NEW PERSPECTIVES ON USING SPATIAL MODELS FOR TRAVEL TIME ESTIMATION

Geostatistical models, used for estimating travel time on an urban transportation network, typically yield a smoothed spatial profile that does not take into account the directional effect of traffic. In reality, area-wide traffic is often highly directional and travel time from A to B could be very different from that from B to A, especially during peak hours. In this paper, we investigate geostatistical models for travel time estimation that specifically considers the directional effect in traffic. An empirical study is performed on the traffic data collected in the city of Tucson, Arizona. Results show that better estimation can be obtained by considering specifically the directional effect.