[Dissertation] Co-Design of Resilient Timing-Constrained Cyber-Physical Systems

Abstract
Acknowledgments
List of Tables
List of Figures
Chapter 1. Introduction
	1.1. Design Challenges
		1.1.1. Architectural Complexity
		1.1.2. Functionality Complexity
		1.1.3. Timing Requirements
		1.1.4. Design Objectives and System Metrics
	1.2. Related Work
		1.2.1. Schedulability Analysis
		1.2.2. System Design and Analysis
		1.2.3. Vehicle Security
	1.3. Our Workflow and Contributions
Chapter 2. Holistic Optimization Framework
	2.1. Software and Hardware Layer Abstraction
	2.2. Optimization Framework
	2.3. System Analysis Models
		2.3.1. Computation Model for Hard Real-time Systems
		2.3.2. Computation Model for Weakly-Hard Real-Time Systems
		2.3.3. Fault-Tolerance Model
Chapter 3. Cross-Layer Codesign for Hard Real-Time System
	3.1. System Model and Formulation
	3.2. Problem Formulation and Optimization
	3.3. Automotive Experimental Results
		3.3.1. Extensibility vs. Latency
		3.3.2. Error Coverage vs. Latency
		3.3.3. Extensibility vs. Error Coverage
		3.3.4. Optimization of All Three Objectives
		3.3.5. Impact of ECU Speed and Number of Tasks
Chapter 4. Co-design of Weakly-Hard System with Timing and Functional Guarantee
	4.1. Security-Driven Codesign for Weakly-Hard Real-Time Embedded Systems
		4.1.1. System Model
		4.1.2. Problem Analysis and Formulation
		4.1.3. Optimization Algorithm for Codesign
		4.1.4. Experimental Results
	4.2. Fault-Tolerance Codesign with Functional and Timing Guarantees
		4.2.1. System Modeling and Problem Formulation
		4.2.2. Meta-Heuristic ALgorithm for Design Space Exploration
		4.2.3. Experimental Results
	4.3. Summary
Chapter 5. End-to-End Analysis and Adaptation For Automated Lane Centering Application
	5.1. Automated Lane Centering System Pipeline
		5.1.1. DNN-based Perception Module
		5.1.2. Planning and Control Modules
	5.2. Attack Model
	5.3. Preliminary Study of Dirty Road Patch Attack
	5.4. Our End-to-end Uncertainty-based Mitigation Approach for Adversarial Attacks
		5.4.1. Perception Confidence as Signal of Attack
		5.4.2. Uncertainty Estimation and Safety Bound
		5.4.3. Uncertainty-aware Trajectory Planner
		5.4.4. Adaptive Controller
	5.5. Experimental Results
		5.5.1. Results for Uncertainty Estimation
		5.5.2. Mitigation Results for Open Dataset
		5.5.3. Mitigation Results in LGSVL-OpenPilot Environment
	5.6. Summary
Chapter 6. Conclusions and Future Works
	6.1. Conclusion
	6.2. Future Directions
References