Detection of Malicious Requests to Protect Web Applications and DNS Servers Against SQL Injection Using Machine Learning

Abstract

According to the Open Web Application Security Project (OWASP), SQL injection is one of the top ten threats associated with web security issues. This type of attack takes advantage of web application vulnerabilities to cause substantial damages such as financial loss and privacy exposure. In addition, it uses the Domain Name System (DNS) recursion process to extract information stored within the DNS-targeted databases. In this paper, Response Policy Zones (RPZ) technique will be applied to protect DNS systems against data exfiltration. On the other hand, the machine learning algorithm AdaBoost will be used to detect this type of attacks. Our proposed tool will allow Web Application Firewall (WAF) systems to automatically recognize and distinguish between normal and malicious SQL queries. Our results have shown improved accuracy score and reduced false and positives alarm rates, and our tool is tested against well-known SQL injection benchmark datasets.