Detecting DDoS Attacks Through AI driven SDN Intrusion Detection System
Francesco Salatino, Mattia Giovanni Spina, Mauro Tropea, Floriano De Rango
Abstract
In this paper, an AI-driven Intrusion Detection System (IDS) in an SDN environment is proposed exploiting the holist view of the SDN controller of the network. In order to detect malicious traffic, the proposed system makes use of Machine and Deep Learning (ML/DL) techniques and, based on a well-known network dataset called CSE-CIC-IDS2018, it trains Decision Tree (DT), Random Forest (RF), and Deep Neural Network (DNN) models. The main task of the proposal is a deep analysis of the network traffic features using two techniques, Pearson's correlation coefficient and the Mean Decrease Impurity (MDI), in order to find the most important features and reduce the dataset size and model complexity guaranteeing optimal system performance in terms of server's response time.