AI-Enabled Threat Detection and Security Analysis for Industrial IoT

Description

This contributed volume provides the state-of-the-art development on security and privacy for cyber-physical systems (CPS) and industrial Internet of Things (IIoT). More specifically, this book discusses the security challenges in CPS and IIoT systems as well as how Artificial Intelligence (AI) and Machine Learning (ML) can be used to address these challenges. Furthermore, this book proposes various defence strategies, including intelligent cyber-attack and anomaly detection algorithms for different IIoT applications. Each chapter corresponds to an important snapshot including an overview of the opportunities and challenges of realizing the AI in IIoT environments, issues related to data security, privacy and application of blockchain technology in the IIoT environment. This book also examines more advanced and specific topics in AI-based solutions developed for efficient anomaly detection in IIoT environments. Different AI/ML techniques including deep representation learning, Snapshot Ensemble Deep Neural Network (SEDNN), federated learning and multi-stage learning are discussed and analysed as well. Researchers and professionals working in computer security with an emphasis on the scientific foundations and engineering techniques for securing IIoT systems and their underlying computing and communicating systems will find this book useful as a reference. The content of this book will be particularly useful for advanced-level students studying computer science, computer technology, cyber security, and information systems. It also applies to advanced-level students studying electrical engineering and system engineering, who would benefit from the case studies. Preface Contents Artificial Intelligence for Threat Detection and Analysis in Industrial IoT: Applications and Challenges 1 Introduction 2 Book Outline References Complementing IIoT Services Through AI: Feasibility and Suitability 1 Introduction 2 IIoT with Edge Intelligence 2.1 The Challenges of EC in IIoT 2.2 Classifications of AI Techniques 2.3 Machine Learning Techniques in IIoT 2.4 Machine Learning Techniques in Edge Computing 2.5 Edge Intelligent IIoT 3 AI-Enhanced Cooperative Computing Architecture 4 Potential Advantages of Learning Techniques in Edge Intelligent IIoT 5 Practical Limitation and Open Issues 5.1 Software Platforms and Middleware 5.2 Task Offloading and Load Balancing 5.3 EI Model Design 5.4 Security Issues 6 Conclusion References Data Security and Privacy in  Industrial IoT 1 Introduction 2 Intrusion Detection in IIoT 3 Authentication Techniques 4 Key Establishment Techniques 4.1 Key Establishment Protocols at Higher Layers 4.1.1 Symmetric Key Establishment Protocols 4.1.2 Asymmetric Key Establishment Protocols 4.2 Key Establishment Protocols at the Physical Layer 4.2.1 Key Establishment Protocols Using channel’s Characteristics 4.2.2 Using Keyless Cryptography 4.3 Cross-Layer Key Establishment Protocols 4.3.1 Cross-Layer Key Establishment Protocols Based on Asymmetric Key Setting 4.3.2 Cross-Layer Key Establishment Protocols Based on Symmetric Key Setting 5 Real IIoT Security Testbeds 6 Conclusion References Blockchain Applications in the Industrial Internet of Things 1 Introduction 2 Industrial Internet of Things 2.1 IIoT Architecture 2.2 IIoT Challenges 3 Blockchain 3.1 Blockchain Structure 3.2 Blockchain Usage in IIoT 4 Blockchain Applications in IIoT 4.1 Smart City 4.2 Manufacturing 4.3 Healthcare 4.0 4.4 Energy Management 4.5 Agriculture 4.0 4.6 Smart Homes 4.7 Autonomous Vehicles 4.8 Multimedia Right Management 5 Analysis 6 Challenges and Open Issues 7 Conclusion References Application of Deep Learning on IoT-Enabled Smart Grid Monitoring 1 Introduction 2 Smart Grid State Estimation 3 Fundamental Concepts for State Estimation Concepts in Active Distribution System 4 State Estimation Problem in Active