Quantifying Measurement Error in Arterial Travel Times Measured by Bluetooth Detectors

Abstract

Travel time is viewed by travelers and road managers as one of the key indicators of quality of service. Bluetooth detectors have recently emerged as a viable technology for the acquisition of vehicle travel times. Several studies have described the application of the technology to freeways and compared the measured travel times with the travel times obtained through other technologies. This paper examines the application of Bluetooth detectors to the acquisition of arterial travel times. The arterial environment is substantially more challenging than the freeway environment because of the frequent interruptions in the traffic flow caused by traffic signals. This paper examines the magnitude of errors in detection time and travel time measurement. It is not feasible to use field data to examine the measurement error because the error is inherent within the observations and cannot be separated. Consequently, a simulation framework is proposed to synthesize measurement errors for a range of arterial traffic conditions. The results show that the mean travel time error is essentially zero for all traffic conditions. However, the variance of the error varies as a function of the traffic conditions. Through multiple regression, the standard deviation of the travel time measurement error is modeled, and it is shown that under some conditions, the 95% confidence interval of this error may reach 25% of the true mean travel time. These results can be used to assess Bluetooth detector deployment plans and provide more insight into the reliability of arterial travel time measurements obtained from Bluetooth detectors.