Investigation on Efficient Machine Learning Algorithm for DDoS Attack Detection

Abstract

Internet of Things (IOT) is a general term for all interconnected devices as well as the technology that enables object-to-object and cloud-to-object communication. However, there are several regular and dangerous threats to the development of this technology. The Distributed DoS (DDOS) attacks are extremely innovative and complex, making them almost inevitable to detect by the existing technology or detection system. Due to their complexity and difficulty, novel types of DDoS attacks are practically impossible for intrusion detection systems to detect or mitigate. Effective DDoS traffic detection is made feasible by Machine Learning (ML) technologies. In this paper, the popular ML methods were tested on the CICDoS2019 dataset to determine the most effective one for DDoS detection. A hybrid MLDDoS detection approach using estimator functions is also proposed. The framework for multi-classifying different DDoS attack types can be improved in future research, and a hybrid algorithm can be tested using updated datasets for DDoS attacks.