), Quadrature Amplitude Modulation (QAM), Quadrature Phase Shift
Keying (QPSK), Network function virtualization (NFV), Software-defined network (SDN), hidden Markov
model (HMM), IDS, scalable and energy-efficient cluster-based anomaly detection (SEECAD) etc. are used to
identify jamming signal problems in fuzzy logic. The article discusses various issues and approaches covered by
many researchers in last few years paying close attention to wireless network-related topics.

Keywords: PCA, RACC, O-LEADM, HMM Model

1. Introduction stands as a critical security concern, capable

of disrupting communication and network
Wireless networks, by their very nature, functionality. These attacks are particularly
introduce vulnerabilities that are distinct relevant in scenarios where devices form net-
from those of traditional wired networks. works without a fixed infrastructure, such as
The inherent openness of wireless commu- disaster recovery. Black Hole Attack involves
nication, coupled with the broadcast nature a malicious node within a wireless network
of radio signals, creates opportunities for that deceitfully attracts and intercepts data
eavesdropping, unauthorized access, and data packets from legitimate nodes [1]. By falsely
interception. The rapid expansion of wireless advertising itself as having the minimum and
infrastructures, including Wi-Fi, cellular net- shortest path to the destination, the malicious
works, and emerging technologies like Inter- node request to other nodes asking for route
net of Things (IoT) devices, has magnified request (R-Req) to route their traffic through
these security concerns, making it imperative it. However, instead of forwarding the data
to explore novel approaches for safeguard- packets to their final destinations, the mali-
ing wireless communications. In the realm cious node drops or hold them, resulting in
of wireless networks, Black Hole Attack data loss and breakdown of communication.

[email protected]; [email protected]; [email protected]

a

DOI: 10.1201/9781003606635-9
 Comparative analysis of wireless routing protocols for security threat—A review 75

The process of a Black Hole Attack follows this research which inspire the researchers
these distinct stages mostly. in this field to propose methods to detect
Advertisement Phase: The malicious node the basic behavior of black-hole nodes and
claims that it has the shortest route for the the route request (RREQ) and RREP which
R-Req by the source and it generate the Route is mostly used by attackers to neutralize the
request reply (RREP) and send it to the target network. O-LEADM [2] is a method used to
node. detect black hole in MANET. It is essential
Attraction Stage: Legitimate nodes, relying to provide safe and secure communication in
on the falsified routing details, unwittingly wireless network, prevent from the network
direct their data packets toward the mali- damage and harm in small and large scale.
cious node. The purpose of this study is to showcase the
Packet Interception: Upon receiving these research conducted in this field and empha-
packets, the malicious node abstains from size the approaches taken to get around secu-
forwarding them to their rightful destina- rity vulnerabilities.
tions, opting to discard or consume them.
Impact and Consequences: As a result,
genuine nodes remain unaware of the com- 2. Theoretical Details
promised communication path, leading to In order to achieve wireless network architectural
data loss, network congestion, and potential complexity, various heterogeneous non-realistic and
service disruption. The detection and elimina- artificial approaches are offered. Communicating in
tion of the Black Hole Attacks present intri- various situations makes the deployment of security
cate challenges due to their deceptive nature. measures and regular packet delivery even more
Nevertheless, researchers have proposed complex. Figure 9.1 illustrates different types of
strategies to tackle this menace. jammers commonly encountered in cybersecurity
Due to highly dynamic and challenging and wireless security. Wireless networks have
open environment of wireless network there revolutionized the way we connect and
are many reasons for packet failure like bat- communicate, but they also come with a range of
challenges that need to be addressed to ensure their
tery drainage in IoT, signal interference and
reliability, security, and efficiency. Some of the key
malicious nodes. A malicious node or faulty
challenges of wireless networks include:
node can enter forcefully or without the prior
knowledge of victim node into the network • Signal Interference and Congestion: Wire-
and damage, disturb the network settings less networks work in shared frequency
and can decrease overall network throughput bands which can lead to potential interfer-
and performance. Black-hole node problem ence from other devices and networks. This
may try to route all the data packets towards interference can degrade the quality of the
itself by falsely claiming of having shortest signal and impact the overall transmission.
path and false unrealistic information. Source • Reduced Bandwidth: The bandwidth of
node chooses the path having higher RREP wireless networks is lower than that of
[2] destination sequence by simply disregard- wired networks. This restriction may
ing other routes and RREPs, comparing all result in slower data transfer in nations
RREPs being trapped in the attacker’s strat- having large population with numerous
egy, and compromising the node connected connected devices.
to the aggressor. The risk of packet drop • Coverage and Range: The range of wire-
exists for nodes that forward data packets less networks can be limited, especially in
along pathways that include rogue nodes, indoor environments or areas with obsta-
which reduces overall network performance. cles. Achieving consistent coverage across
This is the main motivational point behind larger areas can be challenging.
76 Innovations in Computational Intelligence: Trends and Technologies

• Data Security and Privacy: Wireless com-

munication is susceptible to eavesdrop-
ping and unauthorized access due to the
process of broadcast of signals. Ensuring
data security and user privacy is a signifi-
cant challenge.
• Authentication and Authorization: Prop-
erly authenticating and authorizing
devices and users on a wireless network is
essential to prevent unauthorized access.
However, doing so securely and efficiently
can be complex.
• Quality of Service (QoS): Maintaining AQ1:
Please
consistent QoS in wireless networks can be provide
difficult due to factors like signal strength figure
in-text
variations, interference, and network con- citation
gestion. This is particularly important Figure 9.1. Classification of Jammers in caption
for applications like video streaming and Cybersecurity and Wireless Security and
source.
online gaming. Source:
AUTHOR'S SELF GENERATED

• Mobility Management: Wireless networks spectrum management becomes critical.

often serve mobile devices that move Allocating and utilizing frequency bands
between different access points. Seam- effectively to prevent overcrowding and
less handovers from one access point to interference is a challenge.
another without disrupting the connec-
tion can be challenging, especially for 2.1. Security threats to wireless
real-time applications.
• Energy Efficiency: Many wireless devices network
used worldwide, now operate on battery • Traffic Analysis: In this type of problem an
power, and optimizing energy consump- attacker can access information related to
tion is crucial to extend the duration of the identification of activities on the net-
device battery life. Balancing performance work physical location of the user, traffic
with energy efficiency is a significant analysis of the user.
challenge. • Eavesdropping: In this type of attack an
• Heterogeneity: Wireless networks encom- attacker tries to listen to the conversation
pass various technologies such as Wi-Fi, of other person without his/her permis-
cellular, Bluetooth, and IoT protocols. sion. It can be categorized into active and
Managing and integrating these diverse passive forms.
technologies to ensure seamless connec- • Denial of Services (DoS): A Denial of Ser-
tivity can be complex. vice (DoS) attack is a type of cyber-attack
• Location Accuracy: Many applications in which the attacker tries to disturb the
rely on accurate location information. normal functioning of a target node where
Achieving precise location accuracy in the target node is not able to reach the
indoor and urban environments can be resource and the request for the resource
challenging due to signal reflections and is denied.
obstructions. • Brute Force Attack: Brute force attack is
• Spectrum Management: As demand for considered as a passive type of attack in
wireless communication grows, efficient which the attacker will try a number of
 Comparative analysis of wireless routing protocols for security threat—A review 77

efforts or trial and error to encrypt the des-sequence [1] which claims to have the
user’s message. shortest path to deliver the data from the
• Black hole Attack: A Black Hole Attack is source to destination node in the network,
a type of security threat in wireless ad hoc which is not true and hence it will disturb
networks in which a false node will gener- the communication.
ate a request-response, req-sequence and

3. Comparative Analysis of Wireless Routing Protocols for Security

Threat Table 9.1. Comparative Analysis of Wireless Routing Protocols
S. Author Parameters/ Findings
No Method
1. Suma and link quality, • The behavior of any node is determined using control AQ2:
Please
Harsoor [1] residual energy messages destination-sequence (des-Seq) and reply- provide
Higher packet sequence (rep-Seq) table in-
Delivery decision • Every intermediary node within the network delivers text cita-
tion and
Sequence a des-Seq message to every node next to it, and each caption.
control messages node returns to the intermediate node through a rep-
destination- Seq message.
sequence and • If a node’s neighbors’ des-Seq and req-Seq diverge,
reply-sequence it may be considered malicious. Nodes choose
(rep-Seq) forwarding based on behavior and capability to
achieve QoS criteria.
2. Angurala [2] (des-Seq) • Authors differentiated between packet loss caused by
message congestion and malicious nodes.
destination- • The discussed Sheme utilizes an The energy-aware
sequence and dynamic multipath (O-LEADM) routing strategy for
(rep-Seq) reply- MANETs locates black-hole nodes using on-demand
sequence [2] links and a bait plan of action.
3. Grove [3] NS2 simulator • The RACC technique, which is applied at the transport
layer in the OSI model, improves congestion control by
regulating the rate of source nodes at specific hotspot
locations.
• RACC is used to evaluate different modulation
schemes, like Binary Phase Shift Keying (BPSK),
Quadrature Phase Shift Keying (QPSK), and 16
(Quadrature Amplitude Modulation, or QAM), in
order to identify the best modulation scheme for the
method that is being reviewed and suggested.
• The throughput parameter was improved by 17%,
the packet delivery ratio was improved by 8.35%, the
normalized routing overhead was improved by 0.56%,
the MAC overhead was improved by 0.64%, the
average end-to-end delay was improved by 2.04%
(continued)
78 Innovations in Computational Intelligence: Trends and Technologies

