Machine learning models to detect the blackhole attack in wireless adhoc network. (2021)

• Record Type: Journal Article
• Authors: Nagalakshmi, T.J.; Gnanasekar, A.K.; Ramkumar, G.; Sabarivani, A.
• Is Part Of: Materials today. Volume 47:Part 1(2021)
• Page Start: 235
• Page End: 239
• Publication Date: 2021
• Subjects: Wireless adhoc network -- Black hole attack -- Machine learning models -- Random forest method -- Principal component analysis -- Feature selection -- K-means cluster classifier -- SVM -- Decision tree; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2021.04.129
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: An Ad-hoc network is used to set up a wireless connection directly from one device to another computer device. This network is a decentralised network. And it does not have Wi-Fi access point or Router. But, the main drawback of Ad hoc mode is minimal security. And also, without any trouble an attacker can connect with the ad-hoc network. So an efficient intrusion detection system is needed for the digital world. So that, these intrusion detection systems can monitor the network operation and detect the attack. In current digital information world, with the help of machine learning algorithms the intrusion detection systems were built. These IDS perform well and tries to attain better accuracy and speed. In this study six machine learning modelled IDSs were designed and analysed. The IDSs were designed with and without feature selection. The K-means cluster algorithm, SVM, Decision tree and Random forest classifiers are used to build the four IDSs without implementing feature selection. And Random forest and Principal component analysis are used as the feature selection techniques to design the two IDSs with K-means cluster classifier. For each intrusion detection system 19 samples were taken into account for the analysation. From this work it is understood that, in IDSs while implementing the feature selection technique the accuracy and detection rates are improved.