EAROA-MiXNet: An Efficient Deep Learning-Based Intrusion Detection System for Edge-Based IIoT Environment

The growing popularity of edge-based IIoT has created new security challenges due to the distributed and resource-constrained characteristics of edge devices. For this reason, it is crucial to monitor edge physical systems and detect malicious activity using an effective Intrusion Detection System (IDS). In resource-constrained edge networks, existing IDS frequently encounter problems such as excessive latency, scalability problems, and inefficiency because they either employ traditional methods or are not designed for edge-based IIoT deployment. Therefore, this paper presents a lightweight hybrid IDS for Edge-Based IIoT Security using deep learning techniques. In this study, the Enhanced Artificial Rabbit Optimization Algorithm (EAROA) determines the significance value for every input feature in order to determine the optimal feature subset. It makes it possible to eliminate unwanted and redundant attributes. A lightweight Modified MixNet network then analyzes these characteristics to determine the behavior of IIoT networks. In addition, we have addressed the issue of unbalanced data using a Conditional Tabular Generative Adversarial Network (CTGAN)-based data augmentation technique. Two publicly accessible datasets, the WUSTL-IIoT 2021 and Edge-IIoT datasets, are utilized to analyze the model’s effectiveness. The suggested model produced better results and detected attacks more effectively than current methods. These results demonstrate that the suggested approach is effective at detecting cyberattacks and demonstrates flexibility in identifying a wide range of cyberattacks in actual IIoT scenarios.