Energy efficient task partitioning and real-time scheduling on heterogeneous multiprocessor platforms with QoS requirements

We address the problem of partitioning a set of independent, periodic, real-time tasks over a fixed set of heterogeneous processors while minimizing the energy consumption of the computing platform subject to a guaranteed quality of service requirement. This problem is NP-hard and we present a fully polynomial time approximation scheme for this problem. The main contribution of our work is in tackling the problem in a completely discrete, and possibly arbitrarily structured, setting. In other words, each processor has a discrete set of speed choices. Each task has a computation time that is dependent on the processor that is chosen to execute the task and on the speed at which that processor is operated. Further, the energy consumption of the system is dependent on the decisions regarding task allocation and speed settings. (c) 2011 Elsevier Inc. All rights reserved.