SDN-enabled Deep Learning based Detection Mechanism (DDM) to tackle DDoS attacks in IoTs

Abstract

The intelligence-embedded devices also called the Internet of Things, have heterogeneous characteristics and generate enormous data. To oblige the expansion of smart devices in the age of data, software-defined networking (SDN) has emerged as a promising cost-effective, scalable, versatile solution for IoT services. However, the proliferation and pervasiveness of IoT devices also bring some serious security concerns of cyber threats and attacks. This work presents a real-time intrusion detection and mitigation solution for SDN-enabled IoT infrastructure to provide autonomous security in high-traffic IoT networks in the 5G and beyond future. The suggested approach used an ensemble of Convolution Long short-term memory + Bidirectional Long short-term memory (ConvLSTM2D+BLSTM) at the SDN network layer to automate flow feature extraction and classification. For evolving distributed denial of service (DDoS) attacks, the testing is conducted on the CICIDS2017 dataset. The experimental results showed that the proposed security approach could detect and mitigate threats in real-time with high accuracy of 99.69% rate, 99.93% precision, and 99.65% F1-score by performing 10-fold cross-validation.