Modeling and verifying self-adaptive systems: A refinement approach

Abstract

Self-adaptive systems are able to modify their behavior and/or structure to deal with their continuously changing environment and internal dynamics. MAPE control loops, based on these four steps: Monitoring, Analysis, Planning, and Execution, have been identified as crucial elements in realizing self-adaptation of software systems. Adaptive systems are generally more difficult to design, specify and verify due to their high complexity. Ensuring the correctness of the system's adaptation logic is very crucial. In this paper, we propose a refinement approach that aims first to model step-by-step self-adaptive systems based on MAPE control loop. Second, our approach aims to formally specify self-adaptive systems at a high level of abstraction using the Event-B method. This formal specification provides a way to verify several properties for self-adaptive systems such as safety, reachability and temporal constraints. We illustrate our approach by verifying the fire detection system that exhibits a self-adaptive behavior.