Algorithm fuzzy scheduling (AFS) for realtime jobs on multiprocessor systems

The computing in Real-time is rapidly focusing much developments in technologies so that the real-time jobs are to be scheduled and executed on computing systems in particular time frame. The scheduling and load balancing techniques in distributed systems face numerous challenges because of lack of...

Full description

Saved in:
Bibliographic Details
Published in:Indonesian Journal of Electrical Engineering and Computer Science Vol. 25; no. 3; p. 1308
Main Authors: Holagundi, Nirmala, Ashwathsetty, Girijamma Hollalkere, Basthikodi, Mustafa
Format: Journal Article
Language:English
Published: 01.03.2022
ISSN:2502-4752, 2502-4760
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The computing in Real-time is rapidly focusing much developments in technologies so that the real-time jobs are to be scheduled and executed on computing systems in particular time frame. The scheduling and load balancing techniques in distributed systems face numerous challenges because of lack of centralized strategy to dispatch the jobs in multiprocessors systems. In this work, we propose an Algorithm Fuzzy Scheduling (AFS) for real-time jobs that includes of Arrival time, Deadline and Computation time as the scheduling parameters of input. The approach AFS is analyzed and compared with Existing Fuzzy Algorithm (EFA) model for evaluation of performances from the outcome of the simulation. The jobs are scheduled on multiprocessor at higher system load by making use of fuzzy mechanisms in the algorithms. The experimental results prove that the proposed AFS achieves a better performance comparatively to EFA at various system load factors with respect to mean turnaroundtime, mean response time and count of missed deadlines. This is the initial phase of the algorithm, that will be enhanced to consider a greater number of parameters to be associated with jobs for better decision making and to investigate the scope for algorithm level parallelism.
ISSN:2502-4752
2502-4760
DOI:10.11591/ijeecs.v25.i3.pp1308-1319