Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

ORCID

https://orcid.org/0000-0002-8230-6876

Document Type

Open Access Thesis

Embargo Period

5-3-2020

Degree Program

Electrical & Computer Engineering

Degree Type

Master of Science in Electrical and Computer Engineering (M.S.E.C.E.)

Year Degree Awarded

2020

Month Degree Awarded

May

Abstract

The power side-channel attack, which allows an attacker to derive secret information from power traces, continues to be a major vulnerability in many critical systems. Numerous countermeasures have been proposed since its discovery as a serious vulnerability, including both hardware and software implementations. Each countermeasure has its own drawback, with some of the highly effective countermeasures incurring large overhead in area and power. In addition, many countermeasures are quite invasive to the design process, requiring modification of the design and therefore additional validation and testing to ensure its accuracy. Less invasive countermeasures that do not require directly modifying the system do exist but often offer less protection.

This thesis analyzes two non-invasive countermeasures and examines ways to maximize the protection offered by them while incurring the least amount of overhead. These two countermeasures are called clock phase noise (CPN) and voltage noise (VN), and are placed on the same FPGA as an AES encryption module that we are trying to protect. We test these designs against a highly effective algorithm called correlation power analysis (CPA) and a preprocessing technique called the sliding window attack (SW).

We found that the combined effects of the two countermeasures was greater than the impact of either countermeasure when used independently, and published a paper in the 2019 IEEE 30th International Conference on Application-specific Systems, Architectures and Processors (ASAP) on our findings. We found that our best combined countermeasure protected about 76% of the maximum amount of traces that a well-known but invasive competitor, wave dynamic differential logic (WDDL), could with only about 41% of the area and 78% of the power. However, the sliding window attack significantly reduced the amount of protection our combined countermeasure could offer to only 11% of that offered by WDDL. Since then, we updated our methodology and made some adjustments to VN and CPN. Our CPN countermeasure greatly improved, and therefore so did our combined countermeasure, which on average protected up to about 90% of the maximum amount of traces that WDDL could with only about 43% of the area and about 60% of the power. This is remarkable because these results are after the sliding window attack, meaning that our post-proposal countermeasures protect almost as well as WDDL while requiring only about half of the resources.

First Advisor

Wayne Burleson

Second Advisor

Daniel Holcomb

Third Advisor

Russell Tessier

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS