面向异构数据的半分布式联邦学习安全聚合

黄梅; 王玲玲; 张政胤; 刘宇飞; 孙艺桐

doi:10.11959/j.issn.2096-109x.2025029

您当前的位置：

首页 >

文章列表页 >

面向异构数据的半分布式联邦学习安全聚合

学术论文 | 更新时间：2025-06-30

- 面向异构数据的半分布式联邦学习安全聚合
- Secure aggregation for semi-decentralized federated learning under heterogeneous data
- 网络与信息安全学报 2025年11卷第3期页码：175-189
- 作者机构：
  
  青岛科技大学信息科学技术学院，山东青岛 266042
- 作者简介：
  
  [ "黄梅（1998- ），女，陕西安康人，青岛科技大学硕士生，主要研究方向为联邦学习隐私保护。" ]
  [ "王玲玲（1982- ），女，四川巴中人，博士，青岛科技大学副教授、硕士生导师，主要研究方向为应用密码学及联邦学习隐私安全。" ]
  [ "张政胤（2000- ），男，山东枣庄人，青岛科技大学硕士生，主要研究方向为联邦学习隐私保护。" ]
  [ "刘宇飞（2001- ），女，山东聊城人，青岛科技大学硕士生，主要研究方向为联邦学习隐私保护。" ]
  [ "孙艺桐（2000- ），女，山东日照人，青岛科技大学硕士生，主要研究方向为联邦学习隐私保护。" ]
- 基金信息：
  
  国家自然科学基金(61802217);山东省自然科学基金(ZR2023MF082);青岛科技计划重点研发项目(22-3-4-xxgg-10-gx)
- DOI：10.11959/j.issn.2096-109x.2025029
  中图分类号： TP309.2
- 收稿：2024-10-31，
  
  修回：2025-03-30，
  
  纸质出版：2025-06-25
- 稿件说明：
移动端阅览
黄梅, 王玲玲, 张政胤, 等. 面向异构数据的半分布式联邦学习安全聚合[J]. 网络与信息安全学报, 2025, 11(3): 175-189.

Format: HUANG M, WANG L L, ZHANG Z Y, et al. Secure aggregation for semi-decentralized federated learning under heterogeneous data[J]. Chinese Journal of Network and Information Security, 2025, 11(3): 175-189.
黄梅, 王玲玲, 张政胤, 等. 面向异构数据的半分布式联邦学习安全聚合[J]. 网络与信息安全学报, 2025, 11(3): 175-189. DOI： 10.11959/j.issn.2096-109x.2025029.

Format: HUANG M, WANG L L, ZHANG Z Y, et al. Secure aggregation for semi-decentralized federated learning under heterogeneous data[J]. Chinese Journal of Network and Information Security, 2025, 11(3): 175-189. DOI： 10.11959/j.issn.2096-109x.2025029.

摘要

半分布式联邦学习结合集中式和分布式的优势，提高了系统的可扩展性和灵活性。然而，开放异构的联邦学习架构下，半诚实参与者和异构本地数据的存在加剧了模型聚合过程中节点隐私泄露和模型难收敛的问题。一方面，节点间复杂的邻接关系增加了本地数据隐私泄露的风险，节点数据的隐私保护难以得到有效保障。另一方面，本地模型之间的差异随着本地数据异构程度的增大而增大，进而减缓了全局模型收敛速度，降低了全局模型聚合的准确性。为解决上述问题，提出了一种面向异构数据的半分布式联邦学习安全聚合方案，在实现节点隐私保护的同时提升全局模型收敛效率。基于交替方向乘子法设计随机掩蔽机制，在保证模型正确聚合的前提下避免任何实体访问单个节点的本地模型，增强了本地数据的隐私保护。此外，提出了一种双重加权聚合策略，基于全局模型在不同节点样本上的损失设计节点簇内聚合权重，根据不同分簇对全局模型的贡献为其设计全局聚合权重。在3个公共标准数据集上进行的综合实验表明，与先进的方案相比，提升了数据非独立同分布时的模型收敛速度和准确率。

