Research On SQL Injection Attack Detection Method

Nowadays the Internet is closely related to our daily life.We are enjoying the quality of service the provided by Internet at the same time,but also suffering from the multifarious threat of network security.Among the many threats,SQL Injection attacks are ranked in the first place.Now,the existing SQL Injection detection solutions are Static Analysis,Parameterized Query,Dynamic Analysis,Parameter Filtering,Intrusion Detection System and so on.Nevertheless,these methods have a little bit defects.In order to further improve the ability of SQL Injection attack detection.This paper studies the SQL Injection attack detection problem,briefly summarizes the current research status of this problem and analyzes the principle,types and methods of SQL Injection.From the three aspects of SQL Injection behavior and SQL sentence characteristics and running characteristics,we carry out research on SQL Injection attack detection.Finally,we propose two different methods to detect SQL Injection attacks.(1)Proposing a SQL Injection attack detection method based on user behavior analysis: Studying the behavior of normal user and SQL Injection attacker,and analysing the status of Web applications under normal conditions or under different attacks,we extract the behavioral characteristics that can identify SQL Injection attacks and transform them into feature vectors.We build a model to detect SQL Injection attacks by clustering analysis method and algorithm.(2)Proposing a SQL Injection attack detection method based on DNN: We build a URL-SQL mapping model and analyse the sentence and the response of the executed SQL statements.We obtain the ability to identify the SQL statement authigenic structure and the running result feature through reasonable feature selection and eigenvector synthesis.Finally,building a reasonable detection model to detect SQL Injection by DNN(deep neural network)Experiments show that the proposed two methods can effectively detect SQL Injection attacks,and there is further research value in the future.