一种基于增强型空闲时间回收算法的实时系统低功耗策略

A real-time low-power strategy based on extended slack time reclaiming algorithm

针对以往实时系统低功耗算法空闲时间利用率低的问题,本文提出增强型空闲时间回收算法。该算法在离线状态下获得任务的基准频率及运行顺序,生成α-链表,当任务先于其最长运行时间结束时,在α-链表中标注空闲时间,以尽可能多地分配给后续任务使用,同时又满足了实时性的要求。实验表明,该算法与其他低功耗算法相比,至少可以降低10%的功耗。

Against the low efficiency of slack time utilization in anterior real-time low-power schemes, an extended slack time reclaiming algorithm is proposed, which calculates the static optimal frequency and obtains the running sequence of all tasks, and then generates the a-queue off/ine. When tasks terminate before their worst-case execution time, their slack time is marked in the a-queue to be allocated to subsequent tasks as more as possible while guarantee the real-time constraints. The experiment demonstrates that the power cost of this algorithm could reduce at least 10 percent more than other real-time low-power schemes.