ii ACKNOWLEDGMENTS I would like to gratefully acknowledge my wonderful advisor Dr. Christopher Monsere whose support and guidance enabled me to complete this project. It was an honor to work with someone who really cares about his students. I would like to also extend my acknowledgement to Jon Makler, Steve Hansen, Joel Barnett at PSU who contributed to the data collection and analysis of an earlier evaluation on which this research is based on. Dr. Robert Bertini for coordinating PSU's access

more »

... o the INRIX data after ODOTs approval and support. Dr. Robert Fountain and Alexander Bigazzi for their valuable input on time series analysis. Ted Trapanier at INRIX for providing helpful orientation on using the INRIX data for the analysis. Finally, none of this would have been possible without the unconditional love, endless support and extraordinary encouragement of my parents, Fawaz and Ahlam, and my aunt Dr. Faiza. iii ABSTRACT Travel time is one of the most widely used measures of traffic performance monitoring for the transportation systems. It is a simple concept that refers to the time required to traverse between two points of interest. Travel time is communicated and used by a wide variety of audience such as commuters, media reporters, and transportation engineers and planners. Recent developments within the wireless communication area made it possible to collect travel time data at a relatively low cost. These emerging technologies include mobile phone based technologies, in-vehicle navigation technologies and automatic vehicle identification technologies. Although these technologies offer a great collection source for travel time data, they have different levels of accuracy. In this research two sources of travel time data were evaluated. These sources of data were the INRIX travel time data and the Bluetooth travel time data. The granularity of the INRIX and the Bluetooth data were high in which travel time estimates were reported at a one minute interval. A total of 42 GPS vehicle probe surveys were carried out in three different days to evaluate the accuracy of the INRIX and the Bluetooth travel time estimates. Statistical measures such as the mean absolute error (MAE) and the mean absolute percent error (MAPE) were calculated for a total of 6 segments and 3 time periods (midday, pm peak, and weekend). The INRIX estimates during the midday were either within 0.36 minutes or 22% of the ground truth probe runs, while the Bluetooth estimates during the pm peak were either within 1 minute or 24% of the ground truth probe runs. In addition to hypothesis testing for 13,541 matched-pairs observation, correlation testing was carried out to evaluate the behavior of the Bluetooth and INRIX time series.