| Abstract |
: |
A computational grid is a set of large-scale heterogeneous systems. Job scheduling is one of the most important issues in a distributed grid system. Genetic Algorithm as a basic evolutionary algorithm, is a fair way to solve difficult problems that have failed with traditional conventional methods to achieve the desired results. This algorithm has been able to achieve acceptable results for the task scheduling problem in the grid system. In this research, the scheduling of tasks in the grid system is examined in a proposed way. Proposed method to create new evolutionary algorithms resulting from combining genetic algorithm with particle mass optimization algorithm (PSO-GA), genetic algorithm with colonial competition algorithm (GA-ICA), genetic algorithm with refrigeration simulation algorithm, SA-refrigeration algorithm (GA) Genetics deals with ant colony optimization (GA-ACO) to schedule distributed optimizations in the grid system. The main purpose of the hybrid algorithm is improving the local search process of the genetic algorithm (GA) by combining it with the evolutionary algorithms ICA, PSO, SA and ACO, preventing premature convergence and stopping at the local minimums and ensuring global optimization. Considering the time process of workflow and task execution time of works as comparison criteria, finally the proposed hybrid algorithm was able to achieve the desired results for the task scheduling problem in the distributed grid system. |
