Decentralised bottleneck prioritisation strategy for traffic flow improvement

Serok, Nimrod; Havlin, Shlomo; Lieberthal, Efrat Blumenfeld · 2026 · DOAJ

DOI: 10.1140/epjds/s13688-026-00617-6

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper introduces the "Tree Method," a novel decentralized traffic light control strategy designed to mitigate congestion caused by conflicting traffic flows operating on competing cycle times. The research addresses the limitations of existing static and centralized control systems, which often fail to adapt to the dynamic, self-organizing nature of urban traffic or suffer from scalability issues. Specifically, the study aims to resolve the challenge of optimizing signal timing in real-time by prioritizing phases based on the global impact of congestion rather than local conditions alone. The motivation stems from the high economic and environmental costs of traffic congestion and the computational difficulty of achieving optimal network-wide control, which is classified as an NP-hard problem. The methodology relies on identifying "congestion trees"—upstream chains of congested streets originating from a temporal bottleneck at an intersection. The Tree Method calculates the cost associated with each congestion tree, defined by time delays, to determine the duration of traffic light phases for the subsequent cycle. It operates as a "fixed cycle, fixed order" system, maintaining constant cycle lengths and phase sequences while dynamically adjusting phase durations based on calculated congestion costs. The authors evaluated the strategy using the Simulation of Urban Mobility (SUMO), an open-source microscopic traffic simulator. Simulations incorporated both realistic and abstract Origin-Destination matrices across varying traffic conditions to assess performance under different demand levels. The study compared the Tree Method against benchmark techniques, including actuated traffic control and other decentralized approaches, to validate its effectiveness. Results indicate that the Tree Method significantly improves network throughput and reduces average travel times compared to existing control strategies. By prioritizing the resolution of bottlenecks with the most expansive upstream influence, the method effectively mitigates spill-over effects and negative interactions between neighboring intersections. The simulations demonstrated that the approach leads to smoother traffic flow, fewer delays, and shorter queues, even under heavy demand conditions. Furthermore, the method proved superior in improving conditions for the majority of drivers over time, outperforming traditional actuated controls and other decentralized solutions in terms of overall network efficiency. The analytical simplicity of the Tree Method allows for swift real-time adjustments, making it suitable for dynamic feedback loops inherent in traffic systems. The significance of this work lies in providing a scalable, decentralized solution that balances computational efficiency with real-world applicability. Unlike complex centralized systems or computationally intensive reinforcement learning models, the Tree Method offers a straightforward framework that can be deployed in real urban environments. Its ability to automatically activate at specific times or intersections enhances control precision without requiring extensive infrastructure changes. The findings suggest that prioritizing global congestion impacts through tree-based cost calculation is a viable strategy for managing urban traffic, offering a robust alternative to static plans and centralized optimization. This approach supports the development of smart transportation systems that can adapt to fluctuating demand and reduce the external costs associated with congestion.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• traffic density