移动通信2026，Vol.50Issue(2)：2-10,19,10.DOI:10.3969/j.issn.1006-1010.20251130-0003
信道大模型使能的6G环境智能通信及平台实现
Channel Large Model Enabled 6G Environmental Intelligence Communication and Platform Implementation
摘要
Abstract
With the evolution of 6G mobile communications towards integrated sensing and communication as well as ubiquitous connectivity,channel prediction needs to meet the requirements of high dynamism and high precision.However,traditional statistical model paradigms cannot capture the spatiotemporal variations of real propagation environments,and the existing small AI models are poor in multimodal environmental understanding and cross-scenario generalization.To address these challenges,a 6G environment-intelligent communication framework and verification platform enabled by channel large model(ChannelLM)are proposed.Specifically,a multimodal sensing platform is adopted to construct an environment-channel joint dataset and an environmental reconstruction process based on multimodal sensors is designed via combining static mapping and dynamic recognition.Based on this dataset,the proposed ChannelLM is trained to integrate real-time dynamic environmental information for multi-task online channel prediction,enabling a novel paradigm of environment-intelligent communications.Furthermore,a ChannelLM-enabled 6G environment-intelligent communication testbed is developed.Experimental results demonstrate that the platform completes the full pipeline from environmental sensing to channel prediction within a latency of only a few hundred milliseconds while delivering high-accuracy predictions.The results provide essential support for 6G networks to realize environment-adaptive communications,intelligent resource scheduling,and real-time decision-making.关键词
环境智能通信/信道大模型/多模态感知/信道数据集/环境重构/信道预测Key words
intelligent environmental communication/channel large model/multimodal sensing/channel dataset/environmental reconstruction/channel prediction分类
信息技术与安全科学引用本文复制引用
于力,张建华,吴涛,王晶晶,张宇翔,王森..信道大模型使能的6G环境智能通信及平台实现[J].移动通信,2026,50(2):2-10,19,10.基金项目
青年科学基金项目"基于传播环境信息表征的分层信道在线智能预测方法及应用",(A类)"无线信道的建模理论与实验研究"(62401084,62525101) （A类）
国家重点研发计划(2023YFB2904801) （2023YFB2904801）
北京邮电大学-中国移动联合研究院资助项目 （）
