A Hybrid Intrusion Detection System Leveraging XGBoost and RNNs for Enhanced Anomaly Detection in Cloud Data Centers
Asaad Althoubi, Hassan Peyravi · 2023
With the rapid growth of cloud computing, securing cloud data centers against potential security threats has become a critical concern. In this research paper, a hybrid intrusion detection system (IDS) is proposed that combines the power of the gradient boosting algorithm, specifically XG Boost, with the capabilities of Recurrent Neural Networks (RNNs). The objective of the proposed IDS is to effectively predict and detect temporal patterns in network traffic, thereby enhancing the security posture of cloud data centers. By fusing these techniques, the model aims to achieve a holistic view of network behavior, enabling the identification of subtle anomalies that may indicate intrusion attempts. To evaluate the performance of the model, extensive experiments were conducted on real-world network traffic data collected from a cloud data center environment. Evaluation metrics encompass accuracy, precision, recall, and F1-score, providing a comprehensive assessment of the model's effectiveness. The experimental results illustrate the model's efficacy in predicting temporal patterns of network traffic. The model effectively captures sudden shifts in traffic behavior, as well as long-term trends that may indicate evolving threats. Furthermore, the system's ability to adapt to changing network conditions enhances its resilience to dynamic attack strategies.