Dynamic Entropy Constrained Soft Actor-Critic Based Task Scheduling in Cloud Resource Management

Resource scheduling in cloud computing refers to the process of allocating computational resources for efficient task completion, aiming to minimize resource usage while ensuring timely task completion. Traditional approaches of resources allocation face challenges such as energy constraints and strive for dynamic scheduling to meet unpredictable workload demands. To address these limitations, the Dynamic Entropy Constrained Soft Actor-Critic (DESAC) method has been proposed to enhance resource utilization and reduce energy consumption. The DESAC facilitates entropy regularization during training, promoting exploration by adjusting entropy to ensure the strategy remains less deterministic, and facilitating comprehensive exploration of the state-action space. The system's performance and energy efficiency are evaluated using a fitness function based on energy consumption, processing 1,000 tasks (each with a value of 35.47), and comparing the results to those of the existing Actor-Critic (A3C) method.