Energy efficient scheduling for real-time systems with mixed workload

Abstract

In spite of numerous inter-task dynamic voltage scaling (DVS) algorithms of real-time systems with either periodic tasks or aperiodic tasks, few of them were aimed at the mixed workload of both kind of tasks. A DVS algorithm for mixed workload real-time systems should not only focus on energy saving, but also should consider low response time of aperiodic tasks. In this paper, we develop an on-line energy efficient scheduling, called Slack Stealing for DVS (SS-DVS), to reduce CPU energy consumption for mixed workload real-time systems under the earliest deadline first (EDF) scheduling policy. The SS-DVS is based on the concept of slack stealing to serve aperiodic tasks and to save energy by using the dynamic reclaiming algorithm (DRA). Unlike other existing approaches, the SS-DVS does not need to know the workload and the worst case execution time of aperiodic tasks in advance. Experimental results show that the proposed SS-DVS obtains better energy reduction (17% similar to 22%) while maintaining the same response time compared to existing approaches.