Emerging methods in critical infrastructure security and resilience

点击翻译按钮获取中文摘要

With the growing scale and complexity of information and computing systems, security has become one of the cornerstones in the design of every cyber technology. The potential use of Artificial Intelligence (AI) by attackers to counter AI-based cyber-defense is a problem and represents a fundamental question regarding the possibility of an arms race between AI-based defenders and AI-based attackers that needs to be investigated. This arms race could converge in favor of one or the other, or on the contrary create unstable conditions that constitutes in itself a new threat to cybersecurity and our society. In the past, we have developed and published highly intelligent solutions for the planning, design and operation of computer networks running on critical infrastructures. Because of their special nature, improving the security and resilience of emerging cyber systems while ensuring data protection and privacy mandates a high level of innovation in engineering. The purpose of this research is to provide innovative solutions to protect emerging technology-aware critical infrastructures. We plan on enhancing monitoring by developing novel AI-based tools and methods that will automate some of the detection, correlation, and analysis functions while ensuring data protection and privacy. We also aim at modeling sophisticated AI-based cyberattacks targeting the infrastructure to disrupt the control discipline. By exposing the vulnerabilities and understanding the attacks, we plan on designing a process for effectively creating a new class of domain-specific defensive security and privacy solutions. Finally, this will enable us to explore the resilience of the cybersecurity solutions in an attempt to model and understand the potential AI against AI arms race in cybersecurity of critical infrastructures. Due to the complexity of securing large scale systems running on critical infrastructures, big data analysis, machine learning and analytical tools open up new avenues of protection against cyber attacks. We will experiment with deep learning neural networks and reinforcement learning techniques to implement sophisticated AI-based attacks and defenses. We will also experiment with federated learning particularly federated AI and its architectures to address data access challenges. We will use the blockchain technology to address cybersecurity challenges that require a distributed, secure, privacy-preserving, and immutable record-keeping framework to enable trust during information sharing. The cyber physical systems running our critical infrastructures, and on which countless Canadian businesses depend, are vulnerable to attacks by cyber criminals. The innovation of this research aims at developing novel approaches to counter the threat to our Canadian economy and society posed by cybersecurity threats. Our research has the capacity to lead to a theory of AI co-evolution that will allow us to prevent conditions of instability and conflict.