ISO/IEC 20897-2:2022

Information security, cybersecurity and privacy protection — Physically unclonable functions — Part 2: Test and evaluation methods

This document specifies the test and evaluation methods for physically unclonable functions (PUFs). The test and evaluation methods consist of inspection of the design rationale of the PUF and comparison between statistical analyses of the responses from a batch of PUFs or a unique PUF versus specified thresholds. This document is related to ISO/IEC 19790 which specifies security requirements for cryptographic modules. In those modules, critical security parameters (key) and public security parameters (product serial number, identification code, etc.) are the assets to protect. PUF is one solution to avoid storing security parameters, thereby increasing the overall security of a cryptographic module.

Sécurité de l'information, cybersécurité et protection de la vie privée — Fonctions non clonables physiquement — Partie 2: Méthodes d'essai et d'évaluation

General Information

• Status: Published
• Publication Date: 24-May-2022
• ICS: 35.030 - IT Security
• Technical Committee: ISO/IEC JTC 1/SC 27 - Information security, cybersecurity and privacy protection
• Drafting Committee: ISO/IEC JTC 1/SC 27/WG 3 - Security evaluation, testing and specification
• Current Stage: 6060 - International Standard published
• Start Date: 25-May-2022
• Due Date: 19-Jun-2022
• Completion Date: 25-May-2022

Standards Content (Sample)

INTERNATIONAL ISO/IEC
STANDARD 20897-2
First edition
2022-05
Information security, cybersecurity
and privacy protection — Physically
unclonable functions —
Part 2:
Test and evaluation methods
Sécurité de l'information, cybersécurité et protection de la vie
privée — Fonctions non clonables physiquement —
Partie 2: Méthodes d'essai et d'évaluation
Reference number
ISO/IEC 20897-2:2022(E)
© ISO/IEC 2022

---------------------- Page: 1 ----------------------
ISO/IEC 20897-2:2022(E)
COPYRIGHT PROTECTED DOCUMENT
© ISO/IEC 2022
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on
the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address below
or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: [phone]
Email: [e-mail]
Website: www.iso.org
Published in Switzerland
ii
  © ISO/IEC 2022 – All rights reserved

---------------------- Page: 2 ----------------------
ISO/IEC 20897-2:2022(E)
Contents Page
Foreword .iv
Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms, definitions and abbreviated terms . 1
3.1 Abbreviated terms . 1
4 Symbols . 2
5 Tests of PUFs . 2
5.1 General . 2
5.2 Test conditions . 4
5.3 Security tests . 4
5.3.1 General . 4
5.3.2 Test of steadiness . 4
5.3.3 Test of randomness . 5
5.3.4 Test of uniqueness . 5
5.3.5 Test of Tamper-resistance . 5
5.3.6 Test of Mathematical unclonability . . 6
5.3.7 Test of Physical unclonability. 6
Annex A (informative) Tests of the steadiness . 7
Annex B (informative) Tests of the randomness .10
Annex C (informative) Tests of the uniqueness .17
Annex D (informative) Example of the test of the PUF security requirements .19
Bibliography .27
iii
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 3 ----------------------
ISO/IEC 20897-2:2022(E)
Foreword
ISO (the International Organization for Standardization) and IEC (the International Electrotechnical
Commission) form the specialized system for worldwide standardization. National bodies that are
members of ISO or IEC participate in the development of International Standards through technical
committees established by the respective organization to deal with particular fields of technical
activity. ISO and IEC technical committees collaborate in fields of mutual interest. Other international
organizations, governmental and non-governmental, in liaison with ISO and IEC, also take part in the
work. In the field of information technology, ISO and IEC have established a joint technical committee,
ISO/IEC JTC 1.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject
of patent rights. ISO and IEC shall not be held responsible for identifying any or all such patent
rights. Details of any patent rights identified during the development of the document will be in the
Introduction and/or on the ISO list of patent declarations received (see www.iso.org/patents). Any trade
name used in this document is information given for the convenience of users and does not constitute
an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to
the World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT), see
www.iso.org/iso/foreword.html.
This document was prepared by Joint Technical Committee ISO/IEC JTC 1, Information technology,
Subcommittee SC 27, Information security, cybersecurity and privacy protection.
A list of all parts in the ISO/IEC 20897 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www.iso.org/members.html.
iv
  © ISO/IEC 2022 – All rights reserved

