[1]董少华,马新春,樊小超.分层边缘计算中利用联邦学习的差分隐私保护[J].计算机技术与发展,2025,(04):53-58.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0382]
　DONG Shao-hua,MA Xin-chun,FAN Xiao-chao.Differential Privacy Protection Using Federated Learning in Layered Edge Computing[J].,2025,(04):53-58.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0382]
点击复制

分层边缘计算中利用联邦学习的差分隐私保护()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年04期
页码:
53-58
栏目:
移动与物联网络
出版日期:
2025-04-10

文章信息/Info

Title:
Differential Privacy Protection Using Federated Learning in Layered Edge Computing
文章编号:
1673-629X(2025)04-0053-06
作者:
董少华1; 2; 马新春2; 樊小超1
1. 新疆师范大学 计算机科学技术学院,新疆 乌鲁木齐 830054;
2. 新疆电子研究所,新疆 乌鲁木齐 830013
Author(s):
DONG Shao-hua1; 2; MA Xin-chun2; FAN Xiao-chao1
1. School of Computer Science and Technology,Xinjiang Normal University,Urumqi 830054,China;
2. Xinjiang Electronic Research Institute,Urumqi 830013,China
关键词:
分层边缘计算; 联邦学习; 差分隐私; 边缘协作; 数据隐私
Keywords:
layered edge computing; federated learning; differential privacy; edge collaboration; data privacy
分类号:
TP391.1
DOI:
10.20165/j.cnki.ISSN1673-629X.2024.0382
摘要:
分层边缘计算通过在资源受限的边缘设备和云服务器上协作运行,旨在最小化推理延迟和保护数据隐私。 然而,即使来自边缘设备的原始输入数据没有直接暴露于云中,针对边缘数据的最先进的攻击仍然能够从暴露的本地模型的中间输出中重建原始的私有数据,从而引入严重的隐私风险。为了克服这些限制,提出了一种新的算法,该算法合并了边缘阶段和云阶段的模型,提出了确定最佳聚合时间框架的定性指令,以减少计算和通信费用。通过在客户端和边缘服务器级别实现局部差分隐私,增强了本地模型参数更新时的隐私。在 CIFAR-10 和 MNIST 数据集上的实验表明,该算法在训练精度、训练时间和通信-计算权衡方面优于标准的联邦学习方法。 且该算法为分层边缘计算的挑战提供了一个很有前途的解决方案,使内容交付更快、移动服务质量更高。
Abstract:
Layered edge computing, by operating collaboratively on resource - constrained edge devices and cloud servers, aims to minimize inference latency and protect data privacy. However,even if the raw input data from edge devices is not directly exposed to the cloud,sophisticated attacks targeted at edge data can still reconstruct the original private data from exposed local model outputs,introducing serious privacy risks. To overcome these limitations,a new algorithm is proposed that amalgamates models from both the edge and cloud phases,presenting qualitative directives for determining optimal aggregation time frames to reduce computational and com-munication costs. Local differential privacy is implemented at the client and edge server levels to enhance privacy during local model pa-rameter updates. Experiments on the CIFAR-10 and MNIST datasets demonstrate that the proposed algorithm outperforms standard FL methods in terms of training accuracy,training time,and communication-computation trade-offs. It offers a promising solution to the challenges of layered edge computing,enabling faster content delivery and higher mobile services quality.

相似文献/References:

[1]舒志鸿,沈苏彬.在不平衡数据中进行高效通信的联邦学习[J].计算机技术与发展,2021,31(12):33.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 006]
　SHU Zhi-hong,SHEN Su-bin.Communication-efficient Federated Learning from Imbalanced Data[J].,2021,31(04):33.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 006]
[2]杨晓雨,周彩凤*.基于联邦卷积神经网络的鱼类检测系统[J].计算机技术与发展,2023,33(09):155.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 023]
　YANG Xiao-yu,ZHOU Cai-feng*.A Fish Classification System Based on Federal Convolution Neural Network[J].,2023,33(04):155.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 023]
[3]张海超,赖金山,刘 东,等.边缘计算下的轻量级联邦学习隐私保护方案[J].计算机技术与发展,2023,33(09):161.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 024]
　ZHANG Hai-chao,LAI Jin-shan,LIU Dong,et al.Lightweight Federated Learning Privacy Protection Scheme under Edge Computing[J].,2023,33(04):161.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 024]
[4]袁 媛,袁 松*.一种区块链支持的联邦学习认知模型[J].计算机技术与发展,2023,33(11):215.[doi:10. 3969 / j. issn. 1673-629X. 2023. 11. 032]
　YUAN Yuan,YUAN Song *.Federal Learning of Cognitive Model Supported by Blockchain[J].,2023,33(04):215.[doi:10. 3969 / j. issn. 1673-629X. 2023. 11. 032]
[5]王志良,何 刚*,俞文心,等.边缘场景下动态联邦学习优化方法[J].计算机技术与发展,2024,34(02):98.[doi:10. 3969 / j. issn. 1673-629X. 2024. 02. 015]
　WANG Zhi-liang,HE Gang*,YU Wen-xin,et al.Dynamic Federated Learning Optimization Method in Edge Scenarios[J].,2024,34(04):98.[doi:10. 3969 / j. issn. 1673-629X. 2024. 02. 015]
[6]王漫,李培剀,熊勇*.基于YOLO的集装箱锁销分类联邦学习网络[J].计算机技术与发展,2024,34(08):189.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0144]
　WANG Man,LI Pei-kai,XIONG Yong*.A YOLO-based Federated Learning Network for Container Lock Pin Classification[J].,2024,34(04):189.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0144]
[7]王斌,尹晓雅,张磊*.基于横向联邦学习的隐私保护研究[J].计算机技术与发展,2024,34(10):1.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0194]
　WANG Bin,YIN Xiao-ya,ZHANG Lei*.Survey and Analysis of Privacy-preserving Based on Horizontal Federated Learning[J].,2024,34(04):1.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0194]
[8]赖海光,朱邦兵,沈金海,等.基于联邦学习的卫星系统通感波束成形方法[J].计算机技术与发展,2025,(07):48.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0036]
　LAI Hai-guang,ZHU Bang-bing,SHEN Jin-hai,et al.Federated Learning Based ISAC Beamforming Method for Satellite Systems[J].,2025,(04):48.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0036]
[9]石迎澳,李润知,姬怡.基于联邦全局知识蒸馏的异常网络入侵检测方法[J].计算机技术与发展,2025,(07):55.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0087]
　SHI Ying-ao,LI Run-zhi,JI Yi.Anomaly Network Intrusion Detection Method Based on Federated Learning with Global Knowledge Distillation[J].,2025,(04):55.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0087]

更新日期/Last Update: 2025-04-10