Publication Date

Spring 5-28-2020

Document Type

Final Report

Embargo Date

5-28-2021

Abstract

Abstract— Physically unclonable functions leverage process variation in the manufacture of silicon chips and circuit boards to map inputs to outputs in an irreversible and unpredictable but consistent manner. They have many applications as security primitives: they can serve as truly random number generators, create secret keys, and fingerprint specific chips. These primitive functions can then be used to secure confidential information and regulate access to private resources. Current approaches tend to utilize variation in the production of silicon dies as the source of variability in their function’s output. We present a PUF that leverages variation in the entire circuit board’s power delivery network to produce an output to a given input. This allows us to uniquely identify entire boards rather than just individual chips. Because the PUF’s output also depends on the board’s PDN this approach allows us to detect hardware Trojans, malicious chips placed onto circuit boards that carry out hardware-level attacks. These chips’ presence subtly changes the impedance of the PDN allowing us to detect them. The PUF itself as well as its potential to identify hardware Trojans are evaluated here using several different copies of an FPGA development board.

Author's School

McKelvey School of Engineering

Author's Department

Electrical and Systems Engineering

Class Name

Electrical and Systems Engineering Capstone Design Project

Available for download on Friday, May 28, 2021

Share

COinS