Bluetooth data sets allow a direct estimation of travel times across sensor pairs; however, the resulting estimations contain noise because of missed detection rate and alternative paths between sensors. Additionally, although Bluetooth data sets allow tracking of vehicles across sensors, they do not provide an exact path (i.e. a sequence of traffic sections). Estimating vehicle paths from Bluetooth records is not a straightforward task. This paper proposes a joint method to simultaneously infer vehicle paths and travel times using Bluetooth records. The methodology is applied in a case study of Bluetooth measurements in Brisbane, Australia. The results are validated against travel time observations, which are collated by creating a testing set out of the whole data set. Travel time estimates are robust to the changes in the size of the available measurements, and the proposed model significantly outperforms a naive model where travel times are estimated via direct matching.

蓝牙数据集可直接估算传感器对之间的传播时间；但是，由于缺少检测率和传感器之间的替代路径，因此得出的估计值包含噪声。另外，尽管蓝牙数据集允许跨传感器跟踪车辆，但它们没有提供确切的路径（即交通路段序列）。根据蓝牙记录估算车辆行驶路线并非易事。本文提出一种联合方法，使用蓝牙记录同时推断车辆的路径和行驶时间。该方法在澳大利亚布里斯班的蓝牙测量案例研究中得到了应用。通过行程时间观察对结果进行验证，这些时间观察是通过从整个数据集中创建测试集来进行整理的。