具有模糊多目标的网格任务调度算法

A tasks scheduling algorithm for grid computing based on fuzzy multi-objective

在异构的网格计算平台上,网格中有用户、资源管理员、组织管理者等实体,这些实体对网格的管理、使用、维护、安全性、可靠性等目标都提出了要求,并且这些目标有时是不可量化的。针对具有模糊多目标网格计算的任务调度问题,提出模糊多目标网格任务调度模型,使用模糊化等式对多目标进行模糊处理,给出求解该模型的模糊化定理,并对该定理进行证明。利用差分优化算法无需目标函数连续可微的特点,提出使用模糊差分优化算法完成模糊多目标的网格任务调度。实验结果表明,模糊差分优化算法较现有算法在执行时间上处于劣势,但在可靠性、安全性和丢失任务数三个指标上要优于现有算法。

Users, resource administrators, group managers are entities of heterogeneous grid compu- ting in the grid computing platform. They require grid management, use and maintenance, security, relia＇bility, and so on. These objectives are not often quantitative. Aiming at the tasks scheduling problem for grid computing based on multi-objective, a tasks scheduling model for grid computing based on fuzzy multi-objective is proposed. Fuzzy equation is used to process multi-objective, and the theorem of the fuzzy equation is proposed and proved. Utilizing the merit of the differential evolutionary algorithm that the objective function can be not continuous and differentiable, a fuzzy differential evolutionary algo- rithm is proposed to process tasks scheduling for grid computing based on fuzzy multi-objective. In the experiments, the proposed algorithm is compared with the existing algorithms. Although the proposed algorithm is not better on the execution time, it outperforms the existing algorithms in terms of reliabili- ty, security and drooped task numbers.