Convergence of Deep Learning in Cyber-IoT Systems and Security E-Book

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Serie: Artificial Intelligence and Soft Computing for Industrial Transformation
• Sprache: Englisch

Seitenzahl: 572

Veröffentlichungsjahr: 2022

Beschreibung

CONVERGENCE OF DEEP LEARNING IN CYBER-IOT SYSTEMS AND SECURITY

In-depth analysis of Deep Learning-based cyber-IoT systems and security which will be the industry leader for the next ten years.

The main goal of this book is to bring to the fore unconventional cryptographic methods to provide cyber security, including cyber-physical system security and IoT security through deep learning techniques and analytics with the study of all these systems.

This book provides innovative solutions and implementation of deep learning-based models in cyber-IoT systems, as well as the exposed security issues in these systems. The 20 chapters are organized into four parts. Part I gives the various approaches that have evolved from machine learning to deep learning. Part II presents many innovative solutions, algorithms, models, and implementations based on deep learning. Part III covers security and safety aspects with deep learning. Part IV details cyber-physical systems as well as a discussion on the security and threats in cyber-physical systems with probable solutions.

Audience

Researchers and industry engineers in computer science, information technology, electronics and communication, cybersecurity and cryptography.

Leseprobe

Table of Contents

Cover

Series Page

Title Page

Copyright Page

Preface

Part I: VARIOUS APPROACHES FROM MACHINE LEARNING TO DEEP LEARNING

1 Web-Assisted Noninvasive Detection of Oral Submucous Fibrosis Using IoHT

1.1 Introduction

1.2 Literature Survey

1.3 Primary Concepts

1.4 Propose Model

1.5 Comparative Study

1.6 Conclusion

References

2 Performance Evaluation of Machine Learning and Deep Learning Techniques: A Comparative Analysis for House Price Prediction

2.1 Introduction

2.2 Related Research

2.3 Research Methodology

2.4 Experimentation

2.5 Results and Discussion

2.6 Suggestions

2.7 Conclusion

References

3 Cyber Physical Systems, Machine Learning & Deep Learning—Emergence as an Academic Program and Field for Developing Digital Society

3.1 Introduction

3.2 Objective of the Work

3.3 Methods

3.4 Cyber Physical Systems: Overview with Emerging Academic Potentiality

3.5 ML and DL Basics with Educational Potentialities

3.6 Manpower and Developing Scenario in Machine Learning and Deep Learning

3.7 DL & ML in Indian Context

3.8 Conclusion

References

4 Detection of Fake News and Rumors in Social Media Using Machine Learning Techniques With Semantic Attributes

4.1 Introduction

4.2 Literature Survey

4.3 Proposed Work

4.4 Results and Analysis

4.5 Conclusion

References

Part II: INNOVATIVE SOLUTIONS BASED ON DEEP LEARNING

5 Online Assessment System Using Natural Language Processing Techniques

5.1 Introduction

5.2 Literature Survey

5.3 Existing Algorithms

5.4 Proposed System Design

5.5 System Implementation

5.6 Conclusion

References

6 On a Reference Architecture to Build Deep-Q Learning-Based Intelligent IoT Edge Solutions

6.1 Introduction

6.2 Dynamic Programming

6.3 Deep Q-Learning

6.4 IoT

6.5 Conclusion

6.6 Future Work

References

7 Fuzzy Logic-Based Air Conditioner System

7.1 Introduction

7.2 Fuzzy Logic-Based Control System

7.3 Proposed System

7.4 Simulated Result

7.5 Conclusion and Future Work

References

8 An Efficient Masked-Face Recognition Technique to Combat with COVID-19

8.1 Introduction

8.2 Related Works

8.3 Mathematical Preliminaries

8.4 Proposed Method

8.5 Experimental Results

8.6 Conclusion

References

9 Deep Learning: An Approach to Encounter Pandemic Effect of Novel Corona Virus (COVID-19)

9.1 Introduction

9.2 Interpretation With Medical Imaging

9.3 Corona Virus Variants Tracing

9.4 Spreading Capability and Destructiveness of Virus

9.5 Deduction of Biological Protein Structure

9.6 Pandemic Model Structuring and Recommended Drugs

9.7 Selection of Medicine

9.8 Result Analysis

9.9 Conclusion

References

10 Question Answering System Using Deep Learning in the Low Resource Language Bengali

10.1 Introduction

10.2 Related Work

10.3 Problem Statement

10.4 Proposed Approach

10.5 Algorithm

10.6 Results and Discussion

10.7 Analysis of Error

10.8 Few Close Observations

10.9 Applications

10.10 Scope for Improvements

10.11 Conclusions

Acknowledgments

References

Part III: SECURITY AND SAFETY ASPECTS WITH DEEP LEARNING

11 Secure Access to Smart Homes Using Biometric Authentication With RFID Reader for IoT Systems

11.1 Introduction

11.2 Related Work

11.3 Framework for Smart Home Use Case With Biometric

11.4 Control Scheme for Secure Access (CSFSC)

11.5 Results Observed Based on Various Features With Proposed and Existing Methods

11.6 Conclusions and Future Work

References

12 MQTT-Based Implementation of Home Automation System Prototype With Integrated Cyber-IoT Infrastructure and Deep Learning–Based Security Issues

12.1 Introduction

12.2 Architecture of Implemented Home Automation

12.3 Challenges in Home Automation

12.4 Implementation

12.5 Results and Discussions

12.6 Conclusion

References

13 Malware Detection in Deep Learning

13.1 Introduction to Malware

13.2 Machine Learning and Deep Learning for Malware Detection

13.3 Case Study on Malware Detection

13.4 Conclusion

References

14 Patron for Women: An Application for Womens Safety

14.1 Introduction

14.2 Background Study

14.3 Related Research

14.4 Proposed Methodology

14.5 Results and Analysis

14.6 Conclusion and Future Work

References

15 Concepts and Techniques in Deep Learning Applications in the Field of IoT Systems and Security

15.1 Introduction

15.2 Concepts of Deep Learning

15.3 Techniques of Deep Learning

15.4 Deep Learning Applications

15.5 Concepts of IoT Systems

15.6 Techniques of IoT Systems

15.7 IoT Systems Applications

15.8 Deep Learning Applications in the Field of IoT Systems

15.9 Conclusion

References

16 Efficient Detection of Bioweapons for Agricultural Sector Using Narrowband Transmitter and Composite Sensing Architecture

