Scholarlink

<p>A component of the Internet of Medical Things (IoMT) enables seamless communication between interconnected medical devices, facilitating the exchange of sensitive patient information. However, existing IoMT intrusion detection methods face challenges such as low F1-scores, poor adaptability, and high computational complexity. To address these issues, an Efficient Cyber Threat Detection Using a Hybrid Superpixel-Guided Weighted Graph Convolutional Network and Quantum-Enhanced Long Short-Term Memory in IoMT Application (ECTD-Hyb-MSGCNLSM-IoMT) is proposed. Initially, input data is collected from the Canadian Institute for Cybersecurity Internet of Things Dataset for Intrusion Detection and Analysis (CIC IoT DIAD) 2024 dataset. The data undergoes preprocessing using a Range Doppler Matched Filter (RDMF) to convert string elements into numeric variables, remove variations and eliminate inconsistencies. The preprocessed data is then passed to the Adaptive Elite Ant Lion Optimization Algorithm (AEALOA) for optimal feature selection. Subsequently, the selected features are classified using a Hybrid Multiscale Superpixel-Guided Weighted Graph Convolutional Network and Quantum-Enhanced Long Short-Term Memory (HybMSGWGCN-LQLSTM) to identify various cyberattacks, including Distributed Denial of Service (DDoS), Denial of Service (DoS), Reconnaissance (Recon), Message Queuing Telemetry Transport (MQTT) and spoofing attacks. Furthermore, the Brown Bear Optimization Algorithm (BBOA) fine-tunes the parameters of the proposed framework to enhance detection accuracy and reliability. The model is implemented in Python and evaluated using accuracy, precision, and F1-score metrics. The proposed method achieves 21.51%, 12.38%, 21.51% better accuracy, 28%, 22.50%, and 21.51% better precision when compared with existing IDS-CNN-IoMT, MCDS-LSTM-IoMT and EIDS-KNN-IoMT methods 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

Efficient cyber threat detection using a hybrid super pixel guided weighted graph convolutional network and quantum-enhanced long short-term memory in IoMT application

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • G Rekha,
  • T Sudha

摘要

A component of the Internet of Medical Things (IoMT) enables seamless communication between interconnected medical devices, facilitating the exchange of sensitive patient information. However, existing IoMT intrusion detection methods face challenges such as low F1-scores, poor adaptability, and high computational complexity. To address these issues, an Efficient Cyber Threat Detection Using a Hybrid Superpixel-Guided Weighted Graph Convolutional Network and Quantum-Enhanced Long Short-Term Memory in IoMT Application (ECTD-Hyb-MSGCNLSM-IoMT) is proposed. Initially, input data is collected from the Canadian Institute for Cybersecurity Internet of Things Dataset for Intrusion Detection and Analysis (CIC IoT DIAD) 2024 dataset. The data undergoes preprocessing using a Range Doppler Matched Filter (RDMF) to convert string elements into numeric variables, remove variations and eliminate inconsistencies. The preprocessed data is then passed to the Adaptive Elite Ant Lion Optimization Algorithm (AEALOA) for optimal feature selection. Subsequently, the selected features are classified using a Hybrid Multiscale Superpixel-Guided Weighted Graph Convolutional Network and Quantum-Enhanced Long Short-Term Memory (HybMSGWGCN-LQLSTM) to identify various cyberattacks, including Distributed Denial of Service (DDoS), Denial of Service (DoS), Reconnaissance (Recon), Message Queuing Telemetry Transport (MQTT) and spoofing attacks. Furthermore, the Brown Bear Optimization Algorithm (BBOA) fine-tunes the parameters of the proposed framework to enhance detection accuracy and reliability. The model is implemented in Python and evaluated using accuracy, precision, and F1-score metrics. The proposed method achieves 21.51%, 12.38%, 21.51% better accuracy, 28%, 22.50%, and 21.51% better precision when compared with existing IDS-CNN-IoMT, MCDS-LSTM-IoMT and EIDS-KNN-IoMT methods respectively.