A METHOD FOR ESTIMATING TRAFFIC VOLUME USING BLUETOOTH DETECTION

Abstract

This study develops a traffic volume estimation method using Bluetooth MAC addresses collected by several scanners. Bluetooth MAC addresses are randomly detected by a scanner at a detection rate from Bluetooth Equipped Vehicles (BEV) in traffic. The BEVs are also considered as random samples in traffic, which appear at an equipment rate. Therefore, Bluetooth MAC addresses are random samples from traffic retrieved through the two-step sampling. The number of Bluetooth MAC addresses collected by a single scanner is a random variable, not an expected value, and thus it is not appropriate to estimate the traffic volume by simply expanding the number of the MAC addresses. The proposed method in this study estimates traffic volume of a road section by utilising the number of MAC addresses collected by several scanners, each of which has the same detection rate. The method assumes that the number of MAC addresses detected within a time interval follows the Poisson distribution. Then, the traffic volume is estimated so that the likelihood of the observation by the scanners is maximised. The verification by a field data set showed that the proposed method was able to estimate the traffic volume with around 10% error rate.