Timeout-driven adaptive frequency carriers on/off control in wireless access networks

A stochastic timeout threshold based dynamic power management (DPM) strategy is presented for conserving energy of base stations in wireless access networks. It dynamically changes the operation state of multiple frequency carriers with the fluctuation of workloads to guarantee the QoS with minimum power consumption. First, an event-driven continuous-time Markov control processes model is introduced to formulate the DPM problem as a constrained optimization problem. Then, a reinforcement learning algorithm that combines policy gradient estimation and stochastic approximation is proposed for optimizing the DPM policy online. Simulation results demonstrate the effectiveness of the presented approach.