Privacy-Preserving Intrusion Detection in IIoT Using Deep Learning Approaches

Safety and confidentiality pose significant hurdles when developing IIoT applications, despite the technology's many useful industrial uses. An effective intrusion detection system (IDS) architecture may help resolve these issues. The suggested model searches for cyberattacks utilizing the SSAE method. Improving the SSAE variables using the BSA method of optimization improves the accuracy of the classifier. For choosing characteristics, this framework employs the ISSO-FS method. The feature selection procedure optimizes global performance by using the β - GD method. Utilizing real-world data sets such as NSL-KDD and CIC-DDoS, the effectiveness of the suggested IDS model is tested. Utilizing the CIC-DDoS database, the suggested approach yields comparable results in terms of performance metrics. This chapter provides an overview of the BSA-IDS model's computational construction and performance assessment process, with the goal of detecting cyberattack in IIoT systems. With an accuracy of 95.5%, the suggested BSA-IDS model surpasses all of its competitors, including SVM-IDS, RF-IDS, ADABoost, and Density Peak Clustering-based Deep Belief Network (DPC-DBN). Accordingly, BSA-IDS has shown to be the best in accurately categorizing cases.