SQL injection attack detection method using expectation criterion

SQL Injection attack is a kind of attack to a web application that accesses the database of the web application illegitimate. Along with the increasing use of web applications, the database where stores much sensitive information became more and more valuable and vulnerable. Eventually, SQL Injection attack has become rank one in top ten vulnerabilities as specified by Open Web Application Security Project (OWASP). In the other hand, although there was proposed a lot of methods to address the SQL injection attack, the current approaches almost have the limitation to detect full scope of the attack. What is more, the approaches have high precision in detecting pre-existing attacks though, but cannot detect unknown attacks. In this paper, we present an expectation-based solution to address SQL injection attack. Our proposal mainly has two phases. In the first phase, we calculate the occurrence probability of the SQL injection attack special characters in attack dataset and typical dataset respectively, and in the second phase we detect SQL injection attack base on expectation calculating take advantage of the computed occurrence probability.