From offline toward real-time : a hybrid systems model checking and CPS co-design approach for medical device plug-and-play (MDPnP)

摘要

Hybrid systems model checking is a great success in guaranteeing the safety of computerized control cyber-physical systems (CPS). However, when applying hybrid systems model checking to Medical Device Plug-and-Play(MDPnP) CPS, we encounter two challenges due to the complexity of human body: i) there are no good offline differential equation based models for many human body parameters, ii) the complexity of human body can result in many variables, complicating the system model. In an attempt to address the challenges, we propose to alter the traditional approach of offline hybrid systems model checking of time-unbounded (i.e., long-run) future behavior to online hybrid systems model checking of time-bounded (i.e., short-run) future behavior. According to this proposal, online model checking runs as a real-time task to prevent faults. To meet the real-time requirements, certain design patterns must be followed, which brings up the co-design issue. We propose two sets of system co-design patterns for hard real-time and soft real-time respectively. To evaluate our proposals, a case study on laser tracheotomy MDPnP is carried out. The study shows the necessity of online model checking. Furthermore, test results based on real-world human subject trace show the feasibility and effectiveness of our proposed co-design.