网络集群部署约束最大熵模型仿真研究  

Simulation Research on Maximum Entropy Model for Network Cluster Deployment Constraints

作　　者：田时宇 刘岚喆[2] 夏秀云 TIAN Shi-yu;LIU Lan-zhe;XIA Xiu-yun(Hunan University of Information Technology,Hunan Changsha 410151,China;School of Mathematics,Hunan University,Hunan Changsha 410082,China)

机构地区：[1]湖南信息学院,湖南长沙410151 [2]湖南大学数学学院,湖南长沙410082

出　　处：《计算机仿真》2023年第11期274-278,共5页Computer Simulation

基　　金：湖南省2022年普通高等学校教学改革研究重点项目(HNJG-2022-0381)。

摘　　要：由于访问用户数量激增导致网络服务访问速度降低,需部署网络服务集群以提高网络服务可靠性,为此提出基于约束最大熵的网络集群部署数学建模方法。考虑相应约束条件下,在拉格朗日方程中引入乘子,将最大熵函数与阶梯函数结合,测量随机变量最大熵分布,对网络节点部署不确定评定。基于网络节点集群部署分布情况,利用PSO算法优化目标临界值,通过多次迭代消除节点覆盖盲区,将网络节点监测区域划分成多个等份,计算节点间距离,得到节点集群部署的覆盖率。利用PSO权重系数,提高节点部署精度,通过遗传算法调整粒子位置,根据子空间和相邻空间的节点数量的加权值,完成节点集群部署约束数学建模。实验结果表明,所提方法集群部署数学模型效果好,子域划能力强,网络节点有效覆盖率在95%以上。In order to improve the reliability of network services,this paper presented a mathematical modeling method for network cluster deployment based on constrained maximum entropy.Considering the corresponding con-straints,we introduced a multiplier into the Lagrange equation,and combined with maximum entropy function and step function to measure the maximum entropy distribution of random variables and thus evaluate the uncertainty of node deployment.Based on the distribution of node clusters,we used the PSO algorithm to optimize the target critical val-ue,and eliminated the blind spots of node coverage through multiple iterations.Moreover,we divided the monitoring area into multiple equal parts and calculated the distance between nodes,thus obtaining the coverage rate of node cluster deployment.Meanwhile,we used PSO weight coefficients to improve the deployment accuracy,and adjusted particle positions by genetic algorithms.Finally,we completed the mathematical modeling of node cluster deployment based on the weighted values of the number of nodes in subspaces and adjacent spaces.Experimental results show that the proposed method has good performance and strong sub-domain partitioning ability.In addition,the effective cover-age rate of network nodes is more than 95%.

关 键 词：网络服务访问速度 部署网络服务集群 拉格朗日方程 最大熵分布 网络节点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]