Secure SDN-IoT Framework for DDoS Attack Detection Using Deep Learning and Counter Based Approach

The IoT network is unique due to heterogeneous IoT nodes and resource-constrained devices; the approach for securing IoT networks needs to be different from the security measures implemented for traditional network communication. In IoT networks, various security vulnerabilities are exploited by an attacker to generate a variety of DDoS attacks. In this paper, the authors propose a unique approach for securing IoT networks using an SDN-enabled framework that incorporates a dynamic counter-based approach and deep learning models. The aim is to detect and mitigate various security vulnerabilities that attackers exploit to generate DDoS attacks in IoT networks. Specifically, the proposed framework is tested using the CICDDoS2019 dataset to identify reflection attacks and exploitation attacks in TCP, UDP, and ICMP. The framework is also analyzed by varying network parameters such as the number of IoT attack nodes and payload to measure the performance of the SDN controller workload, CPU utilization, and attack detection time. The experimental results demonstrate that the proposed framework can efficiently detect and mitigate DDoS attacks while utilizing CPU resources effectively and in a shorter time compared to existing approaches.