Integrating Federated Learning with Blockchain for Secure and Efficient Vehicle-to-Vehicle (V2V) Communication

Abstract

The future of autonomous vehicles depends on secure, efficient, and trustworthy Vehicle-to-Vehicle (V2V) communication systems. Traditional centralized architectures face critical limitations, including cybersecurity risks, data privacy vulnerabilities, and high latency, which hinder real-time decision-making in dynamic automotive environments. To address these challenges, this work proposes a decentralized framework integrating Federated Learning (FL) with blockchain technology, designed specifically for privacy-preserving, low-latency V2V communication. A custom deep neural network (DNN) contains two hidden layers accommodating 64 followed by 32 neurons while including batch normalization features and dropout regularization for local vehicle training. The FL paradigm allows vehicles to train the model collaboratively without exposing raw data while simultaneously minimizing their network utilization. The global model aggregation combines weighted averaging techniques that utilize local dataset sizes to achieve 88.89% accuracy during synthetic V2V dataset collision prediction tasks. Real-time responsiveness is achieved through this framework which cuts communication delays by 30% as compared to centralized systems by using optimized parameter exchange along with parallel processing. Blockchain integration addresses trust and security gaps: model updates are hashed using SHA-256 and immutably stored via a lightweight Practical Byzantine Fault Tolerance (PBFT) consensus mechanism, enabling auditability while maintaining low overhead. Batch processing of transactions further mitigates blockchain’s inherent latency, ensuring compatibility with time-sensitive V2V operations. Future work will deploy a permissioned blockchain network using smart contracts to automate model validation and mitigate adversarial attacks. The framework will be tested on real-world vehicular datasets to evaluate scalability in urban traffic scenarios. This work advances intelligent transportation systems by enabling secure, decentralized collaboration between vehicles, directly enhancing road safety and traffic efficiency.