16.1 Introduction

16.2 Literature Review

16.3 Properties of Insects

16.4 Working Methodology

16.5 Proposed Algorithm

16.6 Block Diagram and Used Sensors

16.7 Result Analysis

16.8 Conclusion

References

17 A Deep Learning–Based Malware and Intrusion Detection Framework

17.1 Introduction

17.2 Literature Survey

17.3 Overview of the Proposed Work

17.4 Implementation

17.5 Results

17.6 Conclusion and Future Work

References

18 Phishing URL Detection Based on Deep Learning Techniques

18.1 Introduction

18.2 Literature Survey

18.3 Feature Generation

18.4 Convolutional Neural Network for Classification of Phishing vs Legitimate URLs

18.5 Results and Discussion

18.6 Conclusion

References

Web Citation

Part IV: CYBER PHYSICAL SYSTEMS

19 Cyber Physical System—The Gen Z

19.1 Introduction

19.2 Architecture and Design

19.3 Distribution and Reliability Management in CPS

19.4 Security Issues in CPS

19.5 Role of Machine Learning in the Field of CPS

19.6 Application

19.7 Conclusion

References

20 An Overview of Cyber Physical System (CPS) Security, Threats, and Solutions

20.1 Introduction

20.2 Characteristics of CPS

20.3 Types of CPS Security

20.4 Cyber Physical System Security Mechanism— Main Aspects

20.5 Issues and How to Overcome Them

20.6 Discussion and Solutions

20.7 Conclusion

References

Index

Also of Interest

End User License Agreement

List of Illustrations

Chapter 1

Figure 1.1 A comparison of the coding sizes of several formats. With Thinger...

Figure 1.2 Harvard architectural based wireless communication board.

Figure 1.3 NodeMCU microcontroller.

Figure 1.4 Basic circuit of gas sensor.

Figure 1.5 Flowchart of noninvasive detection of oral submucous detection de...

Figure 1.6 Serial monitor.

Figure 1.7 Think Speak Console.

Chapter 2

Figure 2.1 Linear Regression Classification.

Figure 2.2 Random forest regression model.

Figure 2.3 Ada boosting regression.

Figure 2.4 Gradient boosting regression.

Figure 2.5 Support vector regression.

Figure 2.6 Neural network structure.

Figure 2.7 Regression function for ANN.

Figure 2.8 Life cycle of multioutput regression model.

Figure 2.9 Working of multioutput regression.

Figure 2.10 Logistic regression.

Figure 2.11 Decision tree classifier.

Figure 2.12 Structure of random forest classification.

Figure 2.13 Classification based on K-nn algorithm.

Figure 2.14 Feed forward network.

Figure 2.15 Recurrent neural network architecture.

Figure 2.16 Architecture of LSTM network model.

Figure 2.17 Histogram plot (a)(c) and Box plot (b)(d) of the target variable...

Figure 2.18 Histogram plot for the categoric variable (a) Sale condition and...

Figure 2.19 Distribution of categories in neighborhood variable.

Figure 2.20 Categorial variables before and after one hot encoder.

Figure 2.21 Artificial neural network model for regression.

Figure 2.22 Mean_squared_logarithmic_error loss for deep learning models.

Figure 2.23 Precision and recall scores of machine learning methods for clas...

Chapter 3

Figure 3.1 Deep learning with allied stakeholders and their attributes.

Figure 3.2 Possible courses related to the ML & DL programs.

Figure 3.3 Possible job titles and designation from the ML & DL programs.

Chapter 4

Figure 4.1 The flowchart of the proposed work.

Figure 4.2 Bag-of-Words of the two sentences.

Chapter 5

Figure 5.1 Multiple Choice Question based examination system.

Figure 5.2 Descriptive question based examination system.

Figure 5.3 Algorithms used in existing systems.

Figure 5.4 Block Diagram (Student).

Figure 5.5 Block Diagram (Faculty).

Figure 5.6 Block Diagram (Online Assessment System).

Figure 5.7 UML Usecase Diagram for online assessment system.

Figure 5.8 Stages in natural language processing.

Figure 5.9 NLP algorithms used in online examination system.

Figure 5.10 Dataset.

Figure 5.11 Generated syntax tree.

Figure 5.12 Objective question formation.

Figure 5.13 Result generation and storing result.

Chapter 6

Figure 6.1 A schematic representation of e-commerce business process having ...

Figure 6.2 A detailed architecture showing how IoT components are connected ...

Figure 6.3 Implementation of the Q learning algorithm—module and interaction...

Figure 6.4 Azure IoT edge real time data transmission.

Figure 6.5 A real time message flow with stream analytics handled through pu...

Figure 6.6 Deployment using Azure Docker container service.

Chapter 7

Figure 7.1 Difference between boolean logic and fuzzy logic.

Figure 7.2 Block diagram of FIS [9].

Figure 7.3 User temperature membership function.

Figure 7.4 Humidity reference membership function.

Figure 7.5 Heat fan speed reference membership function.

Figure 7.6 Cool fan speed reference membership function.

Figure 7.7 Humidifier reference membership function.

Figure 7.8 Fuzzy base class.

Figure 7.9 Mamdani fuzzy interface system.

Figure 7.10 Fuzzy base rule.

Figure 7.11 Fuzzy rule viewer.

Figure 7.12 Gradient graph of user temperature vs humidity vs heat fan speed...

Figure 7.13 Gradient graph of user temperature vs. humidity vs. cool fan spe...

Figure 7.14 Gradient graph of user temperature vs. humidity vs humidifier.

Figure 7.15 Graph of user temperature vs. heat fan speed.

Figure 7.16 Graph of user temperature vs cool fan speed.

Chapter 8

Figure 8.1 Masked face is divided into 100 blocks of same size and then extr...

Figure 8.2 Feature extraction method using CNN [27].

Figure 8.3 Diagram of convolutional block attention module [30].

Figure 8.4 Curvelet tiling of space and frequency.

Figure 8.5 Block diagram of face recognition using DCT and CS classifier.

Figure 8.6 Block diagram of pre-processing step.

Figure 8.7 Block diagram of fused rule.

Figure 8.8 (a) Used mask, (b) sample face image, (c) generated masked face, ...

Figure 8.9 Sample images of ORL face database.

Figure 8.10 Sample images of AR face database.

Figure 8.11 Performance comparison of recognition rate vs. feature dimension...

Figure 8.12 Performance comparison of recognition rate vs. feature dimension...

Figure 8.13 Performance comparison of recognition rate vs. feature dimension...

Figure 8.14 Performance comparison of recognition rate vs. feature dimension...

Chapter 9

Figure 9.1 Example of CT images of four patients in axial, coronal, and sagi...

Figure 9.2 Chest images of COVID-19 patients presented in [5].

