A DDoS Attack Detection System: Applying A Hybrid Genetic Algorithm to Optimal Feature Subset Selection
Abid Saber, Moncef Abbas, Belkacem Fergani · 2020
The rapid evolution in technology is a great challenge for network security against computer threats. Indeed, distributed denial of service (DDoS) attacks aim to deplete or even cripple target networks with malicious traffic. However, before they can be dealt with, these attacks must be identified through real-time analysis of the NetFlow sent by the routers. A large amount of flow during attacks requires the design of a standalone detector with high capacity to support this load and capable of processing traffic in real-time but with low computation time. For the same purpose, detectors based on machine learning suffer from being uncompetitive because they produce many false positives and above all require a lot of computing resources. In order to overcome these problems, in this article, we propose DDoS-Detector, a new identification and detection system, we identify the most relevant features of malicious traffic and develop a suitable concept for real-time DDoS detection.