Scholarlink

<p>Underwater wireless sensor networks (UWSNs) face significant challenges in accurate node localization due to Non-Line-of-Sight (NLOS) conditions caused by multipath propagation, signal attenuation, and environmental factors. In order to solve these issues, this paper proposes a hybrid deep learning framework which integrates CNN, and LSTM networks. This system is able to identify if it is in NLOS or Line of sight condition, it is capable of classifying the type of NLOS it is experiencing, and is able to correct localization errors, all in a single pipeline. The model outperforms existing methods, as shown by experimental results, achieving 92–95% of accuracy and 0.92<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11227_2025_7942_Article_IEq1.gif" Format="GIF" Height="13" Rendition="HTML" Resolution="72" Type="Linedraw" Width="19" /> </InlineMediaObject> <EquationSource Format="TEX">\(-\)</EquationSource> <EquationSource Format="MATHML"><math> <mo>-</mo> </math></EquationSource> </InlineEquation>0.95 of F1-score, while also being robust in low signal-to-noise ratio (SNR) scenarios (65–70% of accuracy on 0–5dB). It also achieves 8–10% better performance than the baseline in SNR levels on the mid-range, between 7.5 and 12.5 dB. The hybrid architecture is also more energy efficient, with 15–30% lower energy use without sacrificing real-time processing, with inferencing times around 15–20 ms. These findings show the flexibility of this framework in a dynamic underwater environment and its potential for being a solution for UWSN localization.</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

Deeploc: a CNN-LSTM framework for NLOS-aware localization in underwater sensor networks

  • 期刊论文

  • 发表日期：

    2025年10月25日

查看全文
  • Nadia Shamshad,
  • Lei Wang,
  • Danish Sarwr,
  • Syed Agha Hassnain Mohsan

摘要

Underwater wireless sensor networks (UWSNs) face significant challenges in accurate node localization due to Non-Line-of-Sight (NLOS) conditions caused by multipath propagation, signal attenuation, and environmental factors. In order to solve these issues, this paper proposes a hybrid deep learning framework which integrates CNN, and LSTM networks. This system is able to identify if it is in NLOS or Line of sight condition, it is capable of classifying the type of NLOS it is experiencing, and is able to correct localization errors, all in a single pipeline. The model outperforms existing methods, as shown by experimental results, achieving 92–95% of accuracy and 0.92 \(-\) - 0.95 of F1-score, while also being robust in low signal-to-noise ratio (SNR) scenarios (65–70% of accuracy on 0–5dB). It also achieves 8–10% better performance than the baseline in SNR levels on the mid-range, between 7.5 and 12.5 dB. The hybrid architecture is also more energy efficient, with 15–30% lower energy use without sacrificing real-time processing, with inferencing times around 15–20 ms. These findings show the flexibility of this framework in a dynamic underwater environment and its potential for being a solution for UWSN localization.