Route prediction using trip observations and map matching

This paper uses location data traces (from GPS, Mobile Signals etc.) of past trips of vehicles to develop algorithm for predicting the end-to-end route of a vehicle. Focus is on overall route prediction rather than predicting road segments in short term. Researches in past for route prediction makes use of raw location data traces data decomposed into trips for such route predictions. This paper introduces an additional step to convert trips composed of location data traces points to trips of road network edges. This requires the algorithm to make use of road networks. We show that efficiency in storage and time complexity can be achieved without sacrificing the accuracy by doing so. Moreover, its well-known that location traces data has inherent inaccuracies due to hardware limitations of devices. Most of the researches don't handle it. This paper presents the results of route prediction algorithms under inaccuracies in data.