Figure 9.3 Demonstrate measurement of Central-Apnea waveforms of a specific ...

Figure 9.4 Glimpse of alpha satellite which estimates risk factor over diffe...

Figure 9.5 The expected pattern of pandemic presented in Ferguson

et al.

[19...

Figure 9.6 Impact of household and socially targeted policies presented in F...

Figure 9.7 Distributions of patients in the dataset.

Figure 9.8 Gender wise distributions of patients in the dataset.

Figure 9.9 Distributions of patients tested asymptomatic or symptomatic.

Figure 9.10 Gender wise distributions of patients tested asymptomatic or sym...

Figure 9.11 Age wise distribution of patients completing containment.

Figure 9.12 Gender and age wise distribution of patients.

Figure 9.13 Age wise condition of patients.

Figure 9.14 Status of cases.

Figure 9.15 Model mean absolute and squared error.

Chapter 10

Figure 10.1 Detailed flowchart of the methodology used.

Chapter 11

Figure 11.1 Authentication procedure for smart home applications.

Figure 11.2 Authentication procedure between the communication parties.

Figure 11.3 Face expression along with various images which was collected fr...

Figure 11.4 Evaluation with various existing and proposed models.

Figure 11.5 Execution time along with average time and its performance.

Chapter 12

Figure 12.1 IoT cloud-based smart home automation.

Figure 12.2 MQTT architecture.

Figure 12.3 ESP32 NodeMCU.

Figure 12.4 2-input relay switch.

Figure 12.5 DHT 11.

Figure 12.6 Devices connected with MQTT.

Figure 12.7 MQTT broker publish-subscribe.

Figure 12.8 Arduino IDE for development.

Figure 12.9 Controlling home devices from outside.

Figure 12.10 Adafruit dashboard.

Figure 12.11 Feed log of light 1.

Figure 12.12 Temperature feed log (graphical).

Figure 12.13 Temperature feed log.

Figure 12.14 Humidity feed log (graphical).

Figure 12.15 Humidity feed log.

Figure 12.16 Alert mail through IFTTT.

Chapter 13

Figure 13.1 Confidential data sending [4].

Figure 13.2 Integrity [4].

Figure 13.3 Data availability [4].

Figure 13.4 General malware idea [5].

Figure 13.5 Basic diagram of Trojan horse [5].

Figure 13.6 Basic diagram of virus [5].

Figure 13.7 Ransomware attack using Trojan horse [6].

Figure 13.8 Architecture of machine learning [2].

Figure 13.9 Supervised learning [7].

Figure 13.10 Unsupervised learning [7].

Figure 13.11 Basic layer structure of deep learning [8].

Figure 13.12 Combination of CNN and image processing for malware detection [...

Chapter 14

Figure 14.1 System flowchart of the model.

Figure 14.2 Outline of the system models.

Figure 14.3 Separation of all external modules, relationship between those m...

Figure 14.4 Differentiates the modules frontend and backend.

Figure 14.5 Sign up form.

Figure 14.6 Login page.

Figure 14.7 UI interface.

Figure 14.8 (a) Add guardian. (b) Guardian details.

Figure 14.9 Screenshot of current location.

Figure 14.10 Screenshot of alert message.

Chapter 15

Figure 15.1 Previous and predictive data about the usage of IoT from 2018 to...

Figure 15.2 Probable appearance of ML/DL approaches in IoT safekeepings.

Figure 15.3 Basic deep learning concepts.

Figure 15.4 Basic deep learning concepts.

Figure 15.5 Basic architecture of classic three-layer neural network.

Figure 15.6 Convolution neural networks (CNN) working procedure.

Figure 15.7 Recurrent neural network (RNN) structure.

Figure 15.8 Recurrent Neural Network (RNN) structure.

Figure 15.9 Basic model of self-organizing map neural network.

Figure 15.10 Boltzmann machine.

Figure 15.11 Deep reinforcement learning.

Figure 15.12 Basic structure of auto encoder.

Figure 15.13 Different types of autoencoder.

Figure 15.14 Back-prop procedure.

Figure 15.15 Gradient descent with dimensions.

Figure 15.16 An automatic speech recognition (ASR) technique.

Figure 15.17 Traditional vs. DL image recognition technique.

Figure 15.18 Basic structure of natural language processing.

Figure 15.19 Steps of drug discovery using DL.

Figure 15.20 Customer relationship management in DL.

Figure 15.21 Recommendation system structure in DL.

Figure 15.22 Bioinformatics in DL.

Figure 15.23 IoT platform with devices and applications.

Figure 15.24 Different stages of IoT Architecture.

Figure 15.25 IoT programming model.

Figure 15.26 IoT scheduling.

Figure 15.27 IoT memory footprints.

Figure 15.28 Networking in IoT.

Figure 15.29 Some portable devices used in IoT.

Figure 15.30 Energy efficiency in IoT.

Figure 15.31 Example of smart home in IoT.

Figure 15.32 Smart wearables in IoT.

Figure 15.33 IoT applications in a modern car.

Figure 15.34 Industrial Internet of Things (IIoT).

Figure 15.35 Smart city applications of IoT.

Figure 15.36 Agricultural aspects of IoT.

Figure 15.37 IoT in smart retail industry.

Figure 15.38 IoT in energy engagement.

Figure 15.39 Benefits of IoT in healthcare systems.

Figure 15.40 IoT in poultry and farming.

Figure 15.41 DL applications for IoT in healthcare.

Figure 15.42 Key architecture of the DeepSense framework.

Figure 15.43 Performance metrics of heterogeneous human activity recognition...

Figure 15.44 Performance metrics of UserID task with the DeepSense framework...

Figure 15.45 Embedded deep learning.

Chapter 16

Figure 16.1 Flow diagram for proposed work.

Figure 16.2 Life cycle of locust.

Figure 16.3 Schematic steps for locust prevention and destruction.

Figure 16.4 Geographical territory for locust movement.

Figure 16.5 Sensor and relay working procedure with the system.

Figure 16.6 Measurement tracking in different time for a consecutive period....

Chapter 17

Figure 17.1 Snippet of malware dataset.

Figure 17.2 Snippet of intrusion dataset.

Figure 17.3 Malware dataset visualization.

Figure 17.4 Intrusion dataset visualization.

Figure 17.5 Intrusion dataset visualization.

Figure 17.6 Malware detection.

Figure 17.7 Intrusion detection.

Figure 17.8 Malware detection: logistic reg.

Figure 17.9 Malware detection: random forest.

Figure 17.10 Malware detection: neural network.

Figure 17.11 Intrusion detection: neural network.

Figure 17.12 Intrusion detection: logistic regression.

