A Dynamic Power Management Algorithm For Sporadic Tasks In Real-Time Embedded Systems

摘要

Dynamic voltage Scaling (DVS) has been widely studied as an innovative technology in reducing the real-time embedded devices. An efficient performance of DVS algorithm largely depends on how to acquire more slack times to scale voltage down. In this paper, we present a more effective power management algorithm to reclaim online slack time from both higher-priority and lower-priority tasks. Based on the improvement of method, we propose a novel dynamic energy-saving scheduling algorithm for sporadic tasks in real-time embedded systems (DESSAST-RTES) based on EDF. We also introduce the critical speed which was proposed in previous Studies, and the dynamic power management technology to acquire better energy-saving performance. Considering the difference between the real execution time and the worst-case execution time, DESSAST-RTES can scale processor frequencies dynamically with the real execution time, while guaranteeing all task would not miss their deadlines. The experimental results show that the DESSAST-RTES algorithm can reduce more energy consumption than existing DVS algorithms for sporadic tasks.