Research On Detection Technology Of Network Abnormal Behavior

Recent years,with the rapid development of network technology,network has been an inseparable part of people's daily life.Internet does bring a lot of convenience to users,but attack towards internet becomes much more than before.Although many organizations and governmental corporations have established relatively secure protection mechanism,the means of attack become various and consequences are much more serious.Under this circumstance,detection and research on internet abnormal behaviors have been gradually developed.After researching on several currently mature detective techniques of Internet abnormal behaviors,we find that these techniques are still one-side,detection focuses on users' behaviors,not comprehensively analyzing all Internet behavior modes.Moreover,users are easily affected by surroundings,behaviors are unstable,which will disturb the result of detection.This paper introduces two methods for detecting abnormal network behavior:1.In the network traffic anomaly behavior detection although effective but low efficiency of the traditional K-means clustering algorithm,this paper improves the original algorithm by adding a factor to improve the detection efficiency of the algorithm in the algorithm;2.This paper proposes the use of flexible conditional random field model(CRF)to detect the abnormal behavior of network protocol,compared with the traditional method,this method can adapt to various situations that may occur in the network learning and get higher detection efficiency.Finally,after the establishment of the model and the computing system,an application program is designed to detect and analyze the real network data.The results show that these two algorithms have achieved better results than the traditional methods.