Nowadays large amounts of mobility data are being collected ranging from GPS trails to data collected by static or mobile traffic cameras. The management and analysis of this data requires specific treatment for which the traditional database systems are no longer suitable. Therefore most relational database systems have spatio-temporal extensions that come with special-purpose data types, new query primitives, and even extensions at the physical level such spatial indexing techniques. These extensions usually allow to more easily express and more efficiently execute queries such as: “give all cars that passed through a specific region on a specific date“. Recently also spatio-temporal data mining gained attention, addressing questions such as: “given a large collection of trajectories, identify common patterns”, “partition the set of all trajectories into clusters of similar trajectories”, or “find trajectories exhibiting abnormal mobility behaviour”. The goal of this thesis is to study suitable big data architectures, database systems, data structures, indexing techniques, etc. to support spatio-temporal data mining and scale up to real-life collections of trajectory data.

Main Advisor at Université Libre de Bruxelles (ULB)
Co-advisor at Aalborg Universitet (AAU)