Scholarlink

<p>Intra wireless body sensor network (Intra-WBSN) is typically a short range wireless health monitoring network, Intra wireless body sensor network (Intra-WBSN) is typically a short range wireless health monitoring network, Wireless body sensor networks (WBSNs) have emerged as a transformative technology for real-time, non-invasive healthcare monitoring, enabling continuous tracking of vital physiological parameters such as heart rate, blood pressure, and glucose levels. However, their widespread deployment is hindered by critical challenges including limited energy resources, dynamic network topologies due to body movements, and stringent quality of service requirements for reliability and low latency. Conventional routing protocols, which rely on static, rule-based mechanisms, lack the adaptability and foresight needed to operate efficiently in such highly variable environments. To address these limitations, this paper proposes RELIEF-Net (reinforcement learning and federated meta-learning for energy-efficient wireless body sensor networks), a novel AI-driven framework that synergistically integrates reinforcement learning (RL) for adaptive routing, long short-term memory (LSTM) networks for predictive analytics, graph neural networks (GNNs) with attention mechanisms for energy-aware clustering, and federated meta-learning (FML) for privacy-preserving, cross-patient model personalization. At its core, RELIEF-Net employs RL to make context-aware, real-time routing decisions that optimize transmission power, prioritize critical data (e.g., cardiac alerts), and proactively reroute traffic based on LSTM-predicted energy depletion and link instability. GNN-based clustering enhances network organization and ensures efficient, topology-aware data forwarding, while FML enables decentralized learning across patients, preserving data privacy and improving scalability without centralized data aggregation. Extensive simulations demonstrate that RELIEF-Net achieves a 23% improvement in network lifetime, a packet delivery ratio (PDR) ≥ 97%, and end-to-end latency ≤ 85&#xa0;ms for critical data under diverse operational scenarios. By unifying predictive intelligence, adaptive control, and scalable privacy-conscious learning, RELIEF-Net establishes a robust and sustainable solution for intelligent healthcare IoT systems, paving the way for next-generation remote patient monitoring and improved clinical outcomes.</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

Integrating reinforcement learning and federated meta-learning for energy-efficient wireless body sensor networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月8日

查看全文
  • Soufiane Ben Othman

摘要

Intra wireless body sensor network (Intra-WBSN) is typically a short range wireless health monitoring network, Intra wireless body sensor network (Intra-WBSN) is typically a short range wireless health monitoring network, Wireless body sensor networks (WBSNs) have emerged as a transformative technology for real-time, non-invasive healthcare monitoring, enabling continuous tracking of vital physiological parameters such as heart rate, blood pressure, and glucose levels. However, their widespread deployment is hindered by critical challenges including limited energy resources, dynamic network topologies due to body movements, and stringent quality of service requirements for reliability and low latency. Conventional routing protocols, which rely on static, rule-based mechanisms, lack the adaptability and foresight needed to operate efficiently in such highly variable environments. To address these limitations, this paper proposes RELIEF-Net (reinforcement learning and federated meta-learning for energy-efficient wireless body sensor networks), a novel AI-driven framework that synergistically integrates reinforcement learning (RL) for adaptive routing, long short-term memory (LSTM) networks for predictive analytics, graph neural networks (GNNs) with attention mechanisms for energy-aware clustering, and federated meta-learning (FML) for privacy-preserving, cross-patient model personalization. At its core, RELIEF-Net employs RL to make context-aware, real-time routing decisions that optimize transmission power, prioritize critical data (e.g., cardiac alerts), and proactively reroute traffic based on LSTM-predicted energy depletion and link instability. GNN-based clustering enhances network organization and ensures efficient, topology-aware data forwarding, while FML enables decentralized learning across patients, preserving data privacy and improving scalability without centralized data aggregation. Extensive simulations demonstrate that RELIEF-Net achieves a 23% improvement in network lifetime, a packet delivery ratio (PDR) ≥ 97%, and end-to-end latency ≤ 85 ms for critical data under diverse operational scenarios. By unifying predictive intelligence, adaptive control, and scalable privacy-conscious learning, RELIEF-Net establishes a robust and sustainable solution for intelligent healthcare IoT systems, paving the way for next-generation remote patient monitoring and improved clinical outcomes.