Transfer Reinforcement Learning based Framework for Energy Savings in Cellular Base Station Network

Last few years have witnessed an exponential upsurge in data intensive applications over the communication networks. Energy saving is one of the major aspects in such networks wherein the increased traffic load entails deployment of a large number of base stations (BSs). In this paper, a BS switching scheme is proposed which exploits reinforcement learning (RL) for dynamic sectorization of BSs to increase the energy efficiency of cellular networks. Furthermore, previously estimated traffic statistics is exploited through the process of transfer learning for further improvement in energy savings and speeding up the learning process. The superiority of the proposed framework is depicted through simulations and relevant mathematical analysis. Compared to conventional ON/OFF scheme, proposed framework offers around 40% lower average energy consumption for cellular networks with low to moderate loads.