S. Author Parameters/ Findings

No Method
4. NFV and (SDN), NFV, • Connecting their devices to the Internet and other
SDN [4] DNS networks opens up a world of potential ramifications
for consumers.
• It is possible to get what we need without keeping it
permanently on our gadgets. Communication enables
us to coordinate and plan our actions with one another.
• These connected gadgets offer the foundation for the
systems that assist us in running our lives. However,
these links also make our data susceptible to theft and
our equipment defenseless to harm. Network and cyber
security deals with this issue. Many of the companies
spend millions of dollars a year to secure their
computer networks and safeguard their data. Cisco
provides hardware equipped with firewalls, modeling
tools, and many network security protocols; certain
software is required to operate them.
5. Michael IDS and ANN- • The PCA: The authors constructed an IDS system for
Savva [7] based IDS for man-in-the-middle attacks based on two approaches
simulation and compared the outcomes.
they used NS2 • ANN method was used in the second. While the ANN-
simulator and based IDS reported 100% accuracy and detection rate,
they found the the PCA-based IDS claimed 76% accuracy and 74%
accuracy of detection rate.
ANN based IDS • The simulation was conducted using the NS2 simulator.
is more accurate where the 76% accuracy rate and 74% detection rate
with a accuracy of the PCA-based IDS are achieved. When 100% of
of almost 100% detections are made in the case of ANN-based IDS
-based IDSs is accuracy and 100% of PCA- and ANN-based average
0.768 and 1.0, mean accuracy is achieved. and ANN.
respectively.
6. Savva, Three types of • Jammers that are proactive, persistent, deceptive, or
Michael [8] jammer attacks random. First-generation attacks are particularly easy
Proactive to identify
Jammers • Their primary goal is to disrupt and impair node
Reactive connectivity while also keeping the source channel
Jammers occupied. Identifiability of the second sort of attack is
Specific function more difficult than that of continual jammers since it
jammers. focuses on various approaches where valid packets are
provided instead of constant signal transmission.
• The third kind of attack can operate in two ways:
sometimes it acts as a deceptive jammer and other
times it acts as a constant jammer. reactive jammer for
RTS/CTS. In this kind of attack, the RTS request is
destroyed by jammers.
(continued)
 Comparative analysis of wireless routing protocols for security threat—A review 79

S. Author Parameters/ Findings

No Method
7. Simulation AODV routing Channel Frequency 2.4 GHz No. of Nodes 100 nodes
based protocol Terrain File- Digital Elevation Model (DEM)
protocols can yield the Traffic Source- Constant Bit Rate (CBR) traffic load
[9-20] greatest results
in terms of
wireless network
congestion
avidity. They
have also
suggested a task
in which delivery
timing, and
optimizing data
throughput
8. Jeffrey, Cyber-physical The increased access to the public internet has made
Nicholas[21] systems (CPS) incidents targeting essential civilian infrastructure—like oil
are software- pipelines and electrical power grids—alarmingly common.
and hardware- This has increased CPS vulnerability. This article provides
integrated a thorough analysis of the body of research that looks at
integrated the most recent developments. Over the past ten years, CPS
systems. has grown quickly across a variety of industries, including
telemedicine, smart manufacturing, Internet of Things,
autonomous cars etc.

4. Conclusion control techniques in wireless sensor net-

works. International Journal on Recent and
Advancement in the field of computer Innovation Trends in Computing and Com-
and software technology had make a munication, 10(8), 47–54.
remark- able contribution in the growth [3] Grover, Amit, R. Mohan Kumar, Mohit
of wireless network but it has also many Angurala, Mehtab Singh, Anu Sheetal, and
serious issues and challenges as discussed R. Maheswar. (2022). Rate aware conges-
in this paper researcher has given tion control mechanism for wireless sensor
different approach and result in the field networks. Alexandria Engineering Journal,
of security in wireless networks. Table 9.1 61(6), 4765–4777.
presents comparative analysis of wireless routing [4] Suma, S. and Harsoor, B. (2016). An
protocols. However, in my view, the HMM approach to detect black hole attack for
model utilized in the Contiki-based congestion control utilizing mobile nodes in
operating system with the Cooja wireless sensor network. In: Editor, F., Edi-
simulator gives best result although some tor, S. (eds.) CONFERENCE 2016, LNCS,
chances of improvement are there. vol. 9999, pp. 1–13. Springer, Heidelberg. AQ3:
References [5] Kalkha, H., Satori, H. and Satori, Khalid. Please
note
[1] Suma, S. and Harsoor, B. (2022). An (2018). Preventing Black Hole Attack in that the
approach to detect black hole attack for Wireless Sensor Network Using HMM. references
congestion control utilizing mobile nodes in [6] Maria Hanif, Humaira Ashraf, ZakiaJalil, 5-6 is not
cited in
wireless sensor network. Materials Today: Noor Zaman Jhanjhi, MamoonaHumayun,
text.
Proceedings, 56, 2256–2260. Saqib Saeed, and Abdullah M Almuhaideb.
[2] Angurala, M., Bala, M. and Khullar, V. (2022). Aibased wormhole attack detection
(2022). A survey on various congestion
 80 Innovations in Computational Intelligence: Trends and Technologies

techniques in wireless sensor networks. Elec- [14] Premkumar, M., and Sundararajan, T. V. P.
tronics, 11(15), 2324. (2021). Defense countermeasures for DoS
[7] Michael Savva, Iacovos Ioannou. (2022). attacks in WSNs using deep radial basis net-
Fuzzy-logic based IDS for detecting jam- works. Wireless Personal Communications,
ming attacks in wireless mesh IoT net- 120(4), 2545–2560.
works. 20th Mediterranean Communication [15] Prasse, P., Machlica, L., Pevný, T., Havelka,
and Computer Networking Conference J. and Scheffer, T. (2017). Malware detection
(MedComNet). by analyzing network traffic with neural
[8] Sumaiya Thaseen Ikram and Aswani Kumar networks. 2017 IEEE Security and Privacy
Cherukuri. (2016). Improving accuracy of Workshops (SPW), 205–210.
intrusion detection model using pca and [16] Krishnamurthy, M. and Rajashekara, H.M.
optimized svm. CIT, 24(2), 133–148. (2011). Current trends in wireless technolo-
AQ4:
[9] Naga RohitSamineni, Ferdous A. Barbhuiya, gies in academic libraries. DESIDOC Jour-
Please and Sukumar Nandi. (2012). Stealth and nal of Library and Information Technology,
note semi-stealth mitm attacks, detection and 31(1).
that the
references
defense in ipv4 networks. In 2012 2nd IEEE [17] Chen, Z., Zhou, W., Wu, S., and Cheng,
9-20 is International Conference on Parallel. Dis- L. (2020). An adaptive on-demand mul-
not cited tributed and Grid Computing. tipath routing protocol with QoS support
in text. [10] Farhan Ahmad, AsmaAdnane, Virginia Fran- for high-speed MANET. IEEE Access, 8,
queira, FatihKurugollu, and Lu Liu. (2018). 44760–44773.
Man-in-the-middle attacks in vehicular ad- [18] El-Semary, A. M. and Diab, H. (2019). BP-
ATION ADDEDhoc networks: Evaluating the impact of AODV: Blackhole protected AODV rout-
attackers’ strategies. Sensors. ing protocol for MANETs based on chaotic
[11] QuaziWarisha Ahmed, Shruti Garg, Amrita map. IEEE Access, 7, 95185–95199.
Rai, Manikandan Ramachandran, Noor [19] Thebiga, M., and SujiPramila, R. (2020).
Zaman Jhanjhi, MehediMasud, and Moham- A new mathematical and correlation coef-
med Baz. (2022). Ai-based resource alloca- ficient based approach to recognize and to
tion techniques in wireless sensor internet of obstruct the black hole attacks in manets
things networks in energy efficiency with data using DSR routing. Wireless Personal Com-
optimization. Electronics, 11(13), 2071. munications,, 114(2), 975–993.
[12] J. R. Renofio, M. E. Pellenz, E. Jamhour, [20] Singh, J., Goyal, S., Kumar Kaushal, R.,
A. Santin, M. C. Penna, and R. D. Souza. Kumar, N., and Singh Sehra, S. (Eds.).
(2016). On the dynamics of the RPL proto- (2024). Applied Data Science and Smart
col in AMI networks under jamming attacks. Systems (1st ed.). CRC Press. https://doi.
In 2016 IEEE International Conference on org/10.1201/9781003471059
Communications (ICC), pp. 1–6, IEEE. [21] Jeffrey, Nicholas, Qing Tan, and José R. Vil-
[13] David, J., and Thomas, C. (2019). Efficient lar. (2023). A review of anomaly detection
DDoS food attack detection using dynamic strategies to detect threats to cyber-physical
thresholding on flow-based network traffic. systems. Electronics, 12(15), 3283.
Computers and Security, 82, 284–295.
 10 Cascadability analysis of free space
optical wireless data centers
Anand Kumar Dixit1,a, Meenakshi Srivastava2, b, and
Rajiv Srivastava3,c
1
Research Scholar, Amity Institute of Information Technology, Amity University Uttar Pradesh,
Lucknow, India
2
Assistant Professor, Amity Institute of Information Technology, Amity University Uttar Pradesh,
Lucknow, India
3
Ex-Faculty, Indian Institute of Technology, Jodhpur, Rajasthan, India

Abstract: Data centers having optical nature are the heart of modern day communication. The data is rising
continuously and thus increasing the demand for more bandwidth. The current data center relies on the elec-
tronics thus producing lot of heat and for connecting them huge amount of cabling is required. To reduce copper
cabling, use of fiber optic technology with fiber cable was proposed, and optical components produce lesser heat
as compared to electronic devices. However, fiber cabling is still an issue. To deal the cabling and heat dissipa-
tion, free space optical communication can be a better choice. This paper, presents the cascadability analysis
of the free space optical wireless data centers. In this work line of sight links are considered for cascadability
analysis and bit error rate is evaluated for various distances.

