Modeling foundations for executable model-based testing of self-healing cyber-physical systems

Self-healing cyber-physical systems (SH-CPSs) detect and recover from faults by themselves at runtime. Testing such systems is challenging due to the complex implementation of self-healing behaviors and their interaction with the physical environment, both of which are uncertain. To this end, we propose an executable model-based approach to test self-healing behaviors under environmental uncertainties. The approach consists of a Modeling Framework of SH-CPSs (MoSH) and an accompanying Test Model Executor (TM-Executor). MoSH provides a set of modeling constructs and a methodology to specify executable test models, which capture expected system behaviors and environmental uncertainties. TM-Executor executes the test models together with the systems under test, to dynamically test their self-healing behaviors under uncertainties. We demonstrated the successful application of MoSH to specify 11 self-healing behaviors and 17 uncertainties for three SH-CPSs. The time spent by TM-Executor to perform testing activities was in the order of milliseconds, though the time spent was strongly correlated with the complexity of test models.