Boosting Point-Based Trajectory Search with Quad-Tree

The availability of spatial data generated by objects enables people to search for a similar pattern using a set of query points. In this paper, we focus on point-based trajectory search problem which returns top-k results to a set of query points. The primary purpose of this work is to revisit state-of-the-art search algorithms on various indices and find the best choice of spatial index while giving a reason behind it. Furthermore, we propose an optimization on the search method, which is able to find the initial upper bound for the query points, leading to further performance improvement. Lastly, extensive experiments on real dataset verified the choice of the index and our proposed search method.