An Intelligent and Explainable SaaS-Based Intrusion Detection System for Resource-Constrained IoMT
Ahamed Aljuhani, Abdulelah Alamri, Prabhat Kumar, Alireza Jolfaei
Abstract
The Internet of Medical Things (IoMT) has revolutionized healthcare, but its vulnerabilities demand robust security solutions, especially for resource-constrained devices. In this research, we introduce an innovative Software as a Service (SaaS)-based Intrusion Detection System (IDS) designed specifically for the unique challenges of IoMT, deploying at the edge for enhanced efficiency. Our proposed IDS incorporates a multi-faceted approach: Firstly, it leverages the Particle Swarm Optimization (PSO) algorithm for feature engineering, optimizing data representation to reduce computational overhead on resource-constrained devices. Secondly, a diverse ensemble of machine learning and deep learning models is employed to detect a wide array of intrusion attempts within IoMT networks. Thirdly, interpretation is achieved using SHapley Additive exPlanations (SHAP), providing transparency and understanding of the decision-making process. By combining intelligence, efficiency, explainability, and deploying as a SaaS solution at the network edge, our IDS not only bolsters the security of resource-constrained IoMT devices but also empowers healthcare professionals with actionable insights, ensuring patient data privacy and network integrity in this dynamic and critical domain. Finally, the results using a publicly available healthcare dataset namely WUSTL-EHMS-2020 proves the effectiveness of the proposed IDS over some recent state-of-the-art works.