A Review of Graph-Based Models in Urban Transportation Optimization: From Theory to Smart Applications

Urban transportation optimization is a critical challenge in modern cities, where increasing congestion, dynamic traffic conditions, and infrastructure limitations demand intelligent solutions. Graph-based models have emerged as a powerful tool for analyzing and optimizing urban road networks by representing traffic systems as nodes (intersections) and edges (road segments) with weighted attributes such as travel time, traffic flow, and road conditions. This paper provides a comprehensive review of recent advancements in graph-based models for urban transportation optimization, covering traditional algorithms (e.g., Dijkstra's, A*), hybrid approaches integrating machine learning (e.g., Q-learning, GNNs), and real-world applications in smart city initiatives. Key contributions of this review include: (۱) an analysis of graph-based shortest-path algorithms and their adaptations for dynamic traffic conditions, including delay modeling at signalized intersections; (Y) a discussion of AI-enhanced methods, such as reinforcement learning and Graph Neural Networks (GNNs), which improve adaptability to real-time traffic fluctuations; and (✓) case studies demonstrating the effectiveness of these models in cities like Sarajevo and Bengaluru, where optimized routing reduces congestion and improves travel efficiency. Additionally, we highlight emerging trends, such as spatio-temporal graph models for traffic forecasting and physics-guided neural networks for intersection management. The review underscores the potential of graph-based models to enhance urban mobility, offering insights for policymakers, transportation engineers, and AI researchers. Future directions include scalable implementations for large metropolitan networks, integration with IoT and real-time data streams.

Keywords:

Urban transportation , graph theory , GNNs , traffic optimization