Show simple item record

AuthorZubair, Mohammad
AuthorUnal, Devrim
AuthorAl-Ali, Abdulla
AuthorReimann, Thomas
AuthorAlinier, Guillaume
Available date2022-03-24T05:34:01Z
Publication Date2021
Publication NameJournal of Emergency Medicine, Trauma and Acute Care
Resourceqscience
CitationZubair M, Unal D, Al-Ali A, Reimann T, Alinier G. Cybersecurity for next generation healthcare in Qatar, Journal of Emergency Medicine, Trauma & Acute Care 2021:41 http://dx.doi.org/jemtac.2021.qhc.41
ISSN1999-7086
ISSN1999-7094
URIhttps://doi.org/10.5339/jemtac.2021.qhc.41
URIhttp://hdl.handle.net/10576/28978
AbstractBackground: IoMT (Internet of Medical Things) devices (often referred to IoMT domain) have the potential to quickly diagnose and monitor patients outside the hospital by transmitting information through the cloud domain using wireless communication to remotely located medical professionals (user domain). Figure 1 shows the proposed IoMT framework designed to improve the privacy and security of the healthcare infrastructure. Methods: The framework consists of four modules: 1. Intrusion Detection System (IDS) 1 using deep learning (DL) to identify bluetooth-based Denial-of-Service (DoS)-attacks on IoMT devices and is deployed on edge-computing to secure communication between IoMT and edge. 2. IDS 1 is backed up with identity-based cryptography to encrypt the data and communication path. 3. Besides the identity-management system (to authenticate users), it is modeled with aliveness detection using face authentication techniques at the edge to guarantee the confidentiality, integrity, and availability (CIA) of the framework. 4. At the cloud level, another IDS 2 using MUSE (Merged-Hierarchical-Deep-Learning-System-with-Layer-Reuse) is proposed to protect the system against Man-In-The-Middle attacks, while the data is transferred between IoMT-EDGE-CLOUD. Results: These four modules are developed independently by precisely analyzing dependencies. The performance of IDS 3 in terms of precision is 99% and for the identity-management system, the time required to encrypt and decrypt 256-bit key is 66 milliseconds and 220 milliseconds respectively. The true positive rate is 90.1%, which suggests real-time detection and authentication rate. IDS (2) using MUSE (12-layer) the accuracy is >95%, and it consumes 15.7% to 27.63% less time to train than the smaller four-layer model. Conclusion: Our designed models suit edge devices and cloud-based cybersecurity systems and support the fast diagnosis and care required by critically ill patients in the community.
Languageen
PublisherHamad bin Khalifa University Press (HBKU Press)
Subjectdata security
deep learning
intrusion detection system
user authentication
internet of medical things
TitleCybersecurity for next generation healthcare in Qatar
TypeArticle
Issue Number2
Volume Number2021
dc.accessType Open Access


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record