Electromagnetic Physically-Unclonable Functions Generated by Graphene Radio-Frequency Circuits

Rapid advances in wireless technologies have led to a myriad of internet-connected devices, which sense the surroundings and share sensor data along with the identifications of these devices, the so-called Internet-of-Things (IoTs). However, these smart wireless devices with currently affordable authentication techniques based on digital memories are vulnerable to various cyber-attacks and device cloning.

The security challenge posed by counterfeit or malicious hardware has attracted worldwide attention due to the potential significant economic loss to society and cyber threats to individuals. This research aims to develop innovative electromagnetic physical unclonable functions (EMPUFs) that can be used as an ultra-lightweight, attack-resilient authentication module for wireless communication and wireless access control.

Inspired by speech recognition using artificial intelligence to track acoustic signatures of billions of different human voices, the EMPUF exploits imperfections or sample-specific noises in the radio-frequency (RF) circuitry of a wireless device to generate encryption keys. In this scenario, wireless devices can be identified using the device-specific uniqueness derived from the arbitrary waveform and polarization of the radio waves they produce.

In this project, new types of RF devices and circuits will be built using nanomaterials, such as graphene with highly random and somewhat reconfigurable electronic properties, to maximize the uniqueness of RF fingerprints. This interdisciplinary research interfacing hardware security, electromagnetics, RF circuits, nanoelectronics, and artificial intelligence will provide graduate, undergraduate, and K-12 students with a multidisciplinary research experience.

The project will combine research, education, and community outreach activities through a series of programs at the University of Illinois, such as the Early Outreach Program and Early Research Scholars Program, to broaden the participation of students in the STEM fields.

This research aims to develop a new class of strong physical unclonable functions (PUFs) for cryptographic key generation and authentication, as the first barrier fending off cyber-attacks in a network consisting of abundant resource-scarce wireless devices. Manufacturing process variations of the complementary metal-oxide-semiconductor (CMOS) technology have been widely used to implement PUFs and true random number generators based on arbiters, ring oscillators, flip-flops, and static random-access memories (SRAM).

However, it remains challenging to apply this concept to low-cost, small-footprint IoTs and smart devices that have limited available power and memory capacity. This project proposes an ultra-lightweight, energy-efficient EMPUF that identifies signal variations in the RF front-end components such as modulators, synthesizers, mixers, and oscillators built using graphene-based devices.

Specifically, the high entropy inherent in graphene field-effect transistors (GFETs) originating from random strains, defects, and dopant fluctuations will be harnessed to realize high-performance EMPUF instances. Moreover, the reconfigurable and resettable electronic properties of GFETs will be exploited to generate a redundantly large challenge-response pairs (CRPs), enabling the practical realization of strong PUFs and beyond-silicon security primitives.

A computationally efficient machine learning algorithm will also be developed to retrieve important RF footprints from background noises. The outcomes of this research are expected to help establish highly versatile anti-counterfeiting solutions, which are urgently needed in many fields, such as wireless access control, safety in telematics infrastructure, RF identification, encrypted wireless communications, and authentication of merchandise, to name a few.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.