Detection of DDoS Attacks Using Variational Autoencoder-Based Deep Neural Network

Distributed denial of service (DDoS) attacks are one of the most commonly used tools to disrupt web services. DDoS is used by groups of diverse backgrounds with diverse motives. To counter DDoS, machine learning-based detection systems have been developed. Proposed is a variational autoencoder (VAE) based deep neural network (VAE-DNN) classifier that can be trained on an unbalanced dataset without needing feature engineering. A variational autoencoder is a type of deep neural network that learns the underlying distribution of computer network flows and models how the benign and DDoS classes were generated. Because a VAE model learns the distribution of the classes within the dataset, it also learns how to separate them. The variational autoencoder-based classifier can scale to any data size. A deep neural network, quadratic discriminant analysis (QDA), and linear discriminant analysis (LDA) decision boundaries are applied to the latent representation of network traffic to classify the flows. The DNN shows the highest precision and recall of the three classifiers.