A model based monitoring approach for safety critical Cyber-physical systems

Functional safety has become one of the key concerns in the development of adaptive and intelligent cyber-physical systems operating in constantly changing dynamic environments. We propose a general approach for the development of safety-critical cyber-physical systems, which relies on model-based engineering techniques and on the integration of an online simulation module for the verification of relevant safety properties. Having access to formalized safety requirements and to an executable version of the high-level model used for subsystem code generation, this module is able to perform system monitoring and behavioral consistency verification by periodically performing simulation steps based on the observed system states and variables. By verifying the evolution of the system against the expected behavior, this type of verification can detect failures and inconsistent behaviors originating from both systematic and random faults, as is the case with traditional analysis techniques, while also allowing the potential detection of undiscovered issues related to the interaction among different abstraction layers and subsystems. Distributed systems implemented using the IEC 61499 standard for PLC systems are strong candidates for an instantiation of this development approach, given their model of computation and execution semantics. Thus, we discuss an example consisting in the application of this technique to an IEC 61499 supervisory-controlled distributed system.