Robustness and Computational Efficiency of Kalman Filter Estimator of Time-Dependent Origin–Destination Matrices

Abstract

Origin–destination (O-D) trip matrices that describe the patterns of traffic behavior across a network are the primary data input used in principal traffic models and, therefore, a critical requirement in all advanced systems supported by dynamic traffic assignment models. However, because O-D matrices are not directly observable, the current practice consists of adjusting an initial or seed matrix from link flow counts that are provided by an existing layout of traffic-counting stations. The availability of new traffic measurements provided by information and communication technologies (ICT) allows more efficient algorithms, namely for real-time estimation of O-D matrices that are based on modified Kalman filtering approaches to exploit the new data. The quality of the estimations depends on various factors such as the penetration of the ICT devices, the detection layout, and the quality of the initial information. The feasibility of real-time applications depends on the computational performance of the proposed algorithms for urban networks of sensitive size. This paper presents the results of a set of computational experiments with a microscopic simulation of the network of Barcelona's central business district that explore the sensitivity of the Kalman filter estimates in relation to design factor values.