Short-term Wireless Load Indicator Forecasting Method Based On Multi-Model Fusion

Abstract

With the increasing scale of wireless networks, the energy consumption of base stations is also increasing. To minimize the energy consumption of the base station, it is necessary to manage the network communication equipment effectively, such as in some cases, the base station can work or sleep, to reduce the power consumption and achieve the goal of low carbon energy saving. Therefore, a short-term wireless service indicator forecasting method based on the combination model of prophet and LSTM (Long Short-Term Memory) algorithm is proposed. And the comparison experiments with the single model of Prophet and LSTM before combination and two other typical time series forecasting models are designed and realized. The experimental results show that the proposed model has high forecast accuracy, good universality and application prospect.