Schedule-Driven Coordination for Real-Time Traffic Network Control

Xie, Xiao-Feng; Smith, Stephen; Barlow, Gregory · 2012 · Crossref

DOI: 10.1609/icaps.v22i1.13510

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 addresses the challenge of achieving scalable, real-time optimization for vehicle traffic flow across a network of signalized intersections. While adaptive control strategies offer significant potential for reducing congestion, existing methods often fail to balance local efficiency with network-wide coordination. Centralized approaches suffer from scalability issues, while decentralized methods using moving averages or traditional offsets often lack sensitivity to real-time conditions or fail to account for non-local influences from indirect neighbors. The authors propose a decentralized, schedule-driven coordination approach that combines efficient local scheduling with neighbor interaction protocols to minimize cumulative vehicle delay. The methodology centers on a Schedule-driven Intersection Control (SchIC) agent for each intersection. SchIC treats traffic clusters as non-divisible jobs and uses a polynomial-time dynamic programming algorithm to generate near-optimal signal sequences that minimize local cumulative delay. To enable network-wide coordination, agents exchange information exclusively with direct neighbors. A basic protocol allows agents to query upstream neighbors for projected output flows, creating an "optimistic non-local observation" that extends the prediction horizon and incorporates indirect neighbor influences. To address coordination failures, two additional mechanisms are introduced: a "nervousness" prevention mechanism that iteratively splits clusters to ensure maximum green time constraints are met, preventing disruptive schedule changes; and a spillover prevention mechanism that adjusts schedules to avoid blocking upstream traffic due to insufficient road capacity. The approach was evaluated using simulations in the SUMO environment on two distinct networks: a tightly-coupled 5x5 grid with short travel times and high congestion, and a complex real-world network with multi-directional flows. The simulations utilized a fine time resolution of 0.5 seconds. The results demonstrated that the proposed schedule-driven coordination approach established traffic flows with significantly lower average vehicle wait times compared to both simple isolated control strategies and contemporary coordinated strategies that rely on moving average forecasts or traditional offset calculations. The decentralized design proved capable of handling the non-local influences and dynamic instability inherent in tightly-coupled networks. The significance of this work lies in demonstrating that scalable, real-time network optimization is achievable through decentralized, schedule-driven coordination. By limiting interaction to direct neighbors and using optimistic projections of future demand, the system avoids the computational intractability of centralized methods while outperforming existing decentralized techniques. The introduction of specific mechanisms to dampen nervousness and prevent spillover ensures stability in dynamic environments. This approach provides a robust framework for adaptive traffic control that can respond to real-time conditions with high temporal resolution, offering a practical solution for reducing congestion in complex urban networks.

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