Federated learning (FL), as a secure distributed learning framework, has attracted widespread interest in the field of the Internet of Things (IoT) because it can effectively protect the data privacy of participants. However, the traditional FL architecture is, to some extent, centralized and requires a central server to be responsible for model updates and aggregation. This centralized structure is susceptible to single-point attacks, which may cause the entire FL system to crash, and is also vulnerable to attacks from free riders, thereby affecting fairness and safety. To address these challenges, a completely distributed structure that is different from traditional centralized management structures is proposed. The local model from clients is decentralized and managed through a consortium blockchain with undeniable and tamper-resistant features. In addition, the reputation evaluation for clients is introduced in the block-chain to prevent attacks from hitchhikers, and honest participants with different rewards are rewarded based on different reputation values. The experiment results show that this method can achieve high fairness and identify and exclude free riders effectively.
ZHANGC, XIEY, BAIH, et al. A survey on federated learning[J]. Knowledge-Based Systems, 2021, 216: 106775.
[2]
KAIROUZP, MCMAHANH B, AVENTB, et al. Advances and open problems in federated learning[J]. Foundations and Trends in Machine Learning, 2021, 14(1/2): 1-210.
[3]
LIT, SAHUA K, TALWALKARA, et al. Federated learning: challenges, methods, and future directions[J]. IEEE Signal Processing Magazine, 2020, 37(3): 50-60.
[4]
YANGH H, LIUZ, QUEKT Q S, et al. Scheduling policies for federated learning in wireless networks[J]. IEEE Transactions on Communications, 2020, 68(1): 317-333.
[5]
BONAWITZK, IVANOVV, KREUTERB, et al. Practical secure aggregation for privacy-preserving machine learning[C]//Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, 2017: 1175-1191.
[6]
LÜL, YUJ, NANDAKUMARK, et al. Towards fair and privacy-preserving federated deep models[EB/OL]. Computer Science, 2019.
[7]
AICHS, SINAIN K, KUMARS, et al. Protecting personal healthcare record using blockchain & federated learning technologies[C]//2021 23rd International Conference on Advanced Communication Technology (ICACT), 2021: 109-112.
TOYODAK, ZHANGA N. Mechanism design for an incentive-aware blockchain-enabled federated learning platform[C]//2019 IEEE International Conference on Big Data, 2019: 395-403.
[10]
HOUD, ZHANGJ, MANK L, et al. A systematic literature review of blockchain-based federated learning: architectures, applications and issues[C]//2021 2nd Information Communication Technologies Conference (ICTC), 2021: 302-307.
[11]
ABDEL WAHABO, MOURADA, OTROKH, et al. Federated machine learning: survey, multi-level classification, desirable criteria and future directions in communication and networking systems[J]. IEEE Communications Surveys & Tutorials, 2021, 23(2): 1342-1397.
[12]
NGUYEND C, DINGM, PHAMQ V, et al. Federated learning meets blockchain in edge computing: opportunities and challenges[C]//IEEE Internet of Things Journal, 2021: 12806-12825.
[13]
LINJ, DUM, LIUJ. Free-riders in federated learning: attacks and defenses[EB/OL]. Computer Science, 2019.
[14]
FRABONIY, VIDALR, LORENZIM. Free-rider attacks on model aggregation in federated learning[C]//24th International Conference on Artificial Intelligence and Statistics, 2021: 1846-1854.
[15]
WANGJ, CHANGX, RODRÌGUEZR J, et al. Assessing anonymous and selfish free-rider attacks in federated learning[C]//2022 IEEE Symposium on Computers and Communications (ISCC), 2022: 1-6.
[16]
XUJ, XUZ, LINJ, et al. Double security guarantee: protecting user privacy and model security in QoS prediction[C]//2022 IEEE International Conference on Services Computing (SCC), 2022: 140-145.
[17]
NGUYENN, KHANM M H. Context aware data acquisition framework for dynamic data driven applications systems (DDDAS)[C]//MILCOM 2013-IEEE Military Communications Conference, 2013: 334-341.