부산대학교 도서관

DDoS (Distributed Denial of Service) has become a pressing and challenging threat to the security and integrity of computer networks and information systems. The detection of DDoS attacks is essential before any mitigation approaches can be taken. AI (Artificial Intelligence) and ML (Machine Learning) have been applied to the detection of DDoS attacks with satisfactory achievement. However, new types of attacks emerge as the technology for DDoS attacks keep evolving. This study investigates the impact of a new sort of DDoS attack – adversarial DDoS attack. We synthesize attacking traffic using Wasserstein Generative Adversarial Networks with Gradient Penalty (GPWGAN). Experiment results reveal that the synthesized traffic can penetrate the systems, including Random Forest, k-Nearest Neighbor, and Multi-Layer Perceptron, without being detected. This observation is an alarming and pessimistic wake-up call implying the urgent need for countermeasures to adversarial DDoS attacks.