The Classification of DDoS Attacks Using Deep Learning Techniques

Distributed Denial of Service (DDoS) is a wellknown attack with the power of damage. The process of DDoS is to disrupt the normal traffic of a targeted server by overwhelming it with a flood of Internet traffic. This action affects the legitimate users inaccessible to the resources. Moreover, many businesses today have been faced with DDoS attacks derogation that not only take the physical damage of resources, but also cause a massive financial damage. The idea of this research is to find good ways to detect and classify DDoS attacks that aim to avoid the cause of failure due to network attacks by using deep learning techniques. Therefore, our proposed models based on deep neural networks have been provided to perform the capability in the multiclass classification of DDoS. CICDDoS2019 is a new taxonomy of the DDoS attacks dataset that has been utilized as the reference of this framework. Two proposed models have been implemented with the simple DNN structure and the Convolutional autoencoder. The highest accuracies obtained from the proposed models are high up to 87% and 91.9%, respectively. Overall results showed that the proposed networks display the satisfying outcome with the high accuracy, precision, recall, and F1-score. The comparison of the proposed models with other machine learning algorithms, namely, Logistic Regression, and Naïve Bayes are also indicated in this research and the results point out the outperforming efficiency of our proposed models.