Figure 17.13 Malware detection.

Figure 17.14 Malware predicted.

Figure 17.15 Intrusion detection snippet.

Chapter 18

Figure 18.1 Common phases involved in phishing attack.

Figure 18.2 System architecture.

Figure 18.3 Distribution plot of the features.

Figure 18.4 Traditional neural network.

Figure 18.5 Convolution neural network model.

Figure 18.6 CNN model construction.

Figure 18.7 Experimental results obtained using CNN.

Figure 18.8 Model accuracy and loss using CNN.

Chapter 19

Figure 19.1 (a) Represents the coupling of CPS and (b) represents the feedba...

Figure 19.2 Interaction between machine learning and cyber-physical system o...

Figure 19.3 Applications of CPS.

Chapter 20

Figure 20.1 Problem identification in CPS model.

Figure 20.2 Attributes and functionality of CPS.

Figure 20.3 Types of CPS security.

Figure 20.4 CPS security framework features.

Figure 20.5 CPS security threats.

Guide

Cover Page

Series Page

Title Page

Copyright Page

Preface

Table of Contents

Begin Reading

Index

Also of Interest

WILEY END USER LICENSE AGREEMENT

Pages

ii

iii

iv

xvii

1

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

85

86

87

88

89

90

91

92

93

94

95

96

97

99

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

435

436

437

438

439

440

441

442

443

444

445

446

447

448

Scrivener Publishing100 Cummings Center, Suite 541JBeverly, MA 01915-6106

Artificial Intelligence and Soft Computing for Industrial Transformation

Series Editor: Dr. S. Balamurugan ([email protected])

Scope: Artificial Intelligence and Soft Computing Techniques play an impeccable role in industrial transformation. The topics to be covered in this book series include Artificial Intelligence, Machine Learning, Deep Learning, Neural Networks, Fuzzy Logic, Genetic Algorithms, Particle Swarm Optimization, Evolutionary Algorithms, Nature Inspired Algorithms, Simulated Annealing, Metaheuristics, Cuckoo Search, Firefly Optimization, Bio-inspired Algorithms, Ant Colony Optimization, Heuristic Search Techniques, Reinforcement Learning, Inductive Learning, Statistical Learning, Supervised and Unsupervised Learning, Association Learning and Clustering, Reasoning, Support Vector Machine, Differential Evolution Algorithms, Expert Systems, Neuro Fuzzy Hybrid Systems, Genetic Neuro Hybrid Systems, Genetic Fuzzy Hybrid Systems and other Hybridized Soft Computing Techniques and their applications for Industrial Transformation. The book series is aimed to provide comprehensive handbooks and reference books for the benefit of scientists, research scholars, students and industry professional working towards next generation industrial transformation.

Publishers at ScrivenerMartin Scrivener ([email protected])Phillip Carmical ([email protected])

Convergence of Deep Learning In Cyber-IoT Systems and Security

Edited by

Rajdeep ChakrabortyAnupam GhoshJyotsna Kumar MandalandS. Balamurugan

This edition first published 2023 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA © 2023 Scrivener Publishing LLCFor more information about Scrivener publications please visit www.scrivenerpublishing.com.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

Wiley Global Headquarters111 River Street, Hoboken, NJ 07030, USA

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Limit of Liability/Disclaimer of WarrantyWhile the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials, or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read.

Library of Congress Cataloging-in-Publication Data

ISBN 978-1-119-85721-1

Cover image: Pixabay.ComCover design by Russell Richardson

Preface

Deep learning (also known as deep structured learning) is part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised. Deep-learning architectures such as deep neural networks, deep belief networks, deep reinforcement learning, recurrent neural networks and convolutional neural networks have been applied to fields including computer vision, speech recognition, natural language processing, machine translation, bioinformatics, drug design, medical image analysis, material inspection and board game programs, where they have produced results comparable to and in some cases surpassing human expert performance. Artificial neural networks (ANNs) were inspired by information processing and distributed communication nodes in biological systems. ANNs have various differences from biological brains. Specifically, artificial neural networks tend to be static and symbolic, while the biological brain of most living organisms is dynamic (plastic) and analogue. The adjective “deep” in deep learning refers to the use of multiple layers in the network. Early work showed that a linear perceptron cannot be a universal classifier, but that a network with a nonpolynomial activation function with one hidden layer of unbounded width can. Deep learning is a modern variation which is concerned with an unbounded number of layers of bounded size, which permits practical application and optimized implementation, while retaining theoretical universality under mild conditions. In deep learning the layers are also permitted to be heterogeneous and to deviate widely from biologically informed connectionist models, for the sake of efficiency, trainability and understandability, whence the “structured” part.

Deep learning approaches are now used in every aspect of cyber systems and IoT systems. The main goal of this book is to bring to the fore unconventional cryptographic methods to provide cyber security, including cyber-physical system security and IoT security through deep learning techniques and analytics with the study of all these systems.

Thus, the book covers the evolution of deep learning from machine learning in the very first section. It also provides innovative projects and implementation on deep learning-based solutions in Cyber-IoT systems as well as detailed aspect on deep learning-based security issues in Cyber-IoT systems. Finally, this book also covers cyber physical system concepts and security issues. Therefore, we feel the book is well suited for students (UG & PG), research scholars, and enthusiastic readers who wants a good domain knowledge on deep learning.

Part I: Various Approaches from Machine Learning to Deep Learning

Chapter 1 discuss a method using web-assisted non-invasive detection of oral submucous fibrosis using IoHT. Oral cancer is another big threat and disease of human race. The early detection of oral cancer is of the upmost importance. This chapter starts with a detailed literature on oral cancer then it gives all the primary concepts to understand the proposed system of the authors. Then the chapter gives a proposed model to detect noninvasive oral submucous fibrosis which is web-assisted and for IoHT.

Chapter 2 provides a performance evaluation of machine learning and deep learning, the case study used in this chapter is house price prediction. After the introduction, the chapter starts with a detailed literature review of machine learning and deep learning. The authors used a very standard research methodology to carry out this study, which is discussed next. Authors used Gradient Boosting Regression, Support Vector Regression (SVM), Support Vector Regression (SVM), Multi Output Regression, Regression using Tensorflow – Keras and various classification models for their study. The authors concluded with very good performance analysis and results in both tabular and graphical forms.

Chapter 3 gives why it is important to study cyber physical systems, machine learning & deep learning. First, the author discusses cyber physical systems, machine learning & deep learning in detail. Then the author gives a detailed academic program in cyber physical systems, machine learning & deep learning throughout the world and in India. Finally, the author concludes with the importance to study cyber physical systems, machine learning & deep learning.

