Forescating Mobile Network Traffic based on Deep Learning Networks

Abstract

As Internet Service Providers (ISPs) integrate the fifth generation (5G) technology standard for cellular broadband systems, they may face bursts of network traffic due to the future numerous connections. In this sense, our paper is focused on predicting traffic peaks via deep learning techniques, such as Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) of a real mobile core EPC node located in Chile. The results show that LSTM outperforms GRU in terms of traffic prediction by factor of 0.4 and in terms of computational cost, LSTM and GRU have identical behavior.