Image-based Feature Representation for Enhanced Cyber-Attack Detection in IoHT Environments
Abdoualwase M. Al-Azzani, Ahmed Abdullah Al-Shalabi, Mossa Ghurab, Firdaus Alhrazi, Sharaf A. Alhomdy, Malek Nasser Ali Algabri · 2024
The complexity of cybersecurity risks has expanded dramatically with the fast expansion of the Internet of Health Things (IoHT), especially with regard to cyberattacks that target healthcare systems. In healthcare settings, data privacy and integrity are very important, hence developing improved detection techniques that can reliably identify fraudulent activity is urgently needed. In order to improve cyber-attack detection in IoHT, this research presents an image-based feature representation technique that makes use of convolutional neural networks (CNNs), notably the MobileNet architecture. With the use of the ECU-IoHT dataset’s network traffic statistics, the suggested model is able to catch intricate patterns that point to infiltration. Because MobileNet-CNN computation is efficient, it can be deployed in real-time in low-resource situations, which are typical of IoHT installations. Our method outperforms current state-of-the-art methods, with lower false positive and negative rates and better classification accuracy of 99.1%, precision of 99.00%, recall of 97.1%, and F1-score of 98.4%. The findings highlight how image-based deep learning models may improve cybersecurity safeguards in Internet of Things frameworks by offering a powerful instrument for proactive threat identification and removal. This study lays the groundwork for future studies that will use image-based representations of features for real-time threat identification, furthering the area of IoHT cybersecurity.