Deep Learning for Dynamic Slant Channel Modeling in Underwater Wireless Optical Communications

Le résumé fourni par la source

Underwater wireless optical communications (UWOC), with their large bandwidth potential, are a promising solution for the emerging Internet of Underwater Things (IoUT). However, when transmitting data to marine surface vehicles (MSVs) via UWOC links, the signal suffers from multivariate impairments, including seawater absorption, scattering, turbulence, and transceiver misalignment. These challenges complicate channel modeling and performance evaluation, particularly when MSVs experience random posture (RP) variations induced by wind and waves. This study aims to accurately model the combined effects of multiple impairments in dynamic slant UWOC links using parallelized wave optics simulation (WOS). This approach avoids complex mathematical derivations while clearly capturing the interdependencies among depth-related random channel factors. To further improve computational efficiency, we propose a convolutional neural network-based fast channel simulation method (CNN-FCSM) for predicting received optical intensity in integrated slant UWOC channel modeling. This method bypasses computationally intensive simulations, drastically reducing processing overhead. The CNN-FCSM's predictions are then used to evaluate outage probability (OP) and average bit-error rate (ABER) performance under dynamic channel conditions. Through comparative analysis with WOS benchmarks, we demonstrate that the CNN-FCSM achieves superior computational efficiency while maintaining high prediction accuracy.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.