An Efficient Method for Online Detection of DRDoS Attacks on UDP-Based Services in SDN Using Machine Learning Algorithms

Abstract

With advances in mobile devices and systems and the emergence of new ideas such as cloud computing and big data, as well as the tremendous growth in the number of network users, the need to modify the current network architectures has been very much in the foreground in recent years. One of the promising solutions to overcome these challenges is software-defined networking (SDN). SDN is a unique innovative architecture in which network control and traffic flows are independent of each other and planned directly. The SDN’s focused view of networks is more comprehensive than other methods, which is why SDN is more efficient in coping with malicious attacks including amplification attacks. The response to amplification of distributed denial of service (DDoS) attacks is larger than the request. In an amplification attack, the attacker fakes the victim’s address as the source address and the responses are forwarded to the victim instead of the attacker. This is why these attacks are more difficult to discover in traditional networks, while the focused method of SDN can contribute to the detection of such attacks. There are different methods for detecting these attacks, one of which is to use machine learning (ML) algorithms. In line with this, the present paper is aimed at the detection of distributed reflection denial of service (DRDoS) attacks using ML algorithms. Simulation was performed by the use of ML algorithms, and the findings suggest a significant improvement in the detection of DRDoS attacks in comparison with previous methods.