采用离散粒子群算法的网格任务安全级调度

Security Scheduling of Grid Tasks Based on the Discrete Particle Swarm Algorithm

针对异构网格环境中任务调度问题对所面临的安全性研究不足,在考虑了保密性、完整性和真实性等安全性因素的情况下,构造了相应的安全效益函数;依据网格节点的历史行为特点,提出了节点的信誉度动态评估策略;基于行为特点提出了一种离散粒子群算法,由此建立了任务安全级调度新模型.算法基于安全调度的离散空间特征给出了粒子的位置表示方法,从而克服了连续空间解决离散问题所造成的解空间冗余问题.采用分步计算和修改粒子位置的方式重新定义了粒子进化方程,避免了进化过程中速度之间的相互干扰问题.为了防止算法陷入局部最优,引入了均匀扰动速度.实验结果表明,与基于连续空间的粒子群算法和遗传算法相比,所提算法具有较快的收敛速度、较短的调度长度和较高的安全性能.

For the lack of safety studies to the task scheduling under the heterogeneous grid environment,a security benefit function is constructed under the consideration of the performance requirements such as confidentiality,integrity,authenticity,and so on.A dynamic method to evaluate the credibility of nodes and a new discrete particle swarm optimization algorithm are proposed based on the characteristics of historical behavior of grid resource nodes.Then a new model to schedule task security levels is established.The algorithm gives representation for particle locations based on the discrete space features of security scheduling.Hence,solution space redundancy of scheduling based on continuous space is avoided.Particle evolution equation is redefined by means of stepwise calculation and modification of particle positions,which avoids the problem of mutual speed interference during the evolution.A uniform disturbance speed is introduced to prevent the algorithm into a local optimum.Experiments and comparisons with the particle swarm optimization algorithm based on continuous space and the genetic algorithm show that the proposed algorithm has a faster convergence speed,a shorter scheduling length,and a higher safety performance.