A Multiagent Reinforcement Learning-based Energy-Efficient Routing Algorithm for Submerged Optical Wireless Sensor Networks
Abstract
The benefits of a rapid communication rate, ultrawide bandwidth, and reduced delay have made Submerged Optical Wireless Sensor Networks (SOWSNs) a popular topic of study. However, offering a low-consumption and dependable routing in SOWSNs is difficult because of the restricted power resources and highly fluctuating topologies induced by the water stream velocity. An Energy Efficient Routing Algorithm using Multiagent Reinforcement Learning (EERA-MRL) for SOWSNs has been presented as a solution to this problem. To better deal with the changing circumstances and promote the extension of network lifespan, the system is initially represented as a decentralized multiagent system. The remaining power and connection quality have been determined while designing the routing algorithm. A technique for assessing the entire network’s input through a decentralized reward system has been provided. Localized (direct) Reward (LR) is one of the incentive types to express the relationships among individuals or the data about two nearby nodes. A Universal Reward (UR) has been supplied to represent the efficiency of the network or the caliber of data transfer in SOWSNs. Three energy management strategies are suggested to extend the lifespan of SOWSNs. The first strategy uses reinforcement learning to navigate accurately to shorten journeys and minimize energy usage. The second strategy is to enhance node energy usage by utilizing a sleep scheduling algorithm. The last method limits each node’s information transfer depending on the received information variation rate. Simulation findings demonstrate that the suggested strategy significantly outperforms previously published methods regarding mean remaining energy ( Prem) ; hop count; and throughput. The findings demonstrate the low energy usage and elevated throughput (98.5%) of the proposed EERA-MRL-based routing protocol for SOWSNs.
Related
- Optimized energy efficient cluster based routing protocol for real time communication in 5G enabled VANETs using reinforcement learning
- A resource-efficient Q-learning algorithm using Boltzmann policy
- Robust reinforcement learning multi-baseline phase unwrapping algorithm with JPEG-LS
- Efficient Joint Trajectory Optimization and Resource Allocation for Secure UAV Networks: A Deep Reinforcement Learning Approach
- An Improved Escape Algorithm with Reinforcement Learning and Multi-Strategy Synergy for Global Optimization and Real Problems
BibTeX
@article{dinesh2026multiagent,
title = {{A Multiagent Reinforcement Learning-based Energy-Efficient Routing Algorithm for Submerged Optical Wireless Sensor Networks}},
author = {P. M. Dinesh and Y. M. Mahaboob John and P. R. Sukanya Sridevi and M. Balaji},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-75921-2},
url = {https://www.nature.com/articles/s41598-026-75921-2},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - A Multiagent Reinforcement Learning-based Energy-Efficient Routing Algorithm for Submerged Optical Wireless Sensor Networks
AU - P. M. Dinesh
AU - Y. M. Mahaboob John
AU - P. R. Sukanya Sridevi
AU - M. Balaji
PY - 2026
DA - 2026/10/27
JO - Scientific Reports
DO - 10.1038/s41598-026-75921-2
UR - https://www.nature.com/articles/s41598-026-75921-2
AB - The benefits of a rapid communication rate, ultrawide bandwidth, and reduced delay have made Submerged Optical Wireless Sensor Networks (SOWSNs) a popular topic of study. However, offering a low-consumption and dependable routing in SOWSNs is difficult because of the restricted power resources and highly fluctuating topologies induced by the water stream velocity. An Energy Efficient Routing Algorithm using Multiagent Reinforcement Learning (EERA-MRL) for SOWSNs has been presented as a solution to this problem. To better deal with the changing circumstances and promote the extension of network lifespan, the system is initially represented as a decentralized multiagent system. The remaining power and connection quality have been determined while designing the routing algorithm. A technique for assessing the entire network’s input through a decentralized reward system has been provided. Localized (direct) Reward (LR) is one of the incentive types to express the relationships among individuals or the data about two nearby nodes. A Universal Reward (UR) has been supplied to represent the efficiency of the network or the caliber of data transfer in SOWSNs. Three energy management strategies are suggested to extend the lifespan of SOWSNs. The first strategy uses reinforcement learning to navigate accurately to shorten journeys and minimize energy usage. The second strategy is to enhance node energy usage by utilizing a sleep scheduling algorithm. The last method limits each node’s information transfer depending on the received information variation rate. Simulation findings demonstrate that the suggested strategy significantly outperforms previously published methods regarding mean remaining energy ( P_rem) ; hop count; and throughput. The findings demonstrate the low energy usage and elevated throughput (98.5%) of the proposed EERA-MRL-based routing protocol for SOWSNs.
SN - 2045-2322
ER -