Statistical Analysis-Based Feature Selection for Anomaly Detection in AIS Dataset
Gábor Visky, Risto Vaarandi, Sokratis K. Katsikas, Olaf Maennel · 2025
A global trend in the progressive digitalisation of the world is affecting different industries, including the maritime transport sector. Electronic navigation and autonomous sailing heavily rely on sensor data, such as Global Navigation Satellite Systems (GNSS), Light Detection and Ranging (LIDAR), Radio Detection and Ranging (RADAR), or Automated Identification System (AIS) systems. Interference with these systems can endanger the situational awareness of the ship control system and influence navigation-related decisions. Our research focuses on AIS data and seeks possible features for anomaly detection based on transmission timing, reported and calculated speed analysis. We conducted a comprehensive statistical analysis of a 24-hour-long AIS dataset recorded in Tallinn to highlight the special characteristics of such data. Our findings suggest that the use of a single speed-based feature offers limited benefits, leading us to propose the combination of several speed-based features for anomaly detection. The results of this research have the potential to impact the cyber security of ship's navigation systems by the identified properties of AIS data.