Concurrent Architecture and Schedule Optimization of Time-triggered Automotive Systems

摘要

This paper presents a methodology for the concurrent optimization of the architecture and scheduling of upcoming synchronous time-triggered automotive systems. A fully synchronous time-triggered system is highly predictable and therefore the best candidate for safety and drive-by-wire functions with strict real-time constraints. While the architecture of these systems has to be optimized in terms of resource allocation, task mapping, and message routing by taking multiple conflicting objectives into account, the scheduling has to be carried out such that application deadlines are satisfied. In case of stringent real-time constraints, available approaches that treat architecture optimization and scheduling as separate problems become inapplicable as most architectures do not permit a feasible schedule. As a remedy, a novel and efficient approach based on conflict refinement is presented. For a given architecture, either a schedule might be obtained or a conflict refinement is performed to determine and exclude the architecture decision that prevents a feasible schedule. In this paper, an extended architecture model is presented and the scheduling and refinement approaches are given for time-triggered architectures based on the FlexRay protocol. This approach can be extended to other protocols and scenarios. A case study of a realistic time-triggered system gives evidence of the efficiency of the proposed approach, solving a large design problem from the automotive domain.