DEEP LEARNING–ENABLED DATA SECURITY AND INTRUSION DETECTION IN AZURE AND AWS CLOUD SYSTEMS

Abstract

The exponential growth of cloud computing necessitates robust, intelligent security mechanisms to counter increasingly sophisticated cyber threats. This paper presents a deep learning–enabled framework for enhancing data security and intrusion detection within Azure and AWS cloud environments. Leveraging the inherent scalability of cloud platforms, we propose a hybrid model combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to analyze network traffic and system logs in real time. The model autonomously learns spatiotemporal features to detect zero-day attacks, unauthorized access, and anomalous behavior with high accuracy. Implemented and tested across Azure Sentinel and AWS GuardDuty, the framework demonstrates a 98.7% detection rate with a low false positive ratio, outperforming traditional signature-based methods. Additionally, it integrates automated mitigation protocols, enabling adaptive response to emerging threats. This work underscores the efficacy of deep learning in overcoming the limitations of conventional security tools and offers a scalable, cross-platform solution for modern cloud infrastructures. The findings highlight a significant advancement toward self-defending cloud systems, ensuring data integrity, confidentiality, and availability in multi-tenant architectures.