Assessment of benefit and accuracy of traveler information in traffic networks

The success of Advanced Traveler Information Systems (ATIS) depends on several factors, such as correct understanding of the drivers' routing decisions, drivers' compliance rate, reliability of the provided information, accurate percentage of equipped vehicles, etc. Undoubtedly, the most significant element in the success of ATIS is the "quality of information". Quality of information connotes not only the accuracy of information, but also the value associated with it. Accuracy of information is the difference between the travel time that a driver experiences and the information provided. Value of information is a measure of the need for traveler information. Obviously, the need for information arises for the routes that have high uncertainty of travel time.; The first problem addressed in this thesis is the value of traveler information problem. The objective of the first problem is to identify the optimal number of routes in a given highway network that would yield the maximum benefit from traveler information. It is assumed that the benefit (value) of traveler information is directly related to the uncertainty of travel times. Namely, the more uncertain the travel times of a given route over consecutive days, the more traveler information drivers will benefit. For a given route, the coefficient of variation of travel times within a given time period over consecutive days is employed as the indicator of travel time uncertainty. New Jersey Turnpike is used as the study network due to the availability of vehicle-by-vehicle network specific data that covers all origin-destination (OD) pairs during 2004. The problem of identifying the optimal number of subset of routes is modeled as a nonlinear integer-programming problem. The novelty of the analysis is the use of actual vehicle-by-vehicle dataset over a long period of time in a large-scale traffic network.; The second problem addressed in the thesis is the accuracy of traveler information by traffic detectors. The objective of the analysis is to find the optimal number and configuration of roadway segments for detector deployment in order to minimize travel time estimation error. This problem is inherently a space discretization problem regardless of which travel time estimation function is used. The ad-hoc solution to this problem is the equivalent segment configuration such as every half-mile, every one-mile. It is shown in the analysis that the space discretization problem can be expressed as a common clustering problem. The novelty of the proposed approach is the use of probe vehicle trajectory data to obtain statistically significant traffic regime at the study route. Clustering of sample space-time trajectory data is proposed as a viable methodology for solving the optimal roadway segment configuration problem.