Design And Implementation Of Mobile Network Anomaly Detection System Based On Clustering And Correlation Analysis

In recent years,with the continuous development of information technology,mobile Internet has become an indispensable part of people's social life.However,with the explosive growth of the mobile Internet scale,the security threats it faces are also rapidly increasing.Attacks such as malicious code and privacy theft are endless,making it hard to prevent.Mobile Internet security is facing serious challenges.As a technology to identify security threats,anomaly detection has become one of the important means to ensure the security of mobile networks.However,due to the complexity and diversity of mobile network users' behavior,how to build models and detect anomalies in massive amounts of data is still a difficult point.In order to cope with this challenge,this paper applies machine learning to the field of anomaly detection to improve the efficiency of discovering the nature and mode of the attack.Finally,an anomaly detection system based on clustering and association analysis algorithm is designed and implemented.The anomaly detection system mainly includes three modules,and the specific research contents are as follows:1.K-means based abnormal behavior clustering analysis moduleFirstly,an abnormal behavior set is constructed for each user from his or her behavioral data.A weight calculation method for abnormal behaviors and an eigenvalue extraction method for abnormal behavior sets are proposed by using all the behavior sets.Secondly,an improved algorithm is developed,in which we calculate the tightness of all data points and select the initial cluster centers from the data points with high density and low density to improve the clustering effect based on the Kmeans clustering algorithm.Finally,clustering result of the abnormal behavior is got with the input of the eigenvalues of the abnormal behavior set.2.Apriori-based anomaly behavior association analysis moduleFirstly,according to the clustering result of abnormal behavior,an adaptive minimum support calculation method is proposed.The method achieves minimum support by binomial distribution.Then,based on the Apriori correlation analysis algorithm,the mining of association rules between abnormal behaviors is realized.Finally,the results of association rules between abnormal behaviors are output.3.K-means based anomaly detection moduleFirstly,according to the association rules between abnormal behaviors,an eigenvalue extraction method for abnormal behavior is proposed.Then,based on the generated behavior characteristics,a user-oriented eigenvalue extraction method is proposed.Finally,based on K-means clustering algorithm,An improved algorithm is proposed to make anomalous judgments on outliers in the same cluster at each iteration,and finally achieve anomaly detection for the user.