Research on Tactical Communication Network Traffic Prediction Method Based on Deep Learning

Tactical communication networks are very different from civilian networks in terms of network structure, communication behavior, and transmission scenarios. With the rapid development of deep learning methods in network traffic prediction in recent years, good results have been achieved. But the use in the field of tactical communication networks is still relatively rare. We present a multi-source dynamic spatio-temporal graph convolution model (MD-STGCN) for specific analysis of the complex features of tactical communication networks. We incorporate the external environment feature data into the network traffic characteristics and combine the dynamic nature of the network traffic to predict the network traffic spatio-temporal correlation sequence data. Experiments on real datasets show that the method incorporates multiple features of network traffic and enhances the generalization ability of the model, thus improving the prediction accuracy.