Hybrid Discrete Particle Swarm Optimization for Task Scheduling in Grid Computing

Maryam Karimi
2014 International Journal of Grid and Distributed Computing  
Computational Grid is a high performance computing environment that participating machines resources are used through software layer as transparent and reliable. Task assignment problem in Grid Computing is a NP-Complete problem that has been studied by several researchers. The most common objective functions of task scheduling problems are Makespan and Flowtime. This paper gives a classification of meta-heuristic scheduling algorithms in distributed computing that are applicable to grid
more » ... able to grid environment and addresses scheduling problem of independent tasks on Computational Grids. A Hybrid Discrete Particle Swarm Optimization and Min-min algorithm (HDPSO) is presented to reduce overall Completion Time of task. We used DPSO as it has a faster convergence rate than Genetic Algorithm (GA). Also, it has fewer primitive mathematical operators than both GA and Simulated Annealing (SA), making applications less dependent on parameter fine -tuning. It allows us to use the fitness function directly for the optimization problem. Moreover, using discrete numbers, we can easily correlate particle's position to task-resource mappings [2] . We evaluated four scheduling methods with different number tasks and resources based on total Completion Time.
doi:10.14257/ijgdc.2014.7.4.09 fatcat:hic5yf5qgbfq5molwb6l3stfuu