Best Practices for Challenge/Response Authentication

23
Best Practices for Challenge/Response Authentication © 2014 Hitachi ID Systems, Inc. All rights reserved.

description

This document introduces the concept of challenge/response authentication, where users are authenticated by answering a series of personal questions. It then describes a number of best practices for robust, usable deployment of challenge/response authentication techniques.

Transcript of Best Practices for Challenge/Response Authentication

Page 1: Best Practices for Challenge/Response Authentication

Best Practices

for Challenge/Response Authentication

© 2014 Hitachi ID Systems, Inc. All rights reserved.

Page 2: Best Practices for Challenge/Response Authentication

Contents

1 Introduction 1

2 Definitions 2

3 Uses for Challenge/Response Authentication 3

4 Equivalent Authentication Methods 4

5 Search Space, Degree of Randomness 6

6 Social Engineering 8

7 Standardized vs. User-Selected Questions 9

8 Usability 10

9 Privacy Protection 12

9.1 Encryption vs. Hashing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

9.2 Assisted Service versus Self-Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

10 User Enrollment 15

11 Practical Example 17

11.1 Enrollment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

11.2 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

11.3 Authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

APPENDICES 19

A Sample Questions 20

i

Page 3: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

1 Introduction

This document introduces the concept of challenge/response authentication, where users are authenticatedby answering a series of personal questions. It then describes a number of best practices for robust, usabledeployment of challenge/response authentication techniques.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 1

Page 4: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

2 Definitions

An authentication factor is a form of evidence of a user’s identity. It is used by a human user to supportthe claim that he is the legitimate owner of a login account.

Users may authenticate, typically in the context of a login process, using one or more of the following:

1. Something they know – i.e., a secret.

2. Something they have – i.e., a physical possession.

3. Something they are – i.e., a biometric sample.

Passwords and PINs are the most popular authentication technique and are an obvious example of “some-thing a user knows.” Pass-phrases are another example, consisting of multiple words rather than a single,short string of characters.

Challenge/response systems are another example of something a user knows. They typically consist of aseries of personal questions, where the user is expected to know the answer to each question. As withall forms of authentication based on secrets, it is important that people other than the user in question notknow the answers to the user’s question.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 2

Page 5: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

3 Uses for Challenge/Response Authentication

Most computer systems authenticate users using passwords – i.e., users type a secret word or phrase,which is compared against a stored value. Best practices for password management are beyond the scopeof this document. Interested readers can read more about this topic at:

http://Hitachi-ID.com/Password-Manager/docs/password-management-best-practices.html

Some systems may use alternate or supplementary authentication factors – biometric samples (voice print,finger print, iris scan, palm print, etc.); one-time-password (OTP) tokens, smart cards, etc.

In either case, a business problem arises when users have difficulty using their primary authenticationmethod. Problems may include:

1. Forgotten passwords.

2. Inadvertently triggered intruder lockouts.

3. Expired passwords.

4. Lost or damaged OTP tokens or smart cards.

5. Malfunctioning or unavailable biometric sampling devices.

The problem that arises in each of these circumstances is a simple question: How does a self-servicesystem or an IT support analyst reliably authenticate an end user prior to providing assistance? Clearlythe primary authentication method cannot be used, since the user contacted the support organization oraccessed self-help infrastructure precisely because that method did not work.

Most organizations use challenge/response authentication to authenticate users prior to providing assis-tance relating to their primary authentication method. The most common example of this is self-servicepassword reset, where a user:

1. forgets or locks out his password, and is therefore unable to login;2. identifies himself;3. authenticates himself by answering a series of personal questions;4. is able to select a new password; and5. can login again using the new password.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 3

Page 6: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

4 Equivalent Authentication Methods

When a user is able to use one authentication factor to manipulate another, a security equivalence is createdbetween the two factors. If one authentication factor is significantly stronger than the other, this equivalencecan weaken overall security.

This is best illustrated using an example:

1. An organization deploys strong password policies, including minimum length (8 characters), use ofmany character classes (lowercase, uppercase, digits, punctuation marks), mandatory passwordchanges every 60 days, a long password history and composition rules blocking simple passwords,such as dictionary words.

2. The same organization’s help desk authenticates users who forgot their password by asking them toprovide their mother’s maiden name and the last four digits of their social security number.

