A Deep Learning Approach for Multiclass Attack Classification in IoT and IIoT Networks Using Convolutional Neural Networks

The rapid expansion of the Internet of Things (IoT) and industrial Internet of Things (IIoT) ecosystems has introduced new security challenges, particularly the need for robust intrusion detection systems (IDSs) capable of adapting to increasingly sophisticated cyberattacks. In this study, we propose a novel intrusion detection approach based on convolutional neural networks (CNNs), designed to automatically extract spatial patterns from network traffic data. Leveraging the DNN-EdgeIIoT dataset, which includes a wide range of attack types and traffic scenarios, we conduct comprehensive experiments to compare the CNN-based model against traditional machine learning techniques, including decision trees, random forests, support vector machines, and K-nearest neighbors. Our approach consistently outperforms baseline models across multiple performance metrics—such as F1 score, precision, and recall—in both binary (benign vs. attack) and multiclass settings (6-class and 15-class classification). The CNN model achieves F1 scores of 1.00, 0.994, and 0.946, respectively, highlighting its strong generalization ability across diverse attack categories. These results demonstrate the effectiveness of deep-learning-based IDSs in enhancing the security posture of IoT and IIoT infrastructures, paving the way for intelligent, adaptive, and scalable threat detection systems.