Task scheduling for grid computing systems using a genetic algorithm

A grid computing environment is a parallel and distributed system that brings together various computing capacities to solve large computational problems. Task scheduling is a critical issue for grid computing; in task scheduling, tasks are mapped onto system processors with the aim of achieving goo...

Full description

Saved in:
Bibliographic Details
Published in:The Journal of supercomputing Vol. 71; no. 4; pp. 1357 - 1377
Main Authors: Jiang, Yi-Syuan, Chen, Wei-Mei
Format: Journal Article
Language:English
Published: Boston Springer US 01.04.2015
Subjects:
Algorithms
Compilers
Computation
Computational grids
Computer Science
Genetic algorithms
Genetics
Interpreters
Processor Architectures
Programming Languages
Strategy
Task scheduling
Tasks
Task scheduling
Directed acyclic graph scheduling
Genetic algorithms
ISSN:0920-8542, 1573-0484
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A grid computing environment is a parallel and distributed system that brings together various computing capacities to solve large computational problems. Task scheduling is a critical issue for grid computing; in task scheduling, tasks are mapped onto system processors with the aim of achieving good performance in terms of minimizing the overall execution time. In previous studies, there have been several approaches to solving the task-scheduling problem by genetic algorithms, which is a random search technique that is inspired by natural biological evolution. This study presents a genetic algorithm for solving the problem of task scheduling with two main ideas: a new initialization strategy to generate the first population and new genetic operators based on task–processor assignments to preserve the good characteristics of the found solutions. Our proposed algorithm is implemented and evaluated using a set of well-known applications in our specifically defined system environment. The experimental results show that the proposed algorithm outperforms other popular algorithms in a variety of scenarios with several parameter settings.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0920-8542
1573-0484
DOI:10.1007/s11227-014-1368-6