Research On Network Traffic Analysis And Anomaly Detection Based On Alternating Direction Method Of Multipliers

With the rapid development of computer technology and Internet,people enjoy the convenience brought by computer technologies,at the same time,also face the risk of network intrusion and computer crime.Network traffic analysis,as one of the hot topics and the forefront of computer network research,has significant scientific meaning and application value in understanding the network behavior,improving the network performance and protecting the network security.At present,in the computer network operation and management of the vast amounts of traffic data,the urgent need to analyze the complex data processing to ensure the information security of users.At present,in the operation and management of the computer network,a huge amount of network traffic data has been produced.It is urgent to analyze and deal with these complex data in order to ensure the information security of users.In the current study,the method of principal component analysis is widely used in the analysis and detection of network traffic.But it has some existed restriction,which is not entirely accurate and intuitive analysis of the existence of network attack.In addition,a relevant literature has proposed a network traffic matrix decomposition model based on accelerated proximal gradient method,which can effectively detect the abnormal behavior existing in the network.In this paper,a network traffic analysis and anomaly detection model based on alternating direction method of multipliers has been proposed.Using this model,the network traffic is analyzed to determine whether the network is abnormal.In this paper,the C# and Matlab mixed programming is used to implement the network anomaly detection module.Using KDD Cup99 10% data set,the experimental results show that,the proposed ADMM algorithm can effectively detect the network anomaly in the data of large-scale normal traffic,large-scale normal and abnormal traffic and small hybrid traffic.Compared with the acceleration gradient algorithm,this model can reduce the number of algorithm iterations,and improve the computation speed.In addition,the algorithm has been proved to be robust in the detection of network traffic with the abnormal and noise pollution.