MORA: an Energy-Aware Slack Reclamation Scheme for Scheduling Sporadic Real-Time Tasks upon Multiprocessor Platforms

Abstract: In this paper, we address the global and preemptive energy-aware scheduling problem of sporadic constrained-deadline tasks on DVFS-identical multiprocessor platforms. We propose an online slack reclamation scheme which profits from the discrepancy between the worst- and actual-case execution time of the tasks by slowing down the speed of the processors in order to save energy. Our algorithm called MORA takes into account the application-specific consumption profile of the tasks. We demonstrate that MORA does not jeopardize the system schedulability and we show by performing simulations that it can save up to 32% of energy (in average) compared to execution without using any energy-aware algorithm.