Using Big Data to Evaluate Traffic Signal Performance for the State of Tennessee

Abstract

Congestion on roadways has led to increased travel time, waste of fuel, and higher emissions. Compared with other methods such as constructing more lanes and/or roads, traffic signal retiming is a viable alternative way to mitigate congestion. To prioritize retiming and to understand the impact of signal retiming, the performance of signals needs to be evaluated. This thesis details a data-driven and low-cost methodology funded by the Tennessee Department of Transportation for evaluating the performance of traffic signals in Tennessee using third-party segmented probe vehicle data. Intersections were formed by matching the names and geographical information of inbound segments. The average speed, the worst-case travel time, and the bottleneck ranking data during different times of the day for each segment at an intersection were used to develop metrics for a ranking formula that outputs a numerical value from 0-10 for each intersection. The results were appended to an online database accessible via a webpage with search and sorting capabilities. The ranking formula results were compared with the Level-of-Service ranking provided by local traffic engineers in the Cities of Murfreesboro and Franklin. Additionally, testing was conducted on an industry-standard traffic signal controller for its data logging and networking functionalities. An Automated Traffic Signal Performance Measures (ATSPM) server and a website were also set up to display various performance measures of traffic signals. The traffic data was generated by emulating stop bar and advance loop detector events in a lab environment.