Economic Denial of Sustainability (EDoS) attack detection by attention on flow-based in Software Defined Network (SDN)

Abstract

Economic Denial of Sustainability (EDoS) attack, a kind of Denial of Service attack (DoS), is one of the potential threats to cloud computing. Since the characteristics of EDoS attacks such as stealthiness and long-term period make EDoS detection and prevention challenging, various detection mechanisms have been proposed. However, they mainly focused on utilizing the features of EDoS attack traffic to enhance detection performance and overlooked the environment where the defense mechanisms are running over. We realized exiting proposals have limitations, such as unnecessary overhead or performance degradation since they do not exploit the features provided in an SDN-based cloud. To overcome the limitations, we propose an efficient EDoS detection system optimized for SDN-based clouds. It applies the attention technique to calculate the attention scores of the flow network extracted from the SDN controller to detect the EDoS attack. The performance evaluation results show that it achieves better accuracy and processing time than previous work.