多聚类网络异常流量检测方法及应用研究  

Methods and Applications of Abnormal Traffic Detection in Multi-cluster Networks

作　　者：郭春 GUO Chun(Shanxi Police College,Taiyuan Shanxi 030401)

机构地区：[1]山西警察学院,山西太原030401

出　　处：《软件》2024年第3期122-125,共4页Software

摘　　要：随着当前网络服务与应用的增加,网络流量特征边界模糊,网络恶意行为层出不穷,针对网络流量异常的检测及分析成为保证网络安全的首要任务。本文以K-means算法为基础,探究总结网络异常流量检测的方法及应用,首先总结网络异常流量的分类及常见检测方法,其次从初始聚类中心的选取、增加聚类评价函数、其他方法三个角度改进K-means方法,从基于Hadoop和Spark平台下的K-means算法以及“K-means+其他”的集成方法三方面探讨算法的创新性和局限性,做出归纳和总结,并提出解决方案,旨在为网络异常流量检测领域贡献一份力量。With the increase of network services and applications,the boundary of network traffic characteristics is blurred,and network malicious behaviors emerge in an endless stream,and the detection and analysis of network traffic anomalies has become the primary task to ensure network security.Based on the K-means algorithm,this paper explores and summarizes the methods and applications of network anomaly traffic detection,firstly,summarize the classification and common detection methods of network abnormal traffic.Secondly,explore the K-means method improved from three perspectives:selecting initial clustering centers,adding clustering evaluation functions,and other methods.From the innovation and limitations of the K-means algorithm based on Hadoop and Spark platforms,as well as the integration method of"K-means+others",summarize and propose solutions,Intended to contribute to the field of network abnormal traffic detection.

关 键 词：K-means优化 异常检测 HADOOP Spark平台 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]