Abstract

Semi-decentralized federated learning has gained attention for combining the advantages of centralized and decentralized approaches

thereby enhancing system scalability and flexibility. However

in the open heterogeneous federated learning architecture

the existence of semi-honest participants and heterogeneous data has aggravated the problems of node privacy leakage and difficult model convergence during model aggregation. The differences between local models have increased with the degree of local data heterogeneity

which has further slowed down the global model’ convergence and reduced its accuracy. Additionally

the complex network topology among nodes has heightened the risk of local data privacy leakage

making it difficult to ensure local data privacy protection. To tackle these issues

a secure aggregation scheme for heterogeneous data under a semi-distributed federated learning architecture was designed. This scheme aimed to enhance the global model’s convergence performance while safeguarding node privacy. A random masking mechanism was developed based on the alternating direction multiplier method. This mechanism strengthened local data privacy protection and prevented access to individual node models. Moreover

a double-weighted aggregation strategy was proposed. In this strategy

intra-cluster aggregation weights for nodes were determined based on the loss of the global model on different node samples

and global aggregation weights for different clusters were established according to their contribution to the global model. Extensive experiments were carried out on three public standard datasets. The results demonstrate that

compared with advanced schemes

the proposed method improves model convergence speed and accuracy in the context of heterogeneous data.

关键词

Keywords

references

周传鑫 , 孙奕 , 汪德刚 , 等 . 联邦学习研究综述 [J ] . 网络与信息安全学报 , 2021 , 7 ( 5 ): 77 - 92 .

ZHOU C X , SUN Y , WANG D G , et al . A survey of federated learning research [J ] . Chinese Journal of Network and Information Security , 2021 , 7 ( 5 ): 77 - 92 .

LIN F , HOSSEINALIPOUR S , AZAM S , et al . Semi-decentralized federated learning with cooperative D2D local model aggregations [J ] . IEEE Journal on Selected Areas in Communications , 2021 , 39 ( 12 ): 3851 - 3869 .

SATHYA S S , VEPAKOMMA P , RASKAR R , et al . A review of homomorphic encryption libraries for secure computation [C ] // CoRR . 2018 , 1 - 12 .

BOYLE E , GILBOA N , ISHAI Y , et al . Function secret sharing: Improvements and extensions [C ] // ACM Conference on Computer and Communications . 2016 : 1292 - 1303 .

PARASNIS R , HOSSEINALIPOUR S , CHU Y , et al . Energy-efficient connectivity-aware learning over time-varying D2D networks [J ] . IEEE Journal of Selected Topics in Signal Processing , 2024 , 18 ( 2 ): 242 - 258 .

WANG B , FANG J , LI H , et al . Confederated learning: federated learning with decentralized edge servers [J ] . IEEE Transactions on Signal Processing , 2023 , 71 : 248 - 263 .

HARD K , PARTRIDGE , MATHEWS R , et al . Jointly learning from decentralized (federated) and centralized data to mitigate distribution shift [C ] // NeurIPS Workshop on Distribution Shifts: Connecting Methods and Applications . 2021 : 1 – 8 .

BELLET A , KERMARREC A M , LAVOIE E . D-cliques: compensating for data heterogeneity with topology in decentralized federated learning [C ] // International Symposium on Reliable Distributed Systems , 2022 , 1 - 11 .

YEMINI M , SAHA R , OZFATURA E , et al . Semi-decentralized federated learning with collaborative relaying [C ] // 2022 IEEE International Symposium on Information Theory . Piscataway : IEEE Press , 2022 : 1471 - 1476 .

SUN Y , SHAO J , MAO Y , et al . Semi-decentralized federated edge learning with data and device heterogeneity [J ] . IEEE Transactions on Network and Service Management , 2023 , 20 ( 2 ): 1487 - 1501 .

LI X , HUANG K , YANG W , et al . On the convergence of FedAvg on non-IID data [C ] // International Conference on Learning Representations . 2020 : 1 - 26 .

CHEN S , SHEN C , ZHANG L , et al . Dynamic aggregation for heterogeneous quantization in federated learning [J ] . IEEE Transactions on Wireless Communications , 2021 , 20 ( 10 ): 6804 - 6819 .

WANG J , LIU Q , LIANG H , et al . A novel framework for the analysis and design of heterogeneous federated learning [J ] . IEEE Transactions on Signal Processing , 2021 , 69 : 5234 - 5249 .

