Design and evaluation of a neurofuzzy CPU scheduling algorithm

Abstract

CPU scheduling is the basis of multiprocessing operating systems. By switching the CPU among processes, the operating system can make the computer more productive. There are many different CPU scheduling algorithms. Different algorithms have different properties and may favor one class of processes over another and no single one is ideal absolutely for every application. This paper presents an attempt to apply neurofuzzy in the design and implementation of a rule-based scheduling algorithm to solve the shortcoming of well-known scheduling algorithms. A fuzzy-based decision maker has been proposed to compute a new priority of all CPU processes according to the process pre-priority and its execution time. Results given in this paper demonstrate that the average waiting time and the average turnaround time in the proposed algorithm are better than that obtained using preemptive priority scheduling, and closed to that obtained from preemptive shortest-job-first (SJF) scheduling. The new proposed algorithm is a dynamic scheduling algorithm which deals with both process priority and its execution time, while the preemptive SJF scheduling algorithm doesn't. The results obtained, using neurofuzzy, are approximately the choice as for fuzzy but it responds faster than it. On the other hand the functional neurofuzzy is the best algorithm compared with structural and fuzzy scheduling algorithms.