Detecting AIS Transmission Gaps Using Spatio-Temporal Kinematics of Marine Trajectories
27th IEEE International Conference on Mobile Data Management, MDM 2026, Athens, Yunanistan, 29 Haziran - 02 Temmuz 2026, ss.465-470, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/mdm71479.2026.00077
- Basıldığı Şehir: Athens
- Basıldığı Ülke: Yunanistan
- Sayfa Sayıları: ss.465-470
- Anahtar Kelimeler: AIS, anomaly detection, maritime analytics, trajectory mining, vessel monitoring
- Hacettepe Üniversitesi Adresli: Evet
Özet
Detecting Automatic Identification System (AIS) transmission gaps is critical for identifying potential illicit maritime activities. This paper introduces a data-driven framework that reformulates gap detection as a sub-trajectory level anomaly classification problem. Unlike traditional methods that use rigid global cutoffs, we implement trajectory-specific statistical thresholding based on the interquartile range (IQR) to identify intervals that deviate from a vessel's unique reporting behavior. The proposed approach partitions trajectories into fixed-length windows, extracts spatio-temporal kinematics, and utilizes the TabPFN tabular foundation model for efficient classification. Evaluated on a real-world AegeaNET dataset, the proposed model achieves an accuracy of 0.89 and an F1-score of 0.88, significantly outperforming the MiPo algorithm. Furthermore, sensitivity analysis highlights a trade-off: while larger window sizes maximize classification metrics (F1 up to 0.97), smaller windows are essential for maintaining the granular temporal localization required for actionable maritime surveillance.