Spatiotemporal Traffic Prediction Based on Graph Convolutional Network for LEO Satellite Networks

The accurate prediction of traffic load is of critical importance for the design of efficient routing and load balancing schemes for Low Earth Orbit (LEO) satellite networks, which greatly affect the utilisation of inter-satellite links. However, the periodically changing topology of satellite networks leads to significant uncertainty in the traffic dynamics, making it difficult to predict. This paper presents an in-depth analysis of the characteristics of satellite network traffic and proposes a traffic prediction algorithm based on Temporal Graph Convolutional Network (TemGCN). The proposed scheme addresses a gap in the field of satellite network traffic prediction by leveraging the network topological structure and making full use of the highly dynamic but periodically changing topology characteristics of satellite networks. The experimental results demonstrate that the proposed scheme is more accurate and robust than other existing schemes.