Keywords: Bit error rate, data center, free space optical communication, line of sight

1. Introduction
High speed communications switches making the wiring more complex [6]. In addi-
and networks are basically used in tion, we have more consumption of power
data-intensive computing platforms. with ­electrical switches than optical DCNs
These switches and network divides which also have an added advantage of less
the data into computing resources and expensive [7].
therefore the data is processed separately The issue of the limited frequency spec-
delivering output and scalability as trum in classical RF systems is becoming
per the data. A lot of researches are progressively more critical. However, by
going on to deal with these issues such as implementing optical communication free
introductionof new interconnect space technology, such a problem can be alle-
topologies for data center networks [1–4]. viated [8].
In this topology, we put twenty to fifty 60 GHz lines can support multi-Gbps
servers in a rack along with top of rates [9–11]. In contrast, to solve cabling AQ1:
rack (ToR) switch. Now, as a major issues, [1] proposes a totally wireless DC sys- Please
abbre-
portion of the traffic passes through tem based on 60 GHz technology. Despite its viate
some of the hot ToRs to other [5] which promise, 60 GHz technology has its demerit “DCN”,
makes the requirement of fast reconfiguration like, very high attenuation, low bandwidth “ToR”,
and
of Data Center Networks (DCNs). Generally, [10]. Because of the lack of atmospheric limi- “NLoS”
commodity electrical switches are used tations to free space optical (FSO) commu- in full
at first
bynormal DCNs to optimize nication in interior systems, we recommend mention.
the limited bandwidth. But, the problem
with this is the need of numerous links
and switches
a
[email protected]; [email protected]; [email protected]
DOI: 10.1201/9781003606635-10
82 Innovations in Computational Intelligence: Trends and Technologies

employing FSO in Data Center. Furthermore, The impulse response of LoS channel hLoS
light speed in FSO is approximately 1.5 can be expressed as [14, 15]:
times faster than fiber optics, it is ensure less
latency. As a result, we anticipate that FSO (1)
will result in high-performance, low-cost
DC infrastructure. This approach is useful where Ar is area of the photo-detector,
in situations where lines of optical fiber can- ϕ is radiance angle,
not be installed [12, 13]. It has the advan- ψ is light incidence angle,
tages of spectrum restriction and high-speed d is distance between LED and PD.
1
information transmission over other wireless m = - ln ( 2 ) / ln lim ( cos f1/2 ) is order of lam-
x ®¥ n
communication. Above mentioned system is bertian source emission and LED semiangle
predictable to swap other WC methods in half power is denoted by ϕ1/2.
various industries, including last-mile com-
munication. This technology, which is still in
its early phases, is broadly used in network. 3. Mathematical Modelling of
In the Figure 10.1, communication between Cascadability Analysis
the ToRs using line of sight (LoS) and
In this section, mathematical analysis of cacad-
Non-LoS links is shown. Considering four
ability analysis is presented. In ­Figure 10.2,
racks for an as example, in cluster 1, rack
BUS topology is shown where four racks are
1, and rack 2 can communicate using LoS
connected.
communication, similarly in cluster 2, rack
3, and rack 4 can communicate using LoS Power Calculation
model. This type of communication is The power received by the rack 2 when rack
known as intra racks com- munication. The 1 transmit is
rack 2 of cluster 1 and rack 3 of cluster 2 can
also communicate using the LoS model. The (2)
communication between the rack 1 and
racks 3, 4, will takes place using NLoS
model. The rack 1 can also communi- cate
to rack 3 via rack 2 using LoS model. In this
work LOS links are considered for cas-
cadability analysis and BER is evaluated
for various distances.
2. Free Space Optical
Communication AQ2:
Please
provide
In FSO communication systems, free space Figure 10.1. Communication between the ToRs source
serves as the communication channel between using LoS and NLoS links. for all
transceivers, playing a vital role in the effective Source:
figures.
AUTHOR'S SELF GENERATED
transmission of optical signals. The quality of
the received signal is significantly impacted by
the parameters of this channel. FSO enables
optical data transfer with potential data rates
of up to 100 Gbps, making it suitable for high-
speed optical networks. However, atmospheric
conditions can hinder FSO performance, limit- Figure 10.2. LoS communication between the
ing sensitivity and achievable data rates while cascaded racks (BUS topology).
maintaining an acceptable bit error rate (BER). Source:
AUTHOR'S SELF GENERATED
 Cascadability analysis of free space optical wireless data centers 83

The power received by the rack 3 when BER Calculations

rack 1 transmit is In optical WCSNR is given by [14]

(8)

where, Pr(i) is the power received at the ‘ith’
(3) rack.
The shot noise variance will be given by
(4) (9)

The power received by the rack 4 when The thermal noise expression is as follows:
rack 1 transmit is
.(10)
(5)
Finally, BER can be evaluated using:
In Figure 10.3, 16 racks are connected, (11)
where distance among the nodes is shown.
The power received by the rack 14 when Where, Q is error function.
rack 1 transmit is

(6) 4. Results
Table 10.1 contains parameters utilized in the
The power received by the rack 16 when calculation. In the simulation, we used a sin-
rack 1 transmit is gle LED and a single photo detector (PD) [16].

(7) Table 10.1. Simulation parameters [14]

Parameter Value AQ3:
Please
Semi angle 70° provide
source for
LED transmission power 20 dBm
Table.
Photo-detector Area (A) 10−4 m2
Field of View 70°, 10°
Refractive Index 1.5
Room Size 10×10×10 m3
Responsivity (R) 1.28
Avalanche Factor (M) 20
Noise factor 1.5
I2 0.087
Background current (IB) 5 mA
Noise bandwidth factor(I3) 0.52
Fixed capacitance of photo(Cpd) 110 pF/cm2
Light speed 3×108 m/s
Figure 10.3. LoS communication between the Transconductance (gm) 30
racks (16 Racks topology). voltage gain in open loop (Gol) 10
Source: Source:
AUTHOR'S SELF GENERATED
AUTHOR'S SELF GENERATED
84 Innovations in Computational Intelligence: Trends and Technologies

The server room’s volume is estimated to be racks is 3 meter. FoV=70° is applicable for
10 × 10 × 5 m3. In the simulation transmitter the broadcast messages, while in case of point
and receiver pair placed opposite side of the to point connection FoV=10° is applicable.
racks. FSO rack having servers such that the In case of FoV=10°, the maximum received
same side of the rack having all of the serv- power is −3.36 dBm.
ers’ transmitters (receivers). Main aim is to In Figure 10.5, transmitter and receiver
use the intra/inter-rack option. Optical Beams distance is 3√2 meters with FoV=70° and
are scattered to all servers using a beam dis- FoV=10°. The Pma is −35.29 dBm, and the
tributer, allowing for switch-free intra-rack minimum received power is −45.80 dBm,
connectivity. when the transmitted power is 20 dBm. This
In Figure 10.4, transmitter and receiver result is applicable for all the connection
distance is three meters with FoV=70° and where transmitter and receiver racks are align
FoV=10°. The maximum received power to each other, and distance between the adja-
(Pmax) is −32.29 dBm, and the minimum cent racks is 3√2 meter. In case of FoV=10°,
received power is −47.15 dBm, when the the maximum received power is −6.24 dBm.
transmitted power is 20 dBm. This result is In Figure 10.6, transmitter and receiver dis-
applicable for all the connection where trans- tance is 6 meters with FoV=70° and FoV=10°.
mitter and receiver racks are align to each The Pma is −38.29 dBm, and The min power
other, and distance between the adjacent received is −45.18 dBm, when the transmitted

Figure 10.4. LoS communication between the Figure 10.5. LoS communication between the
racks is 3 meters (a) FOV=70° (b) FOV=10°. racks is 3√2 meters (a) FOV=70° (b) FOV=10°.
Source: Source:
AUTHOR'S SELF GENERATED AUTHOR'S SELF GENERATED
 Cascadability analysis of free space optical wireless data centers 85

power is 20 dBm This result is applicable for

all the connection where transmitter and
receiver racks are align to each other, and dis-
tance between the adjacent racks is 6 meters.
In case of FoV=10°, the max power received
is −9.19 dBm.
In Figure 10.7, transmitter and receiver
distance is 6√2 meters with FoV=70° and
FoV=10° The Pma is −41.30 dBm, and The
min power received is −45.48 dBm, when the
transmitted power is 20 dBm This result is
applicable for all the connection where trans-
mitter and receiver racks are align to each
other, and distance between the adjacent
racks is 6√2 meters. In case of FoV=10°, the
max power received is −12.17 dBm.
In Figure 10.8, transmitter and receiver
distance is 9 meters with FoV=70° and
FoV=10°. The max power received is −41.81

Figure 10.7. LoS communication between the

racks is 6√2 meters (a) FOV=70° (b) FOV=10°
Source:

Figure 10.6. LoS communication between the Figure 10.8. LoS communication between the
racks is 6 meters (a) FOV=70° (b) FOV=10°. racks is 9 meters (a) FOV=70° (b) FOV=10°.
Source: Source:
AUTHOR'S SELF GENERATED AUTHOR'S SELF GENERATED
86 Innovations in Computational Intelligence: Trends and Technologies

dBm, when the transmitted power is 20 dBm.

