Machine Learning for Control Systems Security of Industrial Robots: a Post-covid-19 Overview

Abstract

Abstract The Internet of Things has provided people with a seamless, automated home and industrial experience. The concept is now integrated into more domains like Internet of robotic things (IoRT), Internet of medicine Things (IoMT), etc., to improve domain-specific outcomes. For IoRT, which is the robotics implementation of Internet of Things (IoT), poor network security could cause economic and physical damage to both the networked devices and human users of the network. Also, the tendency for data and privacy breaches becomes more prevalent with an increase in the number of devices in the network. Hence, these identified vulnerabilities are the limiting elements for proper IoRT implementation. Various works have proposed security schemes for ensuring the realization of a secure and efficient IoRT network, but with computational time and complexity limitations. However, machine learning methodologies have shown the best promise for identifying malicious traffic in an IoRT network. This work proposes a security architecture using a Deep Neural Network and an ensemble of Decision Trees. This architecture can be implemented online or offline with minimal trade-offs between resources and efficiency. Also, the proposed machine learning models are compared with other commonly implemented schemes using the IoT-23 Dataset. Experimentation and comparison show that the proposed model and architecture are optimal for the malware detection task and security of a typical IoRT network. These contributions are significant for realizing secure and efficient IoRT networks for the future of industrial automation in this post-COVID era.