Vehicular Ad Hoc Networks (VANETs) and Vehicle-to-Everything (V2X) systems are critical enablers of intelligent transportation, autonomous mobility, and smart city ecosystems. However, their highly dynamic topology, decentralized communication, and heterogeneous infrastructure expose them to sophisticated cyber threats such as Sybil attacks, spoofing, distributed denial-of-service (DDoS), and malicious route manipulation. This paper proposes an AI-driven secure vehicular networking framework that integrates Federated Learning (FL), Graph Neural Networks (GNN), and Deep Reinforcement Learning (DRL) for privacy-preserving intrusion detection, trust-aware secure routing, and adaptive cyber defense. The proposed system enables decentralized learning across vehicles and edge nodes while maintaining data privacy, models trust relationships using graph intelligence, and dynamically optimizes defense strategies using reinforcement learning. Experimental evaluation using SUMO and NS-3 simulations demonstrates superior performance, achieving 99.1% detection accuracy, 0.8% false alarm rate, and 18 ms response latency, significantly outperforming traditional machine learning baselines.
Rakesh Kumar Agrawal (Sat,) studied this question.