Enhancing Security in Software-Defined Networks: Hybrid Deep Learning Models for Flooding Attack Detection

Software-Defined Networking (SDN) enhances flex-ibility and centralizes network traffic management, but it also introduces new security challenges, particularly from flooding attacks such as TCP and UDP floods. These attacks can over-whelm network resources, leading to performance degradation or service disruption. This research presents a novel hybrid deep learning approach for detecting flooding threats in SDN utilizing an imbalanced dataset. Balancing is crucial in imbalanced datasets as it reduces bias towards the majority class, allowing the model to reliably detect attacks from minority classes. We have addressed class imbalance through the application of inverse proportional sampling. Our model employs a hybrid methodology combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to proficiently analyze network traffic and identify malicious flooding activities. The proposed model adeptly tackles the temporal dynamics of flooding attacks by utilizing CNN for feature extraction and LSTM for sequence modeling. This research presents real-time threat detection, enhancing SDN security. Our research indicates that the hybrid deep learning model accurately detects flooding-based denial of service attacks and generalizes effectively across various flooding scenarios, positioning it as a promising solution for the defense of SDN networks.