Efficient Scheduling, Mapping and Memory Bandwidth Allocation for Safety-Critical Systems

In safety-critical multi-core architectures, isolation becomes more challenging as multiple components are shared between the cores, which generates delays in execution time called interferences. To reduce buses contention, it is important to limit the bandwidth allocated to each application. Current state-of-the-art solutions for bus budgeting allocation are not scalable and do not provide the correct level of flexibility to model complex systems. To cope with these challenges, we propose a novel heuristic to generate the schedule, mapping and bus budget allocation. We achieve scalable execution times up to 45 times faster compared to the state of the art. We further introduce real-life constraints such as application dependencies, forced mapping and release time. Compared to the state of the art, our approach is applicable in complex systems.