Dependability Modelling and Evaluation of Cyber-Physical Systems: A Model-Driven Perspective

The strong coupling of software and physical processes in the emerging field motivates the development of new methods to respond to dependability in both the cyber and physical domains. Two kinds of definitions of Cyber-Physical System (CPS) from the view of resource, service and process is proposed at first. In addition, the dependable Model of Cyber-Physical Home Control System is presented by taking advantage of Generalized Stochastic Petri Net(GSPN) . Based on the model running and analyzing, the dependency relationships in resource, service and process are analyzed and the dependability parameters (Reliability, Availability, MTTF, etc.) of system which people pay attention to are presented finally. If this model and method are applicable to CPS, the formal methods and verification based on model-driven are appropriate to CPS. The simulation results show that the chances of resource faults are larger than the service error; the effects of a non-detected error perceived from resource level are less than that for the service level. KeywordsDependability; Model-driven; Cyber-physical System; Stochastic Petri Nets