期刊
IEEE INTERNET OF THINGS JOURNAL
卷 10, 期 4, 页码 3071-3078出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JIOT.2021.3097951
关键词
Security; Protocols; Reliability; Computational modeling; Computer architecture; Internet of Things; Genetics; Big data; deep learning (DL); genetic particle swarm optimization (GPSO); Internet of Everything (IoE); security; XGBoost
This article proposes a machine learning-based architecture to identify malicious and benign nodes in an IoE network. Through simulations, it is shown that GPSO-based learning algorithms provide reliable, robust, and scalable solutions, and significantly outperform other security protocols in the classification of nodes.
The exponential growth of the Internet of Everything (IoE), in recent times, has revealed many underlying security vulnerabilities of the nodes forming IoE networks. The extension of conventional security protocol to these devices has been greatly complicated by the prevalence of restricted computational hardware and limited battery life. Modern learning-based algorithms have shown the potential to secure the IoE networks without undue duress on the nodes' limited capabilities. In this article, a machine learning-based architecture has been proposed to identify malicious and benign nodes in an IoE network operating with big data. A novel approach for the cooperation of XGBoost and deep learning models along with a genetic particle swarm optimization (GPSO) algorithm to discover the optimal architectures of individual machine learning models has been proposed. Through simulations, it is shown that GPSO-based learning algorithms provide reliable, robust, and scalable solutions. The proposed model significantly outperforms other security protocols in the classification of malicious and benign nodes forming an IoE network.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据