This repository contains the simulation dataset and code for the submitted paper "Transfer Learning Enabled Transformer based Generative Adversarial Networks (TT-GAN) for Terahertz Channel Modeling and Generating"
The simulation data is provided, which can be generated by the QuaDriGa software in the matlab with the provided statistics in the file "TWC_Indoor_Corridor_LOS.conf".
matlab coeff_generation_3gpp.m
The T-GAN and TT-GAN are implemented in transformer-gan.py and ttgan.py respectively. The ttgan.py is fine-tuned on the pre-trained tgan models. The number of epochs is set as 10000
The tt-gan are compared with the basic gan, t-gan and ray-tracing.