---------------------- Page: 4 ----------------------
ISO/IEC 20897-2:2022(E)
Introduction
This document specifies the test methods for physically unclonable functions (PUFs) for generating
non-stored cryptographic parameters.
Cryptographic modules generate the certain class of critical security parameters such as a secret
key using a random bit generator within the modules. Such modules may store generated security
parameters in embedded non-volatile memory elements. For a higher security, a combination of tamper
response and zeroisation techniques may be used for protecting stored security parameters from active
unauthorized attempts of accessing such parameters. As the reverse-engineering technology advances,
however, the risk of theft of such stored security parameters has become higher than ever.
The rapidly pervading technology called a PUFs is promising to mitigate the above-mentioned risks
by enabling security parameter management without storing such parameters. PUFs are hardware-
based functions providing mathematical unclonability, steadiness and randomness of their outputs and
physical unclonability of the functions themselves, taking advantage of intrinsic subtle variations in
the device’s physical properties, which are also considered objects’ fingerprints. PUFs may be used for
security parameter (e.g. key, initialization vector, nonce and seeds) generation, entity authentication
or device identification in cryptographic modules. More detailed information about the characteristics
and security requirements of the PUF are given in ISO/IEC 20897-1 and this document only describes
test and evaluation methods.
Now, security requirements of PUFs should be considered at system level, meaning that they should
consider many possible attack paths, as detailed further in this document. The purpose of this
document is to specify how to test those security requirements for assuring an adequate level of quality
of the provided PUFs in cryptographic modules. This document is supposed to be used for the following
purposes:
a) In the procurement process of a PUF-equipped product, the procurement body specifies the
security requirements of the PUF in accordance with ISO/IEC 20897-1. The product vendor
evaluates the PUF in accordance with this document whether the PUF satisfies all the specified
security requirements, and reports the evaluation results to the procurement body.
b) The vendors evaluate the security of their PUF in accordance with this document, publicize the
evaluation results and clarify the security of their PUF.
It should be noted that all of the security requirements defined in ISO/IEC 20897-1 are not necessarily
quantitatively evaluable.
v
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 5 ----------------------
INTERNATIONAL STANDARD ISO/IEC 20897-2:2022(E)
Information security, cybersecurity and privacy
protection — Physically unclonable functions —
Part 2:
Test and evaluation methods
1 Scope
This document specifies the test and evaluation methods for physically unclonable functions (PUFs).
The test and evaluation methods consist of inspection of the design rationale of the PUF and comparison
between statistical analyses of the responses from a batch of PUFs or a unique PUF versus specified
thresholds.
This document is related to ISO/IEC 19790 which specifies security requirements for cryptographic
modules. In those modules, critical security parameters (key) and public security parameters (product
