Scholarlink

<p>The growing deployment of Internet of Things (IoT) infrastructures has significantly expanded the attack surface for cyber threats. To address this, we propose ResNet50_CNN1D, a novel deep learning–based intrusion detection framework designed to enhance security in IoT environments. The model synergistically combines the representational power of residual learning (ResNet50) with the feature extraction efficiency of one-dimensional convolutions (CNN1D) to effectively process complex, multivariate network traffic patterns. Evaluated on the comprehensive Edge-IIoTset benchmark dataset, our framework achieves an overall accuracy of 94%. It exhibits robust multi-class detection capabilities, performing strongly against prevalent attacks such as DDoS and ransomware. A detailed class-wise performance analysis confirms the model’s efficacy while also revealing specific challenges in detecting minority attack classes. The findings underscore the proposed framework’s potential as an effective solution for scalable and accurate intrusion detection in IoT networks.</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

Mitigating cyber risks in IoT environments using ResNet50_CNN1D and Edge-IIoTset insights

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月10日

查看全文
  • Hanaa ZainEldin,
  • Samah A. Gamel,
  • Fatma M. Talaat

摘要

The growing deployment of Internet of Things (IoT) infrastructures has significantly expanded the attack surface for cyber threats. To address this, we propose ResNet50_CNN1D, a novel deep learning–based intrusion detection framework designed to enhance security in IoT environments. The model synergistically combines the representational power of residual learning (ResNet50) with the feature extraction efficiency of one-dimensional convolutions (CNN1D) to effectively process complex, multivariate network traffic patterns. Evaluated on the comprehensive Edge-IIoTset benchmark dataset, our framework achieves an overall accuracy of 94%. It exhibits robust multi-class detection capabilities, performing strongly against prevalent attacks such as DDoS and ransomware. A detailed class-wise performance analysis confirms the model’s efficacy while also revealing specific challenges in detecting minority attack classes. The findings underscore the proposed framework’s potential as an effective solution for scalable and accurate intrusion detection in IoT networks.