Chapter 4 proposes a hybrid model using machine learning techniques and semantic attribute to detect fake news, which is one of the important applications in this cyber and social media world. It uses NLP and applied to various datasets from Facebook, Twitter, Whatsapp, etc. The results are finally compared with existing literature and the proposed model is found to be 93% accurate.

Part II: Innovative Solutions based on Deep learning

Chapter 5 gives an online assessment system using Natural Language Processing (NLP) techniques. It starts with the importance of online assessment in the ‘home from work’ scenario due to the COVID-19 pandemic. The chapter then moves to a detailed literature on online assessments. Thereafter, the authors discuss some algorithms for online assessments. In the next section it proposes a system design for online assessment. Finally, implementation is shown and concluded with it’s novelty.

Chapter 6 gives a reference architecture to build deep Q learning-based intelligent IoT edge solutions. The chapter starts with a detailed overview of machine learning and deep learning. Then the chapter moves to dynamic programming features and deep Q learning in IoT and Azure. Thereafter, the authors give a proposed model and detailed result and analysis.

Chapter 7 provides an improved fuzzy logic-based solutions for air conditioning systems. Then the authors give a proposed system which is composed of Fuzzy variables, Fuzzy base class, Fuzzy Rule Base and Fuzzy rule viewer. The chapter then gives the simulated results of the proposed system and conclusion is drawn based on it’s novelty.

Chapter 8 has an important implementation to detect masked face to combat the pandemic situation. The chapter starts with a detailed related work on masked face recognition. Then it gives all the mathematical preliminaries required to understand the proposed system. Thereafter, it moves to the proposed system with algorithms, methods and applications followed by experimental results. It concludes with the novelty of the proposed system.

Chapter 9 is another deep learning approach to encounter COVID-19 pandemic situation. The chapter starts with the introduction of COVID19 situation and the need of deep learning-based solutions to encounter it. Here, the authors propose a medical imaging solution using deep learning where images of lungs are taken and COVID-19 infection positivity is predicted. The chapter also provides the method of COVID-19 variant tracing and biological protein structure. Moreover, this chapter gives selection drugs combination for a particular COVID-19 patient. It ends with detailed result and analysis of the proposed model.

Chapter 10 provides another online question answering system using Bengali language. It starts with the discussion on the existing literature then it moves towards a problem statement. The authors discuss the proposed model in a very structured way with algorithms. Thereafter, a detailed result and analysis is given and compared with existing work.

The chapter ends with analysis of error, some close observations, applications of the proposed model and scope for improvement as future work.

Part III: Security and Safety Aspects with Deep Learning

Chapter 11 gives a secure access mechanism for smart homes using biometric authentication and RFID authentication which can be implemented for IoT systems. The authors give a structured approach and framework for smart home access method with biometrics. The authors then discuss the same using RFID followed by proposed Control Scheme for Secure Access (CSFSC). The proposed system is discussed with mathematical equations and then it provides result of the proposed system.

Chapter 12 is a MQTT-based implementation of home automation system prototype with integrated Cyber-IoT infrastructure and also discusses deep learning-based security issues. After the introduction, literature review and importance of home automation, the author starts with proposed system architecture of home automation. Then it discusses the various security issues in home automation. Thereafter, the author moves to the implementation part of the proposed system and gives the detailed results with discussion.

Chapter 13 gives a malware detection framework using deep learning. Malware is a risk to the privacy of computer users which can cause an economic loss to organizations. Deep Learning is a subfield of machine learning which concentrates on human brains using artificial intelligence result analysis and conclusion.

Chapter 14 gives an application for women safety, namely “Patron for Women”. The authors first give relative research where the first application is a mobile-based women safety application. Secondly, the authors refer to another application which is android-based. Thirdly, it gives another android-based application namely, “Lifecraft” and, finally, “Abhaya and Sakshi”, another two-women safety applications are discussed. Then it provides a new methodology, system and model with deep learning for women safety. The chapter concludes with result and analysis and novelty.

Chapter 15 discusses concepts & techniques in deep learning applications in the field of IoT systems and security. The chapter starts with a detailed introduction and concepts on deep learning. Then it discusses various techniques used in deep learning such as CNN, RNN, GAN, SOM, autoencoders etc. Thereafter, it gives various deep learning applications followed by a detailed concept on IoT system and it’s applications. Next the authors show the amalgamation of deep learning with IoT. The chapter concludes with the author’s finding deep learning applications in the field of IoT systems and security.

Chapter 16 is an implementation of efficient detection of bioweapons for agriculture sector using narrow band transmitter and composite sensing architecture. The authors start with a detailed literature review and understanding of pest and insects to be discovered using deep learning techniques. Then the authors give a structured working methodology and proposed algorithm followed by block diagram of the proposed method.

Chapter 17 gives a deep learning-based malware and intrusion detection framework. After a detailed literature survey author moves to the proposed work with problem description, working model, data set for deep learning application, deep learning algorithms. Then it gives the implementation with python libraries. Thereafter, a detailed result and analysis is provided and concludes with the accuracy of the system and future work.

Chapter 18 gives phishing URL detection-based on deep learning techniques. This chapter focuses on detecting phishing URL using Convolutional Neural Network. The phishtank dataset is considered and the features of the URL are extracted. Finally, the deep learning classifiers is used to detect the URL is phishing or legitimate URL. The performance of the classifier is evaluated based on the accuracy, precision, recall and F1 score.

Part IV: Cyber Physical Systems

Chapter 19 is an overview and understanding of Cyber Physical System (CPS). After a short introduction the authors start with the architecture and design of the CPS and move to reliability and distribution management in CPS. Thereafter, the authors discuss the security issues in CPS with the role of machine learning and deep learning for providing security in CPS. Finally, the authors conclude with various applications of CPS.

Chapter 20 is dedicated to the security issues, threats and solutions in Cyber Physical Systems. With the introduction of CPS, the author gives the motivation of this work. Then the authors discuss the various characteristics of CPS followed by a detailed understanding of threats in CPS. The authors then give various security mechanism aspects to achieve in CPS. The chapter concludes with how to overcome security issues in CPS, with and without deep learning, and discussion with probable solutions.

