Critical node selection method for efficient max-pressure traffic signal control in large-scale congested networks

Decentralized signal control of congested traffic networks based on the Max-Pressure (MP) controller is theoretically proven to maximize throughput, stabilize the system and balance queues for single intersections under specific conditions. Increased implementation cost related to queue monitoring reduces MP applicability. We propose a strategy for identifying the most critical network intersections for MP control, with the aim of reaching high efficiency without a full-network implementation. The strategy is based on node congestion and queue variance data. A modified version of Store-and- Forward model is used to emulate spatio-temporal traffic evolution in a large-scale network with more than 500 intersections and evaluate system performance for different MP node layouts. Results show that more than 90% of the maximum observed improvement can be achieved by controlling only 20% of nodes, selected via the proposed strategy, thus significantly reducing implementation cost. The impact of MP in network traffic characteristics is demonstrated