Performance Improvement of DDoS Intrusion Detection Model Using Hybrid Deep Learning Method in the SDN Environment

Software-Defined Networking (SDN) is a networking paradigm that has the potential to revolutionize the way we develop and operate network infrastructures. SDN enables network engineers to quickly monitor networks, centrally manage networks, and quickly and accurately detect malicious traffic and link failures. In addition to its flexibility, SDN is also susceptible to various attacks like Distributed Denial of Service (DDoS) that can bring down the entire network. In recent years, Deep Learning (DL) approaches have been applied for reliable and highly accurate traffic anomaly detection. Therefore, this paper proposes a method based on DL to classify the benign traffic from the DDoS attack traffic. The major contribution of this paper is to propose a novel hybrid DL method to perform the classification. In addition, we compare different deep learning methods using attack severity, which is integrated into the standard metrics to differentiate the quality of the results of DL methods. The attack severity will be calculated using an appropriate weighting of undiscovered intrusions (FN and FP) discovered during the testing phase. Besides, various DDoS attack detection research projects employ public datasets that are not specific to the SDN environment. We propose a method to determine the adequacy of a selected dataset using two proposed metrics based on quality and quantity conformity evaluation.