Preprint “A Performative Control Approach to Day-to-Day Traffic Signal Retiming” now available on arXiv
Published:
I am pleased to share our preprint, “A Performative Control Approach to Day-to-Day Traffic Signal Retiming”, now available on arXiv! This is the first research paper by lead author Shuchang Jiang, an undergraduate student at Rice University.
Better traffic signal timing can reduce congestion, travel times, and pollution. However, a timing plan optimized for today’s traffic conditions may not work as well tomorrow because travelers adapt their behavior—changing routes, choosing different modes of transportation, or even deciding whether to travel at all. We propose a scalable, gradient-based approach that incrementally adjusts fixed-time traffic signals using empirical traffic observations, without requiring an explicit model of travelers’ behavioral responses. This allows signal policies and travel demand to co-evolve toward a stable operating point, a concept known as performative stability. Simulations on a downtown Denver network demonstrate reduced congestion, improved throughput, and more stable day-to-day traffic dynamics.
This interdisciplinary collaboration combines the expertise of Xinwu Qian from Rice University’s Department of Civil and Environmental Engineering with networked-control methodologies we developed in Santiago Segarra’s group for intelligent wireless networking, extending them to transportation networks.
- Preprint: https://arxiv.org/abs/2610.19022