serial number, identification code, etc.) are the assets to protect. PUF is one solution to avoid storing
security parameters, thereby increasing the overall security of a cryptographic module.
2 Normative references
The following documents are referred to in the text in such a way that some or all of their content
constitutes requirements of this document. For dated references, only the edition cited applies. For
undated references, the latest edition of the referenced document (including any amendments) applies.
ISO/IEC 19790, Information technology — Security techniques — Security requirements for cryptographic
modules
ISO/IEC 20897-1, Information security, cybersecurity and privacy protection — Physically unclonable
functions — Part 1: Security requirements
3 Terms, definitions and abbreviated terms
For the purposes of this document, terms, definitions and abbreviated terms given in ISO/IEC 20897-1,
ISO/IEC 19790 and following apply.
ISO and IEC maintain terminology databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https:// www .iso .org/ obp
— IEC Electropedia: available at https:// www .electropedia .org/
3.1 Abbreviated terms
BER Bit error rate.
iid Independent and identically distributed.
IID
NRBG Non-deterministic random bit generator
1
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 6 ----------------------
ISO/IEC 20897-2:2022(E)
4 Symbols
For the purposes of this document, the following symbols apply.
∀
A math symbol representing “for all” or “for any.”
∈
A math symbol representing set membership.
inter
D A vector representing Inter-HD between PUF responses.
intra
D A vector representing Intra-HD between PUF responses.
The index for the PUF instances. 1≤ i, j ≤ N .
i, j
PUF
The index for the challenge. 1 ≤ k ≤ N .
k
chal
N The sequence size (bit length) of PUF responses.
N The largest number of identical responses (correct responses).
c
N The number of different challenges given to a PUF.
chal
N The number of measurements of responses repeatedly collected for a single challenge.
meas
N The number of PUF instances.
PUF
N The length of PUF response obtained from a single challenge.
res
σ
A standard deviation of a random value.
∑
A sum of all values in the specified range.
The index for the response measurements. 1 ≤ t ≤ N .
t
meas
μ
A mean of a random value.
x A challenge.
y A response.
(t)
y (x) The t-th response of the i-th PUF instance obtained by giving a challenge x.
i
5 Tests of PUFs
5.1 General
In this document, testing a PUF means verifying the security requirements defined in ISO/IEC 20897-1.
As already mentioned in ISO/IEC 20897-1, for the purpose of the ISO/IEC 20897 series, the responses
from multiple PUFs are arranged into a cube as shown in Figure 1. The repetitive calls to a PUF are
illustrated in Figure 2. The single small cube describes a 1-bit response from a PUF. The three axes of
the cube and the time are described hereafter, as directions:
— direction B: “#bits” shows the bit length of the response obtained from a single challenge. In a 1-bit
response PUF, e.g., arbiter PUF, the dimension B collapses.
— direction C: “#challenges” shows the number of different challenges given to a PUF. In a no-challenge
[1]
PUF (or, more rigorously, a one-challenge PUF), e.g., SRAM PUF , the dimension C collapses.
— direction D: “#PUF” shows the number of different PUF devices under test.
— direction T: “#query” shows the number of query iterations under the fixed PUF device and challenge.
2
  © ISO/IEC 2022 – All rights reserved

