Managing traffic in rapidly growing cities remains one of the biggest challenges for urban planners worldwide. Traffic signal lights and other network control systems are crucial for preventing urban chaos, yet testing the effectiveness of new traffic management algorithms has remained a time-consuming and resource-intensive task. In a step towards smarter urban mobility, researchers at the Indian Institute of Technology (IIT)-Bombay, in collaboration with Monash University, Australia, have developed a computationally efficient mathematical framework to evaluate decentralised traffic control policies using fewer computing resources — a move they believe could accelerate the design of next-generation intelligent traffic systems.
The study, led by Namrata Gupta, currently a postdoctoral researcher at University Gustave Eiffel, Lyon, France, and formerly a joint Ph.D. student under an IIT Bombay-Monash University programme, when the research was conducted, along with Professor Gopal R. Patil from IIT Bombay and Professor Hai L. Vu from Monash University, proposes a network-theory-based framework to overcome the limitations of existing evaluation platforms.
Traditionally, traffic signal control (TSC) algorithms are tested through detailed microscopic simulations. These platforms capture vehicle-to-vehicle interactions in rich detail but are computationally expensive, forcing researchers to restrict their experiments to only a handful of scenarios. In contrast, the new framework uses simplified mathematical abstractions to analyse multiple traffic policies quickly, helping researchers evaluate how they might perform under varied conditions.
Speaking to The Hindu, Ms. Gupta explained that the lack of standardised benchmarks in the field was one of the main motivations behind the work. “The development of traffic signal controllers is an active research area, and several new algorithms have emerged in recent years. But there are still no universal benchmarks to evaluate and compare these algorithms,” she said. “We believe that simple abstractions, such as the two-bin model proposed by Dr. Carlos F. Daganzo’s group at UC Berkeley, can provide a low-cost, mathematically tractable platform for testing algorithms systematically. Our vision is that multiple standardised platforms and metrics will eventually exist, allowing researchers to evaluate their algorithms rapidly before conducting richer simulations in tools like PTV VISSIM or SUMO.”
At the heart of the framework is the two-bin model, which groups roads into two broad categories — north–south, and east-west — and mathematically represents vehicle movements between them. This approach uses ordinary differential equations to capture the dynamics of traffic flow and derive the macroscopic fundamental diagram, a tool that relates key variables like vehicle density, speed, and throughput across a road network. By using this abstraction, the researchers can evaluate traffic performance without modelling every single vehicle on the road.
The study introduces two key performance metrics that form the foundation of this framework. The first measures how effectively a traffic policy prevents gridlocks and ensures smooth traffic distribution between directions. The second evaluates the overall flow of vehicles within the network, allowing researchers to determine whether a policy supports efficient mobility. Importantly, these metrics are not tied to a specific control method — they can be applied to traditional algorithms, AI-based controllers, or machine learning-driven strategies, provided they can be adapted to the two-bin framework.
For now, the framework has been validated in simulation environments rather than real-world deployments. “So far, we have tested our proposed metrics using the two-bin abstraction and validated them through PTV VISSIM, which provides a more realistic representation of traffic dynamics,” Ms. Gupta said. While the results have been promising, she cautioned that the model works best in rectangular, grid-like road layouts, such as those found in planned cities, including Chandigarh. Applying it directly to chaotic, irregular urban networks like Mumbai or Delhi is not feasible at this stage. “The two-bin model is not designed to simulate full-scale city networks,” she explained. “Its real value lies in providing researchers with a clean mathematical lens to understand network behaviour under simplified conditions. For more complex, unplanned layouts, the model acts as a benchmarking tool before we move to richer, computationally heavy simulations.”
The team believes the framework could also complement AI-driven and adaptive traffic control systems. Ms. Gupta said one promising area of exploration involves using the two-bin model to design computationally efficient training environments for intelligent controllers. “AI-based traffic signal controllers can certainly be tested within this framework,” she noted. “By providing a simplified environment, we can train these systems more quickly and evaluate their performance before deploying them in real-world or high-fidelity simulation scenarios.”
While the immediate focus of the research is on improving algorithm design, its long-term implications could extend beyond traffic management. Efficient signal control is closely tied to environmental outcomes, since congestion, frequent stopping, and idling contribute significantly to fuel consumption, emissions, and pollution. “Traffic signal control is deeply linked to ecological factors,” Ms. Gupta said. “By providing a structured way to test and improve algorithms, we are contributing to smoother traffic flows, which indirectly reduce fuel wastage and lower urban air pollution.”
Looking ahead, the researchers aim to expand the framework further. Work is underway to develop three-bin and four-bin models to handle more complex layouts, and eventually integrate mixed-traffic systems that include pedestrians, cyclists, and public transport dynamics. “Integrating pedestrian movements and other transport modes is an important step for the future,” Ms. Gupta acknowledged. “At this stage, our work is focused on vehicle traffic, but we plan to explore multi-modal systems as the framework evolves.”
Although the framework is not yet ready for direct deployment by municipal authorities, the researchers believe it lays the groundwork for smarter, more adaptive urban traffic solutions. By providing a faster, lower-cost way to test algorithms and identify promising strategies, the study offers a pathway toward efficient, sustainable, and resilient traffic management systems. As Indian cities grapple with increasing congestion, frameworks like these could help traffic engineers, city planners, and policymakers design better systems for growing urban populations.
Published – September 05, 2025 04:31 am IST