Applications of deep learning algorithms for Supervisory Control and Data Acquisition intrusion detection system

Abstract

Vulnerabilities in the Industrial Control Systems (ICSs) and Supervisory Control and Data Acquisition (SCADA) systems are constantly increasing as these systems incorporate innovative technologies such as the Internet of Things (IoT). As a result of these advancements, the SCADA system became more efficient, simpler to operate, but more exposed to cyber-attacks. A well-planned cyber-attack against SCADA systems can have catastrophic consequences, including physical property damage and even fatalities. To secure these critical infrastructures, security measures should be examined and implemented. These methods could be hardware-based, such as Intrusion Detection Systems (IDS), software-based, or managerial-based. In this paper, we have examined and presented the most recent research on developing robust IDSs using Deep Learning (DL) algorithms, including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Stacked Autoencoders (SAE), and Deep Belief Networks (DBN). For each algorithm, prior works have been identified, examined, and described based on their conceptual similarities. A comparison between different IDS-DL models is provided based on their performance metrics. Because data is such a crucial component of the training and evaluation of IDS-DL models, some of the most utilized network datasets in DL are discussed. The challenges facing DL applications in IDS development are also discussed, as well as future research direction and recommendations.