Security and Resilience in Cyber-Physical Systems: Detection, Estimation and Control

Description

This book discusses the latest advances in cyber-physical security and resilience of cyber-physical systems, including cyber-attack detection, isolation, situation awareness, resilient estimation and resilient control under attack. It presents both theoretical results and important applications of the methods. Security and Resilience in Cyber-Physical Systems begins by introducing the topic of cyber-physical security, covering state-of-the-art trends in both theory and applications, as well as some of the emerging methodologies and future directions for research. It then moves on to detail theoretical methods of attack detection, resilient estimation and control within cyber-physical systems, before discussing their various applications, such as power generation and distribution, autonomous systems, wireless communication networks and chemical plants. Focusing on the detection of and accommodation to cyber-attacks on cyber-physical systems, and including both estimation and artificial-intelligence-based methods, this book will be of interest to researchers, engineers and graduate students within the fields of cyber-physical security and resilient control. Preface Contents Contributors 1 Overview 2 Introduction to Cyber-Physical Security and Resilience 2.1 Introduction 2.2 Cyber-Physical Security and Resilience Functionality Overview 2.3 Cyber-Physical Security Versus Adjacent Fields 2.3.1 Cyber-Physical Security Versus Cyber-Security 2.3.2 Cyber-Physical Security Versus FDII 2.3.3 Cyber-Physical Security Versus Prognostics 2.4 Attack Detection, Isolation, and Identification 2.4.1 Model-Based ADII 2.4.2 Data-Driven ADII 2.5 Attack Resilience 2.6 Resilient Estimation 2.6.1 State of the Art on Resilient and Secure Estimation: A Glimpse on Existing Methods 2.7 Resilient Control 2.7.1 Centralized Secure Control 2.7.2 Distributed Secure Control 2.7.3 Resource-Aware Secure Control References 3 Fundamental Stealthiness–Distortion Trade-Offs in Cyber-Physical Systems 3.1 Introduction 3.2 Preliminaries 3.3 Stealthiness–Distortion Trade-Offs and Worst-Case Attacks 3.3.1 Open-Loop Dynamical Systems 3.3.2 Feedback Control Systems 3.4 Simulation 3.5 Conclusion References 4 Predictive Situation Awareness and Anomaly Forecasting in Cyber-Physical Systems 4.1 Introduction 4.2 Forecasting Framework 4.2.1 Digital Twin Simulation Platform 4.2.2 Anomaly Forecasting Approaches 4.2.3 Dimensionality Reduction 4.2.4 Forecasting Process 4.2.5 Feature Discovery 4.3 Ensemble Forecasting 4.3.1 Ensemble Modeling in Feature Space 4.3.2 Adjusting Cluster Centroids to Physical Points 4.3.3 Dynamic Modeling 4.3.4 Dynamic Ensemble Forecast Averaging 4.3.5 Receding Horizon Anomaly Forecast 4.3.6 Committed Horizon Anomaly Forecast 4.4 Predictive Situation Awareness 4.5 Simulation Results 4.6 Conclusions References 5 Resilient Observer Design for Cyber-Physical Systems with Data-Driven Measurement Pruning 5.1 Notation 5.2 Introduction 5.3 Concurrent Models 5.3.1 Physical Model and Monitor 5.3.2 Threat Model 5.3.3 Data-Driven Auxiliary Measurement Prior 5.3.4 Prior Pruning 5.4 Pruning-Based Resilient Estimation 5.4.1 Unconstrained ell1 Observer 5.4.2 Resilient Pruning Observer 5.5 Simulation Results 5.5.1 Resilient Power Grid 5.5.2 Resilient Water Distribution System 5.5.3 Resilient Wheeled Mobile Robot 5.6 Conclusion References 6 Framework for Detecting APTs Based on Steps Analysis and Correlation 6.1 Introduction 6.1.1 Targeted APT Attack on CPSs 6.1.2 Safety of Cyber-Physical Systems (CPSs) 6.1.3 Organization of Book Chapter 6.2 Advanced Persistent Threats (APTs) 6.2.1 Characteristics of APTs 6.2.2 Life Cycle of APTs Attack 6.2.3 Related Work 6.3 APT Detection Framework 6.3.1 Architectural Design of APT-DASAC 6.3.2 Three Layers of APT-DASAC 6.4 Implementation of APT-DASAC Approach 6.4.1 Implementation Setup 6.4.2 Implementation Dataset 6.5 Experimental Evaluation of APT-DASAC Approach 6.5.1 Result and Discussion 6.6 Conclusion References 7 Resilient State Estimation and Attack Mitigation in Cyber-Physical Systems 7.1 Introduction 7.1.1 Literature Review 7.2 Problem Formulation 7.2.1 Attack Modeling 7.2.2 System Description 7.2.3 Security Problem Statement 7.3 Resilient State Estimation 7.3.1 Multiple-Model State and Input Filtering/Estimation Algorithm 7.3.2 Properties of the Resilient State Estimator 7.3.3 Fundamental Limitations of Attack-Resilient Estimation 7.4 Attack Detection and Identification 7.4.1 Attack Detection 7.4.2 Attack Identification 7.5 Attack Mitigation 7.6 Simulation Examples 7.6.1 Benchmark System (Signal Magnitude , Location Attacks) 7.6.2 IEEE 68-Bus Test System (Mode and Signal Magnitude Attacks) 7.7 Conclusion References 8 State and Attacks Estimation for Nonlinear Takagi–Sugeno Multiple Model Systems with Delayed Measurements 8.1 Introduction 8.1.1 Contributions and Outline 8.1.2 Chapter Organization 8.2 Problem Statement 8.2.1 False Data Injection Attacks on