Scholarlink

<p>The Internet of Things (IoT) ecosystem increasingly relies on the Message Queuing and Telemetry Transport (MQTT) protocol due to its lightweight and efficient communication capabilities. However, MQTT-enabled networks are vulnerable to various security threats, including eavesdropping, weak authentication, and malicious payload injections. Detecting such intrusions is particularly challenging in IoT environments because of constrained resources, high traffic volumes, and heterogeneous attack vectors. To address these challenges, this study proposes an ensemble learning-based Intrusion Detection System (IDS) specifically designed for MQTT-enabled IoT networks. The ensemble integrates Gaussian Naive Bayes (GNB), Kernel-based Support Vector Machines (K-SVM), and Multi-Layer Perceptrons (MLP) as base classifiers, with Logistic Regression serving as the meta-classifier. Among these, MLP plays a pivotal role by leveraging its deep learning capabilities and parallel processing strengths to capture complex, nonlinear attack patterns that are often missed by conventional models. The combination of diverse classifiers in a stacked architecture significantly enhances the system’s precision, scalability, and adaptability. Experimental evaluation of binary and multi-class classification tasks using the MQTT-IoT-IDS2020 dataset demonstrate that the proposed model achieves outstanding detection accuracies of 99.80% and 99.59%, respectively, outperforming traditional machine learning and standalone deep learning approaches. These results highlighted the effectiveness of the ensemble framework, particularly the critical contribution of MLP, in building a robust, high-performance, and scalable intrusion detection solution for secure IoT communications.</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

A novel intrusion detection framework using ensemble learning in MQTT IoT applications

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • Mitesh Solanki,
  • Shilpi Gupta

摘要

The Internet of Things (IoT) ecosystem increasingly relies on the Message Queuing and Telemetry Transport (MQTT) protocol due to its lightweight and efficient communication capabilities. However, MQTT-enabled networks are vulnerable to various security threats, including eavesdropping, weak authentication, and malicious payload injections. Detecting such intrusions is particularly challenging in IoT environments because of constrained resources, high traffic volumes, and heterogeneous attack vectors. To address these challenges, this study proposes an ensemble learning-based Intrusion Detection System (IDS) specifically designed for MQTT-enabled IoT networks. The ensemble integrates Gaussian Naive Bayes (GNB), Kernel-based Support Vector Machines (K-SVM), and Multi-Layer Perceptrons (MLP) as base classifiers, with Logistic Regression serving as the meta-classifier. Among these, MLP plays a pivotal role by leveraging its deep learning capabilities and parallel processing strengths to capture complex, nonlinear attack patterns that are often missed by conventional models. The combination of diverse classifiers in a stacked architecture significantly enhances the system’s precision, scalability, and adaptability. Experimental evaluation of binary and multi-class classification tasks using the MQTT-IoT-IDS2020 dataset demonstrate that the proposed model achieves outstanding detection accuracies of 99.80% and 99.59%, respectively, outperforming traditional machine learning and standalone deep learning approaches. These results highlighted the effectiveness of the ensemble framework, particularly the critical contribution of MLP, in building a robust, high-performance, and scalable intrusion detection solution for secure IoT communications.