Adaptive deep convolutional neural network‐based secure integration of fog to cloud supported Internet of Things for health monitoring system

Abstract

AbstractIn recent years, the healthcare monitoring system plays a significant role in providing early intervention for the people who are under risk. Several advanced technologies including the Internet of Things (IoT) are becoming accessible nowadays due to its high level and ubiquitous monitoring. The IoT also enables a structured and competent technique in handling the healthcare of the patients based on remote patient monitoring and mobile health. In addition to this, the deep learning approaches are employed in health‐based applications so as to achieve promising and satisfactory performances for a sizeable amount of data. While monitoring the health condition of the patients, there occur delays in data transferring to the cloud. So, to overcome such shortcomings, the proposed approach of this paper utilizes four different phases including data acquisition phase (DAP), fog to cloud phase (FCP), decision‐making phase (DMP), and execution phase (EP) in transferring the data to the cloud via the fog layer. The initial phase or DAP phase comprises of data storage and collection. The second phase or the FCP phase comprises of two different types of layers namely fog layer and the cloud layer along with that, secure integration of FCP is described. The third phase or the DMP phase involves feture extraction and classification. Here, Adaptive Deep Convolution Neural Network along with the Levy Flightbased Grey Wolf Optimization (LFGWO) are employed in the classification process to obtain the best optimal solution. The final phase or the EP phase provides notification or information to the doctor or the practitioner soon after the detection of any abnormality case. The experimental results are made by comparing various approaches with the proposed approach and the analysis reveals that the proposed approach provides better results with high accuracy, efficiency, response time with less computational cost in healthcare monitoring applications.