融合安全的网格依赖任务调度双目标优化模型及算法

Integration of Security Grid Dependent Tasks Scheduling Double-Objective Optimization Model and Algorithm

为了解决异构网格环境下依赖任务调度问题面临的安全威胁,综合考虑网格资源节点的固有安全性和行为安全性,分别构建了一个网格资源节点身份可靠性度量函数和行为表现信誉度评估策略.同时,为了确立任务安全需求与资源节点安全属性之间的隶属关系,定义了安全效益隶属度函数,从而建立一个网格任务调度的安全融合模型.以此为基础,提出一个时间-安全驱动的双目标优化网格依赖任务调度模型.为了求解该模型,处理任务间约束关系时引入深度值和关联耦合度的排序定义,再结合网格任务调度问题的具体特点,重新定义和设计新的粒子进化方程.同时,基于均匀分布向量和粒子浓度定义了选择策略,从而提出一种双目标优化的网格依赖任务调度粒子群进化算法,并运用概率论的有关知识证明算法的收敛性.最后,对所提出的离散粒子群进化算法进行了多角度分析和大规模仿真实验,其仿真结果表明,该算法与同类算法相比,不仅具有较好的收敛速度和单目标优化性能,而且在任务调度长度和安全满意度方面具有更好的双目标优化综合性能.

In order to solve the security threats that dependent tasks scheduling problems face under the heterogeneous grid environment,this paper takes into account the inherent safety and behavior of security of the grid resource node and the reliability of measurement functions in the grid resource node.In addition,the behavior in credibility assessment strategies are also constructed.In order to establish the subordinate relationship between the security requirements of the task nodes and resources security attributes,security benefits of the membership functions are defined.Hence,a grid task scheduling model for security integration is established.On this basis,the requirement representation model and the grid resource topology model are defined;thus,the models of doubleobjective optimization of grid task scheduling are proposed.In order to solve this model,the definition of depth values and the sort of coupling is introduced when dealing with the constraints between tasks.A particle evolution equation is re-defined and re-designed to consider the specific characteristics of the grid task scheduling problem.At the same time,a selection strategy is defined,based on the uniformly distributed vector and concentration of particles.Thus,this paper presents a multi-objective optimization of grid task scheduling particle algorithm,and the algorithm is proved to be viable by applying the relevant knowledge of a probability theory.Simulation results show that compared with similar algorithms,under the same conditions,this algorithm has a faster convergence speed and a better performance in double-objective optimization