Abstract

The proliferation of Internet of Things (IoT) devices has created a vast and vulnerable attack surface, making intrusion detection a critical component of IoT security. Traditional intrusion detection systems (IDSs) often struggle with the complexity and dynamism of IoT network traffic. This paper proposes a novel Adaptive Hybrid Deep Learning Framework (AHDL-IDF) for enhanced intrusion detection in IoT networks. Our framework integrates Convolutional Neural Networks (CNNs) for effective feature extraction from network traffic data and Recurrent Neural Networks (RNNs) with attention mechanisms to capture temporal dependencies and prioritize relevant features for improved accuracy. The adaptive nature of the framework allows it to dynamically adjust its parameters based on the characteristics of the incoming traffic, enhancing its resilience to evolving attack patterns. We evaluate the performance of the AHDL-IDF using a publicly available IoT network traffic dataset and compare it against existing state-of-the-art IDS models. The experimental results demonstrate that the AHDL-IDF achieves significantly higher detection accuracy, lower false positive rates, and improved adaptability compared to existing approaches, making it a promising solution for securing IoT networks.