Study on Policy of Dynamic Power Management Based on Stochastic Decision Models

Because of the energy consumption limitation of modern embedded computing terminal,the reasonable policy for the dynamic power management(DPM) must be used.Based on the deeply study of DPM model for the computer system it is proved theoretically with stochastic models that the DPM optimal strategy is a deterministic Markov control strategy.The conclusion provides the theoretical basis for the simplification of DPM algorithm.In example,two idle time distributions are compared.The one is the negative exponential distribution and the others is two parameters Pareto distribution. It is found that the exponential distribution supposition taken by traditional queuing theory is not suitable to DPM.The practical effect of the time-out strategy is well interpreted with the Pareto distribution.