Mapping real-time communicating tasks on a distributed IMA architecture

Current avionics architectures implemented on large aircraft use complex processors, which are shared by many avionics applications according Integrated Modular Avionics (IMA) concepts. Using less complex processors on smaller aircraft such as helicopters leads to a distributed IMA architecture. Allocation of the avionics applications on a distributed architecture has to deal with two main challenges. A first problem is about the feasibility of a static allocation of partitions on each processing element. The second problem is the worst-case end-to-end communication delay analysis: due to the scheduling of partitions on processing elements which are not synchronized, some allocation schemes are not valid. This paper first presents a mapping algorithm using an integrated approach taking into account these two issues. In a second step, we evaluate, on a realistic helicopter case study, the feasibility of mapping a given application on a variable number of processing elements. Finally, we present a scalability analysis of the proposed mapping algorithm.