Actuators/Sensors 8.2.2 Polytopic Modeling of Time-Varying Nonlinear Systems with Delayed Measurements 8.2.3 Polytopic Modeling of Time-Varying Parameters (Malicious Attacks) 8.2.4 LPV Model of Physical Plant Under Data Deception Attacks and Delayed Measurements 8.3 Main Result: Observer Design 8.4 Numerical Simulation 8.4.1 LPV Representation of The Process 8.4.2 Date Deception Attacks Representation on The Actuator/Sensor 8.4.3 Simulation Results 8.5 Conclusions References 9 Secure Estimation Under Model Uncertainty 9.1 Introduction 9.1.1 Overview and Contributions 9.1.2 Related Studies 9.2 Data Model and Definitions 9.2.1 Attack Model 9.2.2 Decision Cost Functions 9.3 Secure Parameter Estimation 9.4 Secure Parameter Estimation: Optimal Decision Rules 9.5 Case Studies: Secure Estimation in Sensor Networks 9.5.1 Case 1: One Sensor Vulnerable to Causative Attacks 9.5.2 Case 2: Both Sensors Vulnerable to Adversarial Attacks 9.6 Conclusions References 10 Resilient Control of Nonlinear Cyber-Physical Systems: Higher-Order Sliding Mode Differentiation and Sparse Recovery-Based Approaches 10.1 Introduction 10.2 Mathematical Modeling 10.2.1 Problem Statement 10.3 Preliminary: Sparse Recovering Algorithm 10.4 Attack Reconstruction When the Number of Potential Attacks is Greater Than the Number of Sensors 10.4.1 System Transformation 10.4.2 Attack Reconstruction 10.5 Attack Reconstruction When the Number of Sensors is Greater Than the Number of Potential Sensor Attacks 10.5.1 State Attack Reconstruction 10.5.2 Sensor Attacks Reconstruction 10.6 Case Study: Cyber Attack Reconstruction in the US Western Electricity Coordinating Council Power System 10.6.1 Mathematical Model of Electrical Power Network 10.6.2 Transformation of DAE to ODE 10.6.3 Parameterization of Mathematical Model of Western Electricity Coordinating Council Power System 10.6.4 Reconstruction of Attacks via Sparse Recovery Algorithm: The Number of Potential Attacks is Greater Than the Number of Sensors 10.6.5 Reconstruction of Attacks and Estimation of States: The Number of Sensors is Greater Than the Number of Potential Sensor Attacks 10.7 Conclusions References 11 Resilient Cooperative Control of Input Constrained Networked Cyber-Physical Systems 11.1 Introduction 11.1.1 Notation 11.1.2 Preliminaries on Algebraic Graph Theory 11.1.3 Preliminaries on Finite-Time Stability 11.2 Input Constrained Robust Consensus Tracking for High-Order NCPS 11.2.1 Problem Formulation 11.2.2 Input Constrained Robust Consensus Tracking with a Static Leader 11.2.3 Input Constrained Robust Consensus Tracking with a Dynamic Leader 11.2.4 Output-Based Input Constrained Robust Consensus Tracking 11.3 Input Constrained Robust Finite-Time Consensus Tracking for High-Order NCPS 11.3.1 Problem Formulation 11.3.2 Input Constrained Robust Finite-Time Consensus Tracking with Relative State Measurements 11.3.3 Input Constrained Robust Finite-Time Consensus Tracking with Relative Output Measurements 11.4 Numerical Examples 11.4.1 Input Constrained Robust Consensus Tracking for High-Order NCPS 11.4.2 Input Constrained Robust Finite-Time Consensus Tracking for High-Order NCPS 11.5 Conclusions References 12 Optimal Subsystem Decomposition and Resilient Distributed State Estimation for Wastewater Treatment Plants 12.1 Introduction 12.2 Model Description of Wastewater Treatment Plants 12.3 Subsystem Decomposition 12.4 Resilient Distributed State Estimator Design 12.5 Simulation 12.5.1 Subsystem Decomposition 12.5.2 Resilient Distributed State Estimator Design 12.6 Conclusion References 13 Cyber-Attack Detection for a Crude Oil Distillation Column 13.1 Introduction 13.1.1 Preliminary 13.1.2 Cyber-Security of Distillation Column 13.2 Distillation Column Design and Modeling 13.2.1 Plant Data 13.2.2 Distillation Column Design 13.2.3 Dynamic Model of the Distillation Column 13.2.4 Control of Distillation Column 13.3 Testbed Design 13.4 Attack Modeling 13.5 Attack Detection Algorithm 13.5.1 UKF Based Attack Detection 13.5.2 Detector Design 13.6 Results 13.6.1 Attack on Distillate Purity Measurement 13.6.2 Attack on Bottoms Impurity Measurement 13.6.3 Attack on Reflux Flow Rate 13.6.4 Attack Case Summary 13.7 Conclusion and Future Work References 14 A Resilient Nonlinear Observer for Light-Emitting Diode Optical Wireless Communication Under Actuator Fault and Noise Jamming 14.1 Introduction 14.2 LED-Based Optical Channel Modeling 14.2.1 System Setup 14.2.2 Luminous Flux Model 14.2.3 Model Calibration 14.2.4 State-Space and Output Measurement Equations 14.3 LED System Model Representation Under Actuator Fault and Noise Jamming Attack 14.4 Resilient Observer-Based Tracking Control Design 14.4.1 Problem Formulation 14.4.2 Unknown Input Observer Design Method 14.4.3 Feasibility of (14.24) for Non-monotonic Outputs 14.4.4 A Switched-Gain-Based Observer Solution 14.4.5 Reference Trajectory Tracking Design 14.5 LED Application Under Actuator Fault and Noise Jamming Attack on the Optical Communication Channel 14.6 Conclusion References Index