Cloud Network Traffic Classification and Intrusion Detection System Using Deep Learning
Kateřina Malá, H S Annapurna · 2023
With the increasing adoption of cloud computing, the security of network traffic and the detection of intrusions within cloud environments have become paramount concerns. This survey paper delves into the realm of cloud network traffic classification and intrusion detection systems, focusing on the application of deep learning techniques. The paper provides an in-depth analysis of the fundamental concepts of cloud network traffic and intrusion detection, highlighting the challenges associated with securing these dynamic and distributed environments. The survey thoroughly explores various deep learning techniques, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, autoencoders, and transfer learning, elucidating their relevance and applicability to cloud security. It delves into their strengths and limitations, paving the way for their integration into cloud security strategies.Two major aspects of cloud security are individually addressed. First, the paper investigates cloud network traffic classification using deep learning models. It dissects different architectural designs, such as CNN-based and RNN-based models, and examines their efficacy in accurately categorizing network traffic. Additionally, the survey delves into intrusion detection systems (IDS) powered by deep learning algorithms, showcasing approaches like anomaly detection with autoencoders and sequence-based intrusion detection with RNNs.The survey goes beyond individual techniques to explore hybrid approaches that combine traffic classification and intrusion detection. Real-world case studies are presented to demonstrate the practical implementation of these techniques in diverse cloud environments, from multi-cloud setups to edge clouds. A comprehensive comparison and analysis of techniques are conducted, focusing on evaluation metrics and factors influencing model performance.As the paper concludes, it highlights emerging trends in deep learning for cloud security and underscores the open challenges in scalability, real-time processing, and ethical considerations. This survey not only serves as a comprehensive guide for researchers and practitioners but also underscores the critical importance of robust cloud security frameworks in an era of evolving cyber threats and cloud-based operations.