Research And Implementation Of Web Attack Detection Technology

Along with the widespread use of the Web site,in parallel with the convenience,Web server attacks are increasing gradually,make the information faces more security threat.Web security has become a emphasis research direction in information security.Relying on a firewall is difficult to ensure the Web server security,Web attack detection model research becomes more and more important.Traditional Web attack detection model not only exhausts a lot of resources,and is limit to variant Web attack,the real-time performance is poor and false positive rate is high.To solve this problem,this paper analyzes and summarizes the most popular Web attack methods,as well as the current Web attack detection techniques.Aim at the disadvantages in the existing Web attack detection technology,this paper proposed a Web attack detection model based on SVM and naive Bayesian.This model can get HTTP request data through the analysis of network traffic,and obtain Web attack signature from the analysis of this data,and use them to build eigenvector.Through learning binary SVM classifier model,the model can classify the attack request and the normal request.And then establish a Naive Bayes model based on the study of attack requests data,to process a further classifications on the attack request data from SVM classifiers.Finally we design a Web attack detection system,by mirroring traffic which flowing to the Web server to the detection server,and then use the analytical program to detect the attack traffic.This paper elaborates the detailed process of each module of this model,the database entity relationship,interface design and so on.Finally the Web attack detection model is implemented,and test the functions of each module of the system by building a test environment.After collecting data for verification and comparison,this system can detect Web attacks effectively,and provides security guarantee to Web site.