Base Station Sleeping Mechanism Based on Traffic Prediction in Heterogeneous Networks

In order to reduce the energy consumption of cellular networks, the possibility of turning off base stations under off-peak time has been focused in recent years. However, traffic in cellular networks is always under fluctuation which leads to the sleeping mechanisms which is based on deterministic traffic variation pattern unsuitable. In this paper, a base station sleeping mechanism based on traffic prediction (BSTP) is proposed to solve the problems caused by traffic fluctuation in heterogeneous networks. Modified Wavelet neural network (MWNN) is used to predict the future traffic of base stations in this paper. Based on the prediction, a base station sleeping mechanism is proposed by making use of Pico Base Stations (PBSs) instead of Macro Base Station (MBS) to provide service under off-peak time. Simulation results show that the MWNN model has a good prediction precision and faster convergence speed, and the sleeping mechanism can significantly reduce the total energy consumption of heterogeneous networks.