Task Scheduling for Multiprocessor Systems Using Queuing Theory

Abstract

This research focuses on comparing different multi-processor task scheduling algorithms. Each algorithm has been simulated using one of queuing theory models in Operations Research (OR) science to evaluate its behavior and efficiency. The comparison includes an analysis of the behavior of central processing unit (CPU) when receiving number of jobs at four random job duration patterns that are; (random, ascending, descending, and volatile low-high). Microsoft Excel 2010 was used to form the data of each case, and the result shows convergence and divergence among the studied algorithms at different patterns. Also it has been found that the Fleischer algorithm is very efficient in enhancing and minimizing the waiting duration for each job at the total job queue of the CPU.

Keywords: Operations Research, Queuing Theory, Multiprocessor, Scheduling Algorithms, Simulation.