Distribution Systems 5 Various State Estimation Methods Used in Smart Grid 5.1 Conventional Approach 5.2 Kalman Filter-Based Approaches 6 Learning Based Applications in SGSE 6.1 Support Vector Machine Approaches 6.2 Bayesian Theorem Approaches 6.3 Regression Analysis Approaches 6.4 Artificial Neural Network Approaches 7 Simulation 7.1 Case Study 1 7.2 Case Study 2 8 Discussion References Cyber Security of Smart Manufacturing Execution Systems: A Bibliometric Analysis 1 Introduction 2 Methodology 3 Findings 3.1 Productivity 3.2 Research Areas 3.3 Institutions 3.4 Authors 3.5 Publishers 3.6 Highly Cited Articles 3.7 Keywords Frequency 4 Conclusions References The Role of Machine Learning in IIoT Through FPGAs 1 Introduction 1.1 Industrial Internet of Things (IIoT) 1.2 Challenges of IIot 1.2.1 Security 1.2.2 Connectivity 1.2.3 IIoT Integration 1.2.4 Data Storage 1.2.5 Analytics Challenges 2 Machine Learning 3 FPGAs 4 Case Study 5 Challenges and Open Issues 6 Conclusion References Deep Representation Learning for Cyber-Attack Detection in Industrial IoT 1 Introduction 2 Cyber-Attack Detection 3 Machine Learning (ML) 3.1 Deep Neural Network (DNN) 3.1.1 Autoencoder 3.1.2 Long Short-Term Memory (LSTM) 3.2 Decision Tree (DT) 3.3 K-Nearest Neighbors (KNN) 3.4 Random Forest (RF) 3.5 Support Vector Machine (SVM) 3.6 Naïve Bayes (NB) 3.7 Challenges of Applying ML on IIoT Data 4 The Proposed ML-Based Detection Method 4.1 Data Engineering 4.2 Data Splitting 4.3 Training the Proposed Method 5 Experimental Setup and Evaluation Results 5.1 Dataset 5.2 Attack Scenario 5.3 Evaluation Metrics 5.4 Evaluation Results 6 Conclusion References Classification and Intelligent Mining of Anomalies in Industrial IoT 1 Introduction 2 Anomaly Detection and Its Challenges in IIoT 3 Literature Review for Anomaly Detection in IIoT 4 Discussion 5 Open Challenges and Future Research Directions 5.1 Lack of Training Data Sets 5.2 Real-Time Anomaly Detection 5.3 Adaptive Learning 5.4 Resource and Energy Constraints 5.5 Privacy and Security Concerns 6 Conclusion References A Snapshot Ensemble Deep Neural Network Model for Attack Detection in Industrial Internet of Things 1 Introduction 2 Previous Works in IIoT Security 3 Methodology 3.1 Dataset 3.2 Preprocessing of Data 3.2.1 Features 3.2.2 Replacing Missing/NaN Values 3.3 Snapshot Ensemble Deep Neural Network 3.4 Evaluation Parameters 4 Implementation and Results 4.1 Software and Hardware 4.2 Results 5 Conclusion and Future Work References Privacy Preserving Federated Learning Solution for Security of Industrial Cyber Physical Systems 1 Introduction 2 Cyber-Physical System (CPS) Security 2.1 Major Attacks on Cyber-Physical Systems (CPS) 2.2 Privacy 3 Federated Learning (FL) 3.1 Architectures of Federated Learning (FL) 3.2 Algorithms of FL 3.3 Challenges and Vulnerabilities of FL 3.4 Countermeasures References A Multi-Stage Machine Learning Model for Security Analysis in Industrial Control System 1 Introduction 2 Background 2.1 Gas Pipeline System 2.2 Water Tank Storage System 2.3 Machine Learning Algorithm 2.3.1 Decision Tree 2.3.2 Random Forest 2.3.3 K-Nearest Neighbors 2.3.4 Logistic Regression 2.3.5 Multi-Layer Perceptron Algorithm 3 Literature Review 3.1 Types of Cyber Attacks 3.2 Detection of Cyber Attacks 3.3 Summary 4 Proposed Models 4.1 Dataset Processing 4.2 Machine Learning Model 4.3 Summary 5 Methodology 5.1 Datasets Collection Methodology 5.2 Feature Selection Methodology 5.3 Machine Learning Classifiers 5.4 Summary 6 Results and Discussion 6.1 Model Performance 6.2 Summary 7 Conclusions References A Recurrent Attention Model for Cyber Attack Classification 1 Introduction 2 Previous Work 3 Proposed Approach 3.1 Data Processing and Visualization 3.2 Recurrent Neural Network (RNN) 3.3 Reinforcement Learning (RL) 3.4 Recurrent Attention Model (RAM) 4 Experimental Analysis 5 Results 5.1 IoT Dataset 5.2 BATADAL Dataset 6 Discussion References