The EditorsAugust 2022

Part IVARIOUS APPROACHES FROM MACHINE LEARNING TO DEEP LEARNING

1Web-Assisted Noninvasive Detection of Oral Submucous Fibrosis Using IoHT

Animesh Upadhyaya1*, Vertika Rai2, Debdutta Pal1, Surajit Bose3 and Somnath Ghosh2

1 Department of Computer Science and Engineering, Brainware University, Kolkata, India

2 Department of Allied Health Sciences, Brainware University, Kolkata, India

3 KSD Dental College & Hospital, Kolkata, India

Abstract

Today’s world is connected with the Internet. In our daily life, we are also connected with many things that are connected through the Internet, and that is the way that “Internet of Things” evolves. In every aspect of life, we have applications of IoT, smart home, smart city, smart transport, and many more. Healthcare is also required to be smart to reach maximum patients in minimum time. In this chapter, we focus on a chronic unbearable disease of the oral cavity characterized by infection and progressive fibrosis. This disease may cause an inability to open the mouth, but it is absolutely curable. The faster the disease is diagnosed the rate of curability is high. Our aim is to develop a model with the help of IoHT device for the detection of oral submucous fibrosis in the early stage.

Keywords: Internet of HealthcareThings, noninvasive, oral submucous fibrosis, ThinkSpeak, MQTT, ESP8266

1.1 Introduction

Early 1926, Nikola Tesla envisioned a “Connected World.” He told Colliers Magazine in an interview: “When Wireless is perfectly applied, the whole world will be converted into a huge brain, which in fact it is, all things particles of a real and rhythmic whole and the instruments through which we shall be able to do this will be amazing simple compared with our present telephone, A man will be able to carry one in his vest pocket” [1].

In 1999, Kevin Asthon coined the phrase “Internet of Things” to refer to supply chain management with RFID-tagged or barcoded item (things) bringing increased efficiency and accountability to the organization. As Ashton said in an RFID journal article (June 22, 2009): “If we had computers that knew everything there was to know about things—using data they gathered without any help from us—we would be able to track and count everything and greatly reduce waste, lost and cost. We would know when things needed replacing, repairing, or recalling and whether they were fresher past their best” [1].

Oral cancer (OC) in today’s world is now a menacing health problem and is now accepted as Indian disease, despite advances in oral cancer therapy, it remains as a disease of later diagnosis [2]. Oral submucous fibrosis (OSF) is the sixth most common cancer with the highest malignant transformation rate. Diagnosis of OSF currently consists of clinical examination, followed by biopsy [3]. There is a requirement for minimal invasive markers for the disease that are specific for risk assessment [4]. To achieve this goal, better understanding of complex molecular events that regulate the progression of this disease is required [5]. Early detection not only decreases the incidence and mortality rate but also improves the survival of oral precancer [6]. With the increasing number of OC patients reported, the need for a continuous and throughout observation system is the need of the hour. Pathologists find it difficult to diagnose OC patients at early stages due to overlapping symptoms. In fact, the information medical practitioners get from a patient’s visit is usually very limited [7]. Till now, there is no gadget available that can be utilized to screen oral well-being at an early stage of the disease. The 21st century has become a century of new digital technologies. Digital technologies have started playing a vital role in the medical healthcare sector [8]. Digital technologies are nowadays are used in dentistry to help in the treatment system, enabling the healthcare practitioners for early diagnosis and detection [9]. While using a wide range of sensors protocols, the IoT-based healthcare system delivers the data collected from cancer patients and monitors them in real-time, it becomes easy for patients and oncopathologists, to screen the OC patients, and differentiate between the stages of cancer for early diagnosis. IoT-based systems reduce the workload demanded by the conventional biopsy [10]. Digital technologies have started playing a vital role in the medical healthcare sector [9]. Digital technologies are now a day are used in dentistry to help in the treatment system, enabling the healthcare practitioners for early diagnosis and detection [10]. Many sensors are easy to install to monitor health and maintain real-time data for further check-up and take necessary action in real-time. IoHT based biosensing systems combine a variety of tiny physiological sensors, transmission modules and also sustain high-processing capabilities like processing the data through machine learning and artificial intelligence techniques over the Cloud to give us advanced ratio through the day-to-day monitoring basis and also the whole system is pocket-friendly, unobtrusive health monitoring solutions. While using a wide range of sensors protocols the IoHT-based healthcare system delivers the data collected from cancer patients and monitors them in real-time, it becomes easy for patients and oncopathologists, to screen the OC patients, and differentiate between the stages of cancer for early diagnosis. IoHt-based systems reduce the workload demanded by the conventional biopsy [9].

In this paper, the canvased device detects inflammation and progressive fibrosis of the submucosal tissue by using the sensor MQ135. It examines the whole surface of the mouth in real-time and collects data from every fragmented section of the region using a low-power Wi-Fi chip named ESP 8266 Microcontroller, and sends it to ThingSpeak which is a web-based service as like as an IoT platform that supports REST API protocol. ThingSpeak is a free and open-source IoT application and API that uses the HTTP protocol to store and retrieve data from sensors via the Internet. the data collected and processed is represented as part of an IoT analytics web-server in the form of a graphical visualization system and is accessible by users as a virtual server, and objects communicate with the cloud via “wireless Internet connections” available to users, with the majority of objects relying on sensors to provide environmental analog data. Scripts, such as JSON, XML, and CSV, may display the measurements that were obtained in this manner. the data collected and processed is represented as part of an IoT analytics web-server in the form of a graphical visualization system and is accessible by users as a virtual server, and objects communicate with the cloud via “wireless Internet connections” available to users, with the majority of objects relying on sensors to provide environmental analog data. Scripts such as JSON, XML, and CSV may display the measurements that were obtained in this manner. the data collected and processed is represented as part of an IoT analytics web-server in the form of a graphical visualization system and is accessible by users as a virtual server, and objects communicate with the cloud via “wireless Internet connections” available to users, with the majority of objects relying on sensors to provide environmental analog data. Scripts such as JSON, XML, and CSV may display the measurements that were obtained in this manner. The data collected and processed is represented as part of an IoT analytics web-server in the form of a graphical visualization system and is accessible by users as a virtual system, and objects communicate with the cloud via “wireless Internet connections” available to users, with the vast majority of items relying on sensors to provide conservational analog data. Scripts, such as JSON, XML, and CSV, may display the measurements that were obtained in this manner.

1.2 Literature Survey

1.2.1 Oral Cancer

Oral cancer [OC] among various types of cancer is the imminent problem nowadays in the Indian population, it is the sixth most cancer reported worldwide in population and is found in both sexes. 200,000 new cases are reported annually worldwide, two-thirds of which occur in developing countries including India with a high mortality rate [11]. OC is increasing rapidly in India due to lack of hygiene, alcohol consumption, tobacco chewing, betyl quid, smoking. OC completely can be defined as a malignant neoplasm in the oral cavity, most frequently occurring oral cancer in India occur with its premalignant conditions like oral squamous cell carcinoma (OSCC), oral submucous fibrosis (OSF), erythoplakia (OE), leukoplakia (OLK), lichen planus (OLP) [12].

