IAVT: Anomalous Vessel Trajectory Detection Using AIS Data

With the mandatory installation of automatic identification system (AIS) equipment on various vessels, a large amount of vessel trajectory data is collected. It has been widely used for maritime data mining in practice. In this paper, we aim to discover anomalous trajectory patterns from AIS-based vessel trajectories, for example, automatically detecting vessel frauds, and recognizing the vessels navigating in the wrong direction. To achieve the objective, we first reconstruct the vessel trajectories using cubic splines interpolation. We then group all vessel trajectories crossing the same source-destination cell and represent each vessel trajectory as a sequence of symbols. The Isolation-Based Anomalous Vessel Trajectory (IAVT) detection method is proposed in this work, which could achieve remarkable detection performance. Finally, we propose to implement visualization experiments on realistic and simulated datasets to illustrate the superior performance of the proposed method.