Energy-efficient multi-granularity resource reservations for multimedia services

Abstract

This work presents the scheduling algorithms to reduce the energy consumption of hard real-time tasks with fixed priorities assigned by a rate monotonic policy and to guarantee that all the I-frames of each video stream meet their deadlines. The proposed scheduling algorithms are called low-power work/limit demand analysis with multi-granularity (lpWDA-MG/lpLDA-MG) and it is an attempt to reduce energy consumption by incorporating the DVS technique while simultaneously supporting multimedia such as MPEG video stream services. We present results to show several features of the proposed algorithms and show the experimental results by comparing with the existing methods that support MPEG4 streams. The experimental results show that the proposed algorithms tremendously reduce energy consumption by 2-50%. © 2012 Springer Science+Business Media B.V.