Computer-aided Architecture Design & Optimized Implementation of Distributed Automotive EE Systems

Complexity of automotive EE systems has grown exponentially during the past decade. Model-based function design and simulation are widely accepted and used by the Automotive EE community of today, to face some of the challenges. Emerging new standards - like AUTOSAR - bring standardized interfaces to low-level software, reducing cost and most importantly introducing a new level of abstraction to function implementers. Ease of software component reuse is another novel benefit brought by AUTOSAR. However, there is little guidance on how to translate the results of model-based function design into robust and efficient system implementations in a highly distributed environment. This paper outlines a proposed system level design methodology, supported by a set of point-tools, to bridge the Implementation Abyss. The flow covers: model transformation, architecture design, scheduling of networks and tasks distributed across multiple ECUs, harness design - followed by metrics and criteria-based evaluation of the resulting architecture. Automated creation of AUTOSAR-compliant configuration files for ECU generation is provided to complete the process.