Power optimization using Markov decision process based on multi-parameter constraint modeling

Abstract

Power optimization based on intelligent algorithm draws more and more attention. This article presents a novel low power optimization strategy based on the high level software power management employing Markov Process for charactering the real running workload. This article formulates workload characterization and selection with stochastic process method, and solves the formula using dynamic voltage frequency scaling base on microprocessor. Based on Markov process, the multi-parameter constraints has been employed to exploit the optimization space. Comparing with existing power optimization algorithm, our proposed power optimization algorithm doesn't need any prior data and maintains a value function representing expected reward. As many hardware events can be effectively captured and modeled, this optimization technique is capable to explore an ideal tradeoff in the constraint space.