TRUST‐based features for detecting the intruders in the Internet of Things network using deep learning. (14th July 2021)

• Record Type: Journal Article
• Authors: Bhor, Harsh Namdev; Kalla, Mukesh
• Is Part Of: Computational intelligence. Volume 38:Number 2(2022)
• Page Start: 438
• Page End: 462
• Publication Date: 2021-07-14
• Subjects: attacks -- deep belief network -- intrusion detection -- IoT -- KDD features; Artificial intelligence -- Periodicals; Computational linguistics -- Periodicals; 006.3
• DOI: 10.1111/coin.12473
• Languages: English
• ISSNs: 0824-7935

Abstract

Abstract: Internet of Things (IoT) is a trending domain and has acquired much interest for various kinds of civilian applications. The purpose of IoT is to make objects accessible and interconnected via internet. Hence, security to IoT devices is a major issue because devices connected to the IoT network are resource‐constrained. In IoT, the nodes exchange information using insecure internet, which makes the network exposed to different attacks. This article proposes a new intrusion detection strategy, namely, Taylor‐spider monkey optimization‐based deep belief network (Taylor‐SMO‐based DBN). The KDD features and the trust factors are employed for intrusion detection. The KDD features are subjected to the classification, which is progressed using a newly devised optimization algorithm, namely, Taylor‐spider monkey optimization (Taylor‐SMO)‐based DBN. The proposed Taylor‐SMO algorithm is designed by integrating the Taylor series and spider monkey optimization (SMO) algorithm and is employed to train the deep belief network (DBN) to achieve accurate intrusion detection. The proposed Taylor‐SMO‐based DBN outperformed other methods with maximal accuracy of 90%, false alarm rate of10%, precision of 90%, and recall of 92%, respectively.