Karataş, Pınar
Due to their low cost, many urban regions are being accessorized by Bluetooth readers (BTR), which can scan any Bluetooth (BT) enabled devices around them. When BT-enabled devices are monitored over an analysis time or across BTRs, traffic state (in terms of speed or travel time) can be estimated in a time-dependent manner. Estimation of these average travel times for urban corridors can further allow generation of input data for Variable Message Signs (VMSs) while average speed for corridor segments helps determination of Level of Service (LOS) values. However, BT data has also some uncertainties and complexities, which need preprocessing and systematic clean-up approach to eliminate outlier values due to either a) very long times caused by slow travel or stopped durations or b) detours often observed in open urban networks. This study focused on development of a BT data analysis framework to eliminate the travel outliers on urban arterials which may stem from vehicles taking detours/subtours or waiting. The cleaned-up data is used to obtain travel time probability dsitrubitions for road links between consecutive BTRs, leading to estimation of travel time confidence intervals. The statistical representation of average link travel time finally allow estimation of urban corridor travel times. Numerical results were obtained from anlaysis of BTR network of 7 intersections along a study corrdior in Mersin. Determination of a 50 seconds threshold for the rescan time allowed estimation of mid-to-mid travel times between 7 BTRs. Use of 4km/hr slow movement filter first eliminated the majority of the extreme values in the travel time data. A second stage of interquartile-range data clean-up further improved the accuracy of BT-based travel time method. Validations was performed via comparisons with the Floating Car Data (FCD) for the same corridor.
Citation Formats
P. Karataş, “USE OF BLUETOOTH DATA FOR MONITORING URBAN TRANSPORTATION,” Ph.D. - Doctoral Program, Middle East Technical University, 2023.