Generative Network-Based Channel Modeling and Generation for Air-to-Ground Communication Scenarios

In this letter, a novel conditional generative adversarial network (CGAN)-based channel modeling method for A2G communication scenarios is proposed. The proposed method can generate channel parameters of each multipath, i.e., the path gain, delay, angle of arrival and departure, and Doppler frequency according to the input of arbitrary location and velocity of the transceiver, so as to realize the A2G channel modeling. The ray tracing (RT) method is used to generate the training data and verify the proposed modeling method in a urban A2G communication scenario. The simulation results demonstrate the proposed CGAN can effectively generate the statistical channel characteristics well consistent with the theoretical ones, and verify the effectiveness and accuracy of the proposed method for A2G channel modeling.