Exploiting Segregation in Bus-Based MPSoCs to Improve Scalability of Model-Checking-Based Performance Analysis for SDFAs

Abstract

Abstract. The timing predictability of embedded systems with hard real-time requirements is fundamental for guaranteeing their safe usage. With the emergence of multicore platforms this task becomes even more challenging, because of shared processing, communication and memory resources. Model-checking techniques are capable of verifying the performance properties of applications running on these platforms. Unfortunately, these techniques are not scalable when analyzing systems with large number of tasks and processing units. In this paper, a modelchecking based approach that allows to guarantee timing bounds of multiple Synchronous Data Flow Applications (SDFA) running on shared-bus multicore architectures will be extended for a TDMA hypervisor architecture. We will improve the the number of SDFAs being analyzable by our model-checking approach by exploiting the temporal and spatial segregation properties of the TDMA architecture and demonstrate how this method can be applied.

State-based real-time analysis of SDF applications on MPSoCs with shared communication resources

“…In the top hierarchy level, a non-preemptive TDMA cluster scheduling mechanism, as described in [37], is also supported. This TDMA scheduler allows clusters of actors (still respecting the lower two hierarchy scheduling levels: SO within the SDFG and SO or RR among SDFGs) to be executed in only specific slots and switches to next slot as soon as the previous expires.…”