智简语义通信的链路设计及关键技术研究

摘要：智简语义通信是一种模型驱动的语义通信新范式，融合了人工智能与通信技术，实现通信对象间高效的语义交互。提出了智简语义通信系统的链路结构和关键技术，从链路结构、语义模型增强、模型传输3 个角度，实现语义通信系统的整体性能增强。提出了4 种智简语义通信的关键技术，为未来6G 赋能多种垂直行业和新场景应用提供了参考：语义知识图谱增强的智简通信技术通过增加语义知识这一信息维度，提升了语义知识恢复准确度和传输效率；语义知识图谱的云-边-端协同预缓存技术可以实现语义知识图谱的高效获取，辅助语义恢复性能；模型传输与部署技术可以实现模型、网络资源与终端能力的有效适配；语义模型传输与恢复级联过程的部署成功率为大规模语义模型传输及资源部署提供理论依据。

关键词：智简语义通信；模型处理；模型传输；语义知识库

Abstract: Intellicise semantic communications is a new paradigm of model-driven semantic communication, which integrates artificial intelligence and communication technologies to realize efficient semantic interaction between communication objects. In general, the link structure and key technologies of the intellicise semantic communication system are proposed from three perspectives: link structure, semantic model enhancement, and model transmission, and the overall performance of the semantic communication system is enhanced. Four key technologies of intellicise semantic communications are proposed, which provide a reference for the future 6G empowerment of various vertical industries and new scene applications: a semantic knowledge graph-enhanced intellicise communication technology improves the accuracy of semantic knowledge recovery and the transmission efficiency by increasing the information dimension of semantic knowledge; the cloud-edgedevice collaborative pre-cache technology of semantic knowledge graph can realize the efficient acquisition of semantic knowledge graph and assist semantic recovery; model transmission and deployment technology can achieve the effective adaptation of the models, network resources and terminal capabilities; the deployment success probability of the semantic model related to the transmission-recovery cascade process is analyzed, which provides a theoretical basis for large-scale semantic model transmission and resource deployment.

Keywords: intellicise semantic communication; model processing; model transmission; semantic knowledge base