Deep learning based system for detecting distributed dental of service (DDoS) attacks in the cloud network

Cloud computing is an efficient technology where businesses and users can provide flexible and scalable services by sharing resources. However, this shared technology allows attackers to carry out various cyber threats such as Distributed Denial of Service (DDoS) attacks. Because of its potential impact, these DDoS attacks are very destructive and violent. As a result of these attacks, servers cannot serve users and crash. In addition, it causes damage to the reputation of organizations and financial losses. Therefore, it is placed almost at the top of the hierarchy of attacks faced by institutions and organizations. In this thesis, it is aimed to detect denial of service attacks that occur in cloud computing environments. Due to the difficulties of obtaining a cloud-based data set, the main motivation of the study was to create a data set in a cloud-based system and then to compare the intrusion detection performances of deep learning algorithms using this data set. First, a cloud-based system was created. After creating a network topology in the OpenStack framework, an HTTP flood attack was made. The performance results of these models were examined by using Artificial Neural Network (ANN), Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) models. The LSTM model was the model that gave the best performance result with an accuracy value of 98%.