A feedback queueing network model for traffic signal control at intersections considering congestion propagation in...

Area of Science:

• Traffic Engineering
• Operations Research
• Queueing Theory

Background:

• Traffic congestion propagation at intersections in dynamic stochastic environments, especially under oversaturated conditions, presents significant challenges.
• Existing models often struggle to accurately capture the complexities of random traffic demand and time-varying network dynamics.

Purpose of the Study:

• To develop a robust feedback fluid queueing network model for analyzing congestion propagation and delay.
• To establish optimization frameworks for traffic signal control aimed at minimizing vehicle delay and total costs.
• To implement a real-time rolling optimization strategy for dynamic traffic management.

Main Methods:

• A feedback fluid queueing network model integrating random traffic demand, time-varying transition probabilities, and state-dependent service capabilities.
• A recursive algorithm for analyzing the queueing network.
• A rolling optimization strategy using the mesh adaptive direct search algorithm for traffic signal control.

Main Results:

• The proposed model and algorithm demonstrate effectiveness across varying traffic intensities with low average absolute (0.5152 vehicles) and relative (6.43%) errors.
• An optimal moderate time step for rolling optimization was identified to minimize delay and costs.
• The method achieved up to a 70% reduction in delay compared to Synchro software under specific conditions.

Conclusions:

• The feedback fluid queueing network model provides an effective approach for understanding and managing traffic congestion.
• The rolling optimization strategy offers significant improvements in traffic signal control, reducing vehicle delay and operational costs.
• Findings highlight the importance of optimal time step selection and demonstrate the potential for substantial efficiency gains in urban traffic management.