Deployment optimization for AUTOSAR system configuration

In AUTOSAR system configuration phase, integrators face many challenges in determining the tradeoffs when mapping hundreds of software components to electronic control units (ECU), so finding a common method for this artificial process is very significant. In this paper, we provide an approach to help system integrators building a correct and efficient system configuration. In our method, nested software components are decomposed iteratively to atomic ones. Then all atomic components are divided into two groups: ECU-dependent components and ECU-independent components. The former is deployed onto ECUs first according to its hardware requirements. The latter is deployed to a suitable ECU using evolutionary algorithm. In particular, we focus on the resource requirements, real-time scheduling, and minimization of inter-ECU communication consumption that has an important impact on the non-functional aspects of real-time control system. Experiment results demonstrate the effectiveness of our approach.