An SRP-based energy-efficient scheduling algorithm for dependent real-time tasks

Abstract

In this paper, we are interested in the energy-efficient scheduling of dependent real-time tasks which may access multiunit resources. Based on the well-known stack resource policy (SRP), we propose the blocking-time stealing (BTS) algorithm to schedule dynamic-priority and fixed-priority tasks on a non-ideal dynamic voltage scaling (DVS) processor. Under the BTS algorithm, tasks are scheduled to be executed at proper processor speeds which are determined according to the sufficient schedulability conditions of the earliest deadline first (EDF) and the rate monotonic (RM) algorithms. The BTS algorithm measures dynamically the slack between the actual blocking time of a task and its estimated worst-case blocking time. The slack is then exploited to lower the processor speed guaranteeing that the feasibility of tasks is not affected. As a result, the energy consumption can be reduced while satisfying the timing constraints of real-time tasks.