基于通信扩展定义的语义通信三层架构

摘要：人工智能的快速发展为Weaver与Shannon所设想的通信扩展定义提供了技术实现基础。基于该通信扩展定义，提出了语义通信三层架构，将语义信息真正融入通信系统设计。发送端可以根据信息的语义重要性进行差异化编码和传输，接收端能够恢复发送端传输数据的语义，并根据语义重要性完成通信的语用目的。进一步地，基于深度学习实现的语义通信三层架构，优化了信息传递的有效性和可靠性，可以更好地满足未来6G新兴应用场景，符合国家战略发展需求。研究了语义通信三层架构搭建方法，系统总结了其中需要面对的挑战与需要实现的关键技术，并探讨了其在物联网、人机交互和智能制造中的广泛应用前景。

关键词：人工智能；通信扩展定义；语义；语用；差异化编码

Abstract: The rapid development of artificial intelligence has provided a technological foundation for the extended definition of communication envisioned by Weaver and Shannon. Based on this extended definition of communication, this paper proposes a three-layer semantic communication architecture that truly integrates semantic information into communication system design. The sender can perform differentiated encoding and transmission based on the semantic importance of the information, while the receiver can recover the semantics of the data transmitted by the sender and accomplish the communication's pragmatic purposes according to semantic importance. Furthermore, this three-layer architecture of semantic communication, implemented using deep learning, optimizes the effectiveness and reliability of information transmission, better meeting the emerging application scenarios of future 6G networks and aligning with national strategic development needs. This paper explores the methods for constructing the three-layer semantic communication architecture, systematically summarizes the challenges faced and the key technologies that need to be implemented, and discusses its wide application prospects in the Internet of Things, human-computer interaction, and intelligent manufacturing.

Keywords: artificial intelligence; extended definition of communication; semantics; objectives; differentiated encoding