OC can be defined as uncontrolled cell growth along with lesions in the oral cavity basically it starts from any tissue of the mouth and invades any other neighboring areas in the mouth. The most common presenting features are prolonged ulceration, which does not heal, referred pain, to the ear, difficulty with speaking, and opening the mouth or chewing and ultimately it leads to death. Most people are diagnosed with premalignant lesions like OLK and OE white and red patches covering the oral cavity [13]. These lesions are further subdivided according to their types, homogenous (flat, thin), nonhomogenous (speckled). According to literature surveys, OLK and OSF are the most common occurring lesions in patients with frequent duration of chewing of betylquid, tobacoo, and it contains arecadine, arecoline, tanin, a harmful substrate when it comes in contact with the oral cavity it leads to activation of T cells, macrophages, IL6, TNF-α and other metabolites under its influence fibroblast differentiate and leads to accumulation and formation of collagen in oral mucousa [14]. According to the World Health Organization [WHO], OLK is defined as a white patch or plaque that cannot be characterized clinically or pathologically as any other disease. Erythroplakia is a condition in which bright red, velvety patches, or plaques are seen in the mouth floor, soft palate, ventral tongue, and tonsillar fauces and its malignant transformation rate is 20% to 68% [15]. OLP lesions are a condition in which white papules or white plaques with painful blisters and fine wavy keratotic lines which indicate lichen planus are seen in the buccal mucosa, tongue, and gingiva [16]. The malignant transformation rate is about 2% to 8%. The malignant potential of OLP is an ongoing controversial matter 25. Malignancy determination is very much difficult because oral lichenoid lesions (OLLs) are the same as OLP, and sometimes, premalignant lesions exhibit lichenoid characteristics [17].

OSF is a condition in which the mouth appears rigid and becomes difficult to open (i.e., trismus). It is mostly seen in buccal mucosa and the oral cavity and pharynx may also be affected. Inflammation and progressive fibrosis of the submucosal tissue is the hallmark of the disease. It has the highest malignant transformation rate in comparison to other precancerous stages [18]. There are various clinical symptoms that the patients show at the primary stage of the disease like halitosis also known as oral malodor is the most commonly notice with a patient suffering from any oral disease. Due to pathological or nonpathological reasons, halitosis is formed by volatile molecules. It is very commonly reported in more than 50% population has halitosis [19]. These volatile compounds are composed of alcohols or phenyl compounds, sulphur compounds, aromatic compounds, nitrogen-containing compounds, and ketones [20]. OC is a disease diagnosis is followed by the old conventional methods of diagnosis based on experts’ clinical observation, cytological analysis, and histological observations are still followed, which is time-consuming and less sensitive and accurate [18].

1.3 Primary Concepts

1.3.1 Transmission Efficiency

Most current IoT systems send data to the cloud via hypertext transfer protocol (HTTP) and use the message queuing telemetry transport (MQTT) conventional transport protocols [21, 22]. A unique payload in the HTTP method is sent by every device that delivers data to the cloud. Like a normal web server, the IoT platform gets these data and saves it in a database. As a consequence of this method of communication, IoT devices are constrained by bandwidth, latency, and battery consumption. Consequently, the establishment of a new cloud connection and an HTTP request with various headers, and it is necessary to provide a payload for each identified location, which is usually in XML or JSON format. Using this method, which relies on conventional HTTP requests, entails a significant cost for delivering tiny and frequent payloads, such as sensor readings, which are often only a few bytes long. Firewire, cloud-based infrastructure for IoT platforms sponsored by the European Union and the European Commission, has used this approach in several iterations [23, 24]. Other open and standard solutions will benefit greatly from this approach. Other options include Xively [25], ThingSpeak [26], and Temboo [27], all of which began by providing just HTTP interfaces. MQTT [28] is an old telemetry system that enables a publish-subscribe–based messaging system that allows bidirectional communication between servers and devices. HTTP-based solutions, as described in Yasumoto et al. [21], are considerably less efficient since they need more bandwidth and have higher latency. AmazonWeb Services, for example, is using this protocol as the foundation for its Internet of Things offerings [21]. The aforementioned companies, as well as others like Kaa, Carriots, and Ubidots [29], have been integrating MQTT into their products in recent years.

For a related strategy, ThinkSpeak suggests raw binary connecting networks instead of HTTP pattern overhead or publishing-subscription protocols. In addition to transparent HTTP compatibility, it also defines the payload encoding for better transmission efficiency, something MQTT does not supply (as opposed to MQTT). PSON (https://github.com/thinger-io/Protoson) is an effective encoding method that uses few resources. For strategies with limited resources, for instance, memory or computational efficiency, Protoson was built from the ground up just for them. This approach makes it possible to encode unstructured data in a compact binary format, similar to JSON. To illustrate this point, Figure 1.1 shows the encoding sizes of numerous prominent platform formats, including BSON (Binary JSON), JSON, and MessagePack, as well as the recommended PSON and XML, which surpasses MessagePack in the IoT sector by a little margin. By using well-organized protocols and an encoding method, the proposed system decreases latency when data is directed to the cloud, conserves bandwidth, conserves battery power, and minimizes memory footprint.

Figure 1.1 A comparison of the coding sizes of several formats. With Thinger.io, memory footprint is reduced, bandwidth is saved, and power consumption is reduced by using a PSON coding technique.

1.4 Propose Model

1.4.1 Platform Configuration

This section demonstrates how to model and set up a device on the ThinkSpeak platform. Registration on the ThinkSpeak platform requires the addition of two resources in the admin panel. Since the GasLevelMonitoring is connected to the platform, we must register the device that provides sensor information (in this instance, it is the GasLevelMonitoring), moreover, there is a data bucket for archiving the collected data. However, the fundamental procedures required for the creation of the case study are given here, in greater detail at https://thingspeak.com/channels/new For this task, we will utilize the devices part of the platform’s side menu, which gives us access to our user account’s devices. As soon as we are in this area, we’ll have to add the device by clicking the add device button in the list of devices, which will open a section for recording basic information about the device, such as its identification and credentials. It is critical to save the identification and password information since you will need them later in the code you write for the device. The device’s identification is GasLevelMonitoring, and its API Key is SKP9YQY2CFVNK919.

Figure 1.2 Harvard architectural based wireless communication board.

