Özet:

Abstract In the field of Industrial IoT area, It produces enormous volumes of data by utilising the power of sensors. The IIoT does, however, confront considerable obstacles, particularly in the form of cyber-attacks that can jeopardise organisations and interrupt operations. Sensitive information is stolen as a result of these attacks, in addition to causing losses in money and reputation.To address these risks, numerous Network Intrusion Prevention Systems (NIDSs) have been developed to protect IIoT systems. But creating a useful and intelligent NIDS is a challenging endeavour, largely because there aren't many large data sets that can be utilised to design and test such systems.In response to these difficulties, this research proposes a novel deep learning-based intrusion detection technique for IIoT systems. To help identify relevant data derived from TCP/IP packets, a hybrid rule-based feature selection mechanism is included in the proposed system. The solution attempts to increase the precision and effectiveness of intrusion detection in IIoT environments by utilising deep learning methods.In this study, deep learning techniques are employed to offer a novel method for industrial internet of things (IIoT) system intrusion detection. The proposed paradigm combines a Deep Feed Forward Neural Network model (DFFNN) with a hybrid rule-based feature selection strategy to quickly train and assess data obtained from TCP/IP packets. The effectiveness of the technique was evaluated on two well-known network datasets, NSL-KDD and UNSW-NB15. This study demonstrated the potential of the provided technique for classifying network attacks in scenarios of IIoT penetration. The trials used a number of evaluation measures to demonstrate the usefulness of the suggested method for precisely identifying and classifying intrusions within IIoT networks.

Anahtar Kelimeler: