Optimal dynamic voltage scaling in power-limited systems with real-time constraints

Dynamic voltage scaling is used in power-limited systems such as sensor networks as a means of conserving energy and prolonging their life. We consider a setting in which the tasks performed by such a system are nonpreemptive, aperiodic and have uncertain arrival times. Our objective is to control the processing rate over different tasks so as to minimize energy subject to hard real-time processing constraints. We prove that the solution to this problem reduces to two simpler problems which can be efficiently solved, leading to a new on-line dynamic voltage scaling algorithm. This algorithm is shown to have low complexity and, unlike similar state-of-the-art approaches, it involves no solution of nonlinear programming problems and is independent of the specific physical characteristics of the system. Both off-line and on-line versions of the algorithm are analyzed.