Distributed Spatiotemporal Distance Join for Trajectory Data

Due to the rapid growth in location-based services throughout the years, extremely large volumes of trajectory data are generated and stored every day. One major challenge is how to efficiently process and analyze this data at scale. One crucial operation in numerous applications, e.g., shipping, transport planning etc., is the join operation on trajectories which takes two or more trajectory datasets and determines where the trajectories are spatially and temporally close.In this paper, we develop a distributed approach to answer trajectory join queries for large trajectory datasets. Our technique differs from other approaches by using a sequence of line segments to represent trajectories rather than a sequence of points. As we show in extensive experiments, doing so improves the efficiency and scalability of our approach by accelerating up to 43x compared to state of the art.