Sensor Data Fusion for Improving Traffic Mobility in Smart Cities

The ever-increasing urban population and vehicular traffic without a corresponding expansion of infrastructure have been a challenge to transportation facilities managers and commuters. While some parts of transportation infrastructure have big data available, so many other locations have sparse data. This has posed a challenge in traffic state estimation and prediction for efficient and effective infrastructure management and route guidance. This research focused on traffic prediction problems and aims to develop novel spatial-temporal and robust algorithms, that can provide high accuracy in the presence of both big data and sparse data in a large urban road network. Intelligent transportation systems require the knowledge of current traffic state and forecast for effective implementation. The actual traffic state has to be estimated as the existing sensors do not capture the needed state. Sensor measurements often contain missing or incomplete data as a result of communication issues, faulty sensors or cost leading to incomplete monitoring of the entire road network. This missing data pose challenges to traffic estimation approaches. In this work, a robust spatio-temporal traffic imputation approach capable of withstanding high missing data rate is presented. A particle-based approach with Kriging interpolation is proposed. The performance of the particle-based Kriging interpolation for different missing data ratios was investigated for a large road network. A particle-based framework for dealing with missing data is also proposed. An expression of the likelihood function is derived for the case when the missing value is calculated based on Kriging interpolation. With the Kriging interpolation, the missing values of the measurements are predicted, which are subsequently used in the computation of likelihood terms in the particle filter algorithm. In the commonly used Kriging approaches, the covariance function depends only on the separation distance irrespective of the traffic at the considered locations. A key limitation of such an approach is its inability to capture well the traffic dynamics and transitions between different states. This thesis proposes a Bayesian Kriging approach for the prediction of urban traffic. The approach can capture these dynamics and model changes via the covariance matrix. The main novelty consists in representing both stationary and non-stationary changes in traffic flows by a discriminative covariance function conditioned on the observation at each location. An advantage is that by considering the surrounding traffic information distinctively, the proposed method is very likely to represent congested regions and interactions in both upstream and downstream areas.