---------------------- Page: 7 ----------------------
ISO/IEC 20897-2:2022(E)
Figure 1 — Cube representation of the response sequences from multiple PUFs
Figure 2 — Responses obtained by repetitive calls to the PUFs.
Among the defined security requirements, the steadiness, randomness and uniqueness may be tested
by measuring responses from actual device(s) (Figures 1 and 2). If the empirical approaches are
insufficient for evaluation, a stochastic model can be applied. If a stochastic model is used, the vendor
shall provide the detailed document which explains the correctness of the model and the validity of the
use of the model.
The tests based on stochastic models are defined in BSI AIS 31 standard and Reference [3]. They refer
to an abstraction of the device where probabilistic aspects are clearly described. The model allows to
derive ideal results, in terms of expectations over distributions (not only estimations based on limited
sampling).
Notice that stochastic models shall be used in these two conditions:
— when a metric is not otherwise testable, owing to the prohibitive number of measurements which
would be required,
— when a predictive (asymptotically) value of a metric is required.
For the stochastic model to be relevant, it shall rely on analogue random properties (such as delays,
voltages, etc.). These measurements would typically be quantified to get the response bits. Some PUF
structures may feature the capability to quantify the analogue properties, for example, when the
response bits are obtained by a logical process. For instance, in the loop-PUF, the number of loops is
counted. The response bit is subsequently computed based on a comparison between two (or more)
3
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 8 ----------------------
ISO/IEC 20897-2:2022(E)
loop numbers. Thus, the information of the number of loops is available for analysis in a stochastic
model.
Some PUF structures could add the capability of measurement quantification in a so-called "test
mode." The quantification capability is thus intentionally added in an artificial manner for the sake of
computing the metric. The addition of the quantification capability can be justified in situations where
the PUF responses might be used only if the PUF itself is trusted.
5.2 Test conditions
The metrics shall be estimated according to several environmental conditions, including temperature,
voltage, and humidity. Extreme values defined from the device standard operating conditions, e.g., the
temperature and supply voltage specified in 7.7.4.3 of ISO/IEC 19790:2012, shall be tested. Also, some
cycles between extreme values shall be performed, so as to check that the device is still functioning as
intended. The accelerated aging techniques in Reference [5] may be used for this purpose.
5.3 Security tests
5.3.1 General
5.3 provides the concrete test and evaluation methods of a PUF. The security tests described in 5.3.2
through 5.3.7 corresponds to the security requirements defined in ISO/IEC 20897-1.
To claim that the PUF satisfies one or more of the security requirements, the vendor shall document the
reason for that based on the conducted tests, evaluation results and/or design rationale. The document
may include the logic diagram of the PUF building blocks, e.g., entropy source, entropy extractor,
pre-processing block, post-processing block, and so forth. For the example of security requirements,
evaluation criteria, and test conditions, see Annex D.
5.3.2 Test of steadiness
Steadiness is the repeatability over T measurements in Figure 2, that is, the estimation of a probability
of failure based on repeated measurements. For devices which require a very high steadiness (e.g.,
for mission critical applications), the experimental approach can be insufficient in terms of error bars
(estimation is too crude). In such a case, a stochastic model may be applied.
A PUF shall meet the steadiness requirement if the application of the PUF requires it (cf. 5.6,
ISO/IEC 20897-1:2020). If a vendor claims that the PUF satisfies the steadiness requirement, the vendor
shall document the reason for that. The steadiness shall be quantified by at least one metric: the intra-
HD, bit error rate (BER), stochastic model, and so forth. The chosen value of challenge bits shall be
precisely described, but is arbitrary. A vendor shall be responsible for determining the criterion of
the steadiness considering the applications and operating environment of the PUF. It is the vendor’s
responsibility, if necessary, to check with the designer, manufacturer, and so forth for detailed
information of the PUF in order to establish the evaluation criteria for steadiness. The concept of the
test of steadiness is illustrated in Figure 3. For the example test procedure for steadiness, see Annex A.
Figure 3 — Principle of the test for steadiness
4
  © ISO/IEC 2022 – All rights reserved

