LSTM-Based Traffic Prediction and SCA-Based Antenna Selection for Energy Efficiency Optimization in Multi-Cell mMIMO System.

The fifth generation mobile communication system can provide higher data rates, larger capacity, and wider connections. However, it cannot be ignored that the continuous growth of energy consumption in wireless access networks and data centers has led to a sharp increase in carbon emissions. This paper proposes an antenna selection (AS) strategy to improve system energy efficiency (EE) in base station (BS). We build the network and channel models, and predict the BS traffic using the long short-term memory (LSTM) network. Based on the predicted results, an optimization model is constructed to maximize EE. Successive convex approximation (SCA) and Dinkelbach algorithm are used to solve the optimization problem, and a three-step local search algorithm (TS-LSA) is proposed to acquire the local optimal value of EE. Simulations show that the antenna selection and EE maximization algorithms proposed in this paper can effectively improve system EE.