RFGB-Hybrid Machine Learning Framework for Secure and Efficient Cloud Resource Allocation

Abstract

In advanced digital infrastructure, cloud computing is the foundation, which provides a cost-effective, scalable, and adaptable solution for various applications. There are several challenges in cloud security and efficient resource allocation due to the increased use of cloud services. The dynamic workload management is lacking in traditional methods, which leads to underutilization, inefficiency, and an increased risk of security threats. These challenges are addressed by the proposed ML-based framework that combines an RF and GB model to improve cloud resource allocation and simultaneously ensure cloud security. The proposed framework also used data preprocessing techniques to reduce the noise, produce a balanced dataset, and extract the key features for workload prediction and anomaly detection. The workload demand can be predicted automatically; intrusion is detected and prevents malicious resources through the proposed hybrid model RF-GB (RFGB). The proposed framework combines anomaly detection and security-aware labelling to eliminate risks like distributed DOS attacks and unauthorized access, ensuring a balance between efficiency and resilience. As an experimental result, the proposed hybrid model attained high-performance results compared to individual RF or GB models in terms of accuracy, recall, and throughput.  Moreover, the adaptive optimization techniques are also used by the proposed framework for enhancing the load balancing and service-level agreements. This research also highlights the potential of ML-based intelligent resource allocation in a secure and adaptive cloud environment. Additionally, for the future digital environment, this research contributes a scalable and reliable methodology that can be utilized in several domain applications like finance, healthcare, and government services.