A blockchain-enabled privacy-preserving and incentive mechanism-driven federated learning scheme for IoV

Vehicle collaboration and data sharing in the Internet of Vehicles (IoV) can significantly improve road traffic efficiency. However, they also introduce risks related to user privacy breaches and data security. Effectively ensuring privacy protection during data sharing remains a critical challenge. In response, this paper proposes CDPFL, a federated learning-based data privacy protection solution driven by blockchain and incentive mechanisms. CDPFL leverages blockchain technology to ensure the transparency, immutability, and traceability of the data sharing process, while differential privacy techniques are employed to add Gaussian noise during local model training, effectively preventing privacy leakage. Additionally, the paper incorporates PBFT consensus and BFT gradient aggregation rules to tolerate Byzantine nodes. A token-based incentive mechanism is designed to encourage vehicles to actively contribute computing resources and high-quality data, while also regulating their behavior. Experimental results demonstrate that CDPFL achieves efficient, secure, and accurate federated learning within the IoV environment, exhibiting strong scalability and robustness.