Feature Ranking using Statistical Techniques for Computer Networks Intrusion Detection

Nowadays with the enlargement of the utility of computer networks, the danger of getting our personal information being misused has also grown rapidly. According to the [1] 2022 Cyber Threat Report released by SonicWall, the health care industry faced a 755% increase in the attacks worldwide and Governments witnessed a 1,885% hike in attacks in 2021. Moreover, the threats due to intrusions are becoming more and more complex and difficult to detect. This research study intends to observe and inspect the network traffic of normal as well as intrusions and to perceive the network traffic features that can distinguish normal traffic from intrusions traffic. In order to extract distinguishing features, the proposed research study utilizes the statistical techniques of ANOVA and the Chi-Square test to order and line up the network traffic features. The experimental results demonstrate that the feature Bytes Sent is ranked at the top when the ANOVA test is applied over the data. In the cases where Chi-Square test is applied over normal and malware data, the features bytes sent and bytes received are ranked at the top respectively. Such ranking can help build an effective intrusion detection system.