Graph Convolutional Neural Network-based Modeling of Ultra-Scale MIMO Channels in 6G Networks

The new generation of information technology is becoming a new engine for high-quality development of the economy and society, and the fifth generation mobile communication system (5G) will gradually evolve and develop into the next generation mobile communication system (6G) with the increase in the demand for communication services. In this paper, a self-learning channel modeling method based on graph convolutional neural network (GCN) is proposed, which enables the parameters and structure of the channel model to be automatically adjusted in complex scenarios through self-learning and self-optimization training. And the validation and channel characterization of the model are carried out after channel modeling to ensure the correctness of the channel model. The temporal correlation between each antenna pair in the model decreases with the increase of the sampling time interval, and the correlation converges to 0.1 when the time interval is larger than 30ms, indicating that the 6G MIMO channel has a long-term dependence on the time series. Meanwhile, experiments show that each antenna array element has an independent set of effective scatterer clusters, which validates the non-smooth property of the 6G MIMO channel. In the S-band and Ka-band experiments, the larger the satellite elevation angle is, the faster the delay extension converges. Under NLOS propagation conditions, the power difference between clusters is more average than under LOS propagation conditions. In this paper, we implement the modeling of the ultra-large-scale MIMO channel for 6G networks, and verify and characterize the channel, which provides a reference for the subsequent research.