Bus Arrival Time Prediction Based on GPS Data

Abstract

As more and more vehicles are introduced into the urban streets every year, traffic congestion has become a serious problem in the urban area all over the world. A desirable strategy to alleviate congestion is to attract more people from private cars to public transit by providing better transit service. Instantaneous and accurate prediction of bus arrival time can reduce waiting times at bus stations and travel time uncertainty, which greatly improves the quality of the transit service. Previously many studies have been devoted to the transit arrival time prediction problem. However, the accuracy of the predictions still needs to be improved. This paper proposes a new methodology that can be used to predict the expected bus arrival times at each bus stop along a service route with an acceptable accuracy. By utilizing global positioning system (GPS) data the authors establish a dynamic model to calculate bus arrival times at downstream stations. This model takes consideration of the variation of travel time during different time periods within a day and different dates. The results indicate that the proposed algorithm can obtain bus arrival times prediction within acceptable accuracy.