Task Scheduling in Grid Computing Environments

• Main Authors: Yi-syuan Jiang, 蔣宜璇
• Other Authors: Wei-mei Chen
• Format: Others
• Language: en_US
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/60998659273566116313

碩士 === 國立臺灣科技大學 === 電子工程系 === 101 === In the grid computing, task scheduling which maps tasks onto a parallel and distributed system is important for achieving good performance in terms of minimizing the overall execution time. Various algorithms that include genetic algorithms (GA) have been proposed for the issue of scheduling. The feature in GA-based algorithm is used the chromosome representation as a solution and transform a population of chromosomes for improving the quality of solutions. This thesis presents a GA-based algorithm to solve this problem by improving the existing genetic algorithm with two main ideas. First of all, a new initialization strategy is used to generate the first population of chromosomes, and classifies the search space into groups for the purpose of accelerating the time to explore the whole solution space. Secondly, the found solution may lose during certain operators unexpectedly in some GA-based algorithms. The modifications, based on problem-specific knowledge, are added to preserve the characteristics of the found solution. Our proposed algorithm is implemented and evaluated using five well-known applications and the specific-defined system environment. The experimental results show that the proposed algorithm outperforms other algorithms within several parameter settings. The more experimental results figure out the relationship between the parameter setting and the performance of the proposed method.