An Advanced Cloud IDS Using Random Forest-Based Ensemble Learning and Feature Reduction Techniques

Abstract

Cloud computing environments are increasingly vulnerable to sophisticated cyberattacks, making efficient intrusion detection systems (IDS) essential for ensuring security. This paper proposes an enhanced cloud-based intrusion detection approach using hybrid ensemble learning and feature engineering techniques. Initially, data preprocessing and visualization-driven feature engineering are applied to reduce irrelevant features and improve data quality. The system integrates multiple machine learning and deep learning models, including Random Forest, Decision Tree, Support Vector Machine, Naïve Bayes, LSTM, and advanced ensemble methods such as Voting Classifier (RF + AdaBoost) and Stacking Classifier (RF + MLP with LightGBM), to improve detection performance. The ensemble approach combines the strengths of individual models to achieve higher robustness and accuracy. The proposed model is evaluated using benchmark datasets such as NSL-KDD and BoT-IoT. Experimental results demonstrate significant improvements in accuracy, precision, recall, and reduced execution time compared to traditional models. This approach provides an efficient and scalable solution for real-time intrusion detection in cloud environments.