生成式语义通信：架构、技术与应用
摘要
本文深入探讨了生成式人工智能 (GAI) 在语义通信 (SemCom) 中的应用，并进行了全面研究。首先介绍了三种由经典 GAI 模型实现的流行 SemCom 系统，包括变分自编码器、生成对抗网络和扩散模型。对于每种系统，阐述了 GAI 模型的基本概念、相应的 SemCom 架构以及近期相关工作的文献综述。随后，提出了一种新型生成式 SemCom 系统，通过融合大语言模型 (LLM) 的最新 GAI 技术实现。该系统在发射方和接收方各部署一个基于 LLM 的 AI 智能体，分别作为“大脑”，以实现强大的信息理解和内容再生能力。这种创新设计允许接收方直接生成所需内容，而无需恢复比特流，这是基于发射方传递的编码语义信息。因此，它将通信思维从“信息恢复”转向“信息再生”，从而 usher in a new era of generative SemCom. 我们 presented a case study on point-to-point video retrieval to demonstrate the superiority of the proposed generative SemCom system, showcasing a 99.98% reduction in communication overhead and a 53% improvement in retrieval accuracy compared to the traditional communication system. Furthermore, four typical application scenarios for generative SemCom are delineated, followed by a discussion of three open issues warranting future investigation. In a nutshell, this paper provides a holistic set of guidelines for applying GAI in SemCom, paving the way for the efficient implementation of generative SemCom in future wireless networks.
引用
@article{arxiv.2412.08642,
title = {Generative Semantic Communication: Architectures, Technologies, and Applications},
author = {Jinke Ren and Yaping Sun and Hongyang Du and Weiwen Yuan and Chongjie Wang and Xianda Wang and Yingbin Zhou and Ziwei Zhu and Fangxin Wang and Shuguang Cui},
journal= {arXiv preprint arXiv:2412.08642},
year = {2024}
}
备注
18 pages, 8 figures