基于动态聚类和信誉评估的抗攻击联邦框架

张唯一; 罗启瑞; 张琳

doi:10.11959/j.issn.2096-109x.2025069

您当前的位置：

首页 >

文章列表页 >

基于动态聚类和信誉评估的抗攻击联邦框架

学术论文 | 更新时间：2026-01-28

- 基于动态聚类和信誉评估的抗攻击联邦框架
- Anti-attack federated framework based on dynamic clustering and reputation evaluation
- 网络与信息安全学报 2025年11卷第6期页码：120-130
- 作者机构：
  
  南京邮电大学计算机学院，江苏南京 210003
- 作者简介：
  
  [ "张唯一（2001- ），女，安徽阜阳人，南京邮电大学硕士生，主要研究方向为联邦学习、隐私保护。" ]
  [ "罗启瑞（2004- ），男，江苏徐州人，南京邮电大学本科生，主要研究方向为联邦学习、隐私保护。" ]
  [ "张琳（1980- ），女，江苏丰县人，博士，南京邮电大学副教授、硕士生导师，主要研究方向为网络安全、可信计算、联邦学习。" ]
- 基金信息：
  
  国家自然科学基金(62372247;61872194);南京邮电大学校级自然科学基金(NY222142)
- DOI：10.11959/j.issn.2096-109x.2025069
  中图分类号： TP309
- 收稿：2025-05-14，
  
  修回：2025-07-25，
  
  纸质出版：2025-12-25
- 稿件说明：
移动端阅览
张唯一, 罗启瑞, 张琳. 基于动态聚类和信誉评估的抗攻击联邦框架[J]. 网络与信息安全学报, 2025, 11(6): 120-130.

ZHANG Weiyi, LUO Qirui, ZHANG Lin. Anti-attack federated framework based on dynamic clustering and reputation evaluation[J]. Chinese Journal of Network and Information Security, 2025, 11(6): 120-130.
张唯一, 罗启瑞, 张琳. 基于动态聚类和信誉评估的抗攻击联邦框架[J]. 网络与信息安全学报, 2025, 11(6): 120-130. DOI： 10.11959/j.issn.2096-109x.2025069.

ZHANG Weiyi, LUO Qirui, ZHANG Lin. Anti-attack federated framework based on dynamic clustering and reputation evaluation[J]. Chinese Journal of Network and Information Security, 2025, 11(6): 120-130. DOI： 10.11959/j.issn.2096-109x.2025069.

摘要

联邦学习在数据异构场景下的实际应用面临着严峻挑战，即恶意客户端的投毒攻击会降低全局模型性能。针对该挑战，提出了一种基于动态聚类和双维度信誉评估的鲁棒联邦学习框架动态聚类联邦学习（dynamic clustering federated learning

DCFL）。首先，DCFL采用无须预设聚类数量的HDBSCAN算法实现客户端自适应聚类，并基于相似性度量对噪声点进行重分配，将具有相似数据分布特征的客户端动态分组，避免将异构的良性客户端误判为恶意客户端。其次，在类簇内部，综合考察客户端模型更新相似度和损失一致性，利用双维度的互补指标综合评估客户端更新质量。最后，设计了动态阈值调整策略，基于信誉加权的簇内聚合策略有效强化了类簇的个性化功能。实验验证了DCFL在4种不同数据分布和安全环境下的性能优势，尤其是在高异构有攻击场景下，相比传统聚类联邦学习算法测试准确率提升了约9%，证明其在复杂联邦环境中具有良好的鲁棒性和适应性。

Abstract

The practical application of federated learning in scenarios with heterogeneous data faces a serious challenge

which is that poisoning attacks by malicious clients can degrade the global model. A robust federated learning framework DCFL (dynamic clustering federated learning) was proposed based on dynamic clustering and dual-dimensional reputation evaluation. Firstly

DCFL employed the HDBSCAN algorithm for adaptive client clustering without predefined cluster numbers

and redistributed noise points based on similarity measurements

dynamically grouping clients with similar data distribution characteristics

preventing misclassification of heterogeneous but benign clients as malicious. Secondly

within clusters

both model update similarity and loss consistency were examined comprehensively as complementary indicators to evaluate client update quality. Finally

a dynamic threshold adjustment strategy was designed

while reputation-weighted intra-cluster aggregation effectively enhances cluster personalization. Experiments validated DCFL's performance advantages across four different data distribution and security environments. Particularly in high-heterogeneity attack scenarios

DCFL improved testing accuracy by approximately 9 percentage points compared to traditional clustered federated learning algorithms

demonstrating its robustness and adaptability in complex federated environments.

关键词

Keywords

references

SATTLER F , MÜLLER K R , SAMEK W . Clustered federated learning: model-agnostic distributed multitask optimization under privacy constraints [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2021 , 32 ( 8 ): 3710 - 3722 .

万伟 , 胡胜山 , 陆建荣 , 等 . 联邦学习在高度数据异构场景下的泛化鲁棒性增强 [J ] . 中国科学: 信息科学 , 2024 , 54 ( 3 ): 566 - 581 .

WAN W , HU S S , LU J R , et al . Enhancing generalization robustness of federated learning in highly heterogeneous environments [J ] . Scientia Sinica (Informationis) , 2024 , 54 ( 3 ): 566 - 581 .

