基于柯西变异粒子群的云服务任务调度算法  

Cloud Service Resource Scheduling Algorithm Based on Cauchy Mutation Particle Swarm Optimization

作　　者：毕洋 倪文龙 BI Yang;NI Wen-long(School of Computer Information Engineering,Jiangxi Normal University,Nanchang Jiangxi 330022,China)

机构地区：[1]江西师范大学计算机信息工程学院,江西南昌330022

出　　处：《计算机仿真》2024年第12期438-443,543,共7页Computer Simulation

摘　　要：云服务中的资源分配与任务调度是一个NP完全问题。基本粒子群算法在解决这类问题时存在种群初始化不合理、收敛速度慢、收敛精度低、容易陷入个体最优解的缺陷。针对以上问题,提出了一种带柯西变异的粒子群算法(CPSO)。首先,利用Tent混沌映射,使得初始种群更符合云服务的应用场景;其次,为了改善粒子群算法面对高维问题时的个体寻优和全体寻优能力,利用非线性函数,使得惯性权重可以自适应调整;最后利用柯西变异粒子改进速度更新公式,实现对粒子位置的扰动,让粒子可以突破个体最优的状态。实验结果表明,CPSO在收敛速度、收敛精度上都优于PSO。从小规模到大规模的云服务场景进一步设计对比实验,结果表明,CPSO都能得出优于PSO的资源分配策略。Resource allocation and task scheduling in cloud services is an NP-complete problem.In solving such problems,the basic particle swarm algorithm has some drawbacks,such as unreasonable population initialization,slow convergence speed,low convergence accuracy,and easy falling into the individual optimal solution.To solve the above problems,a particle swarm algorithm with Cauchy variation(CPSO)is presented.First,the Tent chaotic map is used to make the initial population more suitable for cloud service scenarios.Secondly,the nonlinear adaptive inertial weights are introduced to improve the individual and overall optimization capabilities of particles in different iteration periods.Finally,the Cauchy variant particle is used to improve the speed update formula to disturb the position of the particle so that the particle can break through the optimal state of the individual.The results show that CPSO is superior to PSO in convergence speed and accuracy.Further comparative experiments are designed from small-scale to large-scale cloud service scenarios,and the results show that CPSO can achieve better resource allocation strategies than PSo.

关 键 词：云计算 粒子群算法 帐篷映射 柯西变异 任务调度 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]