5G时代基于机器学习算法的云端业务监测研究
Research on Cloud Service Monitoring Based on Machine Learning Algorithm in 5G Era
徐敏 1郁磊1
作者信息
- 1. 中国移动通信集团广东有限公司广州分公司,广东 广州 510000
- 折叠
摘要
全网dpi已识别的app业务数目高达2万多个,对2万多个业务进行全量异常监控和溯源分析十分耗费资源和人力,往往只针对部分重点小类业务进行保障、监测以及分析,这会导致其它小类业务无法及时发现异常和溯源分析,影响用户上网感知和导致投诉发生.基于此,我们提出了基于机器学习拉依达准则和T值排序算法的云端业务异常监测方法,经试验,本研究能提升监测和溯源效率97％以上,有效挖掘5G云端质差业务.
Abstract
The number of app services identified by dpi in the whole network is up to more than 20,000.It is very costly to monitor and trace the full amount of abnormalities and trace the ori-gin of more than 20,000 services.It is often only for some key small categories of services to guarantee,monitor and analyze,which will lead to the failure of other small categories of serv-ices to detect abnormalities and trace the origin in time,affecting users'online perception and causing complaints.Based on this,we proposed a cloud service abnormal monitoring method based on machine learning Layard criteria and T-value sorting algorithm.After experiments,this study can improve the monitoring and trace the origin efficiency by more than 97％,and effec-tively excavate 5G cloud poor quality services.
关键词
机器学习/云端业务/上网感知Key words
Machine learning/Cloud Service/Internet awareness引用本文复制引用
出版年
2024
NETL
NSTL
万方数据