Traffic flow modeling in highway networks

The emergence of the Advanced Traffic Management System poses new challenges in traffic flow modeling of urban areas. The motivation of this project is to produce working freeway traffic simulations within a reasonable time-scale.

This project. describes a hybrid traffic modeling approach, which is a combination of microscopic and macroscopic traffic modeling techniques. The traffic stream is composed of individual vehicles, while the interactions in the traffic stream are modeled macroscopically using the average speed-density relationship.

All existing freeway simulation models use the time-driven approach which advances the simulation clock after each fixed time-slice. However, this approach has a limitation of capturing the dynamic nature of traffic flow. The project proposes a new assumption of vehicular movement. This assumption leads to an easy implementation of a traffic simulation model using the event-driven approach which advances the simulation clock from one event to the next. The event-driven traffic model provides a new tool for the development of dynamic freeway simulation/assignment models.

A number of experimental results are provided from an empirical comparison of the event-driven approach versus the time-driven approach. These results indicate that the event-driven simulation model is competitive with the time-driven simulation model in both accuracy and efficiency. Finally, specific potential directions for future research are pinpointed.