An Entropy-based DDoS attack Detection and Classification with Hierarchical Temporal Memory

Detecting real-time DDoS attacks is a big challenge for network security. This paper proposes a hybrid machine learning model for the detection and classification of
$D$
DoS attacks. The system consists of a real-time detecting module capable of processing Entropy-based features. In addition, the classification module, based on the Hierarchical Temporal Memory and KNN classifier, is capable of mining changes in Entropy features for the classification of different types of DDoS attacks. Furthermore, it has the incremental learning capability to learn new traffic behavior and recognize new types of attacks. Finally, the simulation is conducted based on the CICDDoS 2019 dataset. As a result, the proposed system can successfully identify different types of attacks with high accuracy and precision.