Kalman filtering estimation of traffic counts for two network links in tandem

Estimating accurately the vehicular density along roads is very important for managing and controlling traffic operations in urban networks. The vehicular density information may be derived from raw counts by loop detectors or other detection devices. However, these counts are subject to errors, which can degrade considerably the density estimates. The extended Kalman filter (EKF) has been applied in the past for obtaining improved density estimates, by coupling the detector counts with independent density estimates, subject to uncorrelated errors. In this paper, the EKF is applied for estimating vehicle counts for two roadway sections in tandem. Because measurement errors at the joint of the two sections are shared by both sections, the resulting count estimates are improved over those obtained by treating the two sections as isolated. This is confirmed by comparing the analytical derivations of the error estimate when treating two tandem sections together with those obtained by treating the sections as isolated ones.