电信科学 ›› 2019, Vol. 35 ›› Issue (8): 158-164.doi: 10.11959/j.issn.1000-0801.2019087

• 运营技术广角 • 上一篇    下一篇

基于决策树的EPG体验异常原因定位

谭晓敏,方艾,梁冰,杨豪杰   

  1. 中国电信股份有限公司广州研究院,广东 广州 510630
  • 修回日期:2019-04-23 出版日期:2019-08-20 发布日期:2019-08-24
  • 作者简介:谭晓敏(1992- ),女,中国电信股份有限公司广州研究院工程师,主要研究方向为移动互联网大数据应用。|方艾(1981- ),男,中国电信股份有限公司广州研究院工程师,主要研究方向为移动互联网与大数据应用、机器学习应用。|梁冰(1978- ),女,中国电信股份有限公司广州研究院工程师,主要研究方向为移动互联网应用平台研发。|杨豪杰(1982- ),男,中国电信股份有限公司广州研究院工程师,主要研究方向为移动互联网应用平台及智能网络软件研发。

Locating causes of abnormality of EPG experience based on decision tree

Xiaomin TAN,Ai FANG,Bing LIANG,Haojie YANG   

  1. Guangzhou Research Institute of China Telecom Co.,Ltd.,Guangzhou 510630,China
  • Revised:2019-04-23 Online:2019-08-20 Published:2019-08-24

摘要:

用户体验的提升一直是IPTV的运维重点。然而,运营商IPTV组网相对复杂,涉及机顶盒、接入网、EPG页面、EPG服务器、媒体服务器等多个环节和调度链路,一旦发生异常,运维人员需排查各环节的监控指标,凭经验定位引起异常的原因。该运维模式专业性强、效率低下,无法快速止损。为解决上述运维痛点,基于大数据技术,采用决策树对EPG体验数据建模,根据决策树模型的决策规则,快速定位异常原因。现网中,决策树模型能够准确快速定位EPG体验的异常原因,为辅助运维人员指明排查方向,提高运维效率。

关键词: IPTV, 决策树, 大数据, 智能运维

Abstract:

Improving user experience has always been the focus of IPTV operation.However,IPTV network is relatively complex,including multiple scheduling links such as set-top box,access network,EPG page,EPG server and media server.Once the abnormality occurs,monitoring indicators of each link will be checked to locate the cause of abnormality by human experience.The operation mode relies on professionals,which is inefficient and unable to stop losses quickly.Thus,to solve the problem above,based on big data technology,decision tree was implied to locate the cause of abnormality according to the classification rule of the tree model.Decision tree model has been used in actual networks of IPTV and helps to find out the cause of abnormality,which improves the efficiency of operation.

Key words: IPTV, decision tree, big data, AIOps

中图分类号: 

No Suggested Reading articles found!