Adaptive Privacy-Preserving Federated Learning Scheme Based on Incentive Mechanism

doi:10.19678/j.issn.1000-3428.0070751

Abstract

Abstract: Federated Learning (FL) leverages client data resources to collaboratively train a global model, whose performance depends on the quality of client data and the level of client participation. Clients expect to receive appropriate compensation after contributing high-quality data to enhance their motivation to participate. In addition, because the local model parameters uploaded by clients contain information about private data resources, they face the risk of privacy leakage. To address these challenges, this paper proposes an incentive-based adaptive privacy-preserving FL scheme. First, a pre-decision game auction mechanism is designed to ensure that clients truthfully report their costs while achieving Nash Equilibrium (NE). Second, a training quality evaluation algorithm is developed based on training time and model loss, which determines client compensation according to the overall training quality evaluation score, thereby incentivizing high-quality data contributors to participate in training. Finally, an adaptive differential privacy technique is employed to perturb the local model parameters and enhance the model utility through dynamic noise allocation. A theoretical analysis demonstrates that the proposed scheme satisfies security and privacy protection requirements, and experimental results validate its effectiveness.

Key words: Federated Learning (FL), Nash Equilibrium (NE), incentive mechanism, privacy leakage, differential privacy

摘要： 联邦学习(FL)利用客户端的数据资源协作训练全局模型，全局模型效果取决于客户端的数据资源质量和其参与的积极性。因此，客户端期望在贡献高质量数据资源后得到相应价格补偿以提升其参与训练的积极性。此外，由于客户端上传的本地模型参数蕴含私有数据资源信息，因而面临隐私泄露的风险。为了解决这些问题，提出了一种基于激励机制的自适应隐私保护联邦学习方案。首先，设计了先决博弈拍卖机制，确保客户端诚实报告成本并实现纳什均衡(NE)。其次，基于训练时间和模型损失值设计了训练质量评估算法，根据总训练质量评估值的高低给予客户端相应的价格补偿，激励高质量数据的客户端参与训练。最后，采用自适应差分隐私技术，对本地模型参数进行扰动，通过动态噪声分配提升模型可用性。理论分析表明，设计方案满足安全性和隐私保护需求，实验结果也验证了其有效性。

关键词: 联邦学习, 纳什均衡, 激励机制, 隐私泄露, 差分隐私

CLC Number:

TP309

GUO Ziyun, TIAN Youliang, LI Mengqian. Adaptive Privacy-Preserving Federated Learning Scheme Based on Incentive Mechanism[J]. Computer Engineering, 2026, 52(11): 269-283.

郭子赟, 田有亮, 李梦倩. 基于激励机制的自适应隐私保护联邦学习方案[J]. 计算机工程, 2026, 52(11): 269-283.

