The Key Technology For The Analysis Of Network Security Behavior Based On Granular Computation Theory

With the rapid development of mobile Internet,the network has reached every aspect of life.Any network malfunctions including equipment failures and attack will result in huge impact on life and work.Compared the traditional media,the spread speed of Internet information is faster,which can form public opinion guide in short time and effect the stability of public opinion.National Cyberspace Security Strategy defines the cyberspace security strategy from network communications infrastructure and information security and transmission,on one hand,which can protect the security of network communication infrastructure,and on the other hand which ensure the safety and reliability of network information to avoid misleading the public.Therefore,in order to maintain the security of cyberspace from the two perspectives proposed above,user behavior of IP layer and application layer(especially Web contents)is analyzed to discover the behavior harming cyberspace security and guarantee the normal operation of the network and related applications.Granular computing has been widely concerned by scholars since its incomplete information processing ability has been recognized.It has been applied in the artificial intelligence,deep learning,and information security and so on.Combined with the application background of network security and based on granular computing,this paper researches on the network traffic anomaly detection and the analysis of micro-blog user behavior.This paper includes three parts:(1)Proposed network traffic anomaly detection algorithm based on dynamic Vague setBased on the dynamic Vague set,the frequent factor model and the correlation factor model are proposed to solve the problem that incomplete data flow in network traffic anomaly detection is intractable to be analyzed.The frequent factor model is computed from frequent factor cognition.Combined with basic cognition,the frequent factor model forms the dynamic cognitive Vague set,which is very helpful to deal with the scene depending on the change of time.Correlation factor model is used to improve the accuracy of anomaly recognition.Experimental results show that the proposed algorithm can maintain high recognition accuracy even the data loss is up to 80%.(2)Constructed micro-blog users' behavior based on dynamic cognitionThe dynamic cognitive process is the process of the change of the granular layer,which constructs the granular layer coherence algorithm by the dynamic cognition of attributes.This dynamic cognition has the characteristics of flexibility,and it can be used as the cognitive granular coherence to meet the specific needs.Based on the dynamic cognition,the corresponding granular layer analysis can be obtained,and then the related granular layer can be used for intelligent classification.On this basis,the method is well applied to the construction of the network of micro-blog users' behavior relationship and the rapid classification of micro-blog users.Given the three important behavior attributes,the classification accuracy is not reduced obviously,achieved good classification results.(3)Designed the Network security behavior analysis platform based on Spark technologyIn order to meet the diversity of data acquisition,processing data source dispersion,effective data processing,the platform exploits the online analysis method based on Spark technology,which can obtain the real-time network flow,Netflow data,and Web firewall log data,and the analysis results from off-line data preprocessing and data analyzing.Based on D-S method and Vague set theory,the multi-source data evaluation method is proposed,which can improve the accuracy of alarm effectively.Meanwhile,the platform based on bus is easy to be expanded.