Available at: https://digitalcommons.calpoly.edu/theses/2749
Date of Award
6-2023
Degree Name
MS in Computer Science
Department/Program
Computer Science
College
College of Engineering
Advisor
Stephen Beard
Advisor Department
Computer Science
Advisor College
College of Engineering
Abstract
Our modern world relies upon large scale industrial plants, that are manged by Industrial Control Systems (ICS), to provide critical services such as electricity, water, or waste processing. Over time, these systems have been increasingly connecting outdated systems to modern internet technology to provide additional functionality such as real-time monitoring while also enabling operators to conduct modern business analytics. Thus, these plants offer a potential target for attackers. Recently, such attacks on Industrial Control Systems have increased in number with well publicized attacks such as Stuxnet and TRISIS. To protect against such attacks while also modernizing their networks, Industrial Control Systems have adopted new technologies. One such technology is the data diode. Replacing the protection formally offered by air gaps, data didoes allow data, often in the form of metrics, to flow outwards while also physically preventing malicious data from flowing into the critical section of an ICS. This thesis presents a series of escalating attacks and defenses corresponding to different threat models based upon the attacker's level of resources and sophistication. These attacks and defenses take the form of three case studies. In case study one, this thesis identifies the naive scenario in which an attacker inserts a bypass. Secondly, in case study two, this thesis discusses how this can be expanded by taking advantage of the existing network design. Finally, in case study three, this thesis demonstrates an invasive scenario in which an attacker executes a man-in-the-middle attack with a network bypass. For each attack, this thesis also offers some possible methods of defense which may not be present in typical Industrial Control Systems due to the assumptions made about the inherent physical defense of industrial networks.