Research On Traffic Prediction Of Base Station Mobile Network Based On Machine Iearning

With the development of Internet technology,various new services and new businesses are emerging,and people’s demand for the network is also increasing.The attendant traffic load problem of the base station also affects the perceived quality and user satisfaction of mobile network users.At the same time,many equipment and facilities are accompanied by high power consumption during the operation of the base station,which is difficult to achieve the goal of energy conservation of the base station.Therefore,under the premise of ensuring the user’s network business experience and service quality,reducing the energy consumption of the wireless communication system becomes the key.Because of the tide phenomenon of the base station,the traffic of the base station will show peaks and valleys at different times.If the base station configuration is run according to the number of carriers in the high-capacity period at the low traffic stage,it will cause a large amount of unnecessary resource waste,and vice versa,it will lead to poor user perception quality.Therefore,we need to establish an effective prediction model for the network traffic of the base station to predict the network traffic of the base station in real time.On the one hand,based on the prediction results,we can set the base station energy saving scheme to improve the overall efficiency of the network.On the other hand,we can also predict the time required for the physical expansion of the base station in advance,so as to guide operators to plan and design the physical expansion more reasonably.This paper first analyzes the characteristics of data,and finds that the traffic data of base station mobile network has the characteristics of selfsimilarity,burst,periodicity and chaos.And the network traffic data has a strong correlation with time,which is time series data.Therefore,the time series data prediction method can be used to model.After that,this paper forecasts the base station network traffic based on ARIMA and LSTM models,and compares the two models,and finds that the LSTM model has better fitting effect.After that,the Prophet model is introduced to improve the LSTM model.It is found that the Prophet LSTM model has better effect in prediction.A base station energy saving scheme based on the Prophet LSTM model prediction is proposed.The traditional energy saving technology is used to select the best energy saving scheme to reduce the energy consumption of the base station.