Scholarlink

<p>Energy efficiency remains the primary operational constraint in wireless sensor networks (WSNs), driven by the limited battery capacity, restricted memory, low computational capability, and narrow bandwidth availability of sensor nodes. To address this, data compression is commonly employed as it reduces the quantity of data that must be sent to the source from each of the sensor nodes, in addition to increasing the network’s operational life, it would reduce energy usage. In this study, a new framework is introduced that integrates both data prediction and compression to improve the precision and effectiveness of data processing in WSN clusters. This architecture aims to decrease communication overhead without impacting data prediction or processing. To minimize mean-square deviation (MSD) in a dual prediction technique based on the Least Mean Square (LMS) algorithm, it employs an enhanced step size. Cluster heads (CHs) can precisely estimate sensor node values. Based on this strategy, the cluster heads and sink use a Hybrid Auto encoder-based Convolutional Neural Network (HAE-CNN) to compress and restore predicted data. This reduces environmental sensor data spatial redundancy and transmission overhead. The recommended design for cluster-based WSNs is successful and cost-efficient, according to real-world dataset analysis.</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

Data Compression and Data Prediction using Hybrid Deep Learning Techniques for Wireless Sensor Networks

  • 期刊论文

  • 发表日期：

    2026年4月4日

查看全文
  • N. Vini Antony Grace,
  • V. Ravichandran,
  • A. N. Duraivel,
  • S. Lekashri

摘要

Energy efficiency remains the primary operational constraint in wireless sensor networks (WSNs), driven by the limited battery capacity, restricted memory, low computational capability, and narrow bandwidth availability of sensor nodes. To address this, data compression is commonly employed as it reduces the quantity of data that must be sent to the source from each of the sensor nodes, in addition to increasing the network’s operational life, it would reduce energy usage. In this study, a new framework is introduced that integrates both data prediction and compression to improve the precision and effectiveness of data processing in WSN clusters. This architecture aims to decrease communication overhead without impacting data prediction or processing. To minimize mean-square deviation (MSD) in a dual prediction technique based on the Least Mean Square (LMS) algorithm, it employs an enhanced step size. Cluster heads (CHs) can precisely estimate sensor node values. Based on this strategy, the cluster heads and sink use a Hybrid Auto encoder-based Convolutional Neural Network (HAE-CNN) to compress and restore predicted data. This reduces environmental sensor data spatial redundancy and transmission overhead. The recommended design for cluster-based WSNs is successful and cost-efficient, according to real-world dataset analysis.