Bridging the gap between practical cases and temporal performance analysis: a models repository-based approach

Despite various analysis models and tests proposed by the real-time community to validate the temporal performance of hard real-time systems, several cases derived from industrial practices do not find their corresponding tests in standard schedulability analysis tools. They need to be transformed and adapted in terms of qualitative and quantitative information while remaining conservative. The current state-of-practice of transformation relies solely on designers expertise. As a result, the adaptation/transformation task tends to be time- and effort-intensive. This paper proposes to capitalize adaptation efforts by storing them in a repository, hence designers can be helped to automatically transform and adapt their practical designs to analyzable models while staying conservative. Thanks to model-driven engineering settings the capitalization is illustrated by the implementation of a repository as a decision support and the automatic adaptation relies on model-based transformations. Also, a case study is presented to stress the importance of our contribution.