面向6G网络的基于语义通信的端到端服务框架  

The End-to-End Service Framework Based on Semantic Communication for 6G Networks

作　　者：傅宇舟 程文驰[1] 陈小军 李赞[1] FU Yuzhou;CHENG Wenchi;CHEN Xiaojun;LI Zan(School of Telecommunications Engineering,Xidian University,Xi'an 710071,China)

机构地区：[1]西安电子科技大学通信工程学院,陕西西安710071

出　　处：《移动通信》2023年第6期35-40,共6页Mobile Communications

基　　金：国家重点研发计划(2021YFC3002102);国家自然科学基金资助项目(61825104);陕西省重点研发计划(2022ZDLGY05-09)。

摘　　要：未来的6G网络需支持更高的通信效率和高效的智能连接。基于AI的语义通信以传递用户意图及语义信息为目标,有望成为6G网络“内生智能”架构的技术支撑。然而,现有的语义通信框架忽略所提取的语义特征中的冗余信息。为研究高效的语义通信,提出基于语义通信的端到端服务框架,该框架将语义通信与AI的语义分析能力深度融合,可对语义特征进一步压缩后再传输,并保证AI服务质量。仿真分析表明,相比传统通信方案,所提方案在目标检测任务和图像重建均取得更优的性能,且取得与全语义特征传输方案相近的性能。Future 6G networks need to support higher communication efficiency and eficient intelligent connection.AI-based semantic communication aims to convey the user intention and the semantic information and is expected to provide technical support for the"endogenous intelligence"architecture of 6G networks.However,the existing semantic communication frameworks ignore the redundant information within the extracted semantic features.To investigate efficient semantic communication,an end-to-end service framework is proposed based on semantic communication,which is the deep integration of the semantic communication and the semantic analysis ability of AI.The proposed framework can further compress the semantic features before transmission while ensuring the quality of AI service.The simulation analysis shows that the proposed scheme can achieve better performance in the object detection task and image reconstruction compared with the traditional communication scheme,and achieves similar performance to the full semantic feature transmission scheme.

关 键 词：语义通信 语义分析模型 端到端服务 

分 类 号：TN915[电子电信—通信与信息系统]