Explainable AI for CPU Resource Scheduling in Cloud Operating Systems

Abstract

Cloud computing environments require intelligent and efficient resource scheduling to manage dynamic workloads and meet service-level objectives. Traditional rule-based scheduling algorithms often fall short in handling the complexity and scale of modern cloud systems. This paper introduces a novel framework that leverages Explainable Artificial Intelligence (XAI) techniques to optimize CPU resource scheduling in cloud operating systems. By integrating interpretable models such as decision trees and SHAP (SHapley Additive exPlanations) values with deep learning-based schedulers, the framework not only enhances scheduling accuracy but also offers transparency in decision-making processes. Experimental results on synthetic and real-world workloads demonstrate the effectiveness of the proposed framework in improving system performance while providing human-understandable insights into scheduling logic.