AI-Driven Adaptive Trust Management Framework for Secure Vehicular Networks in Smart Cities Using NS-3 and SUMO Co-Simulation

Abstract

To address the security challenges of Vehicular Ad Hoc Networks (VANETs) in smart cities, which arise from their high mobility and decentralization nature, robust and adaptive security mechanisms are essential. This paper introduces an AI-Driven adaptive trust management framework. Moving beyond the traditional perimeter-based security, the proposed solution integrates Zero-Trust principles with entropy-weighted trust scoring and edge assisted decision-making. This was developed using integrated SUMO mobility modeling and NS-3 simulations. This framework specifically targets the Sybil, replay and data falsification attacks. Extensive experiments involving 500 to 5000 vehicles and 20 roadside units helps to demonstrate the greater performance over the centralized and the static models. While reducing the authentication latency by 71% the system achieves 94.1% detection accuracy. Furthermore, the framework scales effectively to 10,000 nodes by maintaining high efficiency in dense environments. A one-way ANOVA confirms the statistical significance of these results by validating the system as a robust and a scalable security solution. This approach significantly enhances both the network reliability and the real time responsiveness.