移动通信2024，Vol.48Issue(8)：30-40,11.DOI:10.3969/j.issn.1006-1010.20240725-0002
6G网络中面向AI大模型的联邦学习与协同部署技术综述
A Survey on Federated Learning and Collaborative Deployment for Large AI Models in 6G Networks
田辉 1倪万里 2聂高峰 1孙浩峰 1敖会清1
作者信息
- 1. 北京邮电大学网络与交换技术全国重点实验室,北京 100876
- 2. 清华大学电子工程系,北京 100084
- 折叠
摘要
Abstract
With the gradual advancement of 6G networks and the rapid development of artificial intelligence(AI)technologies,AI models are increasingly used in wireless networks.In this paper,we comprehensively overview the state-of-the-art work and challenges of wireless federated learning(FL)and collaborative deployment technologies for large AI models in 6G networks.Firstly,we compare the differences between large AI models and fundamental models,and various FL paradigms are introduced.Next,we outline several efficient federated fine-tuning schemes for large AI models in wireless networks,including low-rank adaptation,model splitting,user clustering,and cross-silo collaborative fine-tuning.Subsequently,we elaborate on various solutions for deploying large AI models in wireless networks,including lightweight deployment based on model compression,distributed horizontal collaborative model deployment,and cloud-side-end vertical collaborative model deployment.Finally,we discuss the challenges and opportunities of implementing large AI models in 6G networks.关键词
6G网络/AI大模型/联邦学习/协同部署Key words
6G network/large AI model/federated learning/collaborative deployment分类
信息技术与安全科学引用本文复制引用
田辉,倪万里,聂高峰,孙浩峰,敖会清..6G网络中面向AI大模型的联邦学习与协同部署技术综述[J].移动通信,2024,48(8):30-40,11.
