A Novel MLSTM-Based IDS for Cloud Security: Leveraging Jarratt-Butterfly Optimization

Introduction: Cloud Computing (CC) has emerged as a transformative technology offering scalability and flexibility to businesses worldwide. However, security challenges, including unauthorized access, data breaches, and insider threats, remain a significant concern. Traditional Intrusion Detection Systems (IDS) struggle with evolving cyber threats, necessitating advanced approaches leveraging machine learning (ML) and deep learning (DL). Objectives: This study aims to develop a novel IDS model incorporating a Morlet Wavelet Kernel Function and an MLSTM (Modified Long Short-Term Memory) classifier. The primary objective is to enhance intrusion detection accuracy by employing the Jarratt-Butterfly Optimization Algorithm (JBOA) for optimal feature selection. Methods: The proposed IDS model is tested on the BoT-IoT dataset, leveraging deep learning techniques for anomaly detection. Feature engineering techniques, including Recursive Feature Elimination (RFE) and Principal Component Analysis (PCA), are employed for dimensionality reduction. The MLSTM classifier, combined with MWKF-based optimization, is utilized to enhance detection accuracy while minimizing false alarms. Results: Experimental analysis demonstrates that the proposed IDS model outperforms traditional intrusion detection techniques. The model achieves superior accuracy, recall, and precision compared to existing approaches. The integration of JBOA improves feature selection efficiency, leading to reduced computational overhead and enhanced classification performance. Conclusions: The study presents a robust deep learning-based IDS framework capable of effectively detecting cyber threats in cloud environments. The proposed model enhances security in IoT-based cloud infrastructure, ensuring real-time threat mitigation. Future research will focus on addressing data imbalance issues and optimizing computational efficiency further.