TIJER 10(5), pp. 466-472.
ISSN/ISBN: 2349-9249 DOI: Not available at this time.
Abstract: Zero-day attacks are sophisticated harms that take advantage of undisclosed weaknesses in software, making them exceedingly difficult to detect and prevent. Recently, attackers have been anxiously anticipating the discovery of previously unknown vulnerabilities that have not yet been patched or defended against. It is critical to detect and respond to these assaults in a timely way in order to prevent data breaches and secure sensitive information. By analysing network data and identifying aberrant activity, this research provides a machine learning-based solution to detecting zero-day attacks. The suggested system analyses network traffic with machine learning techniques to detect irregularities that might signal a zero-day assault. The NTA is critical for the network intrusion detection system (NIDS) since it monitors and extracts important data from network traffic data. The data is made up of several sorts of attributes that describe network packets, but not all of them are suitable for NIDS. It is critical to choose just those characteristics that have a substantial influence on our system. So, in order to identify the needed features, we apply Benford's Law to the numerical components of the data, such as IP addresses or packet sizes, and Zipf's Law to the non-numerical components, such as protocol headers or payload content. Finally, using ideally chosen features, we employ a semi-supervised ML technique that is successful for identifying zero-day attacks. The system is intended to be adaptable and scalable, with the ability to handle vast volumes of data and react to new attack patterns that keep developing. This project's ultimate purpose is to improve computer system security by identifying and blocking zero-day threats before they do harm.
Bibtex:
@article{,
author = {Sivasankari K. and Adira Nair and Alika Rizvi and Anupama Mahata},
title = {Modern Adaptable Zero Day Attack Detection in Network Traffic: Using Feature Identification and Tree Based Classifiers},
year = {2023},
journal = {TIJER},
volume = {10},
number = {5},
pages = {466--472},
url = {https://www.tijer.org/viewpaperforall?paper=TIJER2305046},
}
Reference Type: Journal Article
Subject Area(s): Computer Science