Multiagent Based Decentralized Traffic Light Control for Large Urban Transportation System

Intelligent traffic control is an important issue of the moderntransportation system. However, in large-scale urbantransportation systems, traditional centralized coordinationmethods suffer bottlenecks in both communication and computation. Decentralized control is hard if there is very limited observationto the whole network as evidences to support joint trafficcoordination decisions. In this paper, we proposed a noveldecentralized, multiagent based approach for massive trafficlights coordination to promote the large-scale greentransportation. Considering that only the traffic from theadjacent intersections may affect the state of a givenintersection one time ahead, the key of our approach is using theobservations of a local intersection and its neighbors asevidences to support the traffic light coordination decisions. Therefore, we can model the interactions as decentralized agentscoordinating with a decision theoretical model. Within a localintersection, constraint optimizing agents are designed toefficiently search for joint activities of the lights. Since thisapproach involves only local intersection cooperation, it is wellscalable and easily implemented with small communication overhead. In the last section, we present our software design on thisapproach and based on our simulation, this approach is feasible toa large urban transportation system