If an intruder wants to compromise a login ID in this organization, he may try to guess a password, but thiswill be hard. He would have to find just the right password in a search space of about (26 + 26 + 10 + 32)8,or about 6 quadrillion (!) possible passwords. Even if he can guess a million passwords every second, thiswould take about 96 years, on average.

A far easier approach would be to spend some time “off-line,” and find out the user’s mother’s maiden nameand social security number. A few phone calls should yield the former, and abuse of the social securityadministration can yield the latter. With these two pieces of data in hand, the intruder can just call the helpdesk, claim to be the victim, correctly answer the two questions and get a new password assigned. Theintruder can now access the user’s accounts.

Even a brute-force attack on these two pieces of data is much simpler. Assuming that the mother’s maidenname is one of the world’s most popular 10,000 surnames, there are only 10 million possible combinations.If a self-service password reset system is available and if it does not enforce intruder lockouts, an intrudermaking 1,000 guesses per second could successfully authenticate and reset the victim’s passwords in about1.5 hours, on average.

To summarize, by implementing a password reset business process using mother’s maiden name and last-4-SSN, the organization has effectively reduced password strength from a search space of 6 quadrillionpossible passwords to one of 10 million possible responses – six hundred million times weaker!

The same effect applies whenever one authentication factor is used to verify user identity prior to manipu-lating another authentication factor:

1. Synchronizing passwords.

2. Single sign-on.

3. Password reset.

4. Clearing or resetting token PINs.

5. Enrolling challenge/response profile data.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 4

Page 7: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

6. Enrolling biometric samples.

BESTPRACTICE

Be sure to formulate a challenge/response policy whose strength iscomparable to the organization’s strongest password policy.

BESTPRACTICE

If possible, combine challenge/response authentication with otherauthentication factors, such as a one-time-password token or a biometricsample.

Multiple authentication factors are always more secure than one.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 5

Page 8: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

5 Search Space, Degree of Randomness

In order to estimate the strength of a challenge/response system, one must estimate the number of possibleanswers for each question.

Consider the following questions, and the estimated (or precise) number of possible answers for each one:

Question Assumptions Possibleanswers

Mother’s maiden name http://www.last-names.net/Articles/Strategies.asp

1,222 commonUS surnames

Last 4 digits of SSN, credit card, phone,etc.

10,000

Month/day of birth, employment, etc. 365

Month/year of birth, employment, etc. Assuming users between 20 and 80years old

720

Make/model/year of first car Assuming 30 year time interval; 300models/year

9,000

Favorite color Assuming most users choose a primarycolor

10

City of birth Depends on how diverse the userpopulation is

50 – 500

Name of friend, spouse, child, etc.http://www.lifesmith.com/comnames.html.Note: very likely to be in top-100 for thesubject’s gender.

100

The probability of any given answer will vary based on parameters such as a user’s age, gender, ethnicorigin, etc. Some examples follow:

• In the US, mother’s maiden name has about a 1% chance of being Smith. A Vietnamese user is justas likely to report Nguyen, and a Chinese user might just as frequently report Chan.

• City of birth depends on where users are physically located – it’s most likely to be the largest metroarea nearest the largest geographical concentration of users. Many users working in an organizationbased in the US Midwest may report Chicago, for example.

• Choice of first car will likewise be impacted by location, as well as the user’s age (most Americanspurchase their first car in their teenage years) and relative popularity of models in the indicated timeperiod.

Just as the strength of a password policy may be estimated by combining the number of possible charac-ters and raising the sum to the power of the password length (see below), so too the strength of a chal-lenge/response policy can be estimated by multiplying out the number of possible answers to the availablequestions:

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 6

Page 9: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

PasswordStrength = (26uppercase + 26lowercase + 10digits + 32punct)8characters (1)

PasswordStrength = 6, 095, 689, 385, 410, 816 (2)

ChallengeStrength = 1222q1 × 10000q2 × 10000q3 × 365q4 × 9000q5 × 10q6 × 100q7 × 100q8 × 500q9 (3)

ChallengeStrength = 20, 071, 350, 000, 000, 000, 000, 000, 000 (4)

This example illustrates that a well crafted challenge/response profile can yield a search space as large asthat of a sound password policy.

BESTPRACTICE

Include as many questions as feasible in the challenge/response profile, tomaximize the search space.

