DDoS Attack Detection and Mitigation in SDN Environment: A Deep Learning Perspective
Monika Dandotiya, Rajni Ranjan Singh Makwana
Abstract
Recent developments in security methods have greatly improved our capacity to detect and neutralize any kind of attack or threat in any network infrastructure, including an SDN, and to defend the internet’s security architecture from any number of potential risks. Distributed Denial-of-Service (DDoS) attacks may happen on any kind of network, but ML and DL are two of the most common methods for protecting against them. In order to identify & counteract DDoS attacks in SDNs, this research offers a hybrid deep learning technique called the stacked LSTM+CNN model. The rising popularity of SDNs, however, makes them an attractive target for cyberattacks because to their central control and adaptability. Due to the complicated structure of network traffic in SDNs, detecting DoS/DDoS attacks is a difficult operation. We created a novel deep learning method by fusing together two distinct deep learning algorithms to solve this issue. On the CICDoS2019 dataset, we attained an accuracy of 99.76 percent using our method. Our work significantly advances state-of-the-art (SOTA) in SDN-based network security. In theory, proposed approach might improve SDN security and protect against DoS/DDoS attacks. This is significant since SDNs are rapidly becoming an integral part of modern network infrastructure, and as such, they must be safeguarded against attacks in order to ensure continued availability & reliability of those resources. Overall, results of this study prove that the proposed method for detecting DoS/DDoS attacks in SDNs works, also point way toward promising directions for further investigation in the area.