SynFlood DDoS Attack Detection with SVM Kernels using Uncorrelated Feature Subsets Selected by Pearson, Spearman and Kendall Correlation Methods
Kishore Babu Dasari, Nagaraju Devarakonda · 2022 Second International Conference on Computer Science, Engineering and Applications (ICCSEA) · 2022
Data availability and protection are getting more important nowadays due to the exponential growth of digital usage around the world. In cyber attacks Distributed Denial of Service (DDoS) is a threat aimed at intercepting admissible clients from penetrating server system stocks by flooding the server with huge amounts of network traffic. This study uses Support Vector Machine (SVM) classification algorithms Linear, RBF, Poly, and Sigmoid kernel function with distinctly Pearson, Spearman, and Kendall uncorrelated feature subsets to evaluate Syn-flood DDoS attack detection. This study used the Syn-flood dataset collected from CIC-DDoS2019 evaluation datasets. On the Syn-flood DDoS attack detection, the results show that Poly and RBF kernels among the VM kernel functions give the best classification results, while Pearson uncorrelated feature subset produces the best results among uncorrelated feature subsets.