The evaluation of freeway service quality is crucial work, and thus, transportation professionals have developed numerous measures including traffic volume, speed, and density. However, recent research efforts have indicated that such traditional measures may not fully reflect the quality of roadway service from the perspective of individual drivers, necessitating the development of alternative approaches that complement or replace the current service quality measures. As an alternative approach, the speed variation of a vehicle has been suggested as a promising indicator of traffic flow quality perceived by individual drivers. In particular, acceleration noise, defined by the standard deviation of the acceleration of a vehicle, has been often studied as a measure of the degree of speed variation. However, previous studies have been limited to the experimental level due to the difficulty in collecting high-resolution vehicle speed profiles for computing acceleration noise. In this dissertation, the characteristics of speed variation, measured by acceleration noise, are investigated using the rich set of GPS data collected from the instrumented vehicles driven by the participants of the Commute Atlanta research program. The employment of the real-world vehicle activity data, composed of every second of vehicle operation, renders this research effort unique and provides an opportunity to investigate the various aspects of acceleration noise in the real-world context. The investigation is performed by relating acceleration noise to its three influential factors: traffic conditions, roadway, and driver/vehicles. In addition, a fuzzy inference system-based methodology, combining vehicle speed and acceleration noise from instrumented vehicles, is proposed as an approach to evaluating traffic flow quality.

Previous research regarding the relationship between speeding behavior and crashes suggests that drivers who engage in frequent and extreme speeding behavior are over-involved in crashes. However, many of these earlier studies relied on estimates of prevailing and pre-crash speeds, and as a result, their conclusions have been questioned. Over the last several years automotive manufacturers have begun installing airbag systems that collect and maintain accurate pre-crash speeds. Though, patterns of driver speeding behavior are also necessary to discern whether drivers who regularly participate in speeding have increased risk of crash involvement. This dissertation presents a framework and methods for quantifying and analyzing individual driver behavior using instrumented vehicles. The goals of the research were threefold: 1) Develop processing methods and observational coding systems for quantifying driver speeding using instrumented vehicle data; 2) Develop a framework for analyzing aggregate and individual driver speeding behavior; and 3) Explore the potential application of behavioral safety concepts to transportation safety problems. Quantitative assessments of driver speeding behavior could be used in combination with event data recorder data to analyze crash risk. Additionally, speed behavior models could aid in the early identification of problem behavior as well as in the development of targeted countermeasure programs. For this research, 172 instrumented vehicles from the Commute Atlanta program were utilized to collect individual driver speeding behavior. Continuous monitoring capabilities allowed the capture of speed and location for every second of vehicle operation. Driver speeds were then matched to road networks and subsequently to posted speed limits using a geographic information system. This allowed differences between the drivers speed and the posted speed. Several processes were developed to assess the accuracy and the completeness of the data prior to analysis. Finally, metrics and analysis frameworks were tested for their potential usefulness in future behavioral risk analysis. The results of the research were both positive and staggering. On average, nearly 40% of all driving activity by the sample population was above the posted speed limit. The amount and extent of speeding was highest for young drivers. Trends indicate that speeding behavior decreases in amount and extent as age increases.