首页 | 官方网站   微博 | 高级检索  
     

基于联邦学习的多用户语义通信系统部署方法
引用本文:涂勇峰,陈文.基于联邦学习的多用户语义通信系统部署方法[J].信号处理,2022,38(12):2486-2495.
作者姓名:涂勇峰  陈文
作者单位:上海交通大学电子信息与电气工程学院,上海 200240
基金项目:国家重点研发计划2020YFB1807700上海市科委基础研究课题20JC1416502国家自然科学基金62071296
摘    要:语义通信是一种有发展潜力的新型通信技术,通过挖掘信源中的语义信息从而减少传输所需要的数据量。语义通信通常采用深度学习的方式建立编解码模型,在收发端共享模型参数的前提下实现端到端的数据传输,但在实际场景中,由于多用户的存在,端到端的传输具有局限性,语义通信系统的部署有更多需要考虑的问题。为了使语义通信能应用于多用户的场景,本文提出了语义通信系统模型的联邦学习部署方式,利用用户端的数据对深度学习模型进行更为有效的训练。从而在不直接使用用户数据的前提下,使模型学习到用户数据的特征,实现了多用户场景下语义通信系统的部署。仿真结果表明,通过联邦学习训练得到的模型可以达到接近于集中训练的效果,并且保护了用户隐私。 

关 键 词:??语义通信    联邦学习    多用户网络    文本传输
收稿时间:2022-03-08

A Deployment Method of Multi-user Semantic Communication System Based on Federated Learning
Affiliation:School of Electronic,Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China
Abstract:? ?Semantic communication is a new communication technology with development potential. It can reduce the amount of data needed for transmission by mining the semantic information in the source. Semantic communication usually adopts the way of deep learning to establish the encoding and decoding model and realize the end-to-end data transmission on the premise that the transceiver shares the model parameters. However, in the actual scene, due to the existence of multiple users, the end-to-end transmission has limitations, and the deployment of semantic communication system has more problems to be considered. In order to apply semantic communication to multi-user scenarios, this paper proposed a federated learning deployment mode of semantic communication system model, which used the user data to train the deep learning model more effectively. Thus, without using user data directly, the model learned the characteristics of user data, and realized the deployment of semantic communication system in multi-user scenario. The simulation results show that the model obtained by federated learning training can achieve the effect close to centralized training, and protect the user's privacy. 
Keywords:
点击此处可从《信号处理》浏览原始摘要信息
点击此处可从《信号处理》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号