Functional feasibility analysis of variability-intensive data flow-oriented applications over highly-configurable platforms

Data-flow oriented embedded systems, such as automotive systems used to render HMI (e.g., instrument clusters, info-tainments), are increasingly built from highly variable specifications while targeting different constrained hardware platforms configurable in a fine-grained way. These variabilities at two different levels lead to a huge number of possible embedded system solutions, which functional feasibility is extremely complex and tedious to predetermine. In this paper, we propose a tooled approach that capture high level specifications as variable dataflows, and targeted platforms as variable component models. Dataflows can then be mapped onto platforms to express a specification of such variability-intensive systems. The proposed solution transforms this specification into structural and behavioral variability models and reuses automated reasoning techniques to explore and assess the functional feasibility of all variants in a single run. We also report on the validation of the proposed approach. A qualitative evaluation has been conducted on an industrial case study of automotive instrument cluster, while a quantitative one is reported on large generated datasets.