A Hybrid Deep Learning Approach for Analyzing and Detecting the Malware in Software Defined Networks

Abstract

The rise of software-defined networking (SDN) has introduced new security challenges, particularly in detecting and mitigating malware threats within network infrastructures. Traditional malware detection techniques often struggle with the dynamic nature of modern cyber threats. This paper presents a hybrid deep learning-based approach for malware detection in SDN environments, leveraging Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Multi-Layer Perceptron (MLP). The proposed CNN-LSTM-MLP model integrates spatial, temporal, and fully connected feature extraction techniques to enhance classification accuracy. The study evaluates multiple LSTM architectures, including Bi-Directional-LSTM, Stacked-LSTM, and LSTM-MLP, demonstrating that the CNN-LSTM-MLP model achieves superior performance. The experimental results, conducted using datasets from the Canadian Institute for Cybersecurity, indicate that our model attains an accuracy of 98%, outperforming existing deep learning-based approaches. Additionally, the study integrates RYU and POX SDN controllers to simulate real-world network environments, ensuring practical applicability. The findings highlight the efficacy of hybrid deep learning models in securing SDN architectures against evolving malware threats.