Small cell sleeping strategy with traffic-aware and high-low frequency resource allocation

With the increase of wireless devices and new applications, highly dense small cell base stations (SBS) have become the main means to overcome the speed bottleneck of the radio access network (RAN). However, the highly-dense deployment of SBSs greatly increases the cost of network operation and maintenance. In this paper, a base station sleep strategy combining traffic aware and high-low frequency resource allocation is proposed. To reduce the service level agreement (SLA) default caused by base station sleep, Long Short-Term Memory (LSTM) algorithm is introduced to predict the traffic flow, based on the predict result, the SBSs sleep and frequency resource allocation are introduced to increase the energy efficiency of the network. Moreover, this paper improves the decision-making efficiency by introducing Kuhn Munkres algorithm (KM) and genetic algorithm (GA). Simulation results show that the proposed strategy can greatly reduce the energy consumption of small cells and the occurrence of SLA default rate.