An efficient one-step-ahead optimal control for urban signalized traffic networks based on an averaged Cell-Transmission Model

Grandinetti, Pietro; de Wit, Carlos Canudas; Garin, Federica · 2015 · OpenAlex-citations

DOI: 10.1109/ecc.2015.7331072

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 optimizing traffic flow in large urban signalized networks, a problem complicated by the computational complexity of global optimization algorithms and the limitations of fixed-time control strategies. The authors propose an efficient, real-time control strategy based on an averaged approximation of the macroscopic Cell Transmission Model (CTM). While traditional CTM extensions handle network dynamics, they often involve combinatorial problems due to binary traffic light states. To overcome this, the study utilizes average theory to transform the system into a linear optimization problem, allowing for computationally efficient control that adapts to continuously changing demands. The methodology begins by modeling urban traffic as a fluid using an extended CTM with First-In-First-Out (FIFO) policies at intersections. The authors define demand and supply functions for each road segment, ensuring physical consistency through mass conservation laws. They then derive an "averaged model" where traffic lights are represented not as binary switches but by their duty cycle (the percentage of green time within a cycle). This approximation smooths the system dynamics, removing fast-switching oscillations while preserving the essential flow characteristics. The control algorithm is formulated as a one-step-ahead linear program that optimizes two performance metrics: Service of Demand (SoD), which measures the number of vehicles served from external inputs, and Total Travel Distance (TTD), which promotes equitable usage of the network infrastructure to prevent congestion disparities. Validation was conducted through software simulations of a network comprising 40 roads connected by standard four-way intersections. The study compared the averaged model against the actual discrete-time system under varying external demands. Results demonstrated that the averaged model accurately captures the operational modes of the network (free vs. congested) with a mean error of approximately 10%. Furthermore, the model closely approximated the density evolution and performance indices (SoD and TTD) of the actual system, with errors remaining below 10%. The proposed control strategy successfully optimized these metrics by adjusting duty cycles at the start of each traffic cycle, avoiding excessive switching while maintaining effective regulation. The significance of this work lies in providing a computationally tractable solution for real-time urban traffic control. By converting the control problem into a linear optimization framework, the authors offer a method that is significantly more efficient than existing model-based algorithms with exponential complexity. This approach allows for dynamic adaptation to real-time traffic conditions, improving network throughput and reducing congestion without requiring prohibitive computational resources. The study establishes a robust foundation for implementing efficient, model-based traffic light control in large-scale urban environments.

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.

• mental model of traffic