Trust Computation Model for IoT devices using Soft Computing and Machine Learning Techniques for Health Care Applications

Abstract

The calculation of trust for online web services is of utmost importance due to the security risks associated with network devices. various cloud platforms offer diverse services for communicating with various IoT devices and generating predictions based on the collected data. Occasionally, data may be compromised by either internal or external groups through low-impact hacking techniques. Investigating assaults and ensuring the elimination of connections can be a laborious task in a sensitive setting. Despite some progress, there are still many unresolved security issues that provide barriers to addressing security problems. This article seeks to conduct a comprehensive examination of existing security solutions for the lowest layer of the Internet of Things (IoT) and promote the development of more inventive IoT designs for those who invest in them. This paper proposes a system for calculating trust in IoT devices using both supervised and unsupervised machine learning algorithms for non-invasive health care applications. Both strategies were assessed based on trust computation weights and compared with their respective machine learning algorithms. The K-means clustering technique is employed for one strategy, while the SVM algorithm is used for the other approach. The aim of this successful implementation is to identify and eradicate a minor threat that occurs during the trusted execution process. An experimental analysis was conducted using NS2 2.35 in combination with an open-source Java platform. At first, a total of 100 nodes are created in the simulation environment. These nodes are then assessed using a clustering method to determine if they are trustworthy or untrustworthy, resulting in the generation of labels. The experimental research demonstrated the superior efficacy of attack detection in comparison to traditional methods of attack detection.