Channel Modeling of Spaceborne Multiwavelet Packet OFDM System Based on CWGAN

Abstract

The physical layer of 5G mobile communication adopts the orthogonal frequency division multiplexing (OFDM) technology based on Fast Fourier transform. High peak-to-average power ratio (PAPR) is one of the main problems of OFDM systems, which will cause the nonlinear distortion of high power amplifiers. Due to the orthogonal characteristics of multi-wavelets in the frequency domain and the integral offset characteristics in the time domain, the use of multiwavelet packet based orthogonal frequency division multiplexing (MWPT-OFDM) can reduce the PAPR by reducing subcarriers, which is very suitable for spaceborne communication. The sixth generation (6G) mobile communication system forms an integrated network by deeply integrating the satellite network and the terrestrial network, and the channel modeling based on the satellite communication has become a research hotspot. A novel Conditional Wasserstein Generative Adversarial Network (CWGAN) framework is proposed to solve the channel modeling problem of spaceborne MWPT-OFDM systems. Specically, replacing the generator and discriminator in Wasserstein GAN with convolutional neural networks and using the pilot information of the MWPT-OFDM signal as a condition to model the spaceborne channel. The experimental results show that the proposed channel modeling framework based on CWGAN can successfully approximate the channel distribution of the spaceborne MWPT-OFDM system and the effect is better than Generative Adversarial Network (GAN). The research results shows that the feasibility of spaceborne channels modeling based on GAN and provide a reference for further research on spaceborne channel modeling.