Mobile Traffic Prediction Based on Densely Connected CNN for Cellular Networks in Highway Scenarios

Abstract

With the explosive growth of communication traffic and the arrival of 5G technologies, wireless big data has become an enabler for operators to manage and improve their wireless communication systems. Although many mobile traffic prediction methods have been proposed in the past few years, few prediction methods combine with the distribution features of base stations to predict the mobile traffic of cellular networks. In this paper, by leveraging on the 4G mobile data collected from one typical city in southeastern China, we propose a mobile traffic prediction approach based on one-dimensional densely connected convolutional neural networks (CNN) to predict the mobile traffic of base stations in highway scenarios. After data acquisition, data analysis and modeling, comparisons are made between the proposed mobile traffic prediction approach and the widely used prediction approaches based on machine learning models like LSTM and SVR, and numerical results show that the proposed mobile traffic prediction approach has outstanding performances.