AI-Based Secure and Energy-Efficient Framework for Multi-Tenant Cloud Systems

This paper proposes an adaptive, energy-aware, and secure architecture for a multi-tenant cloud environment, which aims to solve cloud computing systems' security, energy, and performance problems. Thus, with a vast increase in cloud adoption across markets, industrial sectors like healthcare, IoT, finance, cybersecurity, and efficient resource management have become a high priority. These needs are met by the Hybrid Intrusion Detection System (IDS), the Energy-Aware Resource Management (ERM), and an AI-based Load Balancer within the context of the proposed framework. The proposed Hybrid IDS uses auto encoders for anomaly detection and CNNs for attack classification to identify known and unknown threats. Based on the proposed model, it achieved 98.7% accuracy and a good F1 Score of 20, which is good enough to detect new and common intrusions. The ERM module efficiently uses energy through the RL approach to manage resources dynamically, while ACO is applied for scheduling. This is done in an energy-efficient way, using 22% less energy than traditional schedulers while at the same time boosting CPU utilization. While forward-projecting workloads are based on basic historical measures, the AI-based Load Balancer has 36% lower latency and 28% higher throughput than Round-Robin scheduling. This makes it highly efficient in cases where varied performance may be required in terms of time. The work presented in this paper suggests a novel approach that offers a solution to these problems and supports security, energy efficiency, and high performance, effectively removing the drawbacks of current methods. Future work will include the integration of quantum clouds with others, edge computing support, and multi-cloud support to bring better scalability and resilience. This research shows that cloud infrastructures can be adequate for security and sustainability while meeting today's technological environment requirements.