Scholarlink

<p>The widespread and rapid adoption of Internet of Things (IoT) devices as wireless networks poses challenges in ensuring data security. It is crucial to enhance the security of IoT networks to protect them against a wide range of attacks. Owing to the presence of various vulnerabilities (or attacks) in IoT networks at the same time, the classification of IoT vulnerabilities has become a multiclass problem. This research aimed to provide a deep learning based IoT security framework for the purpose of categorizing and forecasting vulnerabilities. Principal component analysis is employed to reduce the feature set, while sigmoid and ReLU activation is utilized to address the vanishing gradient issue in binary classification. In addition, the performance of binary classification is compared to that of multiclass classification models. The primary goal of detection model is to ascertain whether a sample is benign or harmful. The goal of the classification algorithms is to categorize each sample into one of the following classes: benevolent, tsunami, Mirai, or Gafgyt. The classification of attacks is accomplished through the utilization of a deep neural network (DNN) learning model. The dataset employed is the N-BaIoT multiple vulnerabilities dataset, which exhibits an unbalanced nature. This research proposed the utilization of the oversampling technique synthetic minority oversampling technique (SMOTE) to balance the dataset. The performances of the SMOTE are compared with those of the random oversampling and undersampling approaches. Performance is evaluated with state-of-the-art decision tree and a random forest. The proposed DNN technique achieves an accuracy over 99.99% in all classifier cases.</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

Development of a robust model for identifying vulnerabilities in IoT devices via binary and multiclass deep neural network classification

  • 期刊论文

  • 发表日期：

    2025年6月5日

查看全文
  • Richa Jain,
  • Ashok Kumar

摘要

The widespread and rapid adoption of Internet of Things (IoT) devices as wireless networks poses challenges in ensuring data security. It is crucial to enhance the security of IoT networks to protect them against a wide range of attacks. Owing to the presence of various vulnerabilities (or attacks) in IoT networks at the same time, the classification of IoT vulnerabilities has become a multiclass problem. This research aimed to provide a deep learning based IoT security framework for the purpose of categorizing and forecasting vulnerabilities. Principal component analysis is employed to reduce the feature set, while sigmoid and ReLU activation is utilized to address the vanishing gradient issue in binary classification. In addition, the performance of binary classification is compared to that of multiclass classification models. The primary goal of detection model is to ascertain whether a sample is benign or harmful. The goal of the classification algorithms is to categorize each sample into one of the following classes: benevolent, tsunami, Mirai, or Gafgyt. The classification of attacks is accomplished through the utilization of a deep neural network (DNN) learning model. The dataset employed is the N-BaIoT multiple vulnerabilities dataset, which exhibits an unbalanced nature. This research proposed the utilization of the oversampling technique synthetic minority oversampling technique (SMOTE) to balance the dataset. The performances of the SMOTE are compared with those of the random oversampling and undersampling approaches. Performance is evaluated with state-of-the-art decision tree and a random forest. The proposed DNN technique achieves an accuracy over 99.99% in all classifier cases.