---------------------- Page: 9 ----------------------
ISO/IEC 20897-2:2022(E)
5.3.3 Test of randomness
Randomness is the variability across B-C plane in Figure 1. A PUF shall meet the randomness
requirement if the application of the PUF requires it (cf. 5.6, ISO/IEC 20897-1:2020). If a vendor
claims that the PUF satisfies the randomness requirement, the vendor shall document the reason for
that. A vendor shall be responsible for determining the criterion of the randomness considering the
applications and operating environment of the PUF. It is the vendor’s responsibility, if necessary, to
check with the designer, manufacturer, and so forth for detailed information of the PUF in order to
establish the evaluation criteria for randomness.
The randomness shall be tested by applying a statistical randomness test or entropy estimation to the
B-C plain of the responses (see Figure 1). When the dimension C collapses (e.g., in confined PUFs), the
test is applied to the bit sequence in dimension B. Similarly, when the dimension B collapses (e.g., in
an arbiter PUF), the test is applied to the bit sequence in dimension C. Note that the randomness tests
cannot apply to the cases without enough number of PUF responses. The tests may be the NIST FIPS
[2] [3] [6] [7]
140-1 , BSI AIS 31 , SP800-22 , SP800-90B and so forth. The chosen value of challenge bits shall
be precisely described, but is arbitrary. For the example test procedure for randomness, see Annex B.
NOTE FIPS 140-1 has been withdrawn, but that its entropy tests are still reliable methods which can be
repurposed to evaluate a PUF.
5.3.4 Test of uniqueness
Uniqueness is the variability across B-D plane in Figure 1. A PUF shall meet the uniqueness requirement
if the application of the PUF requires it (cf. 5.6, ISO/IEC 20897-1:2020). If a vendor claims that the PUF
satisfies the uniqueness requirement, the vendor shall document the reason for that. A vendor shall be
responsible for determining the criterion of the uniqueness considering the applications and operating
environment of the PUF. It is the vendor’s responsibility, if necessary, to check with the designer,
manufacturer, and so forth for detailed information of the PUF in order to establish the evaluation
criteria for uniqueness.
The uniqueness shall be assessed by evaluating the inter-Hamming distance of the responses among the
PUF devices, or by the statistical ways similar to the randomness. In practice, it is not always possible
to prepare a sufficient number of devices for the test of uniqueness. In such a case, a stochastic model
may leverage the test (see C.4). If the uniqueness is evaluated by the statistical tests, the applicable
[8] [2]
test includes the NRBG health test in ISO/IEC 18031:2011 which is based on FIPS 140-1 and AIS-
[3] [7]
31 , and NIST SP800-90B . The concept of the test of uniqueness is illustrated in Figure 4. The chosen
value of challenge bits shall be precisely described, but is arbitrary. For the example test procedure for
uniqueness, see Annexes B and C.
Figure 4 — Test of uniqueness
5.3.5 Test of Tamper-resistance
The test of the tamper-resistance verifies whether the PUF does not leak secret data nor lose requisite
properties by invasive, semi-invasive and non-invasive physical attacks. The examples of the physical
attacks include side-channel attacks, reverse engineering, using fault analysis tools such as LVP and
FIB.
5
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 10 ----------------------
ISO/IEC 20897-2:2022(E)
A PUF shall meet the tamper-resistance requirement if the application of the PUF requires it (cf. 5.6,
ISO/IEC 20897-1:2020). If a vendor claims that the PUF satisfies the tamper-resistance requirement, the
vendor shall document the reason for that. A PUF vendor shall be responsible for adducing the rationale
for the tamper-resistance based on the conducted tests, evaluation results, design and implementation
of the PUF. It is the vendor’s responsibility, if necessary, to check with the designer, manufacturer, and
so forth for detailed information of the PUF in order to establish the evaluation criteria for tamper-
resistance.
5.3.6 Test of Mathematical unclonability
The test of the mathematical unclonability verifies whether the PUF’s challenge-response behaviour
is not simulated or emulated by other devices. A mathematically unclonable PUF generates responses
which are difficult to be corelated to the challenge, design and implementation of the PUF. A PUF is
mathematically unclonable if it is impossible to disclose the mapping table or function of the CRPs by
for example dictionary attacks, machine learning attacks, and so on.
A PUF shall meet the mathematical unclonability if the application of the PUF requires it (cf. 5.6,
ISO/IEC 20897-1:2020). If a vendor claims that the PUF satisfies the mathematical unclonability, the
vendor shall document the reason for that. A PUF vendor shall be responsible for adducing the rationale
for the mathematical unclonability based on the conducted tests, evaluation results, design and
implementation of the PUF. It is the vendor’s responsibility, if necessary, to check with the designer,
manufacturer, and so forth for detailed information of the PUF in order to establish the evaluation
criteria for mathematical unclonability.
5.3.7 Test of Physical unclonability
The test of the physical unclonability verifies whether the physical clone of the PUF is impractical to be
manufactured. The physical unclonability ensures that there are no two PUFs that have the same input-
output behaviour. The physical unclonability is assessed by examining whether the PUF surely utilizes
the entropy source derived from the uncontrollable device variation.
A PUF shall meet the physical unclonability if the application of the PUF requires it (cf. 5.6,
ISO/IEC 20897-1:2020). If a vendor claims that the PUF satisfies the physical unclonability, the vendor
shall document the reason for that. A PUF vendor shall be responsible for adducing the rationale for
the physical unclonability based on the conducted tests, evaluation results, design or implementation
of the PUF. It is the vendor’s responsibility, if necessary, to check with the designer, manufacturer, and
so forth for detailed information of the PUF in order to establish the evaluation criteria for physical
unclonability.
6
  © ISO/IEC 2022 – All rights reserved

