SWAF: A Smart Web Application Firewall Based on Convolutional Neural Network
Ines Jemal, Mohamed Amine Haddar, Omar Cheikhrouhou, Adel Mahfoudhi · 2022
Internet network carries a huge stream of HTTP requests and responses between users and servers. Securing server data access is primordial for internet users to enhance confidence in electronic services. Therefore, it is essential to detect and stop malicious HTTP requests arriving on the server quickly and with high accuracy. This paper presents a smart web application firewall (SWAF) based on a convolutional neural network. Using 5-fold cross-validation method, we trained and tested our proposed web application firewall with the CSIC dataset. Our SWAF achieves a high attacks detection rate. It can catch and stop a malicious HTTP request in 2.3ms with an accuracy rate of 99.1%.