DDOS Anomaly Detection using Generative Adversarial Network

Abstract: Software Defined Networks (SDN) were designed to simplify network management by allowing users to monitor and administer the entire network from a single location. In many data centre network environments today, SDN is widely used.On the other side, emerging technologies can result in a slew of vulnerabilities and threats, which manufacturers are currently dealing with. As the number of devices grows, access and backbone layer control becomes more difficult, as a consequence, it raises security worries about network assaults like Distributed Denial of Service (DDoS). DDoS has been a substantial threat to the Internet, because DDoS traffic seems to be normal traffic, it can cause enormous financial loss to businesses and governments. This paper strongly encourages the use of dadgan, a supervised anomaly detection approach built on Generative Adversarial Networks (GANs). DadGAN is trained with generator and discriminator loss to allow for effective anomaly detection in software-defined networks. To demonstrate the effectiveness and generalisation of our technique, various effective performance indicators such as confusion metrics, precision, recall, and F1 score are used. According to the results, our methodology is capable of detecting anomalies, particularly point anomalies.

Link to the paper: https://ieeexplore.ieee.org/document/9683282