Search

The IoT networks are customized to work under various smart environments, utilizing diverse sensors. However, they are vulnerable to several cyberattacks because of their finite resources and limited computing capabilities. Intruders can easily hack the data transmitted across the end nodes. Therefore, it is crucial to protect the privacy of the IoT network against adversarial attacks. For real-time insights from IoT networks, an appropriate networking protocol is essential. In previous studies, several routing protocols have been formulated for intrusion detection with limited performance. Even the application of advanced machine learning (ML) has shown relatively lower accuracy with an increased error rate. To address these issues, an innovative intrusion detection system (IDS) for IoT networks based on medium access control (MAC) protocols with an improved efficient shuffle bidirectional COOT channel attention network (IEsBCCA-Net) method has been proposed. The suggested system uses the MAC protocol to securely transmit the data with low energy consumption and delay, which provides increased throughput. The collected data is stored in the BoT-IoT dataset, which is imbalanced and balanced using the synthetic minority over-sampling technique (SMOTE) and random under-sampling (RUS) techniques. The data is pre-processed using the reformed histogram equalization method, and the optimal features are selected using the modified honey badger algorithm (MHBA). Finally, the intrusions are classified using the IEsBCCA-Net approach, which provides outstanding accuracy compared to state-of-the-art (SOTA) methods.