Adaptive Resource Allocation in Cloud Computing Using Advanced AI Techniques

Managing dynamic and unpredictable workloads in cloud computing remains a critical challenge, particularly in ensuring efficient resource allocation for burst tasks while maintaining elasticity and scalability. This research addresses the issue of adaptive resource management by proposing a multi-VM coordination strategy based on Proximal Policy Optimization (PPO) and Deep Q-Networks (DQN). The proposed method optimizes resource allocation by dynamically adjusting to workload fluctuations, improving the overall task completion rate and reducing execution time. Simulation results validate the effectiveness of the approach, demonstrating that the PPO/DQN-based strategy achieves a significantly higher task completion rate compared to the traditional First-Fit Profit (FFP) bin-packing algorithm, particularly under high resource demand scenarios. Experimental evaluations, conducted on an AMD Opteron and Intel Xeon server environment with 1200 repeated simulations, confirm that the proposed method ensures efficient resource utilization while adapting to elastic workload variations. These findings establish the PPO/DQN-based strategy as a robust solution for real-time cloud resource management, enhancing system responsiveness and reducing operational costs.