A Robust Vehicle to Network Communication using Quantum Blockchain Cryptography and Federated Learning

Vehicular Ad Hoc Networks (VANETs) are direct network communications established between vehicles and roadside infrastructure. This form of communication is designed to provide a seamless autonomous driving experience through constant data sharing among autonomous vehicles and their surrounding infrastructures. However, this massive data exchange creates significant data vulnerability and security threats, such as Denial of Service (DoS) attacks, spoofing, and many others within the V2N communication network. To address this problem, the study proposes a novel quantum-resistant framework integrating blockchain technology and federated learning to enhance the security and privacy of V2N communications. In accordance with this objective, a design science research methodology was adopted, and a framework was developed using VEINS as the primary simulation tool for modelling vehicular networks and communication behaviours. In addition, the study employed the use of TensorFlow Federated and Hyperledger Fabric for client model aggregation and a tamper-proof ledger that manages model updates and access requests, respectively. The results of the study reveal that the proposed framework has a 92.4% rate of malicious node detection during the first three FL rounds, an increased speed of 28%, and a reduced packet overhead of 35% as compared to traditional ECC-based methods. Finally, under high-density traffic conditions, the network communication performance improved, with latency decreased by 18% and throughput increased by 22% due to the improved security framework. Highlighting these results, this study provides a unique contribution to research, as it is the first to combine blockchain, Federated Learning, and Post Quantum Cryptography into a singular V2N security framework and achieves results that surpass traditional approaches. While the result provides a promising security indication, future work must extend testing to real-world vehicular environments and explore lightweight consensus mechanisms.