Traffic-adaptive signal control and vehicle routing using a decentralized back-pressure method

Abstract

The problem of controlling traffic lights under adaptiverouting of vehicles in urban road networks is considered.Multi-commodity back-pressured algorithms, originallydeveloped for routing and scheduling in communicationnetworks, are applied to road networks to control trafficlights and adaptively reroute vehicles. The performance ofthe algorithms is analyzed using a microscopic trafficsimulator. The results demonstrate that the proposed signalcontrol and adaptive routing algorithms can providesignificant improvement over a fixed schedule and asingle-commodity back-pressure signal controllers, in termsof various performance metrics, including queue-length,trips completed, and travel times.