COMBINING A REAL-TIME DOS ATTACK DETECTION WITH CONVOLUTIONAL NEURAL NETWORK

Abstract

Denial-of-service (DoS) is a direct attack method that limits bandwidth and consumes resources on the server. DoS attacks are carried out by sending packets continuously from many different computers to the target server. For example, web servers of high-profile organizations such as banking, commerce and media companies, or government organization, etc,. For the above reason, to control and limit risks to server system resources, in this study, a DoS network attack model on the server has been proposed, and at the same time, a model has been proposed. DoS network attack detection is combined with a Convolutional Neural Network (CNN) model that has been trained on the NSL-KDD dataset previously, the dataset contains 125,972 different attack samples, training and testing results. The investigation accuracy is 99.331%, 99.186% respectively. After detecting the attack, the system is warned and an email is sent to the system administrator.