Enhanced Scheduling for Real-Time Traffic Control

Traffic signal control in road networks is a practical problem which has been widely studied. In this paper, we present an approach for traffic signal control extending ideas of schedule-driven coordination in the system Surtrac. The traffic signal control problem of one intersection is modeled as a parallel machine scheduling problem based on aggregation of traffic flow data. The solution procedure for each parallel machine scheduling problem is based on a forward dynamic programming search. All connected intersections form a distributed system of communicating intersections. The objective is to construct a traffic control sequence for each intersection and minimize the total cumulative delay of all vehicles in the traffic network. Simulation results for a grid network from the SUMO simulator demonstrate the performance of the proposed approach in comparison to the Surtrac system solving the problem using single machine scheduling. The results show a significant improvement in the total cumulative delay given the increase of computational time which is acceptable in real-time processing.