Wrapper Based Linear Discriminant Analysis (LDA) for Intrusion Detection in IIoT

Abstract

The internet has become a part of every human life. Also, various devices that are connected through the internet are increasing. Nowadays, the Industrial Internet of things (IIoT) is an evolutionary technology interconnecting various industries in digital platforms to facilitate their development. Moreover, IIoT is being used in various industrial fields such as logistics, manufacturing, metals and mining, gas and oil, transportation, aviation, and energy utilities. It is mandatory that various industrial fields require highly reliable security and preventive measures against cyber-attacks. Intrusion detection is defined as the detection in the network of security threats targeting privacy information and sensitive data. Intrusion Detection Systems (IDS) have taken an important role in providing security in the field of computer networks. Prevention of intrusion is completely based on the detection functions of the IDS. When an IIoT network expands, it generates a huge volume of data that needs an IDS to detect intrusions and prevent network attacks. Many research works have been done for preventing network attacks. Every day, the challenges and risks associated with intrusion prevention are increasing while their solutions are not properly defined. In this regard, this paper proposes a training process and a wrapper-based feature selection With Direct Linear Discriminant Analysis LDA (WDLDA). The implemented WDLDA results in a rate of detection accuracy (DRA) of 97% and a false positive rate (FPR) of 11% using the Network Security Laboratory-Knowledge Discovery in Databases (NSL-KDD) dataset.