Lightweight Intrusion Detection System Using a Hybrid CNN and ConvNeXt-Tiny Model for Internet of Things Networks
Abstract
The rapid expansion of Internet of Things (IoT) systems across various domains such as industry, smart cities, healthcare, manufacturing, and government services has led to a significant increase in security risks, threatening data integrity, confidentiality, and availability. Consequently, ensuring the security and resilience of IoT systems has become a critical requirement. In this paper, we propose a lightweight and efficient intrusion detection system (IDS) for IoT environments, leveraging a hybrid model of CNN and ConvNeXt-Tiny. The proposed method is designed to detect and classify different types of network attacks, particularly botnet and malicious traffic, while the lightweight ConvNeXt-Tiny architecture enables effective deployment in resource-constrained devices and networks. A real-world dataset comprising both benign and malicious network packets collected from practical IoT scenarios was employed in the experiments. The results demonstrate that the proposed method achieves high accuracy while significantly reducing training and inference time compared to more complex models. Specifically, the system attained 99.63% accuracy in the testing phase, 99.67% accuracy in the training phase, and an error rate of 0.0107 across eight classes, while maintaining short response times and low resource consumption. These findings highlight the effectiveness of the proposed method in detecting and classifying attacks in real-world IoT environments, indicating that the lightweight architecture can serve as a practical alternative to complex and resource-intensive approaches in IoT network security.