Expected energy consumption minimization in dvs systems with discrete frequencies

Energy-efficiency has been an important system issue in hardware and software designs to extend operation duration or cut power bills. This research explores systems with probabilistic distribution on the execution time of real-time tasks for systems with discrete frequen-cies. Most previous studies consider DVS systems with continuous frequencies for the minimization of expected energy consumption un-der timing constraints. However, these approaches cannot guarantee the minimization of expected energy consumption when only discrete frequencies are available. This paper presents new approaches to min-imize the expected energy consumption. By applying intra-task fre-quency scheduling, we develop an efficient algorithm to derive op-timal frequency scheduling for a single task. The algorithm is then extended to cope with periodic real-time tasks with different power characteristics. With inter-task and intra-task frequency scheduling, we present a linear-programming approach to derive optimal solu-tions for frame-based real-time tasks and an on-line algorithm for pe-riodic real-time tasks. Experimental results show that the proposed algorithms can effectively reduce the expected energy consumption