A Blockchain-Based Decentralized Federated Learning Framework with Committee Consensus

Federated learning has been widely studied and applied to various scenarios.\nIn mobile computing scenarios, federated learning protects users from exposing\ntheir private data, while cooperatively training the global model for a variety\nof real-world applications. However, the security of federated learning is\nincreasingly being questioned, due to the malicious clients or central servers'\nconstant attack to the global model or user privacy data. To address these\nsecurity issues, we proposed a decentralized federated learning framework based\non blockchain, i.e., a Blockchain-based Federated Learning framework with\nCommittee consensus (BFLC). The framework uses blockchain for the global model\nstorage and the local model update exchange. To enable the proposed BFLC, we\nalso devised an innovative committee consensus mechanism, which can effectively\nreduce the amount of consensus computing and reduce malicious attacks. We then\ndiscussed the scalability of BFLC, including theoretical security, storage\noptimization, and incentives. Finally, we performed experiments using\nreal-world datasets to verify the effectiveness of the BFLC framework.\n