机器学习在无线信道建模中的应用现状与展望  

Application Status and Prospects of Machine Learning in Wireless Channel Modeling

作　　者：黄鸿清 刘为[2,3] 伍沛然 夏明华 HUANG Hongqing;LIU Wei;WU Peiran;XIA Minghua(School of Electronics and Information Technology,Sun Yat-sen University,Guangzhou 510006,China;CETC Advanced Mobile Communication Innovation Center,Shanghai 200331,China;The 7th Research Institute of China Electronics Technology Group Corporation,Guangzhou 510310,China;Southern Marine Science and Engineering Guangdong Laboratory Zhuhai 519082,China)

机构地区：[1]中山大学电子与信息工程学院,广东广州510006 [2]中国电子科技集团公司新一代移动通信创新中心,上海200331 [3]中国电子科技集团公司第七研究所,广东广州510310 [4]南方海洋科学与工程广东省实验室,广东珠海519082

出　　处：《移动通信》2021年第4期95-104,共10页Mobile Communications

基　　金：国家自然科学基金项目(U2001213);广东省重点领域研发计划项目(2018B010114001);中央高校基本科研业务费项目(19lgjc04)。

摘　　要：为了适应未来6G通信系统的超宽频谱、超大规模天线阵列、高度异构化以及众多新型应用场景,信道建模成为新系统开发必不可少的技术基础。由于6G通信系统将具有典型的大数据特征,基于机器学习的数据驱动型无线信道建模方法已经将成为未来信道模型开发的重要手段。综合分析机器学习在无线信道建模中的应用现状,主要包括确定性信道模型的射线追踪法,随机性信道模型的多径分量聚类与跟踪以及模型参数估计,数据驱动型信道建模,以及信道场景识别,最后,讨论基于机器学习的无线信道建模方法面临的挑战。To adapt the future 6G communication system with ultra-wide spectrum, ultra-massive antenna array, highly heterogeneousness, and numerous new application scenarios, channel modeling is indispensable technical foundation for developing new systems. Since 6G communication system has the typical big data characteristics, the data-driven wireless channel modeling method based on machine learning has become a key method of future channel model development. This paper comprehensively analyzes the state-of-the-art of machine learning in wireless channel modeling, including ray tracing of deterministic channel modeling, the multipath component clustering and tracking as well as model parameter estimation of stochastic channel modeling, data-driven channel modeling, and channel scenario recognition. Finally, the challenges are discussed for the wireless channel modeling based on machine learning.

关 键 词：机器学习 无线信道 信道建模 

分 类 号：TN929.5[电子电信—通信与信息系统]