BESTPRACTICE

Estimate the size of the search space for each question.

BADIDEA

Avoid questions that have very few possible answers.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 7

Page 10: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

6 Social Engineering

In reality, password composition strength and challenge/response policy strength are not quite equivalent.Careful research can yield answers to some of the questions in a challenge/response profile, effectively re-ducing the size of the search space for those questions to 1. In contrast, short of implementation problems,no research will yield password character data.

To avoid this, it makes sense to take into consideration which of the questions in a set are vulnerable tosocial engineering attacks and to discount the strength of those questions.

Some examples:

• Innocent-sounding phone calls may reveal a user’s mother’s maiden name and date of birth.

• Intercepting the user’s (postal) mail over a period of time can likewise reveal the last four digits of theuser’s social security number.

• Calls to the department of motor vehicles may expose the user’s driver’s license number.

Another problem with challenge/response profiles is that the data are typically constant. One’s date of birthdoes not change over time, in the same way as a password might. This means that an intruder has moretime available to defeat the prompts.

These two problems – vulnerable individual questions and an open-ended time period – mean that a chal-lenge/response policy should be constructed with a larger set of possible answers than the password itprotects.

BADIDEA

Avoid questions that have easily-guessed answers.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 8

Page 11: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

7 Standardized vs. User-Selected Questions

Some challenge/response systems use standard questions, while others let users choose their own ques-tions and answer them. There are pros and cons to each approach:

Characteristic Standard Questions User-selected questions

Users are able toprovide answers toevery one of a set ofquestions.

Poor. Some questions may simplynot apply (e.g., “what was your firstcar?” does not apply to someonewho doesn’t drive).

Excellent. Users choose questionsthey can answer.

Difficulty of guessinganswers

A known quantity – usually good. Random. Users often choosequestions that are easilycompromised.

Availability of questionsto intruders

Poor. By creating a new profile, anintruder can see a full set of possiblequestions.

Good. An intruder does notnecessarily know, a-priori, what hisvictim’s questions may be.

The contrasting characteristics of these two approaches lead to some best practices:

BESTPRACTICE

Use both standard and user-selected questions.

Standard questions are used because they are difficult to compromise. Unfortunately, an intruder willing tospend lots of time may spend that time researching answers, and may compromise the standard questionprofile.

User-selected questions are used because the questions themselves are difficult to guess, so a-priori re-search of answers is impossible. This leads to the next best practice:

BESTPRACTICE

When authenticating users, ask the standard questions first and only presentthe user-defined questions after the standard ones have been correctlyanswered.

This protects the user-selected questions from a researching intruder.

In general, it is wise to serialize questions to some degree, in order to obscure subsequent questions fromintruders faced with an initial prompt.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 9

Page 12: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

8 Usability

When designing a challenge/response system, it is important to consider usability. In practice, this meansallowing for several factors:

1. Memorable:

It is important to ask questions whose answers users remember. There is no point asking a user toanswer an obscure question, if the user will either be unable to answer or answer differently everytime.

2. Brief:

Users are impatient, and will not willingly enroll or answer too many questions.

3. Approximate:

Users are often poor typists, and will answer the same question using differently-spelled variations ofthe same answer on different occasions.

4. Constant:

The answers to some questions may change over time. For instance, a user’s favorite movie, actor orbook may change over time. Since a significant time span may elapse between profile enrollment anduse of the profile for authentication, it makes sense to minimize use of such questions and insteadfocus on questions that have concrete answers.

These considerations lead to some practical best practices:

BESTPRACTICE

Make sure that questions in the challenge/response profile are easilyanswered by the intended user and simultaneously hard to answer by anyoneelse.

BADIDEA

Never change the text of a question once it has been deployed and usershave enrolled answers. This could create a mismatch between the newquestion text and answers in user profiles, creating problems when users tryto authenticate.

BESTPRACTICE

During the authentication process, don’t ask the user to answer more than 6or 7 questions.

BESTPRACTICE

During enrollment, don’t ask the user to populate more than 10 or soquestions.

