Big Mobility Data Analytics: Algorithms and Techniques for Efficient Trajectory Clustering

Trajectory clustering is a significant knowledge discovery operation. Especially nowadays, the need for performing advanced analytic operations over massively produced data, such as mobility traces, in efficient and scalable ways is imperative. In this thesis, initially, we address the problem of in-DBMS subtrajectory clustering and in-DBMS incremental and progressive cluster analysis. However, performing such an operation over immense volumes of data in a centralized way is far from straightforward, which calls for parallel and distributed algorithms. Subsequently, after recognizing that the bottleneck of such operations is the underlying trajectory join query, we address the Distributed Subtrajectory Join query by utilizing the MapReduce programming model. Finally, this query is utilized in order to tackle the problem of Distributed Subtrajectory Clustering in an efficient and scalable way.