QBCMVT: An effective quantum based coati-mobilevit model for intrusion detection in IIoT

Abstract

The widespread usage of the Industrial Internet of Things (IIoT) in industrial environments has increased the possibilities of security risks and attacks. To prevent the network from attacks, intrusion detection systems (IDS) are developed to identify hostile activity inside the network flow, in which deep learning and machine learning play a significant role. However, most of the IDS solutions suffer from large feature dimensions, models based on outdated datasets, a lack of attention to the issue of imbalanced datasets, and a lack of comprehensiveness about the types of attacks. As a result, this study introduces a Quantum-based Coati-MobileViT (QBCMVT) model intended for IIoT network analysis and malicious activity detection. In this work, to obtain the best feature subset, the Quantum based Coati optimization algorithm (QCOA) calculates the significance value for each input feature. This allows it to remove duplicate and disruptive characteristics. These features are then processed by a deep learning network based on MobileViT to identify IIoT network behaviour. Moreover, we have utilized a Tabular Generative Adversarial Network (TGAN) based data augmentation method to address the problem of data imbalance. The effectiveness of the QBCMVT framework is evaluated by the Edge-IIoT and WUSTL-IIoT-2021 datasets. The improved performance and efficacy of the suggested model are further illustrated by a thorough comparison with existing ML and DL models and with pertinent studies.