1.4.2 Harvard Architectural Microcontroller Base Wireless Communication Board

We constructed the Harvard Micro-Controller board to construct the strategic model. Transfer the sensed data to the server through the channel port. The proposed model board is shown in Figure 1.2. The major components of the board are:

NodeMCU ESP8266 Microcontroller

MQ135 Gas Sensor

1.4.2.1 NodeMCU ESP8266 Microcontroller

The ESP8266 Wi-Fi System on Chip (SoC) is used by NodeMCU’s software, while the ESP12E module is used by the device’s hardware. The ESP8266 is a low-cost WiFi module with an integrated WiFi module that utilizes ultra-low power technology. The low power consumption is due to the device’s 32-bit TenSilica L 106 microprocessor. Table 1.1 shows the NodeMCU ESP8266 Microcontroller’s basic specs.

In the ESP 8266, the power-saving design functions in three modes. Active mode, sleep mode, and deep-sleep mode with the RTC clock still operating are the three options. It has a built-in Wi-Fi network that can host an app even if the device does not have one. It also functions as a WiFi adaptor. The ESP8266 WiFi unit supports the 802.11 b/g/n protocol and WiFi direct (P2P), which allows devices from various manufacturers to connect and interact without the need for a wireless access point. Figure 1.3 depicts the NodeMCU ESP8266 Microcontroller integration. Furthermore, owing to its simple interoperability with application-specific devices and sensors, the ESP 8266 offers a wide range of applications in the Internet of Things deployments like the new generation home automation systems. By using the GPIO pins provided on the ESP 8266 module, devices or sensors may be readily integrated, and data from sensors can be quickly accessed by the ESP 8266 and processed to make choices based on the circumstance and demand.

Table 1.1 NodeMCU ESP8266 Specifications.

Specifications

Value

CPU

Tensilica L106 32-bit processor

RAM

36Kb

Clock Speed

80MHz/160MHz

Operating Voltage

3.0V 3.6V

Operating Current

80mA (Average)

Available GPIO Pins

10

Frequency range

2.4 GHz ~ 2.5 GHz

Protocols

802.11 b/g/n (HT20)

Security

WPA/ WPA2

Network Protocol

IPv4, TCP/UDP/HTTP

Figure 1.3 NodeMCU microcontroller.

1.4.2.2 Gas Sensor

The MQ135 gas sensor’s sensitive substance is SnO2, which in pure air has a lower conductivity. When the target polluting gas is present, the conductivity of the sensor increases in lockstep with the gas concentration. Through a simple circuit, users may transform the change in conductivity to the corresponding output signal of gas concentration. MQ135 gas sensor has a high sensitivity to ammonia gas, hydrogen sulphide, benzene series steam, and smoke, among other harmful gases. It can detect kinds of toxic gases and is a kind of low-cost sensor for kinds of applications. The illustration is immediately above. The MQ135’s fundamental test circuit is shown in Figure 1.4. Two voltage inputs are required for the sensor: one for the heater (VH) and another for the circuit voltage (VC). VH is used to power the sensor at its typical operating temperature and may be either DC or AC, while VRL is the voltage across the load resistance RL in series with the sensor. VC delivers detectable voltage to the load resistance RL and should operate on direct current.

Figure 1.4 Basic circuit of gas sensor.

1.4.3 Experimental Setup

We deployed the sensor node to the micro-controller and the node is itself connected with the Cloud Server through Bi-directional communication the working mechanism is shown through the Flowchart of the model in Figure 1.5. When the sensor is in into the mouth and examines the whole surface, then the sensor node detects the Gas and sends the data in real time and constructs a datasheet which helps to identify which part is affected by oral cancer at the end onboard LED confirms the sensor properly detects or not. Continuing this process till the whole surface is examined.

Figure 1.5 Flowchart of noninvasive detection of oral submucous detection device.

1.4.4 Process to Connect to Sever and Analyzing Data on Cloud

In this part, we will demonstrate how to simulate physical device behavior such that data are sent to the appropriate data bucket. The device must be programmed so that it can connect to the platform, read sensor information, and write to the data bucket once it has been set up to allow device connections and a data bucket has been established for data storage. When it comes to programming the device, it will utilize the Arduino environment and the GasLevelMonitoring libraries, both of which need some tweaking to work with the ESP8266 microcontroller on which it is built. When you first add the GasLevelMonitoring to the Arduino environment, to add a new board to the Arduino environment, follow the instructions found at http://docs.Thinger.io/hardware/climaStick. The technique is documented in great depth at that location. All that is left is to model the GasLevelMonitoring gadget and then transmit the data to the destination folder on the Thinger.io platform. ThinkSpeak’s basic programming language has been presented in the preceding section as a method to represent.

The following arguments are specified for Connection:

USERNAME It is the username we selected throughout the registration procedure.

DEVICE ID It is the device’s identification, in this case, GasLevelMonitoring, that we utilize in the registration procedure.

DEVICE CREDENTIAL In our case, SKP9YQY2CFVNK919. It is the device’s credential that we utilize in the registration procedure.

SSID With this method, the device will be connected to this Wi-Fi network, which is called “Meeble.”

SSID PASSWORD is the password for the wireless network interface device which identifies for connection to the Internet for communication, this is what we have here. “MeebleLabs.”

A variable named thing of type GasLevelMonitoring is initialized with the USERNAME, DEVICE ID, and DEVICE API KEY parameters, which symbolizes our module and allows us to connect to the sensor through Wi-Fi and get data from it. Following that, inside the setup method (which is only called once at the start of the application), it is done:

Using the device’s SSID settings and SSID PASSWORD to set up the device’s Wi-Fi connection. The add Wi-Fi function is used to connect to a wireless network.

Open the serial monitor after uploading the code to test whether Wi-Fi is working. Ascertain that the baud rate is 115200. In this case, the gas level will be shown as a percentage and relayed to ThingSpeak if Wi-Fi is available. As shown in

Figure 1.6

, the Arduino software’s serial monitor is in use.

In Figure 1.7, the console of Think Speak is shown with the values in the graph, which define the level of gases in every sort of area of the mouth area.

Figure 1.6 Serial monitor.

Figure 1.7 Think Speak Console.

1.5 Comparative Study

There are several methods for detection of Oral Cancer starting from vital staining by the toluidine blue method to brush biopsy, exfoliative cytology, liquid biopsy, CT Scan, MRI, PET CT, HPLC, different types of spectroscopy, and advanced diagnostic tools like molecular markers [30].

Among all these sophisticated methods incisional biopsy is the gold standard for detection and final diagnosis of oral cancer. Biopsy not only determines the disease but also grades the level of cancer (well-differentiated, moderately differentiated, or poorly differentiated squamous cell carcinoma), depth of malignancy, muscular or perineural invasion also be detected.