Usuki, Makoto (University of Tokyo), Yasuda, Shohei (The University of Tokyo), Fuse, Takashi (The University of Tokyo), Seya, Hajime (Kobe University)

Identifying Traffic Congestion Sources Based on Propagation Dynamics in Road Networks

Scheduled for presentation during the Regular Session "S32c-AI-Driven Traffic Monitoring, Safety, and Anomaly Detection" (FR-LA-T32), Friday, November 21, 2025, 16:20−16:40, Southport 2

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

Identifying the sources of traffic congestion is essential for its mitigation. Traffic congestion is known to exhibit chaotic behavior, and severe congestion often spreads across wide areas. However, if the initial sources can be accurately identified, localized interventions targeting those sources may help mitigate large-scale congestion. Despite its importance, most existing studies on traffic congestion have focused on anomaly detection or traffic state estimation, and relatively few have addressed the identification of congestion sources. In this study, we propose a novel method for identifying the sources of traffic congestion by incorporating the propagation dynamics of congestion. Specifically, we introduce a Traffic Propagation Graph (TPG) that explicitly models the upstream-only spread of congestion in road networks, and integrate it into the Non-Negative Matrix Factorization (NMF) framework. Our empirical study, using real-world road networks and connected vehicle data in Kobe, Japan, demonstrates that, compared to conventional NMF, the proposed method can more precisely identify the sources of traffic congestion in a localized manner. The identified sources correspond to major bottlenecks, such as intersections where arterial roads cross or highway exit ramps, further validating the effectiveness of our method.