This result is applicable for all the connection
where transmitter and receiver racks are align
to each other, and distance between the adja-
cent racks is 9 meters. In case of FoV=10°, the
max power received is −12.68 dBm.
In Figure 10.9, transmitter and receiver
distance is 9√2 meters with FoV=70° and
FoV=100. The Pma is −44.81 dBm, when the
transmitted power is 20 dBm. This result is
applicable for all the connection where trans-
mitter and receiver racks are align to each
other, and distance between the adjacent
racks is 9√2 meters. In case of FoV=10°, the
maximum received power is −15.67 dBm. Figure 10.10. Probability of Error for various
In Figure 10.10 shows probability error distances between the racks.
against various distant racks. Because the dis- Source:
tance between the racks across the room will AUTHOR'S SELF GENERATED
vary and referring Figure 10.3, the minimum
distance is 3 meter and the maximum distance
is 12.72 meters. In case of the distance of 3m
BER is very low and of the order of 10–45.
In case of diagonal racks with distance 3√2
meters the BER order is 10–13, and when dis-
tance between the racks is 6 meters the BER
order is 10–4 between the racks is distance 6
√2 the BER order is 10-2. The BER for the
distance of 9√2 is very high and of the order
of 0.2. Therefore, as the distance increases
the BER falls very rapidly, and thus to main-
tain low BER power must be increased. But
increase of power is not possible due to the
LED power limitations. The other possible
choice is decrement of FOV, an in case of
FOV=100, the BER is of the order of 10–200
up to distance of 15 meters.

5. Conclusion
Data centres are the core of modern day
communication, and as the amount of data
kept in DCs grows, the size of the data cen-
tre grows enormously, as does the amount of
heat released. As a result, it is now important
to develop a new communication paradigm
Figure 10.9. LoS communication between the where heat dissipation and amount of cabling
racks is 9√2 meters (a) FOV=70° (b) FOV=10°. required can be controlled. FSO is one such
Source: technology that can considerably reduce
AUTHOR'S SELF GENERATED
 Cascadability analysis of free space optical wireless data centers 87

squandered power. The application of FSO Maxim Podlesny, Md Golam Rabbani, Qi

technology in DC is a relatively new concept, Zhang, and Mohamed Faten Zhani (2012),
with just a few preliminary investigations to Data center network virtualization: A s­ urvey.
date. In this work cascadability analysis of the IEEE Communications Surveys and Tutori-
DC racks is presented and BER is evaluated. als 15, no. 2, 909–928.
[8] H. Davidson et al., (2013), Data center with
It is found that using proper setting param-
free-space optical communications, US Pat-
eters the error free transmission is possible.
ent 8,483,569.
[9] J. Shin et al. (2012), On the feasibility of com-
pletely wireless datacenters. Proc. Archit. for
References Netw. and Commun. Sys, pp. 3–14,
[1] Srivastava R., Bhattacharya P., and Tiwari [10] K. Ramachandran et al. (2008), 60 ghz data-
A.K. (2019), Dual buffers optical based packet center netw.: wireless => worry less?, NEC
switch incorporating arrayed waveguide grat- Laboratories America, Tech. Rep.
ings, Journal of Engineering Research 7, no. 1. [11] S. Kandula et al. (2009), Flyways to de-
[2] Singh A., Tiwari A.K., and Srivastava R. congest data center networks, Proc. ACM
(2018): 19, Design and analysis of hybrid Hotnets-VIII.
optical and electronic buffer based optical [12] N. Riza and P. Marraccin I (2012), Power
packet switch, Sādhanā 43, no. 2. smart indoor optical wireless link applica-
[3] [3] Xu, Maotong, Chong Liu, and Suresh tions, Wireless Comm. and Mobile Comp.,
Subramaniam (2016) PODCA: A passive pp. 327–332N.
optical data center architecture, IEEE Inter- [13] Hamedazimi et al. (2013), Patch panels in
national Conference on Communications the sky: A case for free-space optics in data
(ICC), pp. 1–6. IEEE, 2016 centers, Proc. ACM Hotnets, pp. 23:1–23:7.
[4] Shukla, Manahar Prashant, and Rajiv Sriv- [14] Dixit A.K., Srivastava M., and Srivastava.
astava (2018), Arrayed Waveguide Grating R (2020) Physical layer analysis of opti-
and Re-Circulating Buffer Based Optical cal wireless data centers. Journal of Opti-
Packet Switc., Journal of Optical Communi- cal Communications, vol. 2020, pp. 1–10,
cations 1, no. ahead-of-print. https://doi.org/10.1515/joc-2020-0207
[5] Srivastava, Rajiv, and Yatindra Nath Singh [15] Dixit, Anand Kumar, Meenakshi Srivastava,
(2010) Feedback fiber delay lines and AWG and Rajiv Srivastava. (2023), Analysis of
based optical packet switch architecture, Opti- mimo optical wireless data center networks.
cal Switching and Networking 7, no. 2 75–84. International Journal of Information Tech-
[6] Srivastava, Rajiv, Rajat Kumar Singh, and nology 15, no. 1 (2023): 519–529, https://
Yatindra Nath Singh (2009), Design analy- doi.org/10.1007/s41870-022-01103-8
sis of optical loop memory. Journal of light- [16] Z. Ghassemlooy et al. (2012), Optical Wire-
wave technology 27, no. 21 4821–4831. less Communications: System and Channel
[7] Bari, Md Faizul, Raouf Boutaba, Rafael Modelling with MATLAB, 1st ed. Taylor
Esteves, Lisandro Zambenedetti Granville, and Francis.
11 Wireless network with machine learning
and deep learning techniques
Bhupal Arya1,a, Amrita Kumari2,b, and Jogendra Kumar3,c
1
Research Scholar,Quantum University Uttrakhand, Roorkee, India
2
Assistant Professor,Department of Computer science and Engineering, Quantum University
Uttarakhnad, Roorkee, India
3
Assistant Professsor,Department of computer science and Engineering,
GBPIET Ghurdauri pauri Garhwal,Uttarakhnad, India

Abstract: In the current scenario of wireless technology, we go through the many challenges and issues sur-
rounding the application and usage of tools, techniques of machine learning as well and deep learning algo-
rithms in wireless networks. In heterogeneous networks, designing the machine learning foundation routing
approach is challenging. At the beginning of this work, the fundamental basic concepts of (ML) machine learn-
ing and (DL)deep learning in different wireless networks are presented. Because of the constantly changing
network conditions present in many ad-hoc networks, preparing data collection and training with that data is
a particularly difficult process. In addition, this study gives an overview of several studies that used deep learn-
ing and machine learning techniques in a range of research areas, including networking, communications, and
dissipative settings. Finding possible issues and challenging tasks is the main objective of this survey activity.

Keywords: Deep learning; machine learning; routing algorithms; wireless sensor networks

1. Introduction ratio, etc. to figure out the network dynam-

ics, which include hotspots, incursion distri-
Machine learning in wireless networks, is a bution, congestion zones, traffic bottlenecks,
broader field that encompasses various tech- and spectrum availability.As a result, wireless
niques and algorithms for building models networks with numerous nodes and dynamic
which can trained and learn from data and connection quality may be analyzed using
make some decisions or predictions deep learning.
Deep learning can also be called as a sub-
part of machine learning that focuses on 1.1. Approaches for machine learning
neural networks with multiple layers (deep
neural networks).
in wireless networks
To handle accurate pattern detection i. Reinforcement Learning
from complicated raw data, deep learning ii. Deep Reinforcement Learning
is a promising machine learning technology. iii. K-Nearest Neighbors Algorithm
Deep learning makes use of several layers of iv. Bayesian Net/HMM
neural networks to extract fine, brain-like v. K-Means
properties from high-dimensional raw data. vi. Decision Tree
It is possible to use it to examine a variety of vii. Particle Swarming Routing
network measures, including latency, Apply viii. Deep Q-Learning
metrics like loss rate, link signal-to-noise ix. Q-Learning

[email protected]; [email protected]; [email protected]

a

DOI: 10.1201/9781003606635-11
 Wireless network with machine learning and deep learning techniques 89

2. Machine Learning Techniques in 2D. Rate adaption problem-

Wireless Mesh Network a. Q-Learning
Depending on the suitability of a given b. LA (Learning Automata)
approach and a network issue, machine c. Bayesian
learning techniques may be utilized to handle
different management and design challenges. 2E. Solving the problem of fault
Most often, wireless mesh network design detection a) K-means 2F Routing
optimization issues may be solved using
reinforcement learning (RL). Three different Q-Learning—A machine learning technique
types of learning and their categories - called Q-learning makes use of reinforce-
ment learning to determine the worth of a
A. Supervised learning— given action in a certain state. The “Q” in
a. Support Vector Machine ­Q-learning stands for quality, which is how
b. ANN valuable an action is in maximizing future
c. DT (Decision Tree) rewards.
d. Perception
e. Bayesian 2F. Learning Automata
B. Unsupervised Learning—
a. K-means Learning automata-based processes have been
b. Principal Component Analysis C. applied to route optimization. The learning
Reinforcement Learning- automata principle is used by algorithms like
c. Q-Learning Steiner Connected Dominating Set, and LA
d. LA (Learning Automata) (Learning Automata)-based Multicast Rout-
e. MKP ing to provide an optional path from source
to destination. These problems include con-
Techniques used in solving problems in nection congestion, routing decision delays,
wireless mesh networks- and channel allocation.

