Machine Learning and Deep Learning Methods for Intrusion Detection Systems in IoMT: A survey
Yahya Rbah, Mohammed Mahfoudi, Younes Balboul, Mohammed Y. Fattah, Saïd Mazer, Moulhime Elbekkali, Benaissa Bernoussi
Abstract
The integration of healthcare-related sensors and devices into IoT has resulted in the evolution of the IoMT (Internet of Medical Things). IoMT that can be viewed as an improvement and investment in order to meet patients' needs more efficiently and effectively. It is progressively replacing traditional healthcare systems, particularly after the worldwide impact of COVID. IoMT devices have enabled real time monitoring in the healthcare field, allowing physicians to provide superior care while also keeping patients safe. As IoMT applications have evolved, the variety and volume of security threats and attacks including routing attacks and DoS (Denial of Service), for these systems have increased, necessitating specific efforts to study intrusion detection systems (IDSs) for IoMT systems. However, IDSs are generally too resource intensive to be managed by small IoMT devices, due to their limited processing resources and energy. In this regard, machine learning and deep learning approaches are the most suitable detection and control techniques for IoMT device-generated attacks. The purpose of this research is to present various methods for detecting attacks in the IoMT system. Furthermore, we review, compare, and analyze different machine learning (ML) and deep learning (DL) based mechanisms proposed to prevent and detect IoMT network attacks, emphasizing the proposed methods, performances, and limitations. Based on a comprehensive analysis of current defensive security measures, this work identifies potential open research related challenges and orientations for the actual design of those systems for IoMT networks, that may guide further research in this field.