Network Energy Conservation Scheme Based on Transfer Reinforcement Learning in Cellular Base Stations
In recent years,data-intensive applications have grown exponentially in communication networks,resulting in an increasing workload.In order to ensure network service quality,a large number of base stations need to be deployed,which consumes a lot of energy.How to save energy and increase efficiency is one of the challenges faced by operators.This article proposes a base station switching scheme based on reinforcement learning(RL)for dynamic region partitioning to effectively improve the energy efficiency of cellular networks.In addition,utilizing previously estimated traffic statistics through transfer learning can further improve energy conservation and accelerate the learning process.The superiority of the proposed framework was demonstrated through relevant mathematical analysis and simulation results.Compared with traditional base station switching schemes,the proposed framework reduces average energy consumption by about 40%for cellular networks with low to medium loads.
cellular base stationenergy consumptionreinforcement learningtransfer learning
NETL
NSTL
万方数据
