CDA-PTA: Cognitive Digital Twin-Based Predictive Trust Assessment for Proactive VANET Security

Abstract

Vehicular Ad Hoc Networks (VANETs) require reliable trust assessment because vehicles exchange safety-critical information over highly dynamic communication links. Existing trust mechanisms are largely reactive and depend on current or historical observations, making them less effective when vehicle behavior changes rapidly. This paper proposes Cognitive Digital Twin-Based Predictive Trust Assessment (CDA-PTA), a predictive framework combining continuously updated Digital Twin states, cyber-physical behavioral features, temporal interaction graphs, Temporal Graph Neural Network (TGNN) learning, explainable trust assessment, and trust-aware Software-Defined Networking (SDN) routing. Each vehicle is represented by a Digital Twin, heterogeneous behavioral features are fused, and evolving interactions are modeled as a temporal graph. A Predictive Trust Score (PTS) estimates future trustworthiness over a defined prediction horizon. An explainable AI layer identifies influential behavioral factors, while the SDN controller uses predictive trust during route selection. The manuscript defines a reproducible experimental protocol; numerical result fields are intentionally left for measured values rather than fabricated values.