Scholarlink

<p>Wireless Sensor Networks (WSNs) face critical energy efficiency challenges due to resource limitations, especially in extending network lifetime. This paper presents a reinforcement learning-based solution combining Linear Optimization State-Action-Reward-State-Action (LO-SARSA) for energy-aware Cluster Head (CH) selection and a Proportional Optimization Policy Gradient Method (PO-PGM) for optimal routing. CH selection is modeled as an RL problem, where LO-SARSA learns policies based on residual energy and node distance, while PO-PGM uses a neural network to make adaptive routing decisions. Extensive simulations demonstrate that the proposed approach significantly improves energy consumption, network lifetime, and packet delivery ratio, achieving up to a 20% gain in network lifetime over conventional methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-Faceted Reinforcement Learning Frameworks for Dynamic Cluster Head Selection and Energy-Efficient Routing in Wireless Sensor Networks

  • 期刊论文

  • 发表日期：

    2025年5月26日

查看全文
  • S. Sudhamsu Mouli,
  • T. Veeraiah Talagondapati,
  • M. P. Singh

摘要

Wireless Sensor Networks (WSNs) face critical energy efficiency challenges due to resource limitations, especially in extending network lifetime. This paper presents a reinforcement learning-based solution combining Linear Optimization State-Action-Reward-State-Action (LO-SARSA) for energy-aware Cluster Head (CH) selection and a Proportional Optimization Policy Gradient Method (PO-PGM) for optimal routing. CH selection is modeled as an RL problem, where LO-SARSA learns policies based on residual energy and node distance, while PO-PGM uses a neural network to make adaptive routing decisions. Extensive simulations demonstrate that the proposed approach significantly improves energy consumption, network lifetime, and packet delivery ratio, achieving up to a 20% gain in network lifetime over conventional methods.