ChGBO_QRNN: chronological gooseneck barnacle-based quantum recurrent neural networks for intrusion detection in IoT-SDN

There has been significant growth in the number of heterogeneous components configured within the Internet of Things (IoT) to allow effective data communication. However, IoT components are resource-constrained, and there are no standardized security procedures for communication within IoT platforms. While various solutions have been developed to secure communication in the IoT ecosystem, the existing approaches are not always reliable enough to address cyber threats and unwanted intrusions originating from IoT endpoints. Thus, this research presents a Chronological Gooseneck Barnacle Optimization-based Quantum Recurrent Neural Network (ChGBO_QRNN) for intrusion detection in SDN-IoT. Here, the proposed system contains three phases, which include a sniffer, a detector, and a sensor phase. In the sniffer phase, the features are extracted from the packets, and the extracted features are utilized for intrusion detection. In the detector phase, intrusion detection is carried out by the Quantum Recurrent Neural Networks (QRNN), and ChGBO is employed for training. In the sensing phase, the boundary of the attack is checked. If any concept drift is detected, the identification of a new attack is done using the proposed ChGBO_QRNN. Furthermore, the ChGBO_QRNN has gained weighted Accuracy, weighted True Positive Rate (TPR), weighted True Negative Rate (TNR), weighted F1-score, and weighted precision as 91.868%, 90.134%, 92.894%, 92.072%, and 92.978%.