BESTPRACTICE

Do encode secret answers in manner that allows for approximate matching ofresponses. Allow for answer validation even if the authentication answerdiffers from the profile data by spaces, capitalization or the odd spellingmistake.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 10

Page 13: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

BESTPRACTICE

Do ensure that every question is a complete sentence, rather than a shortphrase. This makes translations easier.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 11

Page 14: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

9 Privacy Protection

Most jurisdictions require organizations to protect the privacy of their employees and customers – i.e., ofboth internal and external users. Privacy protection is often defined as preventing unwarranted release ofany personally identifiable information (PII), and challenge/response profiles do often contain PII.

When designing a challenge/response system, it therefore makes sense to:

BESTPRACTICE

Never display answers, only verify questions.

BESTPRACTICE

Encrypt all sensitive data – especially answers to personal questions.

BESTPRACTICE

Protect the server(s) on which this data is stored.

BESTPRACTICE

Use a closed system, designed for data protection, rather than a directory,designed for data publication.

BESTPRACTICE

Lock out a profile after too many failed authentication attempts.

Intruder lockout for challenge/response profiles should work just like intruder lockout for passwords – pre-venting an intruder from submitting a large number of guessed answers in a short time. Users do maketyping mistakes, so the intruder lockout interval can be high in terms of human guesses while still block-ing automated attacks. For example triggering a lockout after 10 failed attempts in a 10 minute interval isreasonable.

9.1 Encryption vs. Hashing

Encryption refers to a reversible process of protecting sensitive data. In a symmetrical encryption system,data is scrambled using an algorithm that requires a key (the encryption key), and can only be unscrambled(decrypted) using the same key.

In other words, the security of an encryption system depends on how well the encryption/decryption key isprotected.

Hashing refers to a one-way process of mapping plaintext data to a scrambled string of characters. Thereis, by definition, no way to convert the scrambled data (the hash) back to the original plaintext.

When designing a challenge/response system, the question of whether it is better to encrypt or to hash userprofile data often arises. Each approach has pros and cons:

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 12

Page 15: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

Data protection Encrypt profile data Hash profile data

Can be compromised if both the key and scrambled data aresuccessfully compromised?

Yes No

Vulnerable to guessing attacks? Yes Yes

Supports approximate matching of stored profile data againstdata entered during the authentication process?

Yes No

Put another way, encryption is preferable if usability is a key concern, since users are unlikely to answerthe same question in exactly the same way on multiple attempts. For example, when asking a user to typeher mother’s maiden name, she might type “Jones-Smith” in one instance, “Jones Smith” the next time,“jones-Smith” a third time and so on. This example does not even bring spelling and typing mistakes intothe picture.

If the user’s profile data is (reversibly) encrypted, then the matching code can decrypt it at authentica-tion time and apply various approximate matching algorithms (e.g., case insensitive, ignoring spaces andpunctuation marks and even sounds-alike) to validate the user input.

In contrast, using a hashing approach, if the user types an inexact answer, the result is simply an au-thentication failure. This can have a devastating impact on the system’s adoption rate and consequentlysubstantially reduce the return on investment.

On the other hand, in the event that the system where challenge/response data is stored is seriously com-promised, hashed data can only be compromised by a brute-force guessing attack, while encrypted datacan – sooner or later – be extracted by an intruder, since both the data and the key are present on thesystem.

BESTPRACTICE

In most deployments, total system compromise is quite unlikely, so encryptionis strong enough. Usability is always an issue, so hashing of user profile datais usually undesirable.

9.2 Assisted Service versus Self-Service

If challenge/response profiles will be used for both self-service authentication, for instance in the context ofself-service password reset, as well as by help desk staff, for instance in the context of assisted passwordreset, it is often helpful to use different data in each case.

BESTPRACTICE

Use separate challenge/response profile to authenticate the same users in thecontext of self-service vs. assisted service processes.

Using different challenge/response questions to support self-service and assisted processes can promotethe goal of privacy protection, since the use of PII can be limited to self-service only.

The security of assisted service is enhanced by authenticating support staff before allowing them to au-thenticate callers to the help desk. Caller authentication data can be further protected by requiring the helpdesk analyst to enter answers to user questions, rather than displaying them on the screen. An audit trailadds accountability to this process as well.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 13

Page 16: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

If a help desk analyst successfully responds to a user service call, he will have access to part of a user’sprofile. So long as the user has a separate profile for self-service authentication, the help desk analystcannot subsequently, anonymously impersonate the user, since he never had access to the user’s self-service authentication questions.

