AI-Enabled Fast Detection of DDoS and Adversary DDoS Attacks in SDN

Abstract

The importance of network services especially telemedicine services has increased dramatically after the global spread of the Covid-19 virus. Telemedicine services may include live audio and video chats, uploading of images or files etc. in real-time. The efficiency of a network service relies on the dependability and resilience of its foundational infrastructure. The rapid increase of telemedicine services has also resulted in a steady rise in security attack methods, the breadth of their effects, and the ferocity with which they are launched. Distributed denial-of-service (DDoS) attacks have become a major source of worry for Internet service providers. Software Defined Network (SDN), in which the control plane is shifted to a logically centralized location, is widely used for providing large-scale Internet services due to advantages such as programmability, flexibility, etc. But a logically centralized control plane in SDN makes it more vulnerable to DDoS attacks due to a larger attack surface. In this paper, we focus on DDoS attack detection approaches and their adversarial behavior in SDN using the latest Deep Learning techniques. Our technique is able to achieve a detection accuracy of 98.49% for the primary DDoS dataset and 93.19% for the adversary dataset generated by us.