Romdhane, Imene, Rahman, Ziyaur, Mohamed, Nahed Belhadj, Hassan, Md. Zoheb et Kaddoum, Georges.
2026.
« Dynamic resource optimization for a joint max-min fairness and energy-efficiency problem in NOMA-aided underwater optical wireless systems ».
IEEE Open Journal of the Communications Society, vol. 7.
5163–5179.
Prévisualisation |
PDF
Kaddoum-G-2026-33781.pdf - Version publiée Licence d'utilisation : Creative Commons CC BY. Télécharger (3MB) | Prévisualisation |
Résumé
In this paper, we present a dynamic beamforming optimization framework for multi-beam underwater optical wireless communication (UOWC) systems. The UOWC transmitter concurrently transmits multiple beams, and non-orthogonal multiple access (NOMA) is applied within each beam to support multi-user communications. Our goal is to maximize the energy efficiency (EE) of a multi-beam UOWC network while guaranteeing the max-min fairness. Hence, we propose two deep deterministic policy gradient (DDPG)-based beamforming solutions that optimize beam orientations and power allocation while considering the quasi-stationarity of nodes in the underwater environment. The first solution is a single-agent DDPG-based approach, while the second solution is a multi-agent DDPG-based one. We also incorporate sequential learning capabilities into the multi-agent DDPG approach to enhance its optimality, which includes sequential learning of the beam orientation and power allocation tasks. Through extensive simulations, we show that the proposed single and multi-agent DDPG solutions achieve improved fairness and EE as compared to the equal power allocation, weighted minimum mean-square error (WMMSE)-based EE maximization with min-rate constraint (WMMSE-EE-MinRate), and QT-based EE maximization with min-rate constraint (QT-EE-MinRate) benchmarks. Specifically, the sequential multi-agent DDPG model gave at least 68% and 77% higher minimum rate and EE than benchmarks, respectively. Furthermore, the multi-agent DDPG outperforms the single-agent DDPG solution by 20% and more than 28% in terms of minimum rate and EE, respectively.
| Type de document: | Article publié dans une revue, révisé par les pairs |
|---|---|
| Chercheur(-euse): | Chercheur(-euse) Kaddoum, Georges |
| Affiliation: | Génie électrique |
| Date de dépôt: | 03 juin 2026 18:21 |
| Dernière modification: | 12 juin 2026 23:10 |
| URI: | https://espace2.etsmtl.ca/id/eprint/33781 |
Actions (Authentification requise)
 |
Dernière vérification avant le dépôt |