SAINT-IIOT: Elk herd optimized deep learning model for efficient anomaly detection in the IIoT

Abstract

Industrial Internet of Things (IIoT) is an innovative technology that may mitigate manufacturing costs, increase production efficiency, and foster the growth of industrial intelligence. IIoT applications face security and privacy risks as a result of IIoT device abnormalities reveal sensitive information with high authenticity and validity. To address these issues, a novel cascaded Stacked Autoencoder INtegrated aTtention CNN-BiGRU for IIoT (SAINT-IIoT) model has been proposed in this paper to improve the real-time detection of cyber threats in IIoT environments. The proposed methodology employs an Elk Herd Optimization (EHO) algorithm for effectively selecting the features, which address the issue of irrelevant and noisy features. The Deep Learning (DL) technique is used for real-time anomaly classification to handle complex, nonlinear, and time-dependent attack patterns that traditional models often fail to identify. The accuracy of the suggested framework is 7.04%, 12.11%, and 3.26% higher than the existing techniques including DRL-GAN, AIm-ADS, and EPOA-EVAD.