CAO X , FANG X , LIU J , et al . FLTrust: Byzantine-robust federated learning via trust bootstrapping [C ] // Network and Distributed System Security Symposium . 2021 .

ZHAO Z , JOSHI G , A dynamic reweighting strategy for fair federated learning [C ] // ICASSP IEEE International Conference on Acoustics, Speech and Signal Processing . 2022 : 8772 - 8776 .

MA Q , XU Y , XU H , et al . FedSA: A semi-asynchronous federated learning mechanism in heterogeneous edge computing [J ] , IEEE Journal on Selected Areas in Communications , 2021 , 39 ( 12 ): 3654 - 3672 .

ZHAO J , ZHU H , WANG F , et al . PVD-FL: A privacy-preserving and verifiable decentralized federated learning framework [J ] . IEEE Transactions on Information Forensics and Security , 2022 , 17 : 2059 - 2073 .

LIU J , LI Y , ZHAO M , et al . EPFFL: enhancing privacy and fairness in federated learning for distributed e-healthcare data sharing services [J ] . IEEE Transactions on Dependable and Secure Computing , 2024 , 1 - 14 .

CHEN Q , WANG Z , WANG H , et al . FedDual: Pair-wise gossip helps federated learning in large decentralized networks [J ] . IEEE Transactions on Information Forensics and Security , 2022 , 18 : 335 - 350 .

XU M , ZOU Z , CHENG Y , et al . SPDL: Blockchain-secured and privacy-preserving decentralized learning [J ] . IEEE Transactions on Computers , 2023 , 72 ( 2 ): 548 - 558 .

CHEN S , WANG Y , YU D , et al . Privacy enhanced decentralized federated learning at dynamic edge [J ] . IEEE Transactions on Computers , 2023 , 72 ( 8 ): 2165 - 2180 .

GAO W , ZHAO Z , WEI M , et al . Decentralized online bandit federated learning over unbalanced directed networks [J ] . IEEE Transactions on Network Science and Engineering , 2024 , 1 - 14 .

BELL J H , BONAWITZ K A , GASCÓN A , et al . Secure single- server aggregation with (poly) logarithmic overheadC ] // ACM Conference on Computer and Communications . 2020 : 1253 - 1269 .

郑诚波 , 闫皓楠 , 傅彩利 , 等 . 面向智能物联网的双层级联邦安全学习架构 [J ] . 网络与信息安全学报 , 2024 , 10 ( 6 ): 71 - 80 .

ZHENG C B , YAN H N , FU C L , et al . Double layer federated security learning architecture for artificial intelligence of things [J ] . Chinese Journal of Network and Information Security , 2024 , 10 ( 6 ): 71 - 80 .

董业 , 侯炜 , 陈小军 , 等 . 基于秘密分享和梯度选择的高效安全联邦学习 [J ] . 计算机研究与发展 , 2020 , 57 ( 10 ): 2241 - 2250 .

DONG Y , HOU W , CHEN X J , et al . Efficient and secure federated learning based on secret sharing and gradients selection [J ] . Journal of Computer Research and Development , 2020 , 57 ( 10 ): 2241 - 2250 .

SHRESTHA A K , KHAN F A , SHAIKH M A , et al . Enhancing scalability and reliability in semi-decentralized federated learning with blockchain: Trust penalization and asynchronous functionality [C ] // 2023 IEEE 14th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON) . Piscataway : IEEE Press , 2023 : 0230 - 0236 .

JEON B , FERDOUS S M , RAHMAN M R , et al . Privacy-preserving decentralized aggregation for federated learning [C ] // IEEE Conference on Computer Communications Workshops . Piscataway : IEEE Press , 2021 : 1 - 6 .

BLUM M , MICALI S . How to generate cryptographically strong sequences of pseudo-random bits [C ] // Providing Sound Foundations for Cryptography . 2019 : 227 - 240 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于隐私集合求交的大模型双向隐私授权

基于PIR增强的可搜索加密方案研究

面向LBS的ADS-B报文定位数据本地化差分隐私保护方法

FL-IoV：基于联邦学习的综合车联网入侵检测方法

LightLSB：算力受限环境下大规模生物特征图像的轻量级安全传输机制