Scholarlink

<p>The rapid development of the Internet of Things (IoT) has accelerated automation and smart applications, but also heightened vulnerability to cyberattacks due to device heterogeneity, resource limitations and decentralized architecture. Traditional Intrusion Detection Systems often struggle with accuracy and adaptability against evolving threats. To improve IoT security, this work proposes cyber-attack detection using Verifiable Convolutional Neural Network-based Blockchain Technology (CAD-IoT-VCNN-BCT). The framework begins by processing input traffic from the CIC IoT dataset, transforming and reducing data through Spatio-Temporal Principal Component Analysis. Higher-Level Target Navigation Pigeon-Inspired Optimization (HLTNPIO) is used for optimal feature selection. These features are secured via a Software-Defined Networking-based blockchain layer employing Fair Proof of Reputation for decentralized, immutable attack event verification. The secured data are analyzed by a Verifiable Convolutional Neural Network (VCNN) to classify seven attack types: Distributed Denial of Service, Denial of Service, Reconnaissance, Web-based, Brute Force, Spoofing, Mirai. Multiplayer Battle Game-Inspired Optimizer (MBGO) is introduced to fine-tune Verifiable Convolutional Neural Network parameters for better detection. The proposed technique achieves 21.41%, 21.46% and 22.11% higher mathew correlation coefficient; 24.67%, 24.64% and 25.66% higher identification rate compared with the existing techniques.</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

Cyber Attack Detection in Internet of Things using Verifiable Convolutional Neural Network based Blockchain technology

  • 期刊论文

  • 发表日期：

    2025年7月19日

查看全文
  • R. Deepa,
  • Jayaraj Velusamy,
  • N. K. Sakthivel,
  • S. Subasree

摘要

The rapid development of the Internet of Things (IoT) has accelerated automation and smart applications, but also heightened vulnerability to cyberattacks due to device heterogeneity, resource limitations and decentralized architecture. Traditional Intrusion Detection Systems often struggle with accuracy and adaptability against evolving threats. To improve IoT security, this work proposes cyber-attack detection using Verifiable Convolutional Neural Network-based Blockchain Technology (CAD-IoT-VCNN-BCT). The framework begins by processing input traffic from the CIC IoT dataset, transforming and reducing data through Spatio-Temporal Principal Component Analysis. Higher-Level Target Navigation Pigeon-Inspired Optimization (HLTNPIO) is used for optimal feature selection. These features are secured via a Software-Defined Networking-based blockchain layer employing Fair Proof of Reputation for decentralized, immutable attack event verification. The secured data are analyzed by a Verifiable Convolutional Neural Network (VCNN) to classify seven attack types: Distributed Denial of Service, Denial of Service, Reconnaissance, Web-based, Brute Force, Spoofing, Mirai. Multiplayer Battle Game-Inspired Optimizer (MBGO) is introduced to fine-tune Verifiable Convolutional Neural Network parameters for better detection. The proposed technique achieves 21.41%, 21.46% and 22.11% higher mathew correlation coefficient; 24.67%, 24.64% and 25.66% higher identification rate compared with the existing techniques.