Research And Implement Of Enterprise Lan Traffic Anomaly Detection Based On Data Mining

Network has achieved a rapid development in the world since its birthdue to its irreplaceable advantages. It is now applied in many aspects of socialactivities such as production, life, service and education etc, staying with useverywhere but invisible, like the air and water. While brining people manyconveniences, it also deeply influences our behavior habits.TCP/IP protocol cannot be called a perfect product. In the networkenvironment filling everywhere with the viruses and hackers, TCP/IPprotocol’s vulnerability makes no enterprise can luckily escape from variousnetwork attacks. While so long as attacked, the more developed network theenterprise owns, the more severe loss it will get. Therefore, protecting theenterprise’s network has become the primary task of each networkmanagement staff. The research on the network protection technology is alsoprogressing rapidly while no sign of weakening.Under the above mentioned environment, this thesis presents a researchon the network protection. Focus on the analysis and detection of enterprise’s abnormal network flux using data mining technology.Base on the analysis of enterprise’s abnormal network flux due to thevirus infection, hacker attack and equipment failure etc. This thesis researchesthe choice of abnormal network flux attribute, puts forward the method thatuses statistics and cluster analysis to detect gradely the abnormal network flux,then classifies it using abnormity model. Based upon this, this thesis designs adetection and classification system of enterprise’s abnormal growth ofnetwork flux. This system can monitor the network, detect and identify theabnormal network flux, and help the automatic adjustment to the network.Then, we test the system under the environment of enterprise real network,the accuracy about abnormal network flux detection is up to90%.