Dynamic Tracing of DoS Attack Over Software-Defined Networks Using Machine Learning
Amit Chopra, Dinesh Chander Verma · 2022
In this paper, a machine learning approach to detect the Denial-of-Service attack over software-defined networks (SDN) is introduced. It builds the training model using NS-3 traces to detect the Denial-of-Service (DoS) attack that can be launched by an intruder to block access to network resources. This work represents a machine learning approach to detect and prevent the SDN from DoS threat by examining the abnormalities in a given traffic pattern using classifiers. A machine learning model is developed based on the dynamic tracing of log files and traffic classification has been carried out to identify the DoS threat. A dynamic trace log is maintained to filter out the identified nodes from the SDN. Various classifiers viz K-Nearest Neighbors (KNN), Decision tree (DT), Logistic regression(LG), and Support Vector Machine (SVM) are used for traffic classification and their performance is also measured using various parameters Precision, Recall, F1-score, and Accuracy. SDN performance is also analyzed using different simulation scenarios with respect to node densities 50, 100, and 200.