EUVIMEDEUVIMED entdecken
Crossref

A Hierarchical AI-Driven Hybrid Communication Framework for Underwater Swarm Coordination: Achieving Sub-Microsecond Synchronization and Robust Cooperative MIMO Performance

Mandana Mahgoli, Ian F. Akyildiz, Kristinn Andersen, Saemundur E. Thorsteinsson, Magnús Örn Úlfarsson

IEEE Internet of Things Journal · 15.06.2026 · 13 (12) · 26932-26948

DOI: 10.1109/jiot.2026.3678823 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Internet of Things Journal, erschienen am 15.06.2026. Verfasst von Mandana Mahgoli und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: Icelandic Research Fund (Rannís) Project “HAF: Underwater Robotic Sensor Networks with Multimode Devices and Remote Power Charging Capabilities”.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Autonomous underwater vehicle (AUV) swarms face a structural coordination challenge: acoustic channels provide kilometer-scale range but suffer one-way propagation delays on the order of 667 ms/km, while magnetic induction (MI) offers near-instantaneous coupling limited to short range by quasi-static field decay and seawater conductivity. We reframe underwater cooperative multiple-input multiple-output (C-MIMO) from a raw-timing problem into an aggregate-phase-variance problem: the achievable array gain depends on the residual phase variance $\sigma _{\phi } ^{2}$ , not on nanosecond timing in isolation. A seven-term budget yields $\sigma _{\phi } ^{2} \approx 0.629$ rad2 at the design operating point, of which AUV-to-AUV positioning and Doppler residuals contribute $\approx 87$ % and synchronization-related terms (MI sync at 200 ns, oscillator drift, channel-estimation residual) less than 0.4%. The budget yields an analytical beamforming gain of 6.75 dB at $N = 8$ AUVs— $\approx 75$ % of the 9.03 dB ideal coherent limit—tracked by Monte Carlo to within ±0.5 dB. Within this envelope we develop a two-tier delay-coherent Q-learning (TT-DCQL) algorithm with a delay-adjusted target $\gamma _{\text {eff}} = \gamma ^{\tau +1}$ and a synchronization-quality-weighted experience replay buffer. Across 12 training runs (4 ablation conditions $\times 3$ seeds), the full proposal attains evaluation reward $5008 \pm 115$ versus $4864 \pm 70$ for a no-delay-handling baseline (Cohen’s $d = 1.51$ , large effect). An acoustic-only-sync ablation matches reward but sees synchronization quality collapse from 0.82 to 0.01, showing that reward alone hides the sync-quality metric that governs realized C-MIMO performance in a deployment. The hierarchical decomposition reduces per-step coordination cost from $\mathcal {O}(N^{2} M^{2})$ to $\mathcal {O}(N)$ , delivering measured speedups from $1.8\times $ at $N = 8$ to $14.4\times $ at $N = 64$ .

Förderung

  • Icelandic Research Fund (Rannís) Project “HAF: Underwater Robotic Sensor Networks with Multimode Devices and Remote Power Charging Capabilities”

Zitieren

APA

Mahgoli, M., Akyildiz, I. F., Andersen, K., Thorsteinsson, S. E., & Úlfarsson, M. Ö. (2026). A Hierarchical AI-Driven Hybrid Communication Framework for Underwater Swarm Coordination: Achieving Sub-Microsecond Synchronization and Robust Cooperative MIMO Performance. IEEE Internet of Things Journal, 13(12), 26932-26948. https://doi.org/10.1109/jiot.2026.3678823

Vancouver

Mahgoli M, Akyildiz IF, Andersen K, Thorsteinsson SE, Úlfarsson MÖ. A Hierarchical AI-Driven Hybrid Communication Framework for Underwater Swarm Coordination: Achieving Sub-Microsecond Synchronization and Robust Cooperative MIMO Performance. IEEE Internet of Things Journal. 2026;13(12):26932-26948. doi:10.1109/jiot.2026.3678823

BibTeX
@article{Mahgoli2026,
  author = {Mahgoli, Mandana and Akyildiz, Ian F. and Andersen, Kristinn and Thorsteinsson, Saemundur E. and Úlfarsson, Magnús Örn},
  title = {A Hierarchical AI-Driven Hybrid Communication Framework for Underwater Swarm Coordination: Achieving Sub-Microsecond Synchronization and Robust Cooperative MIMO Performance},
  journal = {IEEE Internet of Things Journal},
  year = {2026},
  volume = {13},
  number = {12},
  pages = {26932--26948},
  doi = {10.1109/jiot.2026.3678823}
}
Kostenlos starten →