Machine Learning Based DoS Traffic Analysis on the Testbed Environment

Abstract

Today, malicious users are widespread and are frequently lengthening worldwide. So, network security becomes crucial in the domain of education, government, business, and other sectors with related network connections. The firewall filtering rules itself might cause network vulnerability due to the misconfiguration and order them. The system builds a network testbed using a firewall, and Intrusion Detection System (IDS) and then implements a dataset using DoS traffic and normal traffic from that testbed environment. It is needed to be tested various requirements as features, false positive rates, and accuracy based on datasets apply and built for DoS. The importance of features in the proposed dataset was tested using attribute evaluators and methods. The focus of this work is to improve the performance with two classifiers as Logistic Regression and Support Vector Machine. The system also selects the important features by classifying traffics according to times by machine learning methods.