CAJ | 학술논문

如何充分利用云中资源对任务进行高效调度，是云计算中的重要问题。提出一种基于差分进化的多策略变异差分进化任务调度算法。该算法在标准差分进化算法的基础上加入了基于正态分布的分类和多种变异策略，变异策略中差异向量个体采用基于个体相似度的轮盘赌选择，改善了标准差分算法收敛速度慢和易陷入局部最优的缺点，可有效求解组合优化问题。在仿真平台CloudSim上进行模拟测试，结果表明该算法能得到较短的任务总完成时间，提高了资源利用率。
여하충분이용운중자원대임무진행고효조도，시운계산중적중요문제。제출일충기우차분진화적다책략변이차분진화임무조도산법。해산법재표준차분진화산법적기출상가입료기우정태분포적분류화다충변이책략，변이책략중차이향량개체채용기우개체상사도적륜반도선택，개선료표준차분산법수렴속도만화역함입국부최우적결점，가유효구해조합우화문제。재방진평태CloudSim상진행모의측시，결과표명해산법능득도교단적임무총완성시간，제고료자원이용솔。
How to make full use of cloud resources to dispatch tasks efficiently is an important issue in cloud computing. The MSMDE mentioned in this thesis is an algorithm based on multi-strategy mutation differential evolution of differential evolution. This algorithm is added with category of normal distribution and multiple mutation strategy on the basics of standard Differential Evolution(DE), difference vector individuals in mutation strategy use roulette to choose based on indi-vidual similarity, improve DE algorithm’s shortcomings of slow convergence rate and local optimum tendency, can effec-tively solve combinatorial optimization problem. After running simulation tests on simulation platform CloudSim, the result shows that this algorithm is able to achieve relatively short total task completion time and improve resource utilization.