Blockchain-Based Trust Federated Learning Framework for Iov Security

Recent advancements in intelligent automobiles and artificial intelligence (AI) have sparked significant interest in Internet of Vehicles (IoV) technology. While conventional machine learning methods have been widely used to enhance IoV security, they are not well-equipped to handle the complexities of IoV communications or prevent malicious vehicles from influencing the ML model formation process. These limitations highlight the urgent need for more effective IoV security solutions to ensure the integrity and reliability of vehicular communication networks. To address these challenges, we propose a novel blockchain-based trust-federated learning (FL) framework for IoV attack detection. This framework incorporates a trust-based FL model to enhance the security of IoV communications. We introduce a unique trust value system for vehicles, which improves the reliability of the FL model by selectively using data from trusted vehicles. Additionally, we employ a two-level blockchain approach: the InterPlanetary File System (IPFS) for off-chain local model storage and a dedicated blockchain managed by RSUs for global model aggregation and storage. Experimental results demonstrate the effectiveness of our solution in strengthening IoV communication security.