Statistically optimum virtual trip line for real-time traffic monitoring system

Abstract

The virtual trip line (VTL) has been proposed in conjunction with the floating car data method to minimize the potential breach of the probe vehicle privacy. However, a long VTL, as well as a short one, may have unintended consequences; thus, the VTL length should be optimized. This article intends to propose a method to establish the optimum length of the VTL. The optimal length of VTL is determined empirically based on the geo-location data collected by ten units of smart-phones. A custom-made application is designed and installed to these devices to track and record the movement of the probe vehicle. The recorded geolocation data are projected to the road GIS line to establish the projection distances. We demonstrate that the distances are distributed approximately following the normal distribution, from which the critical region of the optimum VTL length for a given significance level is established. Finally, we empirically evaluate the method, and demonstrate that for a given VTL length, the empirical probability of the probe vehicle crossing a VTL is close to the theoretical probability.