An Efficient Assignment of Voltages and Optional Cycles for Maximizing Rewards in Real-Time Systems with Energy Constraints

There exist real-time applications for which timely results are preferable to perfect results too late. At the same time, many such applications run on battery-powered devices with energy constraints. Assuming that each task T_i of the application has a mandatory part and a prospective optional part, a reward is assigned to the task depending on its deadline. This paper addresses the problem of maximizing rewards subject to time and energy constraints allotted to the application. It is shown that our technique yields better results than previous approaches based on static, dynamic, or quasi-static voltage/optional cycles assignments. Finally, our technique is also valid for real-time applications for which all the parts are mandatory and that are subject to energy constraints.