Searching K-Nearest Neighbor Trajectories On Road Networks

With the proliferation of mobile devices, massive trajectory data has been generated. Searching trajectories by locations is one of fundamental tasks. Previous work such as [3,6,9] has been proposed to answer the search. Such work typically measures the distance between trajectories and queries by the distance between query points and GPS points of trajectories. Such measurement could be inaccurate because those GPS points generated by some sampling rate are essentially discrete. To overcome this issue, we treat a trajectory as a sequence of line segments and compute the distance between a query point and a trajectory by the one between the query point and line segments. Next, we index the line segments by R-tree and match each trajectory to the associated line segments by inverted lists. After that, we propose a k-nearest neighbor (KNN) search algorithm on the indexing structure. Moreover, we propose to cluster line segments and merge redundant trajectory IDs for higher efficiency. Experimental results validate that the proposed method significantly outperforms existing approaches in terms of saving storage cost of data and the query performance.