Randomized DP-DFL: Towards Differentially Private Decentralized Federated Learning via Randomized Model Interaction
IEEE Transactions on Mobile Computing · 2025-12 · 24 (12) · 12802-12817
DOI: 10.1109/tmc.2025.3588537 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2025-12. Verfasst von Weihao Zhu und 6 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 46 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Program of Science and Technology Cooperation of Nanjing with International/Hong Kong, Macao, and Taiwan, Fundamental Research Funds for the Central Universities.
Abstract (Originalsprache)
Traditional federated learning (FL) frameworks rely on a central server for model coordination among distributed mobile terminals (MTs). The centralization faces two critical challenges, i.e., single point of failure and potential privacy leakage. Differentially private decentralized FL (DP-DFL) has been proposed to address these challenges, wherein the MTs exchange models in a decentralized manner and maintain the differential privacy (DP) guarantee by adding noise to local models before model interaction. However, existing DP-DFL frameworks confront difficulty in achieving the expected privacy and convergence performance, simultaneously. To address this issue, we propose a novel DP-DFL framework (called randomized DP-DFL) that employs a randomized model interaction scheme to lower the model exposure frequency and hence reduce privacy budget consumption. Specifically, the scheme includes two sequential steps, i.e., randomized terminal assignment and randomized model transmission. In Step 1), the model interaction phase of DFL is further divided into several sequential sub-stages. MTs are randomly assigned to each sub-stage. In Step 2), each MT sequentially transmits either a model previously received from its neighbors or its own local model according to the assigned sub-stage order. The proposed scheme enhances the MTs’ privacy of DFL since the exposure probabilities of the MTs’ local models are significantly reduced via these two randomized steps. Besides, we theoretically analyze the convergence and privacy performance of randomized DP-DFL. In particular, properly tuning the number of sub-stages in randomized DP-DFL can achieve an optimal balance between privacy and convergence. Experimental results show that randomized DP-DFL consistently outperforms traditional frameworks. Compared with baselines, randomized DP-DFL reduces 40.9% privacy loss under the same target accuracy while improving 9.5% learning accuracy under the same privacy loss on EMNIST and CIFAR-10, respectively.
Förderung
- National Natural Science Foundation of China
- Program of Science and Technology Cooperation of Nanjing with International/Hong Kong, Macao, and Taiwan
- Fundamental Research Funds for the Central Universities
- Key Programs of Ningbo Municipal Natural Science Foundation
- Key Technologies R&D Program of Jiangsu
- Major Natural Science Foundation of the Higher Education Institutions of Jiangsu Province
Zitieren
APA
Zhu, W., Shi, L., Wei, K., Zhou, Y., Wang, Z., Xiong, Z., & Li, J. (2025). Randomized DP-DFL: Towards Differentially Private Decentralized Federated Learning via Randomized Model Interaction. IEEE Transactions on Mobile Computing, 24(12), 12802-12817. https://doi.org/10.1109/tmc.2025.3588537
Vancouver
Zhu W, Shi L, Wei K, Zhou Y, Wang Z, Xiong Z, et al. Randomized DP-DFL: Towards Differentially Private Decentralized Federated Learning via Randomized Model Interaction. IEEE Transactions on Mobile Computing. 2025;24(12):12802-12817. doi:10.1109/tmc.2025.3588537
BibTeX
@article{Zhu2025,
author = {Zhu, Weihao and Shi, Long and Wei, Kang and Zhou, Yipeng and Wang, Zhe and Xiong, Zehui and Li, Jun},
title = {Randomized DP-DFL: Towards Differentially Private Decentralized Federated Learning via Randomized Model Interaction},
journal = {IEEE Transactions on Mobile Computing},
year = {2025},
volume = {24},
number = {12},
pages = {12802--12817},
doi = {10.1109/tmc.2025.3588537}
}
EUVIMED