摘要
当联邦学习应用于车联网时,面临着安全、隐私和通信量大等问题。为了解决此问题,文章提出具有隐私增强的车联网分组联邦学习框架。该框架有效利用路侧单元的算力,对车辆进行分组,然后选择合适的车辆进行联邦学习,以这种方式确保联邦学习的高效性和高准确性,同时减少通信开销,提高系统的可扩展性。本方案不仅减少通信开销,而且有助于防止自恶意车辆的模型参数的收集。安全和性能评估的结果表明,所提出的方案能有效提高隐私和操作效率。
When federated learning is applied to Vehicle-to-Everything(V2X)networks,challenges such as security risks,privacy concerns,and high communication are faced.To address these issues,this paper proposes a privacy-enhanced grouped federated learning framework for V2X environments.The framework leverages the computational power of Roadside Units(RSUs)to dynamically group vehicles and selects qualified participants for federated learning,ensuring high efficiency and accuracy while reducing communication costs and enhancing system scalability.By allowing only randomly selected groups to submit local model parameters after the initial training phase,the framework minimizes communication overhead and mitigates risks from malicious vehicle attacks.Security and performance evaluations demonstrate that the proposed scheme significantly improves privacy protection and operational efficiency.
作者
江荣旺
梁志勇
程亮
杨明
Jiang Rongwang;Liang Zhiyong;Cheng Liang;Yang Ming(Sanya College,Sanya 572022,China;Academician Rong Chumming Workstation,Sanya 572022,China)
出处
《信息通信技术》
2025年第2期76-84,共9页
Information and communications Technologies
基金
海南省自然科学基金高层次人才项目(621RC602)
三亚学院重大专项课题(USY22XK-04)
海南省重点研发项目(ZDYF2023GXJS007)
海南省教育厅重点教改项目(Hnjg2022ZD-42)
海南省高等学校科学研究项目(Hnky2025-33,Hnky2025-38,Hnky2025-93)。
关键词
车联网
联邦学习
分组
高效
V2X
Federated Learning
Grouping
Efficiency
