Reinforcement temporal difference learning scheme for dynamic energy management in embedded systems

Dynamic power management is a technique to reduce power consumption of electronic systems by selectively shutting down idle components. In this paper a novel and non-trivial enhancement of conventional reinforcement learning is adopted to predict the optimal policy out of the existing DPM policies. Reinforcement learning is a computational approach to understanding and automating goal-directed learning and decision-making. The effectiveness of this approach is demonstrated by an event driven simulator which is designed using JAVA for power-manageable embedded devices. Results of the experiments conducted in this regard establish that the proposed DPM scheme enhances power savings considerably.