Design of deep learning models for the identifications of harmful attack activities in IIOT

IICSs, which connect technical systems with physical systems to deliver services to businesses, have become a new field of study since they are susceptible to various cyber attacks that could compromise their ability to continue supply services to businesses. Potential risks result in lost revenue, and reputational risk for businesses. It is challenging to collect information for use in constructing a smart NIDS that can successfully recognize both current and novel assaults, despite the fact that Network Intrusion Detection Systems (NIDSs) have been recommended as a way of protection from them. This paper presents an anomalies recognition system for IICSs using DL algorithms that may train and evaluate data obtained from TCP/IP package in order to address this problem. A well network datasets NSL-KDD & UNSW-NB15 are used to assess the training courses for deep auto-encoder and deep feed-forward (FF) NN designs. The research findings result that this methodology could be used in actual IICS situations because it has a lower false positive rate & a greater detection accuracy compared to previous methods.