Year of Publication

2021

Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation

College

Engineering

Department/School/Program

Electrical and Computer Engineering

First Advisor

Dr. Himanshu Thapliyal

Abstract

The last few decades have seen a large proliferation in the prevalence of cyber-physical systems. Although cyber-physical systems can offer numerous advantages to society, their large scale adoption does not come without risks. Internet of Things (IoT) devices can be considered a significant component within cyber-physical systems. They can provide network communication in addition to controlling the various sensors and actuators that exist within the larger cyber-physical system. The adoption of IoT features can also provide attackers with new potential avenues to access and exploit a system's vulnerabilities. Previously, existing systems could more or less be considered a closed system with few potential points of access for attackers. Security was thus not typically a core consideration when these systems were originally designed. The cumulative effect is that these systems are now vulnerable to new security risks without having native security countermeasures that can easily address these vulnerabilities. Even just adding standard security features to these systems is itself not a simple task. The devices that make up these systems tend to have strict resource constraints in the form of power consumption and processing power. In this dissertation, we explore how security devices known as Physically Unclonable Functions (PUFs) could be used to address these concerns.

PUFs are a class of circuits that are unique and unclonable due to inherent variations caused by the device manufacturing process. We can take advantage of these PUF properties by using the outputs of PUFs to generate secret keys or pseudonyms that are similarly unique and unclonable. Existing PUF designs are commonly based around transistor level variations in a special purpose integrated circuit (IC). Integrating these designs within a system would still require additional hardware along with system modification to interact with the device. We address these concerns by proposing a novel PUF design methodology for the creation of PUFs whose integration within these systems would minimize the cost of redesigning the system by reducing the need to add additional hardware. This goal is achieved by creating PUF designs from components that may already exist within these systems.

A PUF designed from existing components creates the possibility of adding a PUF (and thus security features) to the system without actually adding any additional hardware. This could allow PUFs to become a more attractive security option for integration with resource constrained devices. Our proposed approach specifically targets sensors and energy harvesting devices since they can provide core functions within cyber-physical systems such as power generation and sensing capabilities. These components are known to exhibit variations due to the manufacturing process and could thus be utilized to design a PUF. Our first contribution is the proposal of a novel PUF design methodology based on using components which are already commonly found within cyber-physical systems. The proposed methodology uses eight sensors or energy harvesting devices along with a microcontroller.

It is unlikely that single type of sensor or energy harvester will exist in all possible cyber-physical systems. Therefore, it is important to create a range of designs in order to reach a greater portion of cyber-physical systems. The second contribution of this work is the design of a PUF based on piezo sensors. Our third contribution is the design of a PUF that utilizes thermistor temperature sensors. The fourth contribution of this work is a proposed solar cell based PUF design. Furthermore, as a fifth contribution of this dissertation we evaluate a selection of common solar cell materials to establish which type of solar cell would be best suited to the creation of a PUF based on the operating conditions. The viability of the proposed designs is evaluated through testing in terms of reliability and uniformity. In addition, Monte Carlo simulations are performed to evaluate the uniqueness property of the designs.

For our final contribution we illustrate the security benefits that can be achieved through the adoption of PUFs by cyber-physical systems. For this purpose we chose to highlight vehicles since they are a very popular example of a cyber-physical system and they face unique security challenges which are not readily solvable by standard solutions. Our contribution is the proposal of a novel controller area network (CAN) security framework that is based on PUFs. The framework does not require any changes to the underlying CAN protocol and also minimizes the amount of additional message passing overhead needed for its operation. The proposed framework is a good example of how the cost associated with implementing such a framework could be further reduced through the adoption of our proposed PUF designs. The end result is a method which could introduce security to an inherently insecure system while also making its integration as seamless as possible by attempting to minimize the need for additional hardware.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2021.019

Funding Information

The research in this dissertation was partially supported by grants from Kentucky Science and Engineering Foundation per Grant Agreement KSEF-3998-RDE-020 (2018-2019) and National Science Foundation under Grant No:1738662 (2018-2020).

Share

COinS