An Efficient Wave Propagation Prediction Model Generated by CycleGAN with Ray Tracing Data

The ray tracing technique is an application of optical ray technology in the field of electromagnetic computing, which can accurately consider various propagation paths of electromagnetic waves with high computational accuracy but large computational volume. In this paper, we generate an accurate and efficient prediction model for urban radio propagation using Cycle Generative Adversarial Network (CycleGAN). The color information of red, green and blue (RGB) channels of the planar images of the urban area is used to characterize the power of radio wave propagation, CycleGAN is used to learn this information to achieve the prediction effect. Compared with the traditional modeling methods, the method proposed in this paper can greatly reduce the time required for prediction, and the structural similarity between the prediction and the actual simulation results can reach 0.95. It takes 10.85s to calculate the same number of city area images using Winprop software, while the proposed method only needs 3.08s, which saves a lot of time.