Active Data Routing Based on Reward Backpropagation-Enabled Multi-Agent Q-Learning Toward SDN-Enabled Wireless Buoy Networks
IEEE Transactions on Mobile Computing · 2026-03 · 25 (3) · 3087-3102
DOI: 10.1109/tmc.2025.3611873 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-03. Verfasst von Yuan Liu und 3 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 56 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Frontier Technologies Research and Development Program of Jiangsu, Open Fund of State Key Laboratory of Acoustics and Marine Information.
Abstract (Originalsprache)
Advancements in Wireless Buoy Network (WBN) have significantly accelerated the development of marine exploitation and monitoring, acting as a relay between underwater and surface networks in emerging 6G scenarios. Due to unstable maritime communication environment, it is a challenging issue to deploy the optimal data routing or collection strategies to ensure the collected data to be delivered to the target point. By employing the Software-Defined Networking (SDN) technology, this paper proposes the paradigm of Software-Defined WBN (SDWBN) to improve the network management efficiency and provide a platform to embed the Multi-Agent Reinforcement Learning (MARL) framework (for data routing intelligence), respectively. On account of the proposed SDWBN, this paper proposes a Reward Backpropagation-enabled Multi-Agent Deep Q-learning algorithm (RBMADQ)-based active routing scheme, which aims to assist buoys in making routing decisions and navigating the challenges posed by the dynamic and unstable communication environment. Further, this paper proposes a dual replay buffer-based training method, to enhance the convergence speed of the proposed RBMADQ-based routing scheme. Evaluation results demonstrate that the proposed routing scheme performs better compared with recent research products, with a higher packet delivery rate, lower network latency, and simultaneously, less communication overhead, etc.
Förderung
- National Natural Science Foundation of China
- Frontier Technologies Research and Development Program of Jiangsu
- Open Fund of State Key Laboratory of Acoustics and Marine Information
- Liaoning Provincial Doctoral Research Start-up Fund Award
Zitieren
APA
Liu, Y., Han, G., Lin, C., & Zhu, S. (2026). Active Data Routing Based on Reward Backpropagation-Enabled Multi-Agent Q-Learning Toward SDN-Enabled Wireless Buoy Networks. IEEE Transactions on Mobile Computing, 25(3), 3087-3102. https://doi.org/10.1109/tmc.2025.3611873
Vancouver
Liu Y, Han G, Lin C, Zhu S. Active Data Routing Based on Reward Backpropagation-Enabled Multi-Agent Q-Learning Toward SDN-Enabled Wireless Buoy Networks. IEEE Transactions on Mobile Computing. 2026;25(3):3087-3102. doi:10.1109/tmc.2025.3611873
BibTeX
@article{Liu2026,
author = {Liu, Yuan and Han, Guangjie and Lin, Chuan and Zhu, Shengchao},
title = {Active Data Routing Based on Reward Backpropagation-Enabled Multi-Agent Q-Learning Toward SDN-Enabled Wireless Buoy Networks},
journal = {IEEE Transactions on Mobile Computing},
year = {2026},
volume = {25},
number = {3},
pages = {3087--3102},
doi = {10.1109/tmc.2025.3611873}
}
EUVIMED