实时周期任务的非占先式能耗感知调度
Energy-Aware Scheduling of Nonpreemptive Periodic Tasks in Real-Time Systems
-
摘要: 针对实时周期性任务集在变电压处理器上的非占先式能耗感知调度问题,在简单遗传算法基础上,提出了一种通过构造特定的遗传过程以自然地满足时序和能耗约束的启发式算法. 与贪婪算法相比,该算法在提高系统性能方面具有明显的优势,而且能耗利用效率随着能耗约束值的减小而增大,说明算法具有良好的能耗感知特性.Abstract: The nonpreemptive real-time periodic tasks scheduling on variable processor is mainly concerned in this paper.Based on genetic algorithm (GA),a heuristic algorithm is proposed to make the random search naturally meet the energy and timing constraints,as well as maximize the system reward. Experimental results show that GA has obvious advantages over greedy algorithm on reward improvement. It reveals that energy efficiency increases when the energy budget decreases,which proves the energy awareness of the scheduling algorithm.
下载: