Adaptive classifier-based intrusion detection system using logistic regression and Euclidean distance on network probe vectors in resource constrained networks. (8th November 2021)

• Record Type: Journal Article
• Authors: Saha, Rahul; Kumar, Gulshan; Rai, Mritunjay Kumar; Kim, Hye-jin
• Is Part Of: International journal of information and computer security. Volume 16:Number 3/4(2021)
• Page Start: 226
• Page End: 238
• Publication Date: 2021-11-08
• Subjects: intrusion -- outliers -- learning -- profile -- classification -- Euclidean -- threshold; Computer security -- Periodicals; Information systems management -- Security measures -- Periodicals; Computer networks -- Security measures -- Periodicals; Information technology -- Security measures -- Periodicals; 005.805
• Languages: English
• ISSNs: 1744-1765

Abstract

Intrusion detection system is a second layer of security in network security paradigm. With the progressing wireless technologies, the malicious activities are also increased with a rapid pace. But to secure the data communication in such environment, we need to have intrusion detection mechanism in use. Several mechanisms are introduced for the intrusion detection purpose. These existing algorithms are also capable of incorporating adaptive features but lack in the complexity and usability issues. Moreover, the real time adaptive learning is a missing link in these algorithms. In this paper, we have proposed a model of intrusion detection that deals with the learning mechanism on network probe data and identifies the intrusion by detecting the outliers with logistic regression. We have used Euclidean distance for outlier detection. The results show that our model is less complex in terms of time consumption and efficiently detects the intrusions.