2A. Solving the channel assignment 2G. Artificial Neural Network

problem An Artificial Neural Network (ANN), or
a. K-means simply a neural network, is a computational
b. LA (Learning Automata) model that is modeled after the architecture
and operation of biological neural networks
within the human mental state. Machine
2B. Solving the problem of routine
learning is a subset which involves neural net-
issue works.fundamental building block of deep
a. LA (Learning Automata) learning.
b. MDP
c. ANN
2H. Channel Allocation
2C. Improve the fairness in wireless Bayesian Learning- Using a testing item’s
input characteristics and the whole training
mesh network data set, Bayesian learning attempts to deter-
a. MDP (Markov Decision process) mine the posterior probability distribution of
b. Q-Learning the target attributes.
 90 Innovations in Computational Intelligence: Trends and Technologies

2I. Network Deployment • Big data transport made easier by deep

learning
Mesh routers and mesh gateways (MGs)
• The difficult problem of employing deep
(MRs) must be placed properly during net-
learning for big data transports
work deployment to guarantee the network’s
• to build a thick routing pipe with the abil-
intended performance. The mesh routers and
ity to send a large number of packets per
gateways in wireless mesh networks may be
second.
deployed using Q- Learning and LA (Learn-
• Identifying the hop-by-hop connection
ing Automata) approaches.
failure
• MAC parameters selection to guarantee
2J. Anomaly/Intrusion Detection-[7] service delivery quality.
Decision Trees (DT)—The decision tree is a • Deep learning-based network swarming
well-liked machine learning technique for • to control node movement to achieve the
regression and classification. They’re a fun- appropriate swarming form and effective
damental part of supervised learning and communication architecture
have applications in various domains such • Management of nodes and cluster forma-
as finance, healthcare, and natural language tion in swarming networks
processing. Integrity and Faulty Detection- • Parameters that should be gathered from
design patterns
• Big data architecture selection
2.1. Deep Learning in the Wireless • Deep learning’s suitability in wireless net-
Network: Problems and works for particular Tasks coordination
Challenges among several CP controllers
• Optimization of the transport layer using • Implementation of distributed deep learn-
deep learning ing in wireless nodes
• the multi-queue evolution pattern detection • applying a particular deep learning algo-
• the multi-queue evolution pattern detection rithm component to a particular distrib-
• determining the length of the RED zone uted job
line • using MAC protocols to control wireless
• Mixing several protocols and schemes nodes so they may exchange input param-
with the congestion control scheme eters and output data
• Control of traffic along the whole route • Cross-layer design based on deep learning
from source to destination • To achieve cross-layer optimization
• The fundamental application layer for
deep learning
AQ1:
• Based on the performance objective of the
Please application layer, to design a less-com-
provide plexity deep learning model.
intext
citation • Deep learning leads to DRL
for Figure • applications for controlling cognitive
11.1 and
11.2
radio networks
• Deep learning-based swarming
AQ2: • to control node movement to achieve the
Please
provide appropriate swarming form and effective
source communication architecture
for all the Figure 11.1. Deep learning in wireless network.
figures
• Management of nodes and cluster forma-
Source: tion in swarming networks
AUTHOR'S SELF GENERATED
 Wireless network with machine learning and deep learning techniques 91

• A fundamental application layer for deep suitability in wireless networks for par-
learning ticular tasks
• to specify a less-complexity deep learn- • Coordination among several CP controllers
ing model depending on the performance
objective of the application layer 2.3. Several machine learning
• Based on deep learning Cloud computing approaches forpacket routing
security in the dew
• applications for controlling cognitive q-routing-
radio networks Q-Learning [10] is a reinforcement learning
• DL/DRL implementations that are effec- algorithm that may discover the best course
tive on real-world wireless systems of action in a given situation to maximize
• Collecting Network configurations for the rewards from that situation. Distribu-
the input phases of deep learning may be tive routing of packets ina network is made
challenging. possible by Q-Routing, a modification of
• The wireless devices’ range restrictions Q-Learning.
• incomplete batches of training samples
• Optimization of the transport layer using 2.3.1. Ant-based routing
deep learning
• the multi-queue evolution pattern An innovative use of reinforcement learn-
detection ing called “Ant-Based Routing” is based on
• determining the length of the RED zone basic biological “ants.” These “ants” trav-
line erse the network and quickly pick up the
• Mixing several protocols and schemes best paths using the ant colony stigmergy
with the congestion control scheme model of communication as inspiration. This
• Control of traffic along the whole route method is more durable than conventional
from source to destination routing algorithms in that it is less affected
• Using deep learning to speed up massive by random route corruption while computing
data transfer is difficult problem employ- packet paths.
ing deep learning for big data transports
• To build a thick routing pipe with the 2.3.2. Particle swarm routing
ability to send a large number of pack- Drs. Russ Eberhard and James Kennedy
ets per second. Identifying the hopby-hop invented particle swarm optimization, a
connection failure population-based stochastic optimization
• MAC parameters selection to guarantee approach, in 1995. They learned from the
service delivery quality. Deep learning’s

AQ3:
Please
provide
Figure 11.2. ML using ANN and DNN. Figure 11.3. Layered ANN caption
Source: Source: 11.3–11.7.

AUTHOR'S SELF GENERATED AUTHOR'S SELF GENERATED

92 Innovations in Computational Intelligence: Trends and Technologies

social interactions of flocks of birds and under various circumstances and topolo-
schools of fish. Particle Swarm Routing starts gies. RL-Probe was successful with vari-
with a set of random fixes, followed by iter- ous link topologies and link quality
atively seeks for Optimal solutions through
updating subsequent generations.[3] 2.4. Problems and difficulties with
• The network’s connection may be greatly machine learning in a wifi
improved by using reinforcement learning network
techniques to manage node mobility and Machine learning techniques are now in use
packet routing decisions. widely used in many fields, such as image,
• The four primary categories of contem- speech, video, public safety, and medicine.
porary several types of routing protocols We now have a wealth of data stored in sev-
were developed for wireless networks: eral data centers and these are being analyzed
geographical, on-demand, reactive, and in real-time as a result of the development of
ML/DL-based. Outingtable-based proac- more advanced computer-related processes.
tive protocols. DL-based routing systems While sing machine learning methods and
have received a lot of attention recently techniques, it is possible to construct pre-
due to their enhanced performance in diction or human-like decision-making pro-
complicated networks. cesses. The system should be able to safely
• An innovative method named “RL-Probe” handle the whole planet on its own utilizing
was presented in for improving connec- machine learning techniques. For instance,
tion quality. RL-Probe was crucial in the system gathers all the data generated
boosting RPL’s functionality and capacity. by each person and learns everything about
RPL is a routing protocol for less power the environment on its own. The ability
and dissipative networks. To decrease the of machine learning systems to constantly
overhead, The RL-Probe approach is used learn from data over times one of its major
and is based on the reinforcement advantages. The system may be continually
Learning paradigm. updated while it is in use by using freshly
observed or created data. As a consequence,
• RL-Probe Using the RL-Probe framework, various logics are generated using various
both synchronous and asynchronous training sets using a single machine learning
LQE approaches are compatible. The rel- method. It implies that the system is adapt-
evance of each node is taken into consid- able in terms of how it makes decisions and
eration while making clustering decisions that it can continually learn from experience.
throughout the RPL route maintenance Referring back to the example, the system
and route recovery operation. can sound alert or take any other action that
• Asynchronous probing approach is sug- may be regarded intelligent and beneficial
gested to measure the RSSI and ETX. A when it identifies any harmful or anomalous
MAB model is used to accomplish the situations from the world of humans. Cre-
decision- making process. With the use of ating this ideal system, however, is difficult
the RL-Probe technique, reward function in practice. The difficulty of obtaining trust-
has been utilized to evaluate the trends in worthy training data that covers a variety of
link quality fluctuations. RL-Probe was contexts gives rise to the first problem. To
assessed using three performance assess- build any decision or prediction model with-
ment matrices: Packet loss rate, packet out an appropriate data set, regardless of
overhead on top of the usual energy con- how strong our machine learning algorithms
sumption. Evaluation work was done are at Building robust logic and analyzing
 Wireless network with machine learning and deep learning techniques 93

state may be examined from a broad view-

point using a variety of bits of information,
and problems that could seriously harm net-
work performance can be discovered before-
hand. In sensor networks, several unusual
situations need to be recognized beforehand.
For example, defective application layer sen-
sors may constantly broadcast false detected
Figure 11.4. Anomaly Detection values to a connected server. If we are igno-
Source:
rant of this, we won’t be able to pinpoint
AUTHOR'S SELF GENERATED the exact location of the accident. Ona net-
work, each node makes decisions just based
large data sets in some real-scenario while on its local knowledge, even while RPL cap-
continuously training the model and system. tures the network issue and the topology
Recently, researchers and businesses have is continuously updated while considering
been increasingly curious about the Internet network conditions. The problem might not
of Things. The emergence of several research be identified until it is more significant if
teams as a result of this trend has sped up network traffic is concentrated on a single
the development of critical methods. Com- node. The node with the strangely unequal
munication and Networking techniques, and high workload [11] in this scenario will
particularly for less power and dissipative quickly exhaust its energy before being shut
networks, have drawn a lot of attention in off. On the connection layer, a wireless net-
comparison to other networks because of work interface may malfunction. Due to the
the constrained properties of low-power network’s global view, all of these factors
methods and constrained devices that are may be discovered in this case before the cat-
not the same as those of traditional, anti- astrophic calamity happens. The complete
quated sensor-based networks. picture may be obtained by collecting data
Routing protocols like as RPL and specific and monitoring traffic from the server or
mac protocols such as Slotted time wise and the potent root node. The list of defect cases
Channel Hopping are created and are being that need to be Compatible. The relevance of
widely utilized as standard protocols to boost each node is taken into consideration while
the dependability of communication on less making clustering decisions
power and dissipative networks built with
restricted equipment. Simply said, RPL pre- • Traffic overload due to attack [11]
vents routing loops by producing DODAG • Energy use as a result of an uneven traf-
structures and straightforwardly building fic load RL-Probe. We know that one
routing paths. Additionally, the topology benefit of using ML algorithms is that
created for route management is constantly the method learns diverse conditions,
evolving while taking into account various including inexperienced ones, through-
network data and circumstances. TSCH ena- out time, since the rationale for these
bles the use of several channels as well as decisions is not dependent on estab-
the resolution of Intrusion-related problems. lished static rules. Similar to the above,
However, these processes are carried out it seems advantageous to have a strategy
depending on regional information gath- utilizing machine learning algorithms
ered at every node. Hence, the issues that when we create methods for identify-
are not promptly identified and resolved by ing abnormal circumstances like those
the localized protocols. A network’s current mentioned above on less power and
94 Innovations in Computational Intelligence: Trends and Technologies

