Toward online hybrid systems model checking of cyber-physical systems' time-bounded short-run behavior

摘要

Many Cyber-Physical Systems (CPS) are highly nondeterministic. This often makes it impractical to model and predict the complete system behavior. To address this problem, we propose that instead of offline modeling and verification, many CPS systems should be modeled and verified online, and we shall focus on the system's time-bounded behavior in short-run future, which is more describable and predictable. Meanwhile, as the system model is generated/updated online, the verification has to be fast. It is meaningless to tell an online model is unsafe when it is already out-dated. To demonstrate the feasibility of our proposal, we study two cases of our ongoing projects, one on the modeling and verification of a train control system, and the other on a Medical Device Plug-and-Play (MDPnP) application. Both cases are about safety-critical CPS systems. Through these two cases, we exemplify how to build online models that describe the time-bounded short-run behavior of CPS systems; and we show that fast online modeling and verification is possible.