Validation of Origin–Destination Data from Bluetooth Reidentification and Aerial Observation

Abstract

Vehicle reidentification with Bluetooth signal data has emerged as an effective and economical means for collecting traffic data including origin–destination (O-D) information, which is crucial for transportation planning. Direct vehicle tracking based on time-lapse aerial photography (TLAP) is also increasingly used for O-D studies. Neither technique has previously been validated, so the objective of this study was to validate O-Ds generated by using both techniques against a ground-truth O-D. Traffic volume–, Bluetooth-, and TLAP-based O-D data collected at an interchange in Madison, Wisconsin, were used in this study. Significant variability (2.3% to 7.2%) in Bluetooth match rates was observed for the 12 O-D pairs of the interchange, which were calculated by using a heuristic measure of the difference between traffic volumes (the GEH formula). Uniproportional scaling of the sample Bluetooth O-D (using the average detection rate) resulted in poor fit to the true O-D, but biproportional factoring (Fratar or Furness procedure) resulted in about 85% of the movements’ having a GEH statistic less than 5. Combining Bluetooth data from multiple days resulted in a better fit than the use of a single day's data. The analysis showed that Bluetooth technology could be used to obtain representative O-D information of an access-controlled network for planning applications. TLAP sample O-D data were scaled with an origin-specific factor and resulted in reasonable fit to the true O-D data. Biproportional factoring resulted in a better fit in which 100% of the movements had a GEH value less than 5. Therefore, biproportional factoring is also recommended for TLAP data.