dissipative networks. It is difficult to get s­ ystem—picked up on dissipative, less power

the data set needed to train the model, networks is provided below.
though. Also, compared to other net-
works, the network environment in a • An energy-deficient node
less power dissipative network is quite • a wireless network’s interface is
dynamic. As a result, it is almost hard to malfunctioning
collect a solid data collection that cov- • Intrusion on a specific channel [7]
ers a range of circumstances. • CPU consumption that is too high on a
node
• A node’s memory or buffer is full.
3. Machine Learning in Networking • Strange perceived value
and Communication • both synchronous and asynchronous LQE
approaches are
We provide a number of papers that use
machine learning methods in networking or
communications. These techniques provide 3.2. Data gathering and
a significant contribution to the application administration
of machine learning to solve difficult issues
The authors stress the need to comprehend
brought on by changing and unexpected
the various data kinds that may be gathered
wireless channel status. Despite the fact that
in SDNs as well as the method for extracting
these works have significantly improved per-
information from those data. To take the is
formance when evaluated, some of them only
the beginning of the journey toward network
focus on trace-driven testing or simulation
control based on machine learning. Article
using the same data set as training. Yet, these
outlines a fundamental architecture set up
efforts are important in that they set the way
in a commercial network that leverages the
for leveraging a variety of data to create trust-
Open flow protocol to capture traffic statis-
worthy models to get around the challenge
tics. However, this study mainly focuses on
of foreseeing the future condition of wireless
evaluating the keeping an eye on and catego-
channels.
rization of traffic using data received through
the Open flow protocol and traffic categori-
3.1. Signal classification zation for network.
To operate the network automatically or The signal provided by a sender needs to
autonomously, we want to build a solid be accurately detected at a receiver side in
order to execute dependable wireless com-
munication. In the authors [3] concentrate
on the effects of modulated signal Intrusion
[7] and actual wireless channel circumstances
on classification performance. They suggest
using machine learning to categorize the sig-
nal in a realistic wireless environment. As
our study classifies the signal that has been
altered and is now traveling across a wireless
channel, we consider it to be akin to pattern
recognition does not provide any advice for
operating large-scale MLbased systems or
Figure 11.5. Multi Layered Deep Neural Network networks. However, our study.has provided
Source: clear guidance on how to deploy ML-based
AUTHOR'S SELF GENERATED
 Wireless network with machine learning and deep learning techniques 95

network management and made it practical decide on cloud computing. User input, device
to do so. energy level, and CPU use level are additional
input values that are distinct and steady in
3.3. Network attack forecast contrast to values affected by a dynamic and
unpredictable wireless network.
In the study, a technique based on machine
learning technology for setting security rules
on the SDN controller has been proposed. 3.6. Network of wireless sensor
Security settings for the SDN controller are networks
configured to block the whole subnetwork
to prevent access from possible attackers, 3.6.1. Using machine learning
and machine learning methods are utilized In this area, we present several works that
to identify likely target hosts that can be utilized sensor networks [14] and machine
attacked. The identical datasets were divided learning approaches. These techniques also
into training and test sets in order to evaluate make a significant contribution to the appli-
the suggested strategy. cation of machine learning to solve difficult
issues brought on by dissipative channels and
3.4. Wireless adaptive swimming restricted devices. These studies are important
because they opened the way for the develop-
Network conditions vary greatly depend- ment of trustworthy models that can be used
ing on the situation and alter over time. It to circumvent the challenge of forecasting the
is therefore difficult to predict the network’s future state of dissipative channels.
future condition. Despite the fact that a num-
ber of rate adaptation methods for high-­
quality video streaming have been proposed, 3.7. Diagnostics for channel errors
there is still significant room for advance- Many protocols, including 802.11, 802.15.4,
ment. Numerous research has validated the 802.15.1, etc., share the ISM band. Here,
use of machine learning techniques to video Intrusion between several systems impairs
streaming services. It should be possible to communication quality. In-depth studies were
create a system that selects the best bitrate done by the authors to examine the error pat-
and CDN for the streaming user, learns cru- terns in IEEE 802.15.4, and they discovered
cial parameters, and maximizes QoE. Rein- that there are several patterns for the most
forcement learning was used in addition to common wireless situations. They devel-
taking into account the bandwidth, buffer oped a machine learning technique based on
level, and video rate to automatically create this discovery to categorize wireless channel
the best ABR algorithm. faults into distinct groups and offered the
system that can identify various issues in IoT
3.5. Mobile cloud offloading networks.

The paper presents a dynamic method based

on ML and introduces the usage of cloud 3.8. Spectrum choice
computing for offloading computation from The article also highlights ISM band pollu-
mobile devices. To decide whether to offload tion and restrictions on sensor node power.
computing to the cloud, it keeps an eye on To overwhelm this challenging environment,
network characteristics and device resources. it suggests a ML based channel allocation
Network data, such as available bandwidth, and selection method. Using ML, the sys-
is just one of many input variables employed tem predicts several expected transmission
in this work’s machine learning approach to attempts. This will use performance data like
96 Innovations in Computational Intelligence: Trends and Technologies

RSSI and LQI from the most recent packet test subject had a wearable sensor [14] that
received as well as the following features allowed him to identify himself in the wire-
as input information: the quantity of trans- less sensor network.
mission attempts, the reasons behind each
unsuccessful attempt, and RSSI and LQI. The 3.11. Detection of events
optimal channel from the output is one with
a less projected number of transmission tries. Wireless sensor [12] networks have a wide
range of applications. The initiative focuses
on locating pipeline leaks in the transporta-
3.9. Outlier detection tion networks for gas, oil, and water. Using
Wireless sensor [12] networks with a small algorithm-like pattern recognition, it trains
number of nodes are vulnerable to Intrusion the sensor network to distinguish novel traces
[7], unstable channels, and cyber interven- of events, such as leakages. Here, some scat-
tion, which reduces system performance and tered sensor nodes collaborate to estimate the
increases the possibility that false data may size of the leakage event. Despite the employ-
be relayed to higher management levels. This ment of a wireless sensor network in this
might have major consequences for indus- investigation [11], the difficulties associated
trial automation systems and sensor network with using an unreliable and unsafe wireless
systems used for public safety. The existing channel were scarcely considered.
works for outlier identification, according
to their authors, take a lot of memory, a sig- 3.12. Fault detection
nificant amount of energy, power for com-
Data gathering from the sink node is impre-
putation, transmission overhead, and data
cise due to the less- cost sensors and other
transfer capacity. They suggested employing a
problems caused by unreliable wireless net-
multi-agent framework and a machine learn-
works. Fault data must be observed in order
ing approach for online outlier’s identifica-
to monitor the system and respond quickly
tion to tackle the issue
to incidents. The cause of the issue must also
be identified. The paper proposed a statistical
3.10. Invoremental locatization technique for identify and classify issues in a
wireless sensor [14] network while using the
In general, GPS is one of the most widely used
Machine learning techniques. System faults
methods for object localization. Finding an
and data faults were used to categorize various
object’s specific location is difficult because
fault types. System faults are the other fault
of the limited GPS signal strength inside of
types brought on by less battery, calibration,
buildings. Due of this, other tactics are used.
For example, a large number of nodes are
used as anchor points which estimate the
comparative location of a target object. A
precise interior localization system must be
created in order to increase safety in under-
ground mines or caverns. Unfortunately, the
wireless channel continues to experience
Intrusion, which lowers the estimation’s
accuracy. To find the method with the fewest
errors and evaluate its performance, the study
used two alternative architectures and seven Figure 11.6. Wireless Network Communication Path
different machine learning techniques. The Source:
AUTHOR'S SELF GENERATED
 Wireless network with machine learning and deep learning techniques 97

communication, or connectivity issues. Data collect training data in a variety of settings.

problems are those that are caused by dam- Due to the unpredictable behavior of wireless
aged or inoperable sensors. The authors of networks, creating training data sets may be
have examined machine learning-based flaw challenging.
identification in analogous settings.

