Research On Key Technologies Of Web Site Fingerprint Identification Based On Traffic Features

While the network is developing rapidly,Network security issues are also getting worse,brings a lot of inconvenience to users' online behavior,this poses a huge threat to users' privacy and property.The WEB website fingerprint analysis can identify the websites visited by the users based on the traffic generated by the users' visits,so as to monitor and protect the users'online behavior,and lay the foundation for building a civilized and healthy network environment.At present,the research on fingerprint recognition of WEB websites is still less on the identification of websites of different origins,and the accuracy of the research on fingerprint recognition of encrypted traffic needs to be improved.Therefore,in order to solve the problems existing in the fingerprint analysis technology of WEB websites,this article studies the traffic characteristics of different website types and encrypted websites,and the method of feature selection.First of all,in view of the current research on fingerprint recognition of different types of websites,there is no more suitable feature and feature selection method for fingerprint recognition of different types of websites.This paper presents a feedback-based plaintext web site feature fingerprint recognition technology.By studying the traffic characteristics between different types of websites,cluster analysis is introduced into the process of feature selection,and a feature selection model combining cluster analysis is proposed.The accuracy,recall,accuracy,and modeling time are compared with fingerprint recognition using original features and features selected based on SFFS and SBFS algorithms to verify that the proposed feature selection model combining cluster analysis can be used in basic Colleagues who maintain the accuracy of fingerprint identification on WEB sites,improve the efficiency of identification.Secondly,in view of the problems that encrypted WEB traffic can extract fewer features,and the accuracy of fingerprint recognition of encrypted WEB sites needs to be improved,a feature of fingerprint feature selection for encrypted WEB sites based on automatic feature engineering is proposed.By studying the difference between the characteristics of encrypted WEB website traffic and plain text WEB website traffic,automatic feature engineering is introduced,combined with PCA algorithm,and a fingerprint recognition model combining automatic feature engineering and PCA algorithm is proposed.Finally,the influence of the training data and the amount on the fingerprint recognition effect of the website was analyzed experimentally,and the accuracy of the fingerprint recognition model combining automatic feature engineering and PCA algorithm was verified.