Scholarlink

<p>In recent years, the widespread adoption of Internet of Things (IoT) devices in modern life has been significantly influenced by smart home automation, healthcare, wearable devices, and smart cities have significantly influenced the use of IoT devices in today’s digital era. This has led to a growing concern for data privacy, and the implementation of robust security measures remains a critical challenge. To address these challenges, we propose a novel feature amalgamation technique for mitigating these risks. The proposed method enhances the performance of the kervolutional neural network (KNN) model by detecting various threats in the IoT network and our key contribution lies in integrating the delta and delta-delta features with the input features of the UNSW-NB15 dataset at different levels, achieving a feature fusion that captures both temporal and spatial information. The experimental results demonstrate the superiority of our proposed approach on the UNSW-NB15 dataset, with a decision-level feature detection rate of 99.2% for binary attacks and 96.1% for multiple attacks, thus outperforming the state-of-the-art methods.</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

Kervolutional Neural Network with Feature Fusion for Detecting IoT Security Threats

  • 期刊论文

  • 发表日期：

    2025年11月25日

查看全文
  • D. R. Janardhana,
  • K. Shivanna,
  • Mohamed Ghouse Shukur,
  • C. P. Vijay,
  • H. R. Mahalingegowda,
  • H. V. Nithin

摘要

In recent years, the widespread adoption of Internet of Things (IoT) devices in modern life has been significantly influenced by smart home automation, healthcare, wearable devices, and smart cities have significantly influenced the use of IoT devices in today’s digital era. This has led to a growing concern for data privacy, and the implementation of robust security measures remains a critical challenge. To address these challenges, we propose a novel feature amalgamation technique for mitigating these risks. The proposed method enhances the performance of the kervolutional neural network (KNN) model by detecting various threats in the IoT network and our key contribution lies in integrating the delta and delta-delta features with the input features of the UNSW-NB15 dataset at different levels, achieving a feature fusion that captures both temporal and spatial information. The experimental results demonstrate the superiority of our proposed approach on the UNSW-NB15 dataset, with a decision-level feature detection rate of 99.2% for binary attacks and 96.1% for multiple attacks, thus outperforming the state-of-the-art methods.