Investigating Robustness in Cyber-Physical Systems: Specification-Centric Analysis in the face of System Deviations
Changjian Zhang, Parv Kapoor, Romulo Meira-Goes, David Garlan, Eunsuk Kang, Akila Ganlath, Shatadal Mishra, Nejib Ammar
TL;DR
The paper tackles robustness in cyber-physical systems by introducing specification-based robustness, where a controller must satisfy a high-level STL requirement under parametric deviations. It defines robustness and strict satisfaction, then recasts robustness verification as robustness falsification, proposing a two-layer simulation framework that combines an upper-layer optimization (e.g., CMA-ES, random search) with a lower-layer CPS falsifier across diverse simulation platforms. A CPS robustness benchmark with eight configurable systems and both classic and RL controllers is developed to evaluate the approach, and experiments show that CMA-ES can efficiently locate minimal deviations that breach specifications while random search often uncovers more violations but at larger deviations. The work demonstrates the framework’s ability to compare robustness across controllers and to reveal the structure of falsification search spaces, offering practical insights for controller design, retraining, and runtime monitoring in CPS.
Abstract
The adoption of cyber-physical systems (CPS) is on the rise in complex physical environments, encompassing domains such as autonomous vehicles, the Internet of Things (IoT), and smart cities. A critical attribute of CPS is robustness, denoting its capacity to operate safely despite potential disruptions and uncertainties in the operating environment. This paper proposes a novel specification-based robustness, which characterizes the effectiveness of a controller in meeting a specified system requirement, articulated through Signal Temporal Logic (STL) while accounting for possible deviations in the system. This paper also proposes the robustness falsification problem based on the definition, which involves identifying minor deviations capable of violating the specified requirement. We present an innovative two-layer simulation-based analysis framework designed to identify subtle robustness violations. To assess our methodology, we devise a series of benchmark problems wherein system parameters can be adjusted to emulate various forms of uncertainties and disturbances. Initial evaluations indicate that our falsification approach proficiently identifies robustness violations, providing valuable insights for comparing robustness between conventional and reinforcement learning (RL)-based controllers