A Model Driven-Based Approach for Managing Unanticipated Runtime Adaptation of RTE Systems

Abstract

Nowadays, Real-Time Embedded Systems (RTES) and particularly adaptive ones are gaining more use in several domains. They must be always consistent and available to preserve their usefulness and feasibility, as a response to environmental conditions, resources limitations and user requirements. For this reason, they are subject to anticipated and unanticipated runtime reconfigurations. The design of such systems requires an abstract representation of the execution context independently from low-level details, with lower complexity. In this context, designers adopt runtime models which need to be checked and validated at early stages of development. Besides, these models must be updated according to unpredictable execution changes. In the present paper, we propose an MDE-based approach that allows managing runtime adaptive RTES. Our proposal starts by specifying the behavior and structure of runtime adaptive systems. Then Model-to-Text (M2T) transformations allows us to generate configuration files to simulate adaptive systems behavior and evaluate their real-time constraints. An additional feature offered by our approach consists to monitor the execution context and update runtime models according to unanticipated reconfigurations.