Scholarlink

<p>Wireless Sensor Networks (WSNs), comprising geographically distributed sensor nodes, play a vital role in Internet of Things (IoT) applications through continuous environmental data exchange. However, persistent data sensing and transmission lead to rapid energy depletion, thereby reducing network lifetime. Addressing this challenge, this research proposes an energy-efficient framework named Namib Beetle Algorithm-based Clustering and Jump Gain Integral Recurrent Neural Network for IoT (NBAC–JGTRNN–EE–IoT). In the proposed model, the Namib Beetle Algorithm (NBA) performs optimal clustering, while the Cluster Head (CH) is selected using a Depthwise Separable Convolutional Neural Network integrated with Deep Support Vector Machine (DS-CNN–DSVM) to ensure high precision in CH identification. Subsequently, the JGIRNN-based routing mechanism optimizes data transmission by selecting the shortest and most energy-efficient paths between clusters and the sink node. The primary objectives of NBAC–JGTRNN–EE–IoT are to maximize network lifetime, minimize energy consumption, and enhance data transmission quality. Performance evaluation metrics include energy consumption, packet loss rate, end-to-end latency, live and dead nodes, and packet transmission efficiency. The model was implemented in Python and benchmarked against state-of-the-art methods, namely MTCF–SEE–WSN–IoT, EECRP–IoT–SI, and EEML–SRP–IoT. Simulation results demonstrate that the proposed NBAC–JGTRNN–EE–IoT reduces packet transmission to sink nodes by 26.78%, 31.82%, and 29.31%, respectively, compared to the baselines. Furthermore, it achieves 90% convergence within 125 iterations, outperforming PSO (190 iterations) and GA (210 iterations), while maintaining a low performance variance of ± 2.3%, surpassing PSO (± 5.6%) and GA (± 6.2%). These results validate the proposed framework’s superior energy efficiency, convergence speed, scalability, and robustness, making it a promising solution for sustainable and intelligent IoT-based WSN deployments.</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

Namib Beetle Algorithm Based Clustering and Jump Gain Integral Recurrent Neural Network for Efficient Energy in IoT Networks

  • 期刊论文

  • 发表日期：

    2025年12月19日

查看全文
  • S. Regilan,
  • L. K. Hema,
  • D. Kadhiravan,
  • J. Jenitha

摘要

Wireless Sensor Networks (WSNs), comprising geographically distributed sensor nodes, play a vital role in Internet of Things (IoT) applications through continuous environmental data exchange. However, persistent data sensing and transmission lead to rapid energy depletion, thereby reducing network lifetime. Addressing this challenge, this research proposes an energy-efficient framework named Namib Beetle Algorithm-based Clustering and Jump Gain Integral Recurrent Neural Network for IoT (NBAC–JGTRNN–EE–IoT). In the proposed model, the Namib Beetle Algorithm (NBA) performs optimal clustering, while the Cluster Head (CH) is selected using a Depthwise Separable Convolutional Neural Network integrated with Deep Support Vector Machine (DS-CNN–DSVM) to ensure high precision in CH identification. Subsequently, the JGIRNN-based routing mechanism optimizes data transmission by selecting the shortest and most energy-efficient paths between clusters and the sink node. The primary objectives of NBAC–JGTRNN–EE–IoT are to maximize network lifetime, minimize energy consumption, and enhance data transmission quality. Performance evaluation metrics include energy consumption, packet loss rate, end-to-end latency, live and dead nodes, and packet transmission efficiency. The model was implemented in Python and benchmarked against state-of-the-art methods, namely MTCF–SEE–WSN–IoT, EECRP–IoT–SI, and EEML–SRP–IoT. Simulation results demonstrate that the proposed NBAC–JGTRNN–EE–IoT reduces packet transmission to sink nodes by 26.78%, 31.82%, and 29.31%, respectively, compared to the baselines. Furthermore, it achieves 90% convergence within 125 iterations, outperforming PSO (190 iterations) and GA (210 iterations), while maintaining a low performance variance of ± 2.3%, surpassing PSO (± 5.6%) and GA (± 6.2%). These results validate the proposed framework’s superior energy efficiency, convergence speed, scalability, and robustness, making it a promising solution for sustainable and intelligent IoT-based WSN deployments.