An efficient low-power scheme based on probability distribution of system workloads

An efficient low-power scheme based on probability distribution of system workloads is presented,against low efficiency that the previous studies do not sufficiently explore the opportunities for inter-task scheduling to save more energy.This scheme has two steps:(1) According to the condition that the probability distribution of the system workload is the convolution sum of these of all running tasks',this scheme computes the distribution of the system workload when each task releases.(2) With the worst-case execution time of the system workload obtained by offline,this scheme calculates the clock frequencies required by the system workload and get its lowest statistic power.Furthermore,some modifications to this scheme are done with practical considerations.Simulation results show that it effectively reduces at least 30% energy consumption than previous studies without any negative effect on performance.