This thesis investigates the viability of real-time traffic monitoring based on location information that originates from general mobile phones. The worldwide propagation of mobile phone subscribers and the rapid progress of mobile phone locating technologies in recent years have provided the potential to infer road network traffic conditions area-wide without requiring significant infrastructure implement. While some past research has resulted in various methodologies for collecting traffic data based on mobile phones, none of them has achieved a conclusive scheme that is applicable to practical conditions such as probe types, road network attributes, and computational limitations.; This thesis research has developed a practical scheme of traffic monitoring based on mobile phone location referencing. The proposed scheme includes a novel mathematical formulation for trajectory estimation, an implementation of multi-label shortest path algorithm, probabilistic decomposition of time intervals for travel time estimation, and a new probe filtering approach derived from the mechanism of the travel time estimation. The models and algorithms constitute the major contributions of this thesis research.; The research has also evaluated the implementation of the proposed scheme. A simulation study covers a wide range of traffic conditions, road network topology, and mobile phone locating system characteristics that are representative of those likely to occur in a field implementation. The results of the evaluation suggest that the proposed scheme can provide estimates of average link travel times with average errors of approximately 8% for freeway links and approximately 25% for all links in the network.
展开▼
