EUVIMEDEUVIMED entdecken
Crossref

Adaptive Unscented Kalman Filter for Robust Localization in Ship Ad hoc Networks

Anfal K

Alkadhim Journal for Computer Science · 2026

DOI: 10.61710/edkzny54 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Alkadhim Journal for Computer Science, 2026. Verfasst von Anfal K. Die Arbeit nennt 21 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Reliable ship localization is critical for tracking and collision avoidance in Ship Ad hoc Networks (SANETs), yet traditional localization systems like Global Positioning System (GPS) and Automatic Identification System (AIS) often struggle with noise and signal loss. In this paper, we developed a network-aware Adaptive Unscented Kalman Filter (AUKF) to enhance ship localization. Our method improves the standard UKF by dynamically tuning the process noise covariance matrix using SANETs topological data (node degree and connected component size). This method maintains high accuracy even during irregular ship movements or communications drops. The evaluations using real-world and simulated ship datasets show the proposed framework outperforms the classical UKF. The AUKF achieved a 38.8% drop in Mean Absolute Error (MAE) and a 24.6% reduction in Root Mean Square Error (RMSE), alongside a 42.8% decrease in the 95th percentile error (P95). These findings prove that the network features fused with physical filters are highly effective to enhance localization accuracy in dynamic marine environments.

Zitieren

APA

K, A. (2026). Adaptive Unscented Kalman Filter for Robust Localization in Ship Ad hoc Networks. Alkadhim Journal for Computer Science. https://doi.org/10.61710/edkzny54

Vancouver

K A. Adaptive Unscented Kalman Filter for Robust Localization in Ship Ad hoc Networks. Alkadhim Journal for Computer Science. 2026. doi:10.61710/edkzny54

BibTeX
@article{K2026,
  author = {K, Anfal},
  title = {Adaptive Unscented Kalman Filter for Robust Localization in Ship Ad hoc Networks},
  journal = {Alkadhim Journal for Computer Science},
  year = {2026},
  doi = {10.61710/edkzny54}
}
Kostenlos starten →