Identify the Cyber Vulnerabilities of Remotely Deployed Small Modular Reactors (SMRs) by Designing an AI-assisted Attack

Abstract

In recent years, nuclear power plants (NPPs) have undergone a significant paradigm shift especially due to the wide adoption of digital instrumentation and control (I&C) systems, along with extensive data communication within different entities of the smart grid. Besides adding additional complexity, this digital evolution has exposed nuclear facilities to increased cyber threats, as evidenced by incidents like the Slammer Worm infection and Stuxnet attack. Additionally, the emergence of Small Modular Reactors (SMRs), the smaller counterpart of NPP, may further amplify the cyber vulnerabilities of the nuclear power industry. SMRs might be more susceptible to cyber threats due to their higher reliance on digital communication mainly to support their remote deployment. In this work, we aim to examine the operational infrastructure of SMRs along with its distinct and distributed I&C systems to provide an overview of potential threats associated with their different operational modes. Beside these, we discuss the potential of ML-based intrusion detection systems (IDSs) to identify the cyber threats in SMR ecosystem and analyze the associated risks of adversarial artificial intelligence (AI) to deceive such ML-based IDSs.