Scholarlink

<p>Wireless Body Area Network (WBAN)-based healthcare surveillance systems have expanded rapidly due to recent advances in wireless sensor networks. A small sensor affixed to the human body is capable of measuring a range of psychological parameters. Noise, patient perspiration, and hardware misalignment are common causes of WBAN sensor failures. It is challenging for medical professionals to determine whether the information gathered by these sensors is inaccurate or impacted by malevolent attacks. The classification of physiological parameters as normal or abnormal can result from inaccurate data. This work presents a novel approach utilizing hybrid machine learning algorithms to analyse anomaly identification for wireless devices. The primary objective of this research is to identify anomalous values in the dataset using a hybrid Deep Convolutional Neural Network—Grasshopper Optimization Algorithm (DCNN-GOA). The predicted value of each anomalous record can be used to determine if the identified form is anomalous. The proposed DCNN-GOA algorithm is evaluated regarding accuracy, sensitivity, F1 score, and specificity. The experimental outcomes of the proposed hybrid DCNN-GOA algorithm outperform other existing algorithms in terms of precision (99.8%), F-score (97.9%), recall (99.9%), and accuracy (98.8%). The proposed DCNN-GOA protocol outperforms existing ANN, RNN, CNN-GRU, Conv GRU, Conv LSTM, and CNN-LSTM classifiers regarding accuracy, with 17.4%, 7.83%, 5.12%, 3.09%, and 1.85%, respectively.</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

Enhancing anomaly detection in WBANs using hybrid deep learning and optimization algorithms

  • 期刊论文

  • 发表日期：

    2025年2月21日

查看全文
  • T. Thamaraimanalan,
  • S. Ramalingam

摘要

Wireless Body Area Network (WBAN)-based healthcare surveillance systems have expanded rapidly due to recent advances in wireless sensor networks. A small sensor affixed to the human body is capable of measuring a range of psychological parameters. Noise, patient perspiration, and hardware misalignment are common causes of WBAN sensor failures. It is challenging for medical professionals to determine whether the information gathered by these sensors is inaccurate or impacted by malevolent attacks. The classification of physiological parameters as normal or abnormal can result from inaccurate data. This work presents a novel approach utilizing hybrid machine learning algorithms to analyse anomaly identification for wireless devices. The primary objective of this research is to identify anomalous values in the dataset using a hybrid Deep Convolutional Neural Network—Grasshopper Optimization Algorithm (DCNN-GOA). The predicted value of each anomalous record can be used to determine if the identified form is anomalous. The proposed DCNN-GOA algorithm is evaluated regarding accuracy, sensitivity, F1 score, and specificity. The experimental outcomes of the proposed hybrid DCNN-GOA algorithm outperform other existing algorithms in terms of precision (99.8%), F-score (97.9%), recall (99.9%), and accuracy (98.8%). The proposed DCNN-GOA protocol outperforms existing ANN, RNN, CNN-GRU, Conv GRU, Conv LSTM, and CNN-LSTM classifiers regarding accuracy, with 17.4%, 7.83%, 5.12%, 3.09%, and 1.85%, respectively.