SimilarMove

Abstract

Trajectory prediction has a significant impact on many location-based services such as local search, traffic management, and routing services. Existing trajectory prediction techniques utilize the object's motion history to predict the future path(s). However, these techniques fail when the history is unavailable which realistically happens for multiple reasons such as; history might be difficult to obtain, newly registered user has no past history, or previously recorded data is protected for privacy reasons. This paper introduces a novel system named SimilarMove to predict the future paths of moving objects on road networks without relying on their past trajectories. SimilarMove analyzes the motion pattern of the moving object under investigation and identifies other moving objects that show similar motion patterns. Then, a Markov Model is adopted to digest this set of similar motion patterns and produce the next potential movements of the object under investigation along with their likelihoods. A key aspect of SimilarMove lies in achieving a high quality prediction while being efficient in terms of performance.