Graph Neural Networks (GNNs) in Intelligent Transportation Systems

Abstract

The rapid expansion of urban transportation networks and increasing vehicular traffic have posed significant challenges for traffic management, route optimization, and safety assurance. Intelligent Transportation Systems (ITS) aim to address these challenges by integrating advanced sensing, communication, and data analysis techniques. Recently, Graph Neural Networks (GNNs) have emerged as a powerful tool for modeling complex relational structures inherent in transportation networks. By representing roads, intersections, and vehicles as nodes and edges, GNNs effectively capture spatial and temporal dependencies, enabling precise traffic prediction, anomaly detection, and route optimization. This review explores the theoretical foundations, architectures, and applications of GNNs in ITS, highlighting recent research trends, challenges, and future directions. Emphasis is given to traffic flow forecasting, dynamic routing, and multimodal transport management. Comparative analysis and case studies demonstrate the superiority of GNN-based approaches over traditional machine learning models. The paper concludes with an outlook on integrating GNNs with edge computing and real-time decision-making for sustainable and intelligent urban mobility.

KEYWORDS: Graph Neural Networks, Intelligent Transportation Systems, Traffic Prediction, Dynamic Routing, Spatiotemporal Modeling, Urban Mobility