Towards a Hybrid Deep-Learning SDN-Based Intelligent Attack Detection System for the IoMT

Abstract

The Internet of Medical Things (IoMT) has emerged from integrating medical devices into the Internet of Things (IoT), transforming various healthcare applications, including real-time monitoring and remote patient care. This paper proposes a novel hybrid deep learning framework for intrusion detection within the IoMT environment. The framework leverages the strengths of Long Short-Term Memory (LSTM) networks for sequential data processing and an attention layer to capture both long- and short-term dependencies within the data. This approach is embedded within a Software-Defined Network (SDN) architecture to enhance the efficiency of intrusion detection. The proposed method achieves high accuracy (99.99%) and rapid processing times (<1.85 seconds) on the "IoT-Healthcare security" dataset, demonstrating its effectiveness against prevalent threats. Comparative analysis with benchmark models showcases superior performance in terms of both accuracy and computational complexity.