蜂窝移动网络大数据聚类异常挖掘方法仿真
Simulation of Clustering Anomaly Mining Method in Cellular Mobile Network Big Data
李红艳 1徐寅森 1张子栋2
作者信息
- 1. 商丘学院计算机工程学院,河南 商丘 476000
- 2. 集美大学计算机工程学院,福建 厦门 361021
- 折叠
摘要
蜂窝移动网络可提供话音、数据、视频图像等业务,具有终端移动性、越区切换和跨本地网自动漫游功能,因此上述网络环境中的数据为移动宽带数据流量,数据量巨大,并一直处于自更新状态,导致其聚类过程中易产生异常数据.针对上述问题,提出蜂窝移动网络大数据聚类异常挖掘方法.根据蜂窝移动网络的结构、数据存储结构以及数据特征,对数据属性聚类,并提取异常数据的弱相关特征值.基于以上情况,将提取的弱相关特征值输入至聚类器中,挖掘蜂窝移动网络大数据的异常数据.实验结果表明,蜂窝移动网络大数据量逐渐增多时,研究方法的挖掘准确性仍能保持在98.5%以上,耗时可控制4ms以内,漏检率始终低于 1%,误检率不超过2%.以上所得数据说明该方法的应用可靠性更优.
Abstract
Cellular mobile networks can provide voice,data,video and image services,and have terminal mobility,hand-off function and auto-roaming network.However,the amount of broadband data is huge,and it is al-ways in the state of self-updating,leading to abnormal data.Therefore,a method of mining clustering anomalies of big data in cellular mobile networks was proposed.Based on the structure,data storage and data characteristics of cellular mobile networks,the data attributes were clustered,and then weak relevance features of abnormal data were extracted.On this basis,the extracted feature values were input into the cluster to mine the abnormal data in cellular mobile net-works.Experimental results show that when the big data in cellular mobile networks increases in numbers gradually,the mining accuracy of the proposed method can still remain more than 98.5%,and the time can be controlled within 4ms.Meanwhile,the missed detection rate is always less than 1%,and the false detection rate is not less than 2%.The above data proves the application reliability of the method.
关键词
蜂窝移动网络/大数据/异常挖掘方法/数据聚类/弱相关特征Key words
Cellular mobile network/Big data/Abnormal mining method/Data clustering/Weak correlation fea-ture引用本文复制引用
基金项目
河南省高等学校精品在线开放课程建设项目(教高[2019]671号)
教育部高教司产学合作协同育人项目(202002269002)
出版年
2024
NETL
NSTL
万方数据