A Novel Framework for Adversarial DoS Attack Generation on UAVs

Denial of Service (DoS) attacks have been more significant than ever for unmanned aerial vehicles. Detecting DoS attacks is a significant challenge since there is a lack of enough data about such attacks. In this research, we propose a new framework to generate adversarial DoS attacks that can be used to create more accurate machine learning-based intrusion detection systems. The proposed framework uses Conditional Tabular Generative Adversarial Networks to generate synthetic data. Machine learning-based intrusion detection systems are applied to validate the synthetic data. Experimental results show that the synthetic attack data affects the accuracy of machine learning-based intrusion detections. All machine learning-based intrusion detection models have close accuracy results for both real data and synthetic data. Experimental evaluations prove that the proposed framework generates synthetic DoS attack data that will help to create more accurate machine learning based intrusion detection systems for unmanned aerial vehicles.