文献类型：期刊文献

中文题名：基于机器学习的复杂网络数据流均衡调度仿真

英文题名：Complex Network Data Flow Equilibrium Scheduling Simulation Based on Machine Learning

作者：陈建彪[1]

第一作者：陈建彪

机构：[1]新乡学院计算机与信息工程学院

第一机构：新乡学院计算机与信息工程学院

年份：2019

卷号：36

期号：11

起止页码：264-267

中文期刊名：计算机仿真

外文期刊名：Computer Simulation

收录：CSTPCD;;北大核心:【北大核心2017】；

语种：中文

中文关键词：机器学习;复杂网络;数据流;均衡调度

外文关键词：Machine learning;Complex network;Data flow;Balanced scheduling

摘要：针对传统方法下复杂网络数据流均衡调度中,普遍存在着调度准确率较低、完成时间较长和能量消耗较大等问题,提出了一种基于机器学习的复杂网络数据流均衡调度方法。通过对网络数据流进行分析,获取数据流节点相关参数,利用机器学习中的KNN算法对数据流节点进行分类,数据流节点和存储节点之间不同属性相应的重要程度不同,考虑到数据流任务公平性问题,根据数据流公平性获取数据流负载均衡度,以负载均衡度构建数学模型,考虑最短时间和最优数据流负载构建猫群算法的适应度函数,获取最优调度方案,完成数据流均衡调度。实验结果表明,所提方法调度准确率较高、完成时间较短、能量消耗较低。
Aiming at the problems of low scheduling accuracy, long completion time and high energy consumption in traditional data flow balancing scheduling for complex networks, a machine learning based data flow balancing scheduling method for complex networks is proposed. Through the analysis of network data flow, the related parameters of data flow nodes are obtained. The KNN algorithm in machine learning is used to classify data flow nodes. The importance of different attributes between data flow nodes and storage nodes is different. Considering the fairness of data flow tasks, data flow nodes are obtained according to the fairness of data flow. Flow load balancing degree is used to construct a mathematical model based on load balancing degree. Considering the shortest time and the optimal data flow load, the fitness function of cat swarm algorithm is constructed to obtain the optimal scheduling scheme and complete the data flow balancing scheduling. The experimental results show that the proposed method has higher scheduling accuracy, shorter completion time and lower energy consumption.