Mobile Traffic Prediction Method Based on Spatio-Temporal Characteristics

Abstract

The 5G era is coming, and mobile network traffic will explode once again. For network operators, accurate and timely traffic prediction is very important. It can perform resource allocation and congestion control in a timely manner to improve user experience. This paper proposes a mobile traffic prediction method based on spatio-temporal characteristics. The traffic matrix is regarded as a single channel image. The Densely Connected Convolutional Neural Network (DenseNet) is used to extract the spatial characteristics of the traffic, and the Bidirectional Gated Recirculation Unit (Bi-GRU) is used to extract the temporal characteristics of the traffic, spatio-temporal characteristics are integrated to predict mobile traffic. Compared with the existing methods, the experimental results show that the prediction performance of the method in this paper has improved significantly.