Towards in-memory sub-trajectory similarity search

Spatial-temporal trajectory data contains rich information aboutmoving objects and has been widely used for a large numberof real-world applications. However, the complexity of spatial-temporal trajectory data, on the one hand, and the fast collectionof datasets, on the other hand, has made it challenging to ef-ficiently store, process, and query such data. In this paper, wepropose a scalable method to analyze the sub-trajectory simi-larity search in an in-memory cluster computing environment.Notably, we have extended Apache Spark with efficient trajectoryindexing, partitioning, and querying functionalities to supportthe sub-trajectory similarity query. Our experiments on a realtrajectory dataset have shown the efficiency and effectiveness ofthe proposed method.