物联网学报2024，Vol.8Issue(1)：1-16,16.DOI:10.11959/j.issn.2096-3750.2024.00369
面向6G的生成对抗网络研究进展综述
Survey on the research progress of generative adversarial networks for 6G
摘要
Abstract
The deep integration of artificial intelligence(AI)and communication technology is the typical feature of the 6G network.On the one hand,AI injects new vitality into the development of the 6G network,which can effectively use the data generated by the historical operation of the network.It enables the network to be self-maintained and self-optimized,and accelerates the process of network intelligence.On the other hand,the rich scenarios and IoT devices of the 6G network provide a large number of application fields and massive data for AI.These can enable the better deployment of AI,fully demonstrate the performance advantages of AI,and provide high-quality services for users.However,in prac-tice,it is difficult to give full play to the performance advantages of AI due to the difficulty of sample collection,high cost of the collection,and lack of universality which caused by the complexity of the environment.Therefore,academia and in-dustry introduce generative adversarial network(GAN)into the design of wireless networks.The powerful feature learn-ing and feature expression ability of GAN can generate a large number of generated samples,which realizes the expan-sion of the wireless database.The introduction of GAN can effectively improve the generalization ability of AI models for wireless networks.Owing to the excellent performance of GAN,the generative model represented by GAN has attracted increased attention in the field of wireless networks,and rapidly became the new research hotspot of 6G networks.Firstly,the principle of GAN and its different versions of improved derived models were summarized.Then,the framework,ad-vantages and disadvantages of each model were analyzed.Secondly,the research and application status of these models in wireless networks were reviewed.Finally,the research trends of GAN were proposed for the 6G network requirements,which provided some valuable exploration for future research.关键词
生成对抗网络/无线网络/信道估计/物理层安全/无线感知/零和博弈Key words
generative adversarial network/wireless network/channel estimation/physical layer security/wireless sens-ing/zero-sum game分类
信息技术与安全科学引用本文复制引用
孟婵媛,熊轲,高博,张煜,樊平毅..面向6G的生成对抗网络研究进展综述[J].物联网学报,2024,8(1):1-16,16.基金项目
国家自然科学基金项目(No.62071033) （No.62071033）
中央高校基本科研业务费资助项目(No.2022JBGP003) The National Natural Science Foundation of China(No.62071033),The Fundamental Research Funds for the Central Universities(No.2022JBGP003) （No.2022JBGP003）
