REINFORCED HAWK INTELLIGENCE: A CNN-AHHO FRAMEWORK FOR SCALABLE CLOUD RESOURCE MANAGEMENT

Abstract

Cloud computing workloads are projected to grow by 23.1% annually, with over 80% of enterprises adopting multi-cloud strategies. This trend creates a pressing need for optimal virtual machine (VM) resource allocation to ensure cost efficiency and performance reliability. However, existing allocation strategies often suffer from static optimization limitations and an inability to adapt efficiently to dynamic workloads, leading to frequent resource underutilization and service delays.To address these challenges, this work proposes a novel Multi-Agent Deep Reinforcement Learning-Based Adaptive Harris Hawks Optimization (MADRL-AHHO) algorithm for intelligent cloud resource allocation, utilizing the VM Resource Allocation dataset (VM-0 to VM-5 classes). Initially, the dataset undergoes preprocessing through normalization and feature selection to reduce dimensionality and noise.Feature extraction is enhanced using a Convolutional Neural Network with Firefly Optimization (CNN-FFO), which is benchmarked for its learning capacity and convergence behavior. However, to address performance limitations under dynamic load conditions, CNN-FFO is further improved through integration with Adaptive Harris Hawks Optimization (CNN-AHHO). This approach dynamically adjusts exploration and exploitation capabilities based on multi-agent reinforcement feedback.Through continuous interaction with the environment, agents learn optimal VM allocation policies that maximize resource utilization and minimize Service-Level Agreement (SLA) violations. Experimental results demonstrate that the proposed CNN-AHHO significantly outperforms CNN-FFO and conventional machine learning methods in terms of allocation accuracy, convergence rate, and computational efficiency, thus offering a robust and adaptive solution for modern cloud infrastructure management