Hybrid Deep Learning Based DDoS Detection System Using Software Defined Networking

Computer Science and Engineering

The Software-Defined Networking (SDN) paradigm exhibits substantial potential, as it enables the centralized controller to dynamically configure and manage networks. The SDN paradigm's programmable and open characteristics have led to the emergence of novel security challenges, such as Distributed Denial of Service (DDoS) attacks. The impact of DDoS attacks can impede network services, diminish network efficiency, and result in significant harm to the overall network infrastructure in the realm of SDN. The utilization of machine learning and deep learning algorithms has become a prevalent method for detecting DDoS attacks, owing to their efficacy in analyzing voluminous data sets and detecting patterns or anomalies that signify the presence of an ongoing attack. The aim of this study is to create a dataset that is customized for SDN by incorporating the acknowledged features into a CSV file. This study introduces a novel technique for detecting DDoS attacks: the Hybrid Deep Learning Model-Based DDoS Detection System (HDL3DS). This model employs Autoencoder and Deep Neural Network techniques to detect DoS attacks with reliability. HDL3DS's performance was assessed and compared to that of other DL models, such as MLP, LSTM, and DNN. In accordance with the experimental findings, the proposed HDL3DS outperformed other models, obtaining a 99.98% accuracy rate.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.