物联网下分布式在线学习资源个性化调度仿真
|
摘要
针对当前方法进行在线学习资源调度中,存在用户效用函数值计算不准确、调度完成时间过长、成本消耗较大等问题。提出基于遗传算法的在线学习资源个性化调度方法。通过分析资源调度任务执行时间和调度执行费用之间的关系,计算出任务调度时间效用和费用效用的权重值,根据权重计算结果构建资源调度效用模型,利用该模型给出资源调度子任务完成最短、成本最少的调度目标函数。采用贪心算法对目标函数进行求解,通过子任务资源调度时间—成本来约束适应度函数,以适应度函数值的调整完成在线学习资源个性化调度。结果表明,所提方法在学习资源调度中所用完成时间较短、成本消耗较低,整体资源利用率较高。
Due to inaccurate calculation of avail function value,long completion time of scheduling and high cost,a method for scheduling online learning resource based on genetic algorithm was proposed. By analyzing the relationship between the execution time of resource scheduling task and the execution cost,the weight values of task time utility and cost utility were calculated. Then,the resource scheduling utility model was built based on the weighted result. Moreover,this model was used to give the scheduling objective function for completing resource scheduling subtask with the shortest completion time and the least cost. In addition,the greedy algorithm was used to solve the objective function,and the fitness function was constrained by the scheduling time-cost of sub-task resource. Thus,the personalized scheduling of online learning resources was completed by adjusting the fitness function value. Simulation results prove that the proposed method has shorter completion time,lower cost,and higher overall resource utilization rate during the learning resource scheduling.
Due to inaccurate calculation of avail function value,long completion time of scheduling and high cost,a method for scheduling online learning resource based on genetic algorithm was proposed. By analyzing the relationship between the execution time of resource scheduling task and the execution cost,the weight values of task time utility and cost utility were calculated. Then,the resource scheduling utility model was built based on the weighted result. Moreover,this model was used to give the scheduling objective function for completing resource scheduling subtask with the shortest completion time and the least cost. In addition,the greedy algorithm was used to solve the objective function,and the fitness function was constrained by the scheduling time-cost of sub-task resource. Thus,the personalized scheduling of online learning resources was completed by adjusting the fitness function value. Simulation results prove that the proposed method has shorter completion time,lower cost,and higher overall resource utilization rate during the learning resource scheduling.
引文
[1]李媛祯,等.一种Hadoop Yarn的资源调度方法研究[J].电子学报,2016,44(5):1017-1024.
[2]魏蔚,刘扬,杨卫东.一种通用云计算资源调度问题的快速近似算法[J].计算机研究与发展,2016,53(3):697-703.
[3]杨继君,佘廉.面向多灾点需求的应急资源调度博弈模型及优化[J].中国管理科学,2016,24(8):154-163.
[4]陈晓旭,等.基于最小费用最大流的大规模资源调度方法[J].软件学报,2017,28(3):598-610.
[5]张凯.云计算平台服务资源均衡调度仿真研究[J].计算机仿真,2017,34(9):424-427.
[6]赵昱,等.基于改进QPSO算法的云计算资源调度策略研究[J].火力与指挥控制,2017,42(4):14-17.
[7]黄坚.基于物联网技术的特种设备信息化平台设计[J].电子设计工程,2017,23(25):81-84.
[8]邓志龙,等.一种基于深度强化学习的调度优化方法[J].西北工业大学学报,2017,35(6):1047-1053.
[9]高太光,黄敏,陈培友.物联网环境下煤矿救援资源调度多Agent自动协商模型[J].管理工程学报,2016,30(3):134-140.
[10]封岸松,王宏,张彦武.无线HART网络节点变速率资源调度算法实现[J].电子技术应用,2016,42(3):95-97.
[2]魏蔚,刘扬,杨卫东.一种通用云计算资源调度问题的快速近似算法[J].计算机研究与发展,2016,53(3):697-703.
[3]杨继君,佘廉.面向多灾点需求的应急资源调度博弈模型及优化[J].中国管理科学,2016,24(8):154-163.
[4]陈晓旭,等.基于最小费用最大流的大规模资源调度方法[J].软件学报,2017,28(3):598-610.
[5]张凯.云计算平台服务资源均衡调度仿真研究[J].计算机仿真,2017,34(9):424-427.
[6]赵昱,等.基于改进QPSO算法的云计算资源调度策略研究[J].火力与指挥控制,2017,42(4):14-17.
[7]黄坚.基于物联网技术的特种设备信息化平台设计[J].电子设计工程,2017,23(25):81-84.
[8]邓志龙,等.一种基于深度强化学习的调度优化方法[J].西北工业大学学报,2017,35(6):1047-1053.
[9]高太光,黄敏,陈培友.物联网环境下煤矿救援资源调度多Agent自动协商模型[J].管理工程学报,2016,30(3):134-140.
[10]封岸松,王宏,张彦武.无线HART网络节点变速率资源调度算法实现[J].电子技术应用,2016,42(3):95-97.