Secure IoT Healthcare Architecture with Deep Learning-Based Access Control System

• Published in: Journal of nanomaterials Vol. 2022; no. 1
• Main Authors: Thilagam, K., Beno, A., Lakshmi, M. Vanitha, Wilfred, C. Bazil, George, Santhi M., Karthikeyan, M., Peroumal, Vijayakumar, Ramesh, C., Karunakaran, Prabakaran
• Format: Journal Article
• Language: English
• Published: New York Hindawi 2022; John Wiley & Sons, Inc
• Subjects: Access control; Algorithms; Artificial intelligence; Artificial neural networks; Big Data; Cloud computing; Data analysis; Data augmentation; Data integrity; Data processing; Data storage; Deep learning; Health care; Industrial applications; Internet of Things; Interoperability; Leakage; Machine learning; Medical equipment; Medical research; Modules; Neural networks; Privacy; Radio frequency identification; Remote monitoring; Social networks; Wearable technology; Wireless networks
• DOI: 10.1155/2022/2638613
• ISSN: 1687-4110

The existing healthcare system based on traditional management involves the storage and processing of large quantities of medical data. The incorporation of the Internet of Things (IoT) and its gradual maturation has led to the evolution of IoT-enabled healthcare with extraordinary data processing capability and massive data storage. Due to the advancement in the Industrial Internet of Things (IIoT), the resulting system is aimed at building an intelligent healthcare system that can monitor the medical health of the patient by means of a wearable device that is monitored remotely. The data that is gathered by the wearable IoT module is stored in the cloud server which is subject to privacy leakage and attacks by unauthorized users and attackers. To address this security issue, an IoT-based deep learning-based privacy preservation and data analytics system is proposed in this work. Data is collected from the user, and the sensitive information is segregated and separated. Using a convolutional neural network (CNN), the health-related information is analyzed in the cloud, devoid of users’ privacy information. Thus, a secure access control module is introduced that works based on the user attributes for the IoT-Healthcare system. A relationship between the users’ trust and attributes is discovered using the proposed work. The precision, recall, and F1 score of the proposed CNN classifier are achieved at 95%. With the increase in the size of the training set, higher performance is attained. When data augmentation is added, the system performs better without data augmentation. Further, the accuracy of around 98% is achieved with an increased user count. Experimental analysis indicates the robustness and effectiveness of the proposed system with respect to low privacy leakage and high data integrity.