Model-based Construction and Verification of Cyber-Physical Systems

Abstract

By combining communication, computation, and control (3C), Cyber-Physical Systems (CPS)0tightly couple the physical world with the cyber-world, to enable more applications, enhance performance, increase dependability and etc. Among these goals, as CPS are widely used in the safety-critical area, guaranteeing the basic dependability/safety is after all the prerequisite and often the top concern. However, the behavior of CPS is extremely complex. First of all, due to the existence of both discrete control modes transition and continuous real-time behavior in CPS, the behavior of CPS is a complex hybrid state space, which is difficult to understand and handle. Secondly, most CPS applications are working in the open environment and acquiring real-time data from the environment intensively to adjust their own behavior. The dynamic environment makes the behavior space more complex to reason. When a system is too complex to analyze directly, building an abstract model of the system and then conducting analysis on the model to answer questions about the original system is an important and widely-used method. Meanwhile, a reasonable model also plays important roles in the phase of specification, design, development, testing, monitoring and so on. Therefore, it is an important topic of investigating how model-based methods can be applied in the context of CPS to increase the quality and dependability of the system. During the past decade, our research group at Nanjing University has devoted a lot of efforts into this mission. We conducted comprehensive research in a wide spectrum of CPS including model-driven design, verification, control, monitoring, and testing. In this paper, we will make a general review of the progress we made on these directions recently.