Generative adversarial networks based digital twin channel modeling for intelligent communication networks

• Published in: China communications Vol. 20; no. 8; pp. 32 - 43
• Main Authors: Zhang, Yuxin, He, Ruisi, Ai, Bo, Yang, Mi, Chen, Ruifeng, Wang, Chenlong, Zhang, Zhengyu, Zhong, Zhangdui
• Format: Journal Article
• Language: English
• Published: China Institute of Communications 01.08.2023; State Key Laboratory of Rail Traffic Control and Safety,Beijing Jiaotong University,Beijing 100044,China%Institute of Computing Technologies,China Academy of Railway Sciences Co.,Ltd.,Beijing 100081,China
• Subjects: channel modeling; Channel models; Delays; digital twin; Digital twins; Generative adversarial networks; Generators; intelligent communication networking; Training; Wireless communication; generative adversarial networks; digital twin; channel modeling; intelligent communication net-working
• ISSN: 1673-5447
• DOI: 10.23919/JCC.fa.2023-0206.202308

Integration of digital twin (DT) and wireless channel provides new solution of channel modeling and simulation, and can assist to design, optimize and evaluate intelligent wireless communication system and networks. With DT channel modeling, the generated channel data can be closer to realistic channel measurements without requiring a prior channel model, and amount of channel data can be significantly increased. Artificial intelligence (AI) based modeling approach shows outstanding performance to solve such problems. In this work, a channel modeling method based on generative adversarial networks is proposed for DT channel, which can generate identical statistical distribution with measured channel. Model validation is conducted by comparing DT channel characteristics with measurements, and results show that DT channel leads to fairly good agreement with measured channel. Finally, a link-layer simulation is implemented based on DT channel. It is found that the proposed DT channel model can be well used to conduct link-layer simulation and its performance is comparable to using measurement data. The observations and results can facilitate the development of DT channel modeling and provide new thoughts for DT channel applications, as well as improving the performance and reliability of intelligent communication networking.