Kernel-based machine learning intrusion detection systems for ICMPv6 DDoS detection
Abeer Abdullah Alsadhan, Noora Al Roken, Sam Ansari, Bilal Khan, Sadiq H. Abdulhussain, Abir Jaafar Hussain · Results in Engineering · 2025
This study presents a novel intrusion detection framework that leverages flow-based representations of ICMPv6 traffic and kernel-based machine learning algorithms to enhance detection accuracy. The proposed system integrates three classifiers—support vector machine (SVM), random forest (RF), and logistic regression (LR)—with nine distinct kernel functions to construct a robust and adaptive detection architecture. Experimental evaluations, conducted on a purpose-built ICMPv6 flow dataset, reveal that the SVM model employing the radial basis function (RBF) kernel outperforms other configurations, achieving a detection accuracy of 92.67% and weighted precision and recall values of 93.00%. These findings underscore the efficacy of kernel-enhanced machine learning models in accurately identifying and mitigating ICMPv6-based DDoS attacks, and establish the proposed approach as a viable and high-performing solution for next-generation intrusion detection systems. • Utilized flow-based IPv6 packets for NIDS attack detection. • Applied various kernels on three ML classifiers. • Introduced a novel framework with enhanced performance.