A Deep Learning-Based Intrusion Detection Technique for a Secured IoMT System

• Published in: Informatics and Intelligent Applications Vol. 1547; pp. 50 - 62
• Main Authors: Awotunde, Joseph Bamidele, Abiodun, Kazeem Moses, Adeniyi, Emmanuel Abidemi, Folorunso, Sakinat Oluwabukonla, Jimoh, Rasheed Gbenga
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2022; Springer International Publishing
• Series: Communications in Computer and Information Science
• Subjects: Healthcare data and information; Internet of Medical of Things; Intrusion detection; Machine learning; Networks vulnerability; Security and privacy
• ISBN: 3030956296; 9783030956295
• ISSN: 1865-0929; 1865-0937
• DOI: 10.1007/978-3-030-95630-1_4

The emergence of medical sensors in smart healthcare has brought about an intelligent Internet of Medical Things (IoMT) system for detecting life-threatening ill-ness globally. The IoMT-based system has been used to generate a huge amount of data that experts can use for various purposes like diagnosis, prediction, and real-time monitoring of patients. However, patients’ health data must be transferred to cloud database storage and external computing devices for processing due to the limited storage capability and calculation capability of IoMT-based devices. This can result in security and privacy issues due to a lack of control over the patient’s health information and the network’s vulnerability to numerous forms of assaults. Therefore, this paper proposes a swarm-neural net-work-based model to detect intruders in the data-centric IoMT-based system. The proposed model can be used to detect intruders during data transfer, allowing for efficient and accurate analysis of healthcare data at the network’s edge. The performance of the system was tested using a real-time NF-ToN-IoT dataset for IoT applications that collected telemetry, operating systems, and network data. The results of the proposed model are compared over the standard intrusion detection classification models that use the same dataset using various performance metrics. The experimental results reveal that the proposed model attains 89.0% accuracy over the ToN-IoT dataset.