Power-Aware Runtime Scheduler for Mixed-Criticality Systems on Multicore Platform

In modern multi-core Mixed-Criticality (MC) systems, a rise in peak power consumption due to parallel execution of tasks with maximum frequency, specially the overload situation, may lead thermal issues, which affect reliability and timeliness MC systems. Therefore, managing has become imperative this regard, we propose an online management heuristic for This reduces system as much possible during runtime by exploiting dynamic slack per-cluster Dynamic Voltage Frequency Scaling (DVFS). Specifically, our approach examines multiple ahead determine most appropriate one assignment, that impact on temperature. However, changing frequency selecting proper task assignment core re-mapping at can be time-consuming cause deadline violation is not admissible high-criticality tasks. analyze then optimize run-time scheduler evaluate it various platforms. The proposed experimentally validated ODROID-XU3 (DVFS-enabled heterogeneous platform) embedded real-time benchmarks. Results show achieves up 5.25% reduction 20.33\% temperature compared existing method while meeting constraints different criticality modes.