A distributed learning framework with blockchain and privacy-preserving for IoV

The Internet of Vehicles (IoV), as a critical component of the Internet of Things (IoT), constructs a distributed system comprising vehicles, roadside units (RSUs), and cloud servers. In the IoV environment, the secure sharing of data and the protection of privacy are of paramount importance, as they directly impact the decision-making processes of intelligent vehicles and overall road safety. Given the openness of IoV, it faces risks of privacy leakage and poisoning attacks during data exchange and model training, which threaten the integrity and reliability of the data. To address these challenges, this study proposes a privacy protection framework that integrates blockchain technology and differential privacy. This framework incorporates dual differential privacy techniques within federated learning to enhance data privacy protection and designs a dynamic gradient aggregation mechanism to defend against data poisoning attacks, thereby ensuring the security of the data. Experimental results demonstrate that this framework maintains high model accuracy even under attack rates of up to 30%, exhibiting remarkable resilience against such attacks. Overall, this study emphasizes the significance of data security and privacy protection in the IoV domain and illustrates the potential of blockchain and differential privacy technologies in enhancing the security of IoV data and safeguarding user privacy. This research provides robust support for the sustainable development of IoV.