AI-Driven Energy-Efficient Resource Management in Cloud Computing

Cloud computing has come out as one of major platforms of providing scalable and on demand services but steady development of cloud infrastructures has added pressure on issues of resource inefficiency and high-energy consumption. To resolve these problems, this paper suggests an AI-based framework of Energy-Efficient Resource Management Framework (AI-EERM) of cloud data centers. The framework combines workload prediction through machine learning and task scheduling through reinforcement learning to dynamically distribute resources with minimum energy consumption. A Bidirectional Stacked Long Short-Term Memory (Bi-SLSTM) model is utilized to predict future workload requirements, a Dual Deep Q-Network (DDQN) scheduler is used to allocate tasks to virtual machines depending on the load, execution time, and energy usage. As experimental studies demonstrate proposed AI-EERM framework performs better in energy efficiency (0.95), cost efficiency (0.90), SLA compliance (0.98) and a task completion rate (0.97) compared to existing methods like CNN, LSTM, FRU, and RNN.