Real-time data-driven traffic simulation for performance measure estimation

Abstract

Congestion is a major issue in transportation sector. As professionals in the transportation field are increasingly exploring new solutions to alleviate traffic congestion, interest in the use of on-line simulation as a tool for estimating metrics of the traffic network for use in real-time operations has grown. The goal of the on-line simulation is to provide traffic information to facilitate more informed travel decisions and enable improved active traffic management. Performance estimation of arterials is a particularly challenging problem because it includes complexities not present in highways. The results of a sequence of experiments are presented to evaluate the effectiveness of a dynamic, data-driven, simulation-based system for estimating arterial performance measures in real-time. The envisioned system is comprised of a microscopic traffic simulation model driven by point sensor data. The conceptual framework of the system is presented, highlighting its key components. Four iterative applications of the framework are then presented, including a proof of concept experiment, two field tests and, a pseudo-field test involving origin-destination pairs from the Federal Highway Administration (FHWA) next generation simulation dataset. The results of the four applications demonstrate the feasibility of employing point sensor data to drive a microscopic traffic simulation and estimate arterial performance measures in real-time.