Detection of Distributed Denial of Service Attack on Controller in Software Defined Network
Raghul Kannan A, Naveen Sundar G, D. Narmadha · 2024
A network which is said to be controlled based on a centralized controller falls under the category of software-defined network which is intelligent too. Contrary to conventional networks, SDN is easier to monitor and more vulnerable to breaches of security. Early attack susceptibility in SDN is directly dependent on the planes including application plane, data plane, and control plane. The contemporary issues on the plane include DDoS, unauthorized access, DoS, and etc. There are certain techniques consisting of HTTPS, tunneling the connection are available to address the issue in SDN context. When it comes to controller monitoring, a secure coding model is necessary, as generic controller monitoring solutions do not utilize deep learning policies to secure the model during its training phase. Henceforth, Convolution Neural Network for Anomoly Detection(CNN-AD) leverages the CICDDOS-2019 dataset to identify and mitigate DDoS attacks, offering a robust solution that incorporates deep learning policies to enhance the security of SDN controllers during their training phase. CNN-AD method shows 97.51 % of accuracy. This research not only provides a critical advancement in SDN security but also introduces an innovative approach to utilizing deep learning for real-time attack detection and prevention.