Separating question sets used for self-service and assisted service also makes it possible to ask users toanswer questions in the context of the self-service profile which would be inappropriate to answer, due tocompromise of user privacy, in an assisted service scenario.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 14

Page 17: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

10 User Enrollment

Challenge/response systems must get the answers to personal user questions somewhere. When a chal-lenge/response system is first deployed, existing data may be inadequate (does not cover all users or tooeasy to compromise). This situation leads to a need for user enrollment:

1. A subsystem is required to identify users who must enroll. Inside an organization, this may be assimple as “every Active Directory user.” On an Extranet, this may be “every new user.”

2. A subsystem is required to prompt un-enrolled users to complete their profiles:

(a) It should not invite too many users at any one time. This could overload the e-mail (or otherinvitation) system. Moreover, some fraction of the users will not understand the process, or wantto verify that the invitation is legitimate and will consequently call the help desk. Too many userscalling the help desk at one time would overwhelm the help desk.

(b) It should not invite any one user too often, even if that user does not respond. Too-frequentinvitations are perceived as nothing more than a nuisance.

(c) It should be able to force enrollment if a user has been invited many times, and failed to respond.

3. The enrollment system must authenticate users before allowing them to update their profile.

4. The enrollment system should collect more data than will be used in a single authentication attempt.This makes randomization of the questions posed to users possible.

These ideas can be restated as best practices:

BESTPRACTICE

Collect enough data from a user to support random question selection duringa subsequent authentication process.

BESTPRACTICE

Employ an automated enrollment system, which invites users to completetheir challenge/response profile.

BESTPRACTICE

Roll out invitations gradually (example: 1000 users per day), not all-at-once.

BESTPRACTICE

Do not invite the same user too often (say no more than once weekly).

BESTPRACTICE

Force users to enroll when they repeatedly ignores invitations to completetheir profile.

BESTPRACTICE

Authenticate users prior to enabling them to see and update their profile.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 15

Page 18: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

BESTPRACTICE

Use an existing authentication method – e.g., a current network password –prior to enrollment.

BADIDEA

Never use a weak authenticator, such as a PIN, a non-expiring password oran e-mailed secret, to authenticate users.

Using a weak authentication would make the entire system weak, since an intruder could break the initialauthentication, complete a challenge/response profile and subsequently reset a user’s password. This alsoleads to a corollary:

BESTPRACTICE

Enroll users before they forget or lock out their password. Do not startenrollment only after a user needs service. Users experiencing anauthentication problem cannot reliably authenticate themselves and are in noposition to complete their challenge/response profile in a trustworthy manner.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 16

Page 19: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

11 Practical Example

Following is a practical example showing how a fictitious organization could combine many of the ideaspresented in this document:

11.1 Enrollment

Every user in the organization that has a login ID on Active Directory, and whose last login date is less than30 days old (i.e., the account is not obviously dormant) is invited to complete a challenge/response profile:

1. Users are invited to enroll on Monday, Tuesday, Wednesday and Thursday. No user is invited on otherdays.

2. Up to 500 users are invited to enroll daily.

3. Users are invited at most once per week.

4. Any user can volunteer to update their profile any time. The controls above only impact invitations.

5. Users whose profiles are incomplete and who are about to be sent a fourth invitation are insteadattached to a special AD security group called ForcedEnrollment. A group policy object (GPO) isapplied to this group to change their Windows shell from EXPLORER.EXE to a program that launchesa kiosk-mode web browser, where they must complete enrollment.

6. Users who complete their profiles are removed from ForcedEnrollment (if they were members) andare not invited to enroll again.

11.2 Questions

Three sets of questions are enrolled:

1. Standard questions that users may be asked to answer during self-service authentication:

(a) Twenty questions are provided.

(b) Users are asked to answer at least eight.

2. Standard questions that users may be asked to answer when they call the help desk:

(a) Ten questions are provided.

(b) Users are asked to answer at least four.

3. User-defined questions:

(a) Users are asked to provide two question/answer pairs.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 17

Page 20: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

11.3 Authentication

When a user calls the help desk, they are asked to provide answers to a random 3 of the at-least-4 helpdesk questions. The help desk analyst must type the answers that the user provides - they cannot see theanswers.

