Decentralised bottleneck prioritisation strategy for traffic flow improvement

Abstract We introduce a novel decentralised traffic light control strategy, termed the Tree Method, designed to mitigate the challenges posed by conflicting traffic flows operating on competing cycle times during specific phases at traffic intersections. This methodology hinges on the precise identification and subsequent prioritisation of congestion bottlenecks, assessed through their expansive influence on the entire road network. The Tree Method calculates the cost associated with each congestion tree and advances a prioritisation scheme that emphasises the global, rather than local, impact of traffic flow. To evaluate the effectiveness of this approach, we utilised the Simulation of Urban Mobility (SUMO) to conduct a series of simulations incorporating both realistic and abstract Origin-Destination (OD) matrices across varying traffic conditions. The Tree Method demonstrated a significant capability in identifying the principal contributors to congestion and their upstream effects, leading to major improvements in throughput and average travel times. Comparative analysis of the Tree Method against other, traffic light control techniques revealed superior performance also in improving conditions for the majority of drivers and across time. This means that traffic moves more smoothly through junctions, with fewer delays and shorter queues, even under heavy demand. Additionally, the simplicity of the Tree Method’s analytical framework supports real-time operational adjustments, aligning well with the dynamic feedback loops inherent in traffic flow systems. As cities face growing congestion challenges, our findings highlight a control strategy that is both effective and simple enough to be deployed in real urban environments.