LSTM Network Based Traffic Flow Prediction for Cellular Networks

Abstract

The traffic flow prediction of cellular network requires low complexity and high accuracy, which is difficult to meet using the existing methods. In this paper, we propose an long short-term memory (LSTM) network based traffic flow prediction in which we consider temporal correlations inherently and nonlinear characteristics of cellular network traffic flow data. We use Back Propagation Through Time (BPTT) to train the LSTM network and evaluate the model using mean square error (MSE) and mean absolute error (MAE). Simulation results show that the proposed LSTM network based traffic flow prediction for cellular network is superior to the stacked autoencoder network based algorithm.