5. Conclusion
3.13. Routing
Our study has uncovered a wide range of
Multi-hop routing systems [19] could utilize methods and protocols that might be used to
less energy with the use of machine learning. improve wireless networks and communica-
The work provided a Machine Learning- tions. Based on the literature, we examined
based clustering technique to efficiently place a variety of problems and difficulties with
the nodes(sensor) in the closest cluster. The applying machine learning and deep learn-
researchers applied ML to the wireless sensor ing algorithms to improve various opera-
network’s routing protocol. The suggested tions in networking in various ways. For
routing technique aims to extend network tasks like network administration, network
life and move data packets as quickly as fea- optimization, signal, Data management,
sible. This suggests that ML approach adop- channel assignment, network security, route
tion on WSN is advantageous for resource selection, etc. Machine learning and Deep
management. learning algorithms perform incredibly well.
Our study has uncovered a wide range of The two primary learning methods that per-
methods and protocols that might be used to form well for network operations are deep
improve wireless networks and communica- reinforcement learning and deep learning.
tions. Based on the literature, we examined However, it could be challenging to collect
a variety of problems and difficulties with training data in a variety of settings and it
applying deep learning and machine learning is not possible in this research to complete
algorithms to improve networking operations all the algorithms and methods of ML and
in various ways. For tasks like network admin- Deep learning which we can cover in future
istration, network management, optimization scope Due to the unpredictable behavior of
signal management, network security, opti- wireless networks, creating training data
mum route selection. Machine learning and sets may be challenging
Deep learning algorithms perform incred-
ibly well. The two primary learning methods
AQ4:
that perform well for network operations References Please
are deep reinforcement learning and deep change
[1] Johnson, A. and Smith, B. (2016). Enhancing
learning. However, it could be challenging to the refer-
Data Delivery Efficiency in AODV-Based Ad ences into
Hoc Networks. Wireless Communications Num-
bered
and Networking Conference (WCNC), 1–6. references
[2] Lee, C., and Park, D. (2017). Performance as per
Evaluation ofAODV in Dynamic Mobil- style,
ity Scenarios. IEEE Transactions on Mobile
Computing, 16(8), 2134–2147.
[3] Emilio Ancelotti, CarVallati, Raffaele Bruno,
Enzo Mingozzi, A reinforcement learning-
based link quality estimation strategy for
RPL and its impact on topology manage-
Figure 11.7. ment, Computer Communications, Volume
Source: 112, 2017, Pages 1–13.
AUTHOR'S SELF GENERATED
98 Innovations in Computational Intelligence: Trends and Technologies

[4] Martinez, F., and Rodriguez, G. (2018). Large- Scale MANETs. Wireless Personal
Throughput Optimization in AODV Net- Communications, 113(1), 1–17.
works for IoT Applications. Ad Hoc Net- [16] S.A. Haque, M. Rahman, S.M. Aziz, Sensor
works, 75, 60–72. anomaly detection in wireless sensor net-
[5] Simi S, Sruthi Ann Varghese,“Enhance QoS works for healthcare, Sensors15 (4) (2015)
by LearningData flow rates in Wireless Net- 8764–8786.
works using Hierarchical Docition”, 4th [17] Zhang, W., and Chen, X. (2021). Analyz-
International Conference on Eco-friendly ing Mobility Models and Their Impact on
Computing and Communication Systems, AODV Performance. Journal of Network
ICECCS 2015, Procedia Computer Science and Computer Applications, 185, 102968.
70 (2015) 708–714, Elsevier [18] Garcia, M., and Lopez, A. (2023). Secure
[6] Kim, H., and Jung, I. (2019). Enhancing Routing with AODVin MANETs: A Crypto-
Security in AODV Routing Protocol for Ad graphic Approach. International Journal of
Hoc Networks. Computer Communications, Network Security, 25(4), 692–704.
135, 87–98. [19] Rahman, A., and Khan, M. (2023). Exploit-
[7] Wu, J., and Li, K. (2020). Performance Eval- ing Cross-Layer Design for Improved AODV
uation of AODV in Heterogeneous Ad Hoc Performance in VANETs. Ad Hoc Networks,
Networks. Wireless Personal Communica- 123, 102377.
tions, 112(3), 1519–1532. [20] Wei Wang; Xiaohong Guan; Xiangliang
[8] A. Garofalo, C. Di Sarno, V. Formicola, Zhang,“Profiling program and user behav-
Enhancing intrusiondetection in wireless sen- iors for anomaly intrusion detectionbased
sor networks through decision trees,Depend. on non-negative matrix factorization” 43rd
Computing. (2013) IEEE Conference on Decision and Control,
[9] Rahman, S., and Khan, T. (2022). Perfor- 2004. Issue Date: 14–17 Dec. 2004, On
mance Enhancement of AODV Through page(s): 99–104 Vol.1
Dynamic Parameter Adjustment. Interna- [21] Hazem M. El-Bakry, Nikos Mastorakis A,
tional Journal of Wireless Networks and- “Real-Time Intrusion Detection Algorithm
Communications, 14(1), 53–64. for Network Security,WSEAS Transactions
[10] Zhang, Y., and Wang, Q. (2016). A Survey of on communications, Issue 12,Volume 7,
Energy-Efficient Routing Protocols in Wire- December 2008.
less Sensor Networks. Journal of Internet [22] HR, M., MV, A., S, G. and S. V., “Devel-
Technology, 17(3), 471–482. opment of anti-phishing browser based on
[11] S. Kosunalp, A new energy prediction algo- random forest and rule of extraction frame-
rithm for energy- harvesting wireless sen- work” Cybersecurity, 3(1), 2020
sor networks with Q-learning, IEEEAccess 4 [23] Saxena, S., Shrivastava, A. and Birchha, V.,
(2016) 5755–5763. “A Proposal on Phishing URL Classification
[12] Wang, H., and Li, Q. (2018). An Improved for Web Security,” International Journal of
AODV Algorithm for Load Balancing in Computer Applications, 178(39), pp.47–49,
Mobile Ad Hoc Networks. Wireless Per- 2019.
sonal Communications, 99(3), 1987–2002. [24] G. B. Huang, H. Zhou, X. Ding and R.
[13] Nayak, p. and G.K.Swetha (2021) “Routing Zhang, “Extreme Learning Machine for
in wireless sensornetworks using machine Regression and Multiclass Classification,”
learning techniques”, measurement Volume in IEEE Transactions on Systems, Man, and
178, June 2021, 108974 Cybernetics, Part B (Cybernetics), vol. 42,
[14] Qian Mao, Fei Hu, and Qi Hao, “Deep no. 2, pp. 513–529, April 2012.doi: 10.1109/
Learning for Intelligent Wireless Networks: TSMCB.2011.2168604.
A Comprehensive Survey”, IEEE communi- [25] A. Qayyum, A.S. Malik, N.M.Iqbal, M.
cations surveys and tutorials, vol. 20, no. 4, Abdullah, M.F. Rasheed, W., Abdullah,
fourth quarter 2018, pp. 2595–2621 T.A.R. and M.Y.B. Jafaar,., Image classifi-
[15] Li, X., and Wu, Y. (2020).On the Per- cation based on sparsecoded features using
formance of AODV Routing Protocol in sparse coding technique for aerial imagery:
 Wireless network with machine learning and deep learning techniques 99

a hybrid dictionary approach. Neural Com- AODV Performance. Journal of Network

puting and Applications, 2017 pp.1–2 and Computer Applications, 185, 102968.
[26] Yin C, Zhu Y, Fei J, He X. A deep learn- [28] Altaber, A., “Phishing Websites Classification
ing approach for intrusion detection using using Hybrid SVM and KNN Approach,”
recurrent neural networks. IEEE Access International Journal of Advanced Com-
2017;5:21954–61. puter Science and Applications, 8(6), 2017.
[27] Zhang, W., and Chen, X. (2021). Analyz- [29] KSII Transactions on Internet and Informa-
ing Mobility Models and Their Impact on tion Systems, Robust URL Phishing Detec-
tion Based on Deep Learning, 14(7), 2020
12 Assessing technology innovation of
cellular health
M. Vasumathi Devia, Y. Muvvanab, Y. Sai Likhitha, M. Gowri
Bhavanic, and M. Vishnu Priyad
Computer Science and Engineering, Vignan’s Nirula Institute of Technology and Science for Women,
Guntur, India

Abstract: The conventional landscapes therapeutic treatment approach has changed in recent years due to the
arrival of cell phone applications, or heeltaps. The present heeltaps are designed to investigate the skills of scien-
tific specialists to determine those who are reliable, better, and well-liked ones. We can talk about the method of
integrating mHealth advances into clinical practice and figuring out whether professionals want to accept such
through the application of the procedure of normalization hypothesis. We looked at app attributes and perfor-
mance data, completed assessment reports on characteristics, capabilities, and feedback from the audience, and
used data analysis to evaluate feature cross-correlation and market alternatives. There is a possibility of devel-
opment regarding the which are already accessible the marketplace, as we discover unsatisfactory specifications
for workmanship and dependability. Strong ties exist between them.

