State graph modeling and simulation framework for urban traffic systems

Abstract

The modeling and simulation of urban traffic systems (UTS) have been widely used for solving traffic congestion problems. In this paper, we suggest a new approach for modeling UTS using the state graph, which is a state-based modeling formalism of discrete event systems (DES). We show the formal procedure to de-velop a mesoscopic traffic simulation model in the following order, which is based on the integrated proce-dure for DES modeling: source traffic system, reference model, formal model, and simulation model. The modeling scope in this paper is limited to UTS including signalized intersections, left and right turn pockets, split and merge points of lanes, traffic signals for left, right, and through movements. We illustrated the validity of the modeling procedure by modeling a real traffic system in Daejeon, South Korea, and implemented the simulation model using $State Graph Simulator^®$ ($SGS^®$). We verified the traffic model by comparing with the real data collected from the source traffic system. We also modeled the same traffic system using $SGS^®$ and VISSIM, and compared the models. We expect that researchers who are interested in mesoscopic traffic simulation can simulate a real traffic system including signalized intersections by following the suggested procedures and applying given state graph models.