# US 2022/0045871 A1 · Correcting physical unclonable function errors based on short integers solutions to lattice problems

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2022/0045871 A1
- Document ID: US20220045871A1
- Status: Published application
- Kind: A1
- Application number: 16/986,127
- Filed: 2020-08-05
- Published: 2022-02-10
- Earliest claimed: 2020-08-05
- Assignees: Analog Devices, Inc.
- Applicants: Analog Devices, Inc.
- Inventors: Chiraag Juvekar
- Web page: https://patentgrep.com/patent/US20220045871A1

## CPC

- H04L 9/3278: {using physically unclonable functions [PUF]}
- H04L 9/0643: {Hash functions, e.g. MD5, SHA, HMAC or f9 MAC}
- H04L 9/0866: {involving user or device identifiers, e.g. serial number, physical or biometrical information, DNA, hand-signature or measurable physical characteristics}
- H04L 9/0894: {Escrow, recovery or storing of secret information, e.g. secret key escrow or cryptographic key storage}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Physical unclonable functions (PUFs) are described. The PUFs utilize intrinsic information to determine the confidence level of comparison values. The information about confidence levels may be used to simplify the process of recovering the PUF secret. Since the information about confidence levels may be intrinsic, and not know outside the PUF, the PUF may be secure.

## Claims

### Claim 1

A physical unclonable function (PUF), comprising:
a plurality of physical structures;
comparison circuitry configured to produce a bit stream representing comparison values of a plurality of comparisons of values of a first property of the plurality of physical structures;
enrollment processing circuitry configured to receive the bit stream and generate both a secret and a sketch; and
recovery processing circuitry configured to process in combination a bit stream, the sketch, and information indicative of a confidence level of bits in the bit stream.

### Claim 2 (depends on 1)

The PUF of claim 1, wherein the plurality of physical structures are ring oscillators, and wherein the first property is a resonance frequency.

### Claim 3 (depends on 1)

The PUF of claim 1, wherein the plurality of physical structures are resistors, and wherein the first property is a resistance.

### Claim 4 (depends on 1)

The PUF of claim 1, wherein the plurality of physical structures are resistors, and wherein the first property is a capacitance.

### Claim 5 (depends on 1)

The PUF of claim 1, wherein the recovery processing circuitry is configured to process the bit stream, the sketch, and the information indicative of a confidence level of bits in the bit stream to solve a learning with errors (LWE) problem.

### Claim 6 (depends on 5)

The PUF of claim 5, wherein the processing circuitry is configured to reduce a hardness of the LWE problem by applying the information indicative of a confidence level of bits in the bit stream.

### Claim 7 (depends on 1)

The PUF of claim 1, wherein the enrollment processing circuitry is configured to generate the secret as a hash of the bit stream.

### Claim 8 (depends on 7)

The PUF of claim 7, wherein the enrollment processing circuitry is configured to generate the sketch by multiplying the bit stream by a matrix.

### Claim 9 (depends on 8)

The PUF of claim 8, wherein the matrix is a random matrix.

### Claim 10 (depends on 1)

The PUF of claim 1, wherein the enrollment and/or recovery processing circuitry is configured to generate the matrix from a seed.

### Claim 11 (depends on 1)

The PUF of claim 1, wherein the processing circuitry is configured to calculate the secret without using error correction.

### Claim 12

A physical unclonable function (PUF), comprising:
a plurality of physical structures having respective values of a first property;
readout circuitry configured to compare pairs of values of the first property of respective pairs of the plurality of physical structures and to generate a digital stream of bits from a result of the comparisons; and
first processing circuitry configured to generate a secret and a sketch from the digital stream of bits; and
second processing circuitry configured to recover the secret by processing in combination the digital stream of bits, the sketch, and information indicative of a confidence level of bits in the digital stream of bits.

### Claim 13 (depends on 12)

The PUF of claim 12, wherein the plurality of physical structures are ring oscillators, and wherein the first property is a resonance frequency.

### Claim 14 (depends on 12)

The PUF of claim 12, wherein the plurality of physical structures are resistors, and wherein the first property is a resistance.

### Claim 15 (depends on 12)

The PUF of claim 12, wherein the first processing circuitry is configured to determine a susceptibility to flipping of bits in the digital stream of bits.

### Claim 16 (depends on 12)

The PUF of claim 12, wherein the second processing circuitry is configured to solve a learning with errors (LWE) problem using the digital stream of bits, the sketch, and the information indicative of the confidence level of bits in the digital stream of bits.

### Claim 17 (depends on 16)

The PUF of claim 16, wherein the first processing circuitry is configured to generate the sketch by multiplying the digital stream of bits by a matrix.

### Claim 18

A method of operating a physical unclonable function (PUF) to recover a secret, comprising:
receiving a response bit stream resulting from application of a plurality of challenge values to the PUF;
receiving a sketch;
receiving information indicative of a confidence level of bits in the response bit stream; and
solving a learning with errors (LWE) problem using the response bit stream, the sketch, and the information indicative of the confidence level of bits in the response bit stream.

### Claim 19 (depends on 18)

The method of claim 18, wherein solving the LWE problem comprises reducing a hardness of the LWE problem using the confidence level of bits in the response bit stream.

### Claim 20 (depends on 18)

The method of claim 18, wherein the response bit stream represents a hash of a raw bit stream.
