Advanced queueing and scheduling techniques in cloud computing using AI-based model order reduction

Abstract Resource management in computing presents significant challenges that require innovative solutions. This paper proposes a novel architecture integrating artificial intelligence (AI) with model order reduction (ROM) and advanced queueing theory models to enhance resource allocation and task scheduling efficiency. The study demonstrates substantial improvements in critical performance parameters, including response time optimization, resource utilization, and energy consumption management through comprehensive mathematical modeling and machine learning frameworks. The methodology incorporates predictive analytics for resource demand forecasting, intelligent scheduling algorithms for automatic workload adaptation, and calendar queueing techniques for real-time decision-making. Extensive simulations and analyses across multiple queueing scenarios validate the theoretical framework, establishing a robust foundation for efficient computation in large-scale distributed environments. Results indicate a 50% reduction in response time, a 50% increase in throughput, and a 15% improvement in resource utilization. The ROM implementation achieved a 65–80% reduction in processing overhead while maintaining 95–98% accuracy compared to full-scale models. Energy efficiency improved by 20% through intelligent workload distribution, with system reliability reaching 99.99% uptime. This research contributes to computing advancement by demonstrating the effectiveness of integrating AI-driven resource management with traditional queueing theory, providing a scalable solution for modern infrastructure optimization. The proposed framework’s ability to automatically adapt to varying workloads while maintaining optimal performance parameters represents a significant step forward in resource management technology.