EUVIMEDEUVIMED entdecken
Crossref

Learning Ocean Channel Cartography: A Digital Twin Approach to Communication-Aware Autonomy for Maritime 6G

Mandana Mahgoli, Ian F. Akyildiz, Kristinn Andersen, Teresa Sofia Giesta da Silva, Saemundur E. Thorsteinsson

IEEE Transactions on Communications · 2026 · 74 · 11534-11549

DOI: 10.1109/tcomm.2026.3712573 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Communications, 2026. Verfasst von Mandana Mahgoli und 4 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 51 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Autonomous underwater vehicle (AUV) swarms are fundamentally constrained by communication channels whose quality varies sharply with position and depth. Conventional path planning treats connectivity as a reactive constraint, causing mission failures when vehicles encounter dead zones from multipath fading, thermal stratification, or acoustic shadowing. This paper presents Ocean Channel Cartography (OCC), a framework that constructs persistent 3D channel quality maps via cooperative sensing and Gaussian Process (GP) interpolation with an oceanographically-structured composite kernel. Contributions include: 1) a composite GP kernel combining squared-exponential, periodic, and Matérn components to encode large-scale stratification, thermocline-related vertical periodicity, and bathymetric irregularities; 2) a hierarchical deep reinforcement learning (DRL) controller jointly optimizing mission coverage, communication reliability, and energy through uncertainty-aware rewards; 3) an adaptive multi-modal switching mechanism for acoustic, Magnetic Induction, and optical transceivers; and 4) a collaborative multi-mission learning protocol refining a shared, persistent map across deployments. Relative to an acoustic-only baseline, the multi-modal scheme yields an aggregate throughput gain of $3.87\times \pm 0.21\times $ through spatial offloading. Monte Carlo simulations show that the composite GP kernel reduces channel-map RMSE to 2.65 dB versus 3.52 dB (SE-only) and 3.57 dB (Matérn-only) averaged across eight environments (a $\approx 25\%$ reduction), with the largest environment-specific gain reaching $3.3\times $ in thermocline-dominated regimes. Under the simulated evaluation setting, the hierarchical DRL controller retained 48.2% of the initial energy budget compared with 9.9% for a Flat MARL baseline (Table III), at 82.4% vs. 86.4% mission progress. Multi-mission learning across three environments improves mission completion from 36.0% to 70.0%. These results suggest OCC as a direction worth further study for maritime 6G, with field validation identified as the essential next step.

Zitieren

APA

Mahgoli, M., Akyildiz, I. F., Andersen, K., Silva, T. S. G. D., & Thorsteinsson, S. E. (2026). Learning Ocean Channel Cartography: A Digital Twin Approach to Communication-Aware Autonomy for Maritime 6G. IEEE Transactions on Communications, 74, 11534-11549. https://doi.org/10.1109/tcomm.2026.3712573

Vancouver

Mahgoli M, Akyildiz IF, Andersen K, Silva TSGD, Thorsteinsson SE. Learning Ocean Channel Cartography: A Digital Twin Approach to Communication-Aware Autonomy for Maritime 6G. IEEE Transactions on Communications. 2026;74:11534-11549. doi:10.1109/tcomm.2026.3712573

BibTeX
@article{Mahgoli2026,
  author = {Mahgoli, Mandana and Akyildiz, Ian F. and Andersen, Kristinn and Silva, Teresa Sofia Giesta da and Thorsteinsson, Saemundur E.},
  title = {Learning Ocean Channel Cartography: A Digital Twin Approach to Communication-Aware Autonomy for Maritime 6G},
  journal = {IEEE Transactions on Communications},
  year = {2026},
  volume = {74},
  pages = {11534--11549},
  doi = {10.1109/tcomm.2026.3712573}
}
Kostenlos starten →