Classification of DoS/distributed DoS threats in software defined networks using advanced deep belief network-long short term memory architecture

Abstract

With the evolution of telecommunication core and access networks, the next generation networks leverages software defined networks (SDN) to provide flexi bility, scalability and centralized control. Denial of service (DoS)/distributed DoS (DDoS) attacks have been a major threat to next generation networks especially to the centralized architecture of SDNs. The ever-changing and dynamic nature of the DoS/DDoS attacks makes it challenging to detect and resolve them. The existing models to handle DoS/DDoS attacks often suffer from false positive rates and adaptability. In order to solve these problems, this study aims to create and apply sophisticated deep learning framework namely adversarial DBN-LSTM to accurately detect and classify various DoS/DDoS attack types. The proposed adversarial DBN-LSTM model is based on the generative adversarial networks. The proposed model uses generator to generate the adversarial attack and discrim inator to detect the attacks. The adversarial DBN-LSTM model is evaluated using a dataset specifically generated in a Mininet-based SDN controller environment to ensure relevance and practical applicability. The performance of the adver sarial DBN-LSTM is compared with other prevalent models. The adversarial DBN-LSTMmodelachieves accuracy about 99.4%. The proposed work achieves a breakthrough in identifying and preventing DoS/DDoS threats in relation to SDNenvironment.