实时系统最大化奖赏值的能耗感知调度

Energy-aware scheduling of reward maximum in real-time systems

针对实时系统中周期任务在可变电压处理器上的能耗感知调度问题,提出了一种贪婪人工蜂群算法。在人工蜂群算法的基础上,引入贪婪算法的思想,按照最大价值密度优先原则,为任务分配执行速度,以保证价值密度大的任务能被执行,从而提高系统的奖赏值。仿真结果表明:能耗利用率随着能耗约束的减小而增大,说明算法具有良好的能耗感知特性;相比贪婪算法,当能耗约束因子为0.1时,系统的奖赏值平均提高了78%,随着任务数的增加、能耗约束越强,其优势更加明显。

A novel algorithm is proposed for energy-aware scheduling of periodic tasks on variable voltage processors in the real-time systems.Based on the combined idea of artificial bee colony algorithm with greedy algorithm,different speed is allocated for different tasks to ensure the execution of tasks with higher value density.Simulation results show that the system reward of energy per unit is increased with the decreasing of energy constraint,therefore the scheduling algorithm is of the energy awareness.Compared with the greedy algorithm,the rewards of the real-time system using the greedy artificial bee colony algorithm are averagely improved by 78% while energy constraint factor is 0.1.Mmore system rewards can be obtained with the increasing number of tasks or energy constraints.