MA X S , ZHANG J , GUO S , et al . Layer-wised model aggregation for personalized federated learning [C ] // Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . Piscataway : IEEE Press , 2022 : 10082 - 10091 .

ZHANG J Q , HUA Y , WANG H , et al . FedALA: adaptive local aggregation for personalized federated learning [C ] // Proceedings of the AAAI Conference on Artificial Intelligence . Vancouver : AAAI Press , 2023 : 11237 - 11244 .

应作斌 , 方一晨 , 张怡文 . 动态聚合权重的隐私保护联邦学习框架 [J ] . 网络与信息安全学报 , 2022 , 8 ( 5 ): 56 - 65 .

YING Z B , FANG Y C , ZHANG Y W . Privacy-preserving federated learning framework with dynamic weight aggregation [J ] . Chinese Journal of Network and Information Security , 2022 , 8 ( 5 ): 56 - 65 .

GHOSH A , CHUNG J , YIN D , et al . An efficient framework for clustered federated learning [J ] . IEEE Transactions on Information Theory , 2022 , 68 ( 12 ): 8076 - 8091 .

LONG G D , XIE M , SHEN T , et al . Multi-center federated learning: clients clustering for better personalization [J ] . World Wide Web , 2023 , 26 ( 1 ): 481 - 500 .

朱素霞 , 顾玢珂 , 孙广路 . 基于相似度加速的自适应聚类联邦学习 [J ] . 通信学报 , 2024 , 45 ( 3 ): 197 - 207 .

ZHU S X , GU B K , SUN G L . Adaptive clustering federated learning via similarity acceleration [J ] . Journal on Communications , 2024 , 45 ( 3 ): 197 - 207 .

LIU X Y , LI H W , XU G W , et al . Privacy-enhanced federated learning against poisoning adversaries [J ] . IEEE Transactions on Information Forensics and Security , 2021 , 16 : 4574 - 4588 .

康海燕 , 冀源蕊 . 基于本地化差分隐私的联邦学习方法研究 [J ] . 通信学报 , 2022 , 43 ( 10 ): 94 - 105 .

KANG H Y , JI Y R . Research on federated learning approach based on local differential privacy [J ] . Journal on Communications , 2022 , 43 ( 10 ): 94 - 105 .

YANG Z Q , ZHANG Y G , ZHENG Y , et al . FedFed: feature distillation against data heterogeneity in federated learning [C ] // Proceedings of Advances in Neural Information Processing Systems . Vancouver : NeurIPS , 2023 : 60397 - 60428 .

刘艳 , 王田 , 彭绍亮 , 等 . 基于边缘的联邦学习模型清洗和设备聚类方法 [J ] . 计算机学报 , 2021 , 44 ( 12 ): 2515 - 2528 .

LIU Y , WANG T , PENG S L , et al . Edge-based model cleaning and device clustering in federated learning [J ] . Chinese Journal of Computers , 2021 , 44 ( 12 ): 2515 - 2528 .

ZHANG Z X , CAO X Y , JIA J Y , et al . FLDetector: defending federated learning against model poisoning attacks via detecting malicious clients [C ] // Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . New York : ACM , 2022 : 2545 - 2555 .

CHEN Z Y , LIANG J , YU Z X , et al . Resilient collaborative caching for multi-edge systems with robust federated deep learning [J ] . IEEE Transactions on Networking , 2025 , 33 ( 2 ): 654 - 669 .

GU H L , FAN L X , TANG X X , et al . FedCut: a spectral analysis framework for reliable detection of Byzantine colluders [J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024 , 46 ( 9 ): 5905 - 5920 .

徐明迪 , 李正潇 , 王梓行 , 等 . Robust-PFedproto: 基于个性化层的鲁棒联邦原型学习方法 [J ] . 网络与信息安全学报 , 2025 , 11 ( 3 ): 67 - 80 .

XU M D , LI Z X , WANG Z H , et al . Robust-PFedproto: robust federated prototype learning based on personalized layers [J ] . Chinese Journal of Network and Information Security , 2025 , 11 ( 3 ): 67 - 80 .

TAHVILI S , HATVANI L , FELDERER M , et al . Automated functional dependency detection between test cases using Doc2Vec and clustering [C ] // Proceedings of the 2019 IEEE International Conference on Artificial Intelligence Testing (AITest) . Piscataway : IEEE Press , 2019 : 19 - 26 .

雷诚 , 张琳 . 基于更新质量检测和恶意客户端识别的联邦学习模型 [J ] . 计算机科学 , 2024 , 51 ( 11 ): 368 - 378 .

LEI C , ZHANG L . Federated learning model based on update quality detection and malicious client identification [J ] . Computer Science , 2024 , 51 ( 11 ): 368 - 378 .

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

DP-FedDRC：缓解维度坍塌的差分隐私个性化联邦学习

基于风险权值聚合联邦学习的金融欺诈检测模型

FL-IoV：基于联邦学习的综合车联网入侵检测方法

FLWD:基于联邦学习的Webshell检测方法

Robust-PFedproto:基于个性化层的鲁棒联邦原型学习方法