Performance monitoring for arterial traffic control and management system is an area of emerging focus in the United States. To properly study traffic flow at signalized intersections, both arrival/departure traffic flow data and associated signal status data are required. Although many existing signal control systems are capable of generating data to support performance assessment, most do not make it "easy" for the managing agencies to prioritize improvements and plan for future needs. Indeed, the 2005 Traffic Signal Operation Self Assessment Survey indicated that the majority of agencies involved in the operation and maintenance of traffic signal systems do not monitor or archive traffic system performance data in an effort to improve their operation. Therefore, despite studies having shown that the benefits of investments in improved signal timing outweigh the costs by 40:1 or more, signal retiming is often not repeated frequently enough to account for rapidly changing traffic patterns, largely due to the expense of manual data collection and performance measurements.
The need to address the above problems inspired this research. The goal is to develop a real-time arterial performance measurement system, which can automatically collect and archive high-resolution traffic signal data, and build a rich list of performance measures. The objectives of this doctoral research are two-fold: (1) to develop a system for high-resolution traffic signal data collection, archival, and preprocessing; and (2) to develop a set of methodologies that can measure traffic signal performance, including queue length, delay and level of service (LOS) for individual intersections and travel time and number of stops for an arterial corridor. In this research, a system for high resolution traffic signal data collection is successfully built. The system, named as SMART-SIGNAL (Systematic Monitoring of Arterial Road Traffic and Signals), is an arterial data collection and performance measurement system, which simultaneously collects "event-based" high-resolution traffic data from multiple intersections and generates arterial performance measures in real time. In the SMART-SIGNAL system, a complete history of traffic signal control, including all signal events such as vehicle actuations on detectors and signal phase changes, is archived and stored.
Using the collected "event" data, mathematical models are built to calculate intersection and arterial performance measures. A time-dependent queue length estimation model is proposed that can handle long queues under both under-saturated and over-saturated conditions. The model examines the changes in signal detector's occupancy profile within a cycle, and derives queue length by identifying traffic flow pattern changes during the queue discharging process. A turning movement proportion estimation model is also offered in this thesis. Detector counts from surrounding intersections are used to calculate right turning traffic for the subject intersection.
An innovative algorithm is proposed in this research for arterial performance measurement by tracing virtual probe vehicles from origin to destination. One of three maneuvers: acceleration, deceleration or no-speed-change, is selected based on the current traffic states of the virtual probe. The step-by-step maneuver calculation stops until the virtual probe "arrives" at the destination, and various arterial performance measures, including travel time, can thus be estimated. An interesting property of the proposed model is that travel time estimation errors can be self-corrected with the signal status data, because the differences between a virtual probe vehicle and a real probe can be reduced when both of them meet the red signal phase. The virtual probe mimics regular travel behaviors of arterial drivers and thus can be treated as a representative vehicle traversing the arterial.
The SMART-SIGNAL data collection system has been installed on an 11-intersections arterial corridor along France Avenue in Hennepin County, Minnesota since February 2007. Event-based signal data are being collected in a 24/7 mode and then immediately archived in the SMART-SIGNAL system, thus yielding a tremendous amount of field data available for research. The field study shows that the proposed mathematical models can generate accurate time-dependent queue lengths, travel times, numbers of stops, and other performance measures under various traffic conditions.