Enhancing Fault Diagnosis in Safety-Critical Cyber-Physical Systems

Cyber-Physical Systems (CPSs) present increasingly complex challenges, particularly within safety-critical domains. Ensuring the correctness of safety-critical CPSs is crucial for maintaining their viability and full operational capacity, as undetected failures can be not only costly but also life-threatening. The timely detection and diagnosis of faults are essential in the CPS development cycle, given the prohibitively high costs of rectifying faults post-hardware integration. Consequently, thorough testing of CPSs becomes paramount, underscoring the necessity for advanced fault diagnosis procedures that can accurately detect, explain, and rectify faults. In this thesis, we develop innovative tools and methods tailored for CPSs, aimed at facilitating swift and accurate fault diagnosis. Our contributions encompass fault injection, global optimization, search-based testing for fault localization, mutation testing against formal properties, and signal feature coverage-based testing. This holistic approach offers significant advantages over existing methods, ensuring thorough testing across various scenarios. The proposed solutions have been evaluated using publicly available benchmarks from diverse domains, demonstrating the applicability and effectiveness of our research. The findings of this thesis advance the knowledge in CPS testing, paving the way for more reliable and safe CPSs in the future.