Survey of Machine Learning Based Intrusion Detection Methods for Internet of Medical Things

The Internet of Medical Things (IoMT) allows the collection of physiological data using sensors, then their transmission to remote servers, permitting physicians and health professionals to analyze these data continuously and permanently. However, on the one hand, this technology faces security risks ranging from violating patient's privacy to their death due to wireless communication exposing these data to interception attacks. Moreover, these data are of particular interest to attackers due to their sensitive and private nature. On the other hand, adopting traditional security, such as cryptography on medical equipment suffering from low computing, storage and energy capacity with heterogeneous communication, represents a challenge. Moreover, these protection methods are ineffective against new attacks and zero-day attacks. Security measures must be adopted to guarantee the integrity, confidentiality and availability of data during collection, transmission, storage and processing. In this context, using Intrusion Detection Systems (IDS) based on Machine Learning (ML) can bring a complementary security solution adapted to the characteristics of IoMT systems. This paper performs a comprehensive survey on how IDS based on ML addresses security and privacy issues in IoMT systems. For this purpose, the generic three layers architecture of IoMT and the security requirement of IoMT systems are provided. Then, the various threats that can affect IoMT security and the advantages, disadvantages, methods, and datasets used in each solution based on ML are identified at the three layers composing IoMT. Finally, some challenges and limitations of applying IDS based on ML at each layer of IoMT are discussed, which can serve as a future research direction.