基于改进FP-Growth算法的5G跨专业网络根因分析研究  

Root Cause Analysis of 5G Cross-Domain Network Based on Improved FP-Growth Algorithm

作　　者：黄兵明 乔治[2] 黄朝明 马瑞涛[1] 韩莹莹 Huang Bingming;Qiao Zhi;Huang Chaoming;Ma Ruitao;Han Yingying(China Unicom Research Institute,Beijing 100048,China;China United Network Communications Group Corporation Limited,Beijing 100045,China;Ultrapower,Beijing 100012,China)

机构地区：[1]中国联通研究院,北京100048 [2]中国联合网络通信集团有限公司云网运营中心,北京100045 [3]北京神州泰岳软件股份有限公司,北京100012

出　　处：《信息通信技术》2024年第6期65-72,共8页Information and communications Technologies

基　　金：国家自然科学基金资助项目(62032013)。

摘　　要：针对经典关联规则挖掘算法在5G无线、传输、动环等跨专业网络运维中遇到的挑战,通过分析跨专业故障告警数据的业务和拓扑关联特点,提出一套基于关联规则挖掘算法的优化思路和改进方法,通过对干扰告警的过滤、提升小概率发生故障告警的关注度、对时间序列因果关系再挖掘等优化改进算法,提升算法准确率和运算效率。最后通过应用案例效果表明,该方法在大量5G网络跨专业告警中可有效快速完成根因告警判断,极大提高运维人员根因定位分析的效率。Analyzing the challenges of applying association analysis based algorithms to the multiple domains,such as wireless network,transmission network and dynamic environments,of Telecommunication operation and maintenance,considering the characteristics of the data set,it proposes a set of optimization on association analysis based algorithms.Through filtering noise alarm,escalating the attention metric to fault related alarm data set,whose probability might be too low,and applying casual inference algorithm to time series events,it improves the algorithm accuracy and computing efficiency.Through the result posted at the end of this article,it shows that the method can help to accelerate the root cause analysis,improve the efficiency of locating the fault position in 5G network through-cross-domain alarms.

关 键 词：5G网络 关联分析 根因分析 因果模型 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TN929.5[自动化与计算机技术—计算机科学与技术]