Mixed Harmonic Runnable Scheduling for Automotive Software on Multi-Core Processors

The performance of automotive electronic control units (ECUs) has improved following the development of multi-core processors. These processors facilitate fast computing performance without increasing clock speed. System developers partition automotive application runnables to have parallelizability and avoid interference between various software modules. To improve the performance of such systems, an efficient scheduler is necessary. In this regard, for multi-core ECUs, the automotive open system architecture (AUTOSAR) suggests partitioned static priority scheduling for parallelized software. In the AUTOSAR approach, clustering and partitioning of runnables for specific cores becomes difficult, but there is no exact criterion followed for partitioning the runnables. Consequently, cores are not balanced against loads, and under contingency conditions, there is a chance that tasks will miss deadlines. In this study, we address this problem by exploring a mixed harmonic runnable scheduling algorithm that includes partitioned scheduling. We tested this algorithm using high load conditions under contingency consequences, and we evaluated it using models of periodic runnables, periodic interrupts, and event-triggered interrupts. The performance parameters considered in this paper are balancing performance and the deadline missing rate. Our results indicate that the proposed algorithm can contribute toward improving the functional safety of vehicles.