An AI Enabled Framework for Reliable V2X Communication in Intelligent Transportation Systems

This work suggests an AI based reliability enhancement mechanism for V2X communication in ITS. Traditional IEEE 802.11p and 5G NR V2X systems are suffering from performance loss with high mobility and dense traffic. To enhance efficiency, the system model adopts a LSTM based reliability prediction, DRL driven cross layer optimization, FL assisted with attacker collaboration attack in hierarchical edge–cloud structure and AI assisted anomaly detection. The LSTM is utilized to predict link reliability based on mobility, SINR, vehicle density and historical packet delivery information and the DRL dynamically optimizes power control and resource allocation. In SUMO–NS 3 coupled simulations, remarkable performance gain are exhibited: reliability reaches 97.3% at 140 km/h (9% better as compared to NR V2X), PDR achieves 95.6% in crowded traffic scenario, throughput increases by 18–22% with high interference level, latency jitter lowers to +8% during topological changes, energy cost witnesses a drop at the level of 2.12 mJ per packet. This decision latency of in total 10.6 ms also meets requirements for URLLC, demonstrating real time and scalability feasibility.