Keywords: app features, new technology uptake, mobile health, and fitness app (health app), normalization
process (NPT), reliability (UX), and ease-of-use (UI)

1. Introduction States increased by 123%, from $2.6 trillion

to $3.2 trillion [3]. Twenty-one to forty-seven
Concerning healthcare Clinicians and percent of this enormous expense [4] can be
researchers alike are interested in the vast attributed to inefficiency in the form of non-
amounts of patient demographic, treatment fee induced duties such unfit assignments, as
plan, cost, and insurance coverage data being appropriate missions’ usage of antimicrobial
generated by the booming healthcare indus- agents, as well as deception. Researchers dis-
try. Recent years have seen an increase in the covered that 251,454 patients perish in the
number of scholarly articles focusing on the US every year. annually because of physician
out-of-the-ordinary applications of health- malpractice [5], so some of these costs may
care information mining technologies. Never- be attributable to subpar care. The move to
theless, the absence of a clear and reasonable a value-based healthcare company is fraught
explanation prompted us to create an over- with obstacles, but it may be smoothed down
view of the literature Investigation into the with better decision-making in accordance
subject. with accessible data [4]. Statistics creation as
Many countries, the healthcare industry is part of a control system is becoming increas-
thriving [1]. Costs, inefficiency, low quality, ingly popular in the healthcare industry [6].
and increasing complexity are only some of the Every day, this device collects a tremendous
difficult outcomes of this expansion [2]. From amount of data. Analytics provides methods
2010 to 2015, healthcare costs in the United and tools for deciphering these intricate and

[email protected]; [email protected]; [email protected];

a

[email protected]
d

DOI: 10.1201/9781003606635-12
 Assessing technology innovation of cellular health 101

copious records [2] to produce numbers that for the Introduction with Health Data
can inform negotiating choices. Recently, health data has received increasing
attention for both healthcare and data
mining. PARK, 2018 [4] Abstract In this
2. Literature Survey paper we discuss the challenges in analyzing
and mining health data to find clinical insight.
Somayyeh Zakerabasali, Seyed Mohammad, This paper reviews major challenges, the
Tayebeh Banias Adi, Azita Yazdani and Sha- sources of data, meth- odologies and
habuddin Abhari, 2021 others [1]. Many technologies along with open issues in big
mobile health (mHealth) apps are being data analytics in health care.
AQ2:
integrated into the healthcare system, but Yujun Ma1, (Member, IEEE), Yulei Please
their potential success depends on uptake by Wang1, Ju Yang2, Yiming Miao2, And Wei check the
sentence
healthcare providers. Some of the important Li2 (2017) [ 5]]
is correct.
benefits associated with mHealth technology The paper talks about the challenges of a
are increased efficiency, cost breakdowns and medical and health care system, i.e. uneven
enhanced health care service delivery pro- distribution of medical resources in rural
cesses. Future research is needed to explore areas, explosive growth in chronic diseases,
additional barriers at the healthcare system and rising medical expenses require medical
level i.e. legal, reimbursement/ accountable reform approach to a good solution. This rec-
care organizations, economic and financial ommends us Health Internet of Things (IoT)
factors, health system policy and standardisa- along with big data to deal with such prob-
tion so that systematic interventions can be lems. This paper demonstrates a large-scale
implemented to improve access to mHealth health application system that utilizes the
technology. [2] Ayan Banerjee Member, health IoT, big data architecture, key technol-
IEEE, and Sandeep K. S. Gupta Senior Mem- ogies and typical application of health IoT.
ber, IEEE~~2014~ Abstract: While smart The paper proposes to examine the health
mobile medical computing systems (SMDC- status of non-hospitalized patients using a
Ses) exploit context information to provide Wireless Body Area Network (WBAN) sys-
healthcare services, unexpected changes of tem with load-sharing processing capabilities
the clinical context may result in incorrect on both private and public clouds. [SHAKER AQ3:
or incomplete service provision, generating EL-SAPPAGH 1,2, FARMAN ALI 2, SAMIR Please
check the
faults and violating requirements. To mitigate EL-MASRI3, KYEHYUN KIM4, AMJAD sentence
this issue, this paper presents an approach to ALI 2,5, AND KYUNG-SUP KWAK 2, is correct.
analyze SMDCSes that accounts for dynamic (Member,...2019 [6]] Establishment of smart
changes in context and the interaction with environment and integration of used compo-
the physical world which may exist. nents is the main research issue in a newly
Achilleas Papageorgiou, Michael Strikes, developed network as well standardization
Eugenia Politou, Efthimios Alepis, Constan- network must have a single architecture,
tinos Pataki’s [1], 2018 [3]. The paper seeks protocol, and application program interface
to offer broad feedback to the developers of to connect between heterogeneous smart
these apps to shore up their security/privacy objects. To deliver MH apps and solutions
features while pointing out regular mistakes to address serious medical problems that are
made in the application development life easier to use with increased precision.
cycle that can threaten user privacy rights. [Sabyasachi Dash1†, Sushil Kumar Shakya AQ4:
AQ1: [Gaspard Harerimana, (Student Member, war 2,3†, Mohit Sharma 4,5 and Sandeep Please
Please check the
check the
IEEE), Beakcheol Jang, (Member, IEEE), Jong Kaushik6*,2019 [7]] The paper analyses the sentence
Wook Kim, (Mem- Ber, Ieee), And Hung Kook] data in health care through their necessity of
sentence is correct.
the paper provides a deep understanding
is correct. collectiveness.
 102 Innovations in Computational Intelligence: Trends and Technologies

[8] Sohail Imran, Tariq Mahmood, Ahsan All the way through the innovation imple-
Morshed and Timos Sellis (Fellow, IEEE), mentation process, the NPT comprises the
“Enabling Big Data Analytics In TheHealth- following four key additives:
care Sector: Challenges And Neuro-Fuzzy
Based Solutions”, 2021 1. “Coherence” requires knowing the cost,
Mobile Health (mHealth) apps for moni- importance, and uniqueness of the action
toring health status of patients with chronic or by itself “identity the experience-­
medical conditions are increasingly becoming creating the desire.” because the method
prevalently used [Gerson, Rodriguez, Devasena, called “experience-making” “paintings
Inupakutika, Sahak, Kaghyan, David Akopian, that creates and organizes an exercise as
Palden Lama, Patricia Chalela, Amelie Ramirez a cognitive category, it’s seen as an assort-
et al.,2020 [10]]. In this paper, the authors dem- ment of competencies this includes sepa-
onstrate the methodology of assessing perfor- ration from other hobbies and a common
mance of a developed prototype model mHealth experience of intent [56].
app for breast cancer patients on EHT. 2. “Cognitive involvement” requires that the
main players in the intervention design
and upkeep.
3. Conceptual Structure 3. “Collective motion” influences kinship
In facts structures (IS) studies and notably in between expert groups and conforms to
technology adoption studies, knowledge of the overarching organizational contextual
the goals or dreams for phone clients is cru- content of objectives, morale, manage-
AQ5:
Please cial to clarify the purpose of usage, satisfac- ment, and assets.
note tion, involvement [53].
that the
Implementing new clinical practices and Fourth, “reflexive tracking” involves eval-
reference
numbers implementing innovations in healthcare uating the new intervention for its efficacy
[53]
requires sophisticated organizational strate- and practice.
onwards
is not gies and sociological understanding of change. Desk I provide an empirical explanation of
occurred So, we took a sociological concept from the NPT methods and components available
in refer-
ence list. the field of generation called the “NPT” [54]
and applied it to the field of technology to
explain how healthcare systems have incor-
porated technological and organizational
changes. This theory sheds light on the
human and organizational linkages behind
the spread of innovative interventions into
everyday practice [55]. Structured inter-
views and questionnaires have been utilized
with the NPT represents the core concept to
examine adoption within health. Us utilize
the NPT to understand value for medical
experts apply a novel way to practice the use
of medicinal drugs, as well as the importance
of app features in attracting and retaining
experts. Engagement, or user engagement, is
the worth of the consumers consider follow AQ6:
interacting app, which is tied to longevity for Please
good application characteristics this stimu- Figure 12.1. Conceptual framework. provide
source
late it is usage. Source: for all the
AUTHOR'S SELF GENERATED figures
 Assessing technology innovation of cellular health 103

Table 12.1. Mechanisms of NPT of mHealth in medicine

AQ7:
Please Contracts of NPT[57] Definition [58], [55| Mechanisms of Implementation of
provide mHealth apps for professionals
in-text
citation Coherence Participants’ understanding of the mHealth App Categorization
for table Sense-making intervention and its differentiation
12.1
Cognitive Others’ influence and Users’ Reviews and App Quality
Participation Effort participants’ commitment to Evaluation
work with the intervention
Collective Action Fit with overall context including Trialability and App Feature Relevance
AQ8:
Please Commitment goals, morale, trust and resources
provide Reflective Monitoring Participants’ evaluations and Users’ Engagement and App Popularity
source
for all the Appraisal appraisals of the intervention
tables Source: AUTHOR'S SELF GENERATED

during mHealth app deployment. As a result, Table 12.2. Research questions and
the NPT details how to embed (implement) methodology
novel artistic routines in existing (sustain- Research Questions Research Approach based
able) societal frameworks (integration). on XPT
You can see how this three-degree NPT RQ1: What are NPT constructs
method works in practice in Figure 12.1. The the app categories, • Implementation
4 NPT mechanisms are defined by the various features, quality stage: app Categories
categories of mHealth apps and the require- and popularity that (coherence)
ments they satisfy. Positive app categories are affect the adoption • Embedding stage:
developed primarily based on their abilities intention of mobile app Feature relevance
(the need for intervention). Customers can app innovation (collective action)
verify their hopes that the app will meet their by health • Integration stage: app
needs by testing it out before they fully accept professionals? Quality (cognitive
the innovation, which is part of the process of participation), app
“implementing” the innovation. The second Popularity (reflexive
monitoring)
level of the system is called “embedding,” and
Findings: Tables 3. 4 and 5
it outlines the mechanisms that can be uti-
lized to carry out new practices; The systems RQ2: What app • Mean differences of
in the instance represent the abilities., that metrics are affected app “integration"
by the inclusion of metrics based on
need to be met, of the apps. Like the idea of
app features? feature inclusion
a social capital, “embeddedness” consists of Findings: Table 7
relational characteristics like norms and trust,
RQ3: To what • Correlations between
levels can be vital for organizations like medi-
extent are the the NPT dynamics
cal care that incorporates social improves this
attributes of Findings: Tables 8a and 8b
frequently necessitate deep comprehension of innovative
intricate societal issues [59]. adoption of
inHealth by health