基于多边协商的分布式网格资源调度

基于多智能体的理论，提出了一种智能网格系统资源调度方法，通过网格节点自主选择任务来实现网格系统中的资源优化调度.由于各节点的自主性，对于完成不同的任务将存在不同的支持度.采用模糊认知图建立了对任务分配方案的支持度协调模型，并提出了标准支持度的概念，同时保证节点支持度协调的可行性和有效性.网格节点在自身利益最大化目标的驱动下，无需固定的上层资源调度单元，通过多边的竞争和协商组成任务组，实现分布式网格资源的优化调度.提出的调度策略适用于分布式计算，具有较好的实时性和鲁棒性.

Distributed resource scheduling based on multi-node negotiation for grid systems

An intelligent resource scheduling strategy for grid systems was proposed based on the theory of multi-agent. Grid nodes independently chose subtasks in grid systems. Due to the autonomous characteristics of grid nodes, different support degrees were assigned to each grid node for different tasks. The fuzzy cognitive map was employed to construct a support degree negotiation model. Considering that each grid node possesses different types of resources, definition of the standard support degree was presented to ensure feasibility and validity of the support degree negotiation model. Without the upper level resource scheduling unit, task teams were constituted by multiple grid nodes through competition and negotiation to maximize interests of each grid node. The proposed scheduling method is applicable for distributed computation, which makes the grid computation system possess better real-time performance and robustness.