MAPE-K patterns for self-adaptation in cyber-physical systems

Cyber-physical systems (CPS) are characterized with their concurrency, heterogeneity and time sensitivity. In this context, it is crucial to have self-adaptive CPS systems in order to manage changing in their internal and external environment and supporting new requirements. Also, CPS systems are known with their restricted connectivity, and consequently, they must perform a decentralized adaptation. This is added to the general complexity of modeling the adaptation process. In fact, the MAPE-K (Monitoring, Analysis, Planning, Execution and Knowledge) control loop model has been identified as a crucial element for realizing self-adaptation in software systems. To respond to the design challenge of self-adaptive CPS, we propose a set of decentralized MAPE-K design patterns to help designer in building such complex systems thanks to software design patterns which provide a general reusable solution to a commonly occurring problem. In this paper, we provide a set of decentralized patterns to model CPS using the refinement technique. For this purpose, we proposed a standard notation based on the UML modeling language to describe the different MAPE-K patterns for decentralized control in self-adaptive CPS. We illustrate our approach by modeling the smart parking application using the coordinated control pattern.