Modeling massive AIS streams with quad trees and Gaussian Mixtures

Pressing issues related to the movement of people and goods can be tackled today thanks to improvements in tracking and communications technology that have made it possible to collect movement data on a big scale. Maritime data from the Automatic Identification System (AIS) is one of the fast growing sources of movement data. Existing approaches for AIS data analysis suffer from scalability issues. Therefore, scalable distributed modelling and analysis approaches are needed. This paper presents a novel scalable movement data model that takes advantage of an adaptive grid based on quad trees. Our data model supports anomaly detection in massive movement data streams by combining advantages of both grid and vector-based approaches. We demonstrate the applicability of this approach for anomaly detection in AIS datasets comprising 560 million location records.