Towards Quantum Artificial Intelligence Electromagnetic Prediction Models for Ladder Logic Bombs and Faults in Programmable Logic Controllers

Abstract

Industrial control systems rely on Programmable Logic Controllers (PLCs) for the communication with sensors and actuators. On the other hand, vulnerabilities in Ladder logic and in turn PLCs being vulnerable to malware written in ladder logic or any other IEC 61131-3-compliant language is a new far more dangerous challenge in industrial control systems. Attackers install malware into a PLC's existing control logic to either modify the behavior continuously or wait for specific triggers. We consider Ladder Logic Bombs (LLBs) which are an extension of attacks like Stuxnet. These include stealthy LLBs, which are difficult to discover by humans manually evaluating PLC programs. We propose a monitoring technology for PLC's via Hybrid Quantum-Classical modelling of electromagnetic emissions from the Programmable Logic Controllers even in the case of stealthy LLBs. Our goal is to discover Ladder logic bombs from typical faults utilizing Electromagnetic (EM) spectrum domain-based control system processor fault analysis. We investigated various different types of ladder logic bomb activation, triggering modes, and payloads and discovered that our hybrid classical-quantum deep learning model can operate decoupled from the network and is very accurate at detecting logic layer vulnerabilities and faults.