Empirical Evaluation of the Accuracy of Technologies for Measuring Average Speed in Real Time

A federal mandate challenges states to acquire and to disseminate reliable travel time–speed information with limited sensor infrastructure and resources; the mandate also opens the opportunity to look beyond traditional sensor technologies. Some of these new and promising travel data technologies include various deployments and combinations of GPS, probe vehicles, cellular devices, Bluetooth devices, radio frequency identification, automated license plate recognition (LPR), and even social media. To take on this challenge, the objective of this study was to provide several key considerations for evaluation of travel speed data for general cases. The key items included obtaining reliable ground truth data, transforming and comparing data sets, and evaluating data accuracy. Along with the explanation of these considerations, the results of a case study are provided to help illuminate the issues. The case study, which was performed in the vicinity of downtown Nashville, Tennessee, along Interstate 40 and Interstate 65 evaluated real-time travel time–speed data from Bluetooth sensors, from probes supplied by two major vendors, and from remote traffic microwave sensors. These data were compared with ground truth data from an LPR-based vehicle tracking system as well as video footage collected simultaneously. The paper discusses the reliability of ground truth, the advantages and shortcomings of different technologies, the evaluation of data accuracy methodologies, and future research directions.