A Dynamic Modelling Framework of Real-Time Guidance Systems in General Urban Traffic Networksby R. Jayakrishnan, Univ of California at Irvine, United States,
Hani S. Mahmassani, Univ of California at Irvine, United States,
Abstract: There is considerable interest around the world in developing and implementing realtime driver information and/or guidance systems to reduce congestion in the urban traffic networks. However most of these attempts have proceed without much insight into the significant influence of many of the key elements and phenomena involved in a traffic system under information that could determine the effectiveness of alternative designs of such systems. These include: the context, extent and form of the information provided to the drivers, the response and compliance of the drivers to the information, the system-wide implications of particle equipping of the vehicle population, frequency of information update, etc. In this paper a simulation-based framework that incorporates these elements is presented along with illustrative results on its application to realistically sized networks. The traffic simulation is performed using macroscopic local speed-flow relations in discretized segments, while individual drivers' route choice decisions are modelled at the network nodes. The network path processing is based on k-shortest path tree-building that allows the drivers to select among several competing route. Efficient data structures are utilized for storing the network and the k-shortest paths are found using efficient binary heap-sorting procedures. The program can simulate networks with only a fraction of the drivers receiving information. The driver route switching behavior is based on candidate mechanisms of choosing to stay on a route unless an alternative route becomes sufficiently attractive. In addition, the model can also be used to evaluate the effectiveness of guidance instructions provided by a central controller. Simulations are carried out for different behavioral model parameters as well as for different fractions of drivers with information, the results of which are provided with a discussion of substantive conclusions and key insights. The simulations are carried out on a supercomputer, and the advantages of vector processing are also discussed.
Subject Headings: Driver behavior | Traffic models | Traffic congestion | Information systems | Transportation networks | Information management | Dynamic models | Traffic analysis
Services: Buy this book/Buy this article
Return to search