Towards DoS/DDoS Attack Detection Using Artificial Neural Networks

The extensive and complex network attacks on smart devices and computers trigger a need for robust and adaptive intrusion detection systems (IDSs). Most of the existing intrusion detection systems are implemented on certain rules where new attacks are not detectable and used outdated datasets such as KDD Cup 99. This paper presents an Artificial Neural Networks (ANN) as a machine learning approach to intrusion detection system for Denial of Service Attacks (DoS) and Distributed Denial of Service Attacks (DDoS). The proposed method used the Bayesian Regularization (BR) backpropagation and scaled conjugate gradient (SCG) descent backpropagation algorithm. The network was trained and tested with subset of the CICID2017 dataset that meets the real world criteria. The features that best characterize each attack and normal network traffic in the input dataset was selected carefully. The result revealed the proposed method successfully detect DoS/DDoS attacks with an accuracy of 99.6% using Bayesian Regularization and 97.7 % in scaled conjugate gradient descent.