When a user accesses the self-service system, they must answer a random 5 of the at-least-8 self-servicequestions. If they can do this successfully, they are subsequently presented with a second authenticationstep, presenting both of the self-defined questions.

In all cases, user answers are compared in a case-insensitive, punctuation-insensitive and spacing-insensitivemanner to stored, encrypted data. This allows users who are poor typists to successfully sign into the self-service system.

In all cases, the system (not the user) chooses which questions to randomly select. This prevents anintruder from choosing to answer questions to which he has acquired answers.

Ten consecutive authentication failures in the space of 30 minutes triggers an intruder lockout for 1 hour.This blocks systematic guessing attacks by an intruder, without impacting most legitimate users.

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 18

Page 21: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

APPENDICES

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 19

Page 22: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

A Sample Questions

Following are a set of sample questions that can be used to authenticate users. Users should be instructedto only answer questions if they are confident the answers will never change.

Sample security questions, which may have alpha-numeric questions and so are suitable for a text userinterface, include:

• Which bank branch do you live closest too?• What car do you wish you owned?• What is your favorite food?• Who is your favorite book character?• What is your favorite game or sport?• What is your favorite movie?• What is your favorite pizza topping?• What is your favorite restaurant?• What is your favorite season of the year?• What is your favorite sports team?• In which department did you first work?• What was your first position in the company?• What was your first car?• Who is the person you admire the most?• What was the most memorable day in your life?• Who was your childhood hero?• What is the nickname of your sibling?• Who was your first boss?• What award are you proudest of?• What city were you born in?• What is the farthest from home you have traveled?• What is the name of the first school you attended?• What is the name of the first person you were romantically interested in?• What is your astrological sign?• What is your father’s middle name?• What is your mother’s’ middle name?• Who is your favorite actor, actress or celebrity?• What is your favorite musical band?• What is your favorite beverage?• What is your favorite board game?• Who is your favorite book character?• What is your favorite dessert?• What is your favorite hobby or pastime?• What is your favorite ice cream topping?• What is your favorite song?• What is your favorite television show?• What is your favorite vacation spot?• What is your mother’s maiden name?• What is your place of birth?• What is your school team’s mascot name?• What was the breed of your first pet?• What was the color of your first automobile?

© 2014 Hitachi ID Systems, Inc.. All rights reserved. 20

Page 23: Best Practices for Challenge/Response Authentication

Best Practices for Challenge/Response Authentication

• What were the make and model of your first car?• What was the name of a favorite childhood pet?• What was the name of your first girlfriend/boyfriend?• What was the street name of your childhood home?• What was your favorite toy when you were a child?• What did you do on your first job?• What was your first phone number as a child?• What year did you purchase your first car?• What was the name of your first pet?• Who is your favorite politician?• Who is your most disliked politician?• Who is a famous, living person you would most like to meet?• Who was a famous, now deceased person you would have liked to meet?• Who is your favorite artist?• Who is your favorite author?• With whom did you share your first romantic kiss?• Who was your favorite elementary school teacher?

The following questions may also be used. They differ from the above in that they have numeric answers,making them suitable for use with a telephony-based solution that uses touch-tone input:

Sample security questions, that have numeric answers and so are suitable for authentication using a touch-tone phone, include:

• What is your favorite radio station (number on the dial - NNNN)?• In what year did you start with your company?• On what date were you hired?• What is your parents’ wedding anniversary date?• Type a significant date in your life (YYYYMMDD)?• What are the last 4 digits of your SSN?• What are the last 4 digits of your home phone?• What is a birth date of a family member?• What is a relative’s telephone number that is not your own?• What is the date of your anniversary [mm/dd/yyyy]?• What is the mileage distance you live from your furthest relative?• What is your Country or employee identification number?• What is your date of birth (MM/DD/YYYY)?• What is your driver’s license number?• What is your favorite dial number of radio station?• What is your favorite or lucky number?• What is your passport number?• What is your social security number?• On what year you will be eligible to retire?

www.Hitachi-ID.com

500, 1401 - 1 Street SE, Calgary AB Canada T2G 2J3 Tel: 1.403.233.0740 Fax: 1.403.233.0725 E-Mail: [email protected]

File: /pub/wp/documents/bp-challenge-response/best-practice-challenge-response-1.texDate: 2007-05-21