Uncertainty Modeling and Quantitative Evaluation of Cyber-physical Systems

Abstract

Cyber-physical System (CPS) represents a system that tightly integrates computation, communication, and physical processes. As an effective modeling language, AADL is often applied for real-time and embedded systems. However, AADL has limitations in modeling stochastic events because the interaction between the system and an uncertain external environment is often complex and unpredictable. In this paper, we propose a stochastic hybrid modeling language based on AADL, called SHML. SHML supports both continuous behavior analysis and probabilistic modeling of CPSs. To achieve the verification objective, we present a set of mapping rules to transform the SHML design into networks of stochastic hybrid automata (NSHA). By using statistical model-checking techniques, the obtained NSHA model and performance queries are jointly applied to evaluate the quantitative performance of SHML designs. Experiments on traffic collision avoidance systems are conducted, and the results demonstrate the usability and effectiveness of our approach.