Hybrid Machine Learning Algorithm to Detect DDoS Attacks in Software Defined Networks

Abstract

The distributed denial of service (DDoS) attack can be considered one of the most dangerous and costly digital threats on computer networks. The network is monitored by software-defined networking (SDN) and emerging technology that provides a network architecture to separate the control plane from the data plane. This research presents a proposed solution aimed at addressing this type of attack by effectively monitoring and detecting the targeted traffic. The detection of DDoS attacks is performed by hybridizing two machine learning algorithms, support vector machine, and K-nearest neighbors call (HKSVM). This research was applied on InSDN and SDN datasets with 18 features. The experiments showed that the average accuracy rate (ACC) of the HKSVM is approximately 99% with a small amount of total flux. The results show that the HKSVM was an effective method for detecting DDoS in SDN as performed the best.