Blockchain combined with Federated Learning technology research

Federated learning has garnered significant attention and been widely explored in various domains as a means to train models while preserving data privacy through local data sharing. However, several challenges persist within federated learning, including privacy information leakage, the reliability of uploading model parameters to servers, and communication costs. In recent years, researchers and practitioners have made substantial progress in addressing these challenges by integrating blockchain technology with federated learning. This fusion has resulted in enhanced data privacy protection, incentivized data sharing, and improved verification accountability. To provide a comprehensive understanding of the latest research findings and development trends in blockchain-based federated learning, this review paper aims to present a detailed analysis. Initially, the paper examines the inherent issues of federated learning and identifies potential security and privacy attacks that blockchain can mitigate. Furthermore, the distinct characteristics of blockchain-based federated learning are elucidated. Subsequently, an in-depth exploration of state-of-the-art approaches in blockchain-based federated learning is conducted, focusing on security and privacy mechanisms, recording and rewarding systems, as well as verification and accountability protocols. Additionally, the paper delves into pertinent topics concerning the integration of blockchain and federated learning, addressing associated challenges and considerations.