Abstract—Transportation modes are aplenty in today’s urban environment. Commuters use public transport such as buses, trains, taxis, personal motor vehicles, walking, bicycles, etc. to travel between places. One of the major concerns for the people who rely on public transportation is the unavailability or inaccuracy of systems that predict the estimated arrival time or the schedule based on the current location of vehicles and the traffic situation. With the advent of technology, a large set of urban transportation operators have begun to use location reporting systems such as GPS devices on-board their fleet, with the primary purpose of monitoring and managing their fleet. This paper describes methods for predicting the arrival time taking advantage of the location reports from such devices. The system pipeline developed is based on a complex event processing engine within which an algorithm is implemented to continuously predict in real-time the estimated arrival time in an online fashion. The developed system in first phase is evaluated using a vehicle simulator that generates vehicle trajectories along real public transportation routes.
Index Terms—Complex event processing, data stream mining, real-time distributed systems, spatial data mining.
Naveen Nandan is with SAP Research & Innovation, Singapore (e-mail: naveen.nandan@sap.com).
Cite:Naveen Nandan, "Online Grid-Based Dynamic Arrival Time Prediction Using GPS Locations," International Journal of Machine Learning and Computing vol.3, no. 6, pp. 516-519, 2013.