/ Recommend / Download Citations

URL: https://www.ecice06.com/EN/10.19678/j.issn.1000-3428.0070751

https://www.ecice06.com/EN/Y2026/V52/I11/269

References

[1] YANG Q, LIU Y, CHEN T J, et al. Federated machine learning: concept and applications[J]. ACM Transactions on Intelligent Systems and Technology, 2019, 10(2): 1-19.
[2] BLANCHARD P, MHAMDI E M, GUERRAOUI R, et al. Machine learning with adversaries: Byzantine tolerant gradient descent[J]. Advances in Neural Information Processing Systems, 2017, 30: 119-129.
[3] GUERRAOUI R, ROUAULT S. The hidden vulnerability of distributed learning in Byzantium[C]//Proceedings of International Conference on Machine Learning. [S. l.]: PMLR, 2018: 3521-3530.
[4] KANG J W, XIONG Z H, NIYATO D, et al. Incentive mechanism for reliable federated learning: a joint optimization approach to combining reputation and contract theory[J]. IEEE Internet of Things Journal, 2019, 6(6): 10700-10714.
[5] ZHU L, LIU Z, HAN S. Deep leakage from gradients[J]. Advances in Neural Information Processing Systems, 2019, 32: 14774-14784.
[6] ZHAO B, MOPURI K R, BILEN H. iDLG: improved deep leakage from gradients[EB/OL]. [2024-10-11]. https://arxiv.org/abs/2001.02610.
[7] HAO M, LI H W, XU G W, et al. Towards efficient and privacy-preserving federated deep learning[C]//Proceedings of 2019 IEEE International Conference on Communications (ICC). Shanghai, China: IEEE Press, 2019: 1-6.
[8] MCMAHAN H B, RAMAGE D, TALWAR K, et al. Learning differentially private recurrent language models[EB/OL]. [2024-10-11]. https://arxiv.org/abs/1710.06963.
[9] DENG Y H, LYU F, REN J, et al. AUCTION: automated and quality-aware client selection framework for efficient federated learning[J]. IEEE Transactions on Parallel and Distributed Systems, 2022, 33(8): 1996-2009.
[10] WANG Z B, LI J X, HU J H, et al. Towards privacy-driven truthful incentives for mobile crowdsensing under untrusted platform[J]. IEEE Transactions on Mobile Computing, 2023, 22(2): 1198-1212.
[11] JIAO Y T, WANG P, NIYATO D, et al. Toward an automated auction framework for wireless federated learning services market[J]. IEEE Transactions on Mobile Computing, 2021, 20(10): 3034-3048.
[12] 刘艺璇, 陈红, 刘宇涵, 等. 联邦学习中的隐私保护技术[J]. 软件学报, 2022, 33(3): 1057-1092.LIU Y X, CHEN H, LIU Y H, et al. Privacy-preserving techniques in federated learning[J]. Journal of Software, 2022, 33(3): 1057-1092. (in Chinese)
[13] LYU L J, CHEN C. A novel attribute reconstruction attack in federated learning[EB/OL]. [2024-10-11]. https://arxiv.org/abs/2108.06910.
[14] GU Y H, BAI Y B, XU S B. CS-MIA: membership inference attack based on prediction confidence series in federated learning[J]. Journal of Information Security and Applications, 2022, 67: 103201.
[15] HU R, GUO Y X, GONG Y M. Federated learning with sparsified model perturbation: improving accuracy under client-level differential privacy[J]. IEEE Transactions on Mobile Computing, 2024, 23(8): 8242-8255.
[16] MCMAHAN B, MOORE E, RAMAGE D, et al. Communication-efficient learning of deep networks from decentralized data[C]//Proceedings of the International Conference on Artificial Intelligence and Statistics. [S. l.]: PMLR, 2017: 1273-1282.
[17] WEI K, LI J, DING M, et al. Federated learning with differential privacy: algorithms and performance analysis[J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 3454-3469.
[18] ZHAO L C, WANG Q, ZOU Q, et al. Privacy-preserving collaborative deep learning with unreliable participants[J]. IEEE Transactions on Information Forensics and Security, 2020, 15: 1486-1500.
[19] 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. (in Chinese)
[20] LI Y, DU W, HAN L Q, et al. A communication-efficient, privacy-preserving federated learning algorithm based on two-stage gradient pruning and differentiated differential privacy[J]. Sensors, 2023, 23(23): 9305.
[21] VARUN M, FENG S Y, WANG H, et al. Towards accurate and stronger local differential privacy for federated learning with staircase randomized response[C]//Proceedings of the 14th ACM Conference on Data and Application Security and Privacy. New York, USA: ACM Press, 2024: 307-318.
[22] TIAN D X, ZHOU J S, WANG Y P, et al. Channel access optimization with adaptive congestion pricing for cognitive vehicular networks: an evolutionary game approach[J]. IEEE Transactions on Mobile Computing, 2020, 19(4): 803-820.
[23] CHIVUKULA A S, YANG X H, LIU W, et al. Game theoretical adversarial deep learning with variational adversaries[J]. IEEE Transactions on Knowledge and Data Engineering, 2021, 33(11): 3568-3581.
[24] 刘亮, 毛武平, 李汶蔚, 等. 空天地一体化边缘计算网络中基于博弈论的任务卸载策略[J]. 计算机工程, 2025, 51(2): 238-249.LIU L, MAO W P, LI W W, et al. Task offloading strategy based on game theory in the space—air—ground integrated edge computing networks[J]. Computer Engineering, 2025, 51(2): 238-249. (in Chinese)
[25] XU J L, WANG S G, ZHANG N, et al. Reward or penalty: aligning incentives of stakeholders in crowdsourcing[J]. IEEE Transactions on Mobile Computing, 2019, 18(4): 974-985.
[26] HU R, GONG Y M. Trading data for learning: incentive mechanism for on-device federated learning[C]//Proceedings of 2020 IEEE Global Communications Conference. Taipei, China: IEEE Press, 2020: 1-6.
[27] SUN P, CHE H X, WANG Z B, et al. Pain-FL: personalized privacy-preserving incentive for federated learning[J]. IEEE Journal on Selected Areas in Communications, 2021, 39(12): 3805-3820.
[28] YI Z N, JIAO Y T, DAI W T, et al. A Stackelberg incentive mechanism for wireless federated learning with differential privacy[J]. IEEE Wireless Communications Letters, 2022, 11(9): 1805-1809.
[29] WANG D, REN J, WANG Z B, et al. PrivAim: a dual-privacy preserving and quality-aware incentive mechanism for federated learning[J]. IEEE Transactions on Computers, 2023, 72(7): 1913-1927.
[30] ZHENG Z H, HONG Y M, LI K Q, et al. An incentive mechanism based on AoU, data quality, and data quantity for federated learning[C]//Proceedings of the 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD). Tianjin, China: IEEE Press, 2024: 1832-1837.
[31] ZHANG L F, ZHU T Q, XIONG P, et al. A robust game-theoretical federated learning framework with joint differential privacy[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(4): 3333-3346.
[32] 王勇, 李国良, 李开宇. 联邦学习贡献评估综述[J]. 软件学报, 2023, 34(3): 1168-1192.WANG Y, LI G L, LI K Y. Survey on contribution evaluation for federated learning[J]. Journal of Software, 2023, 34(3): 1168-1192. (in Chinese)
[33] DWORK C, ROTH A. The algorithmic foundations of differential privacy[J]. Foundations and Trends^® in Theoretical Computer Science, 2014, 9(3/4): 211-487.
[34] CHEN Z L, NI T J, ZHONG H, et al. Differentially private double spectrum auction with approximate social welfare maximization[J]. IEEE Transactions on Information Forensics and Security, 2019, 14(11): 2805-2818.
[35] MYERSON R B. Optimal auction design[J]. Mathematics of Operations Research, 1981, 6(1): 58-73.
[36] DENG L. The MNIST database of handwritten digit images for machine learning research[J]. IEEE Signal Processing Magazine, 2012, 29(6): 141-142.
[37] XIAO H, RASUL K, VOLLGRAF R. Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms[EB/OL]. [2021-10-11]. https://arxiv.org/abs/1708.07747.

Please choose a citation manager

Content to export