Abstract:

To meet the changeable reality of task sets for the runtime system and needs of non-ignorable switching overhead for processor state, we propose an overhead-aware energy-efficient real-time scheduling algorithm called a dynamic slack reclamation based overhead-aware energy-efficient real-time scheduling in multiprocessor systems (DSROM) for periodic tasks deployed on multi-core and multiprocessor systems. The main idea of the algorithm is to implement energy-efficient scheduling for real-time tasks at the initial time of each TL plane, and to reclaim dynamic slack time at the earlier completion time of a periodic task in each TL plane. Consequently, the algorithm can obtain a reasonable tradeoff between real-time constraint and energy-saving while guaranteeing the optimal feasibility of periodic tasks. Extensive simulation results demonstrate that the DSROM can guarantee the optimal feasibility of periodic tasks and save more energy on average than the existing algorithms when the total workload of the system exceeds a threshold, saving energy by about 20% at most.

Key words: dynamic slack time, overhead, energy saving, real-time scheduling