AIS-based Anomaly Detection for IUU Fishing Activities

Omkar Hadawale; Siddharth Shinde; Kaushal Kadu; Nikita Saindane

Abstract

The future of the Earth’s fish businesses is seriously threatened by the continuous use of illegal or illicit, unreported, and unregulated (IUU) fishing methods, a growing global demand, and deteriorating ocean ecosystem health. The livelihoods of legal fishing are also harmed by IUU fishing. The ongoing efforts to develop sustainable fisheries policies are also hampered by this. In order to manage fishery resources and ensure the safety of maritime traffic, fishing activity must be tracked and predicted. Ocean traffic situation awareness depends on the static and dynamic data that the automatic identification system (AIS) reports about a ship. AIS systems can, however, be disabled to cover up prohibited or unlawful activity, including as piracy or illicit fishing. To clearly differentiate between deliberate and accidental AIS transmission switching anomalies, we suggest a Multiclass Supervised Machine Learning based anomaly detection system. The multi-class anomaly framework collects AIS communications that failed for a variety of reasons, such as power outages or purposeful AIS shut-off. The location, name, and message timestamp of the vessel are transmitted via AIS. To forecast the ship's direction, speed, and course throughout its anomalous time, we employ Random Forest Classification.

Abstract

Keywords