On Service Traffic Prediction of Base Stations along High-speed Railway

This paper provides a novel feature extraction algorithm for the service traffic prediction of base stations along high-speed railways. Considering that the traffic of base stations has both time and space characteristics, a feature extraction algorithm is proposed to extract the spatial characteristics of neighboring base stations. Long Short-Term Memory neural network (LSTM) with the feature extraction algorithm are compared with the traditional Recursive Neural Network (RNN) algorithm and Autoregressive Integrated Moving Average algorithm (ARIMA), the results prove that the LSTM algorithm is better than the RNN and ARIMA algorithms, and the prediction performance of the LSTM is better after adding the feature extraction algorithm.