A New Power-Aware Scheduling Algorithm for Distributed System

By dynamically adjusting CPU speed and voltage with DVS technique, it is possible to save significant amounts of energy with no or little performance loss. There are numerous algorithms published with this goal, most of which concern the power-aware management in real-time system, embedded system, wireless sensor work etc. In this paper, we present a heuristic algorithm, applying the DVS technique in distributed system, where we place high importance on communication time between processors. At the same time, we concentrate on a large number of tasks. In this situation, we could divide a dependent task into many subtasks, which could be concurrently executed on processors in distributed system. After a shortest deadline has been made, our power-aware scheduling algorithm will reduce the energy consumption without affecting its performance. The experimental results show that our algorithm can reduce the CPU energy consumption by up to 40 percents, or even more, with an average of nearly 36 percents.