Towards Systematic Model-based Testing of Self-adaptive Software

Self-adaptive software reconfigures automatically at runtime to environment changes in order to fulfill its specified goals. Thereby, the system runs in a control loop which includes monitoring, analysis, adaptation planning, and execution. To assure functional correctness and non-functional adequacy, testing is required. When defining test cases, the control loop’s tasks have to be validated as well as the adapted system behavior that spans a much more complex decision space than in static software. To reduce the complexity for testers, models can be employed and later be used to generate test cases automatically—an approach called Model-based Testing. Thereby, a test modeler has to specify test models expressing the system’s externally perceivable behavior. In this paper, we perform a Failure Mode and Effects Analysis on a generic perspective on self-adaptive software to figure out the additional requirements to be coped with in test modeling. Keywords—self-adaptive software; problem statement; model-based testing; failure mode and effects analysis