MSTGAN: A Multi-slot Conditional Generative Adversarial Network Based on Swin Transformer for Channel Estimation

• Published in: IEEE communications letters 2023-07, Vol.27 (7), p.1-1
• Main Authors: Cheng, Lujie, Zhang, Zhi, Dong, Chen, Liu, Sirui
• Format: Article
• Language: English
• Subjects: Channel estimation; Communications systems; conditional generative adversarial network; Convolution; Data augmentation; deep learning; Feature extraction; Generative adversarial networks; Generators; Machine learning; multi-slot; SISO (control systems); Swin Transformer; Three-dimensional displays; Training; Transformers
• ISSN: 1089-7798; 1558-2558
• DOI: 10.1109/LCOMM.2023.3271872

Accurate channel estimation is the key to enhance the transmission performance of communication systems. In this letter, we propose a multi-slot conditional generative adversarial network (cGAN) based on Swin Transformer called MSTGAN for channel estimation in a single-input single-output (SISO) scenario. Specifically, the proposed MSTGAN could learn the temporal correlation feature from continuous channel slots by the 3D convolution and extract the deep feature well by the Swin Transformer to improve the accuracy of channel estimation. The model is trained with data augmentation to reduce the computational cost for offline training. The simulation results demonstrate that the proposed method outperforms the linear minimum mean square error (LMMSE) method and other deep learning methods. Furthermore, extension schemes of the MSTGAN to the multiple-input multiple-output (MIMO) case are also provided.