Error Modeling and Analysis for Travel Time Data Obtained from Bluetooth MAC Address Matching

Travel time has been very difficult to collect until recently. Current attempts at exploiting short-range communication protocols that rely on unique identifiers, primarily Bluetooth, have significantly simplified the travel time collection task. Many transportation agencies are now considering using Bluetooth travel time estimates to feed a variety of applications, such as user information systems. As Bluetooth-based travel time data collection increases in popularity, investigating the errors that are characteristic of this detection type becomes more important. 

A Bluetooth sensor, called the Media Access Control Address Detection (MACAD) system, was developed for travel time data collection in this study to facilitate testing system configurations and allow for future deployments. Three types of antennae and three different sensor arrangements were tested to determine the effects of these variables on travel time error. The collected travel time data were compared to license plate reader data, which, because of their relatively small detection zone for vehicle license plate recognition, were taken as the ground truth travel time. A regression model was used to investigate whether travel time error can be predicted with observable explanatory variables. Descriptive statistical analysis was also employed to evaluate the impacts of individual variables on the travel time error. The results suggested that a combination of sensors is desirable, despite the potential loss of accuracy, as the higher matching rates obtained by the system will improve sample size and reduce random error rates. Findings of this study are helpful to transportation professionals attemptimg to understand the errors associated with the Bluetooth-based travel time data collection technology and to configure the sensors to mitigate the errors.

Publication Date: 
Saturday, December 31, 2011
Publication Number: 
WA-RD 782.1
Last modified: 
01/18/2019 - 16:12
Yinhai Wang, Yegor Malinovskiy, Yao-Jan Wu, Un Kun Lee
Transportation Northwest (TransNow), Washington State Transportation Center (TRAC-UW)
Number of Pages: 
Travel time, Data collection, Bluetooth technology, Sensors, Testing, Error analysis, Regression analysis, Recommendations.