---------------------- Page: 11 ----------------------
ISO/IEC 20897-2:2022(E)
Annex A
(informative)

Tests of the steadiness
A.1 General
The steadiness should be evaluated by at least one of the following metrics: Intra-HD, the bit error
rate (BER), the stochastic model, and so forth. This Annex provide the detail procedure to evaluate the
steadiness based on the BER, intra-HD, and stochastic model.
A.2 Bit error rate
Steadiness may be quantified by the average bit error rate of the response sequence arranged in T
direction (see Figure 2). The bit error rate is the number of differences obtained during multiple queries
divided by the number of queries. It is optimally expressed in logarithmic scale. A vendor should be
responsible for determining the criterion of the steadiness considering the applications and operating
environment of the PUF.
NOTE For extensive PUFs, some challenges are devoted to challenge-response authentication. The steadiness
test is preferable to be performed on the challenges which are not used for the challenge-response authentication.
The test consists in taking N (=×NN ) measurements. N denotes the largest number of identical
reschal c
responses among the N measurements. All other responses (subtraction N-N in total) are considered
c
incorrect and are thus counted as errors. Therefore, the steadiness rate is quantified by the ratio N /N.
c
Let  be BER of a PUF in which a wrong response is obtained (owing to noise) at maximum once every
ˆ
1/ queries. The steadiness test is thus to check that the expected value of pN= /N over n queries
c
(assuming the responses are iid) is such that
11−≤ pˆ ≤
with high probability.
7
© ISO/IEC 2022 – All rights reserved

---------------------- Page: 12 ----------------------
ISO/IEC 20897-2:2022(E)
ˆ
Let p be the population rate of p, then the following inequation is obtained with the confidence level of
z:
ˆˆ ˆˆ
pp11− pp−
() ()
ˆ ˆ
pz− ≤≤pp + z .
n n
Thus, the constraint is on the error interval length
ˆˆ
pp1 −
()
2z ≤.
n
Using the relationship
2
1
ˆˆ
pp1 − ≤≤ ,
()
 
2
 
the error interval can be bounded to
1
2z ≤.
4n
Therefore, the required number of CRPs under the bit error rate  is given by the following inequation:
2
z
n≥ .
2

A.3 Intra-Hamming distance
[9][10]
Steadiness is often quantified by the intra-HD, especially in academic papers .
intra
Let D be the 3D vector that consists of all measured responses for N , N and N .
puf chal meas
j j
intra () ()
12
 
D = HD yx() , yx()
()
i ki k
 
∀≤11iN≤∀,,≤≤kN ∀≤1 jj≠≤N
pufchalm12 eas
intra
The mean of the D is calculated as follows:
2
intraintra inttra
 
μ =E DD= ∑∑∑∑
 
NN⋅⋅NN⋅−()1 ⋅N
pufchalmeas meas res
intra
The standard deviation of the D is calculated as follows:
2
2
intraintra intra
σμ= ∑∑∑∑ D −
(()
NN⋅⋅NN⋅−()12⋅−N
pufchatmeas meas res
intra intra
Generally, when both μ and σ are around zero, the reproducibility of the response of the PUF
is high, and therefore, the steadiness is hi
...