Requirement-driven multi-workflow scheduling based on improved evolutionary multitasking embedded bi-level optimization

Abstract

When multiple users share the same cloud service resources, cloud computing makes it more difficult for workflow applications to schedule tasks. Therefore, it is important to design an appropriate scheme that benefits both users and cloud service providers. We create a bi-level optimization model to explain cloud client cooperation in order to address this issue. In order to ensure fairness while vying for resources across several processes, users combine execution time and cost as a goal for user satisfaction. They also coordinate resource allocation to minimize the error between the time and cost of a single workflow execution. Cloud service providers maximize their profits by rationally and dynamically adjusting prices. In the solution method to maintain the diversity of the population in the optimization process, adaptive cross-variance probability and population local replacement strategy are proposed, which reduces the poorly adapted individuals to play the game and accelerates the convergence of the population. The experimental findings demonstrate that the model’s algorithm’s validity is confirmed by various datasets and that the user’s service quality and the cloud service provider’s interests are balanced.