This page is to share and discuss various standards for identity assurannce, which is often defined as:
"the ability to determine, with some level of certainty, that the person presenting themselves in an online transaction is who they say they are."
Virginia Tech Standard for Personal Digital Identity Levels of Assurance
Identity assurance includes identity proofing, the credential issuance process, and credential strength. Erv recently approved the Standard for Personal Digital Identity Levels of Assurance, which is posted at www.it.vt.edu/administration/policies.html. Virginia Tech's LOA's reflect (but may not completely conform to) standards from NIST Special Publication 800-63, Version 1.0.2, Electronic Authentication Guide http://csrc.nist.gov/publications/nistpubs/index.html .
While there may not be a mandate that the soft PDC follow any national or industry standard, past conversations with universiy Legal counsel have indicated that standards-based credentialing processes are desirable. The soft PDC may fit into Virginia Tech's LOA 3. The credential itself should be able to meet the criteria for NIST 800-63 LOA 3. The identity proofing would likely determine whether the soft PDC could qualify as NIST LOA 2 or NIST LOA 3. Similarly, the soft PDC credential should meet the criteria for an InCommon Silver credential. The identity proofing will likely determin whether the soft PDC would qualify for InComon Bronze or Silver.
InCommon Bronze and Silver
The InCommon Identity Assurance Assessment Framework at http://www.incommonfederation.org/assurance/ states (in Section 2.1), "InCommon Bronze and Silver are intended to be compatible with Federal NIST 800-63 Levels 1 and 2."
The Identity Proofing section (4.2.2.3) for the Silver profile specified at http://www.incommonfederation.org/docs/assurance/InC_Bronze-Silver_IAP_1.0.1.pdf , specifies In Person Proofing and Remote Proofing criteria that are very close (but identical) to those specified for In Person and Remote Proofing for LOA 2 in NIST 800-63 (section 6.3.1). However, the InCommon Silver profile (section 4.2.2.3.1) includes criteria for identity proofing using an Existing Relationship, which is not specifally categorized in Table 3 in NIST 800-63, but seems to be derived from the a paragraph immediately following Table 3.
NIST 800-63:
At Level 2, employers and educational institutions who verify the identity of their employees or students by means comparable to those stated above for Level 2 may elect to become an RA or CSP and issue credentials to employees or students, either in-person by inspection of a corporate or school issued picture ID, or through on-line processes, where notification is via the distribution channels notmally used for sensitive, personal communications.
InCommon Silver section 4.2.2.3.1:
Employers and educational institutions which verify the identity of their employees, students or other affiliates by means comparable to those stated for In Person Proofing or remote Proofing may be designated an RA by the IdP operator. The IdP operator shall confirm that the applicant is a person with a current relationship to the organization, record the nature of that relationship and verify that the relationship is in good standing. If the IdP operator's IdMS directory or database is separate from the institution's or RA's database, the IdP operator shall confirm that the applicant's name and address are consistent in both places.
Perhaps Virginia Tech's identity proofing practices would show that we qualify for Silver in the "Existing Relationship" category.
Sample Silver in-person Registration Process from Penn State
Sample Silver Remote Registration Process from Penn State
Requirements for recording information from identity documents
InCommon Silver Section 4.2.3.2 item #2 requires that the RA record from the ID:
- ID number
- Date of issuance
- Date of expiration
- Address, if available
- Date of birth
NIST 800-63-01 (Table 3 in section 6.3.1) requires the RA to record from the ID:
- ID number
- Address
- Date of birth
I-9 procedures require recording:
- Document title
- Issuing authority
- Document #
- Expiration date (if any)
NIST 800-63 section 8.2.3 states the following regarding Level 3 and the Soft cryptographic token:
Soft cryptographic token: a cryptographic key stored on a general-purpose computer. Hardware tokens validated at FIPS 140-2 Level 1 or higher may also be used to hold the key and perform cryptographic operations. The claimant shall be required to activate the key before using it with a password or biometric, or, alternatively shall use a password as well as the key in an authentication protocol with the verifier. If a password is employed to unlock the soft token key, the key shall be kept encrypted under a key derived from a password meeting the requirements for Level 2 authentication, and decrypted only for actual use in authentication. Alternatively, if a password protocol is employed with the verifier, the use of the password shall meet the requirements for Level 2 authentication assurance.
I think this means that a "soft cryptographic token" stored on a general-purpose computer is considered a LoA 3 credential if either of the following is true:
1) a password or biometric is required to activate the key. The password would need to meet NIST LoA 2 requirements.
2) the claimant is required to use a password as well as the key in an authenticated protocol with the verifier. In this case the password must meet NIST LoA 2 requirements, which includes password strength described in 8.2.2.4 and maybe 8.2.2.3.
Regarding option # 1, after the soft PDC enters the hands of the user, we cannot technically enforce use of a password that would meet NIST LoA 2, to activate the key.
Question (probably for Internal Audit): If we created a policy that required the user to maintain a password on a soft PDC (or P12 key store) that met NIST LoA 2 strength requirements, and issued a soft PDC that initially met the LoA 3 requirements, would that justify our recognizing that soft PDC as a Virginia Tech LoA 3 credential according to the LoA defined in the VT standard for Personal Digital Identity Levels of Assurance? http://www.it.vt.edu/publications/pdf/Standard_for_Personal_Digital_Identity_LOA_Final-09June2010.pdf
Regarding option #2, according to Randy Marchany, http://computing.vt.edu/accounts_and_access/pickinggoodpasswords.html has the strength rules and they are close to the NIST criteria. I think Table A-1 shows that with our password strength rules and "dictionary" check, we're in the acceptable range.
Question: If a soft PDC were issued that initially me the LoA 3 requirements, and if it were subsequently used in combination with, say the PID and its password, would that justify our recognizing that soft PDC as a Virginia Tech LoA 3 credential according to the LoA defined in the VT standard for Personal Digital Identity Levels of Assurance? http://www.it.vt.edu/publications/pdf/Standard_for_Personal_Digital_Identity_LOA_Final-09June2010.pdf
- Edit Oct 21
Greg Kroll says:
Link to pdf of NIST 800-63 http://csrc.nist.gov/publications/nistpubs/800-63/SP8...
Link to pdf of NIST 800-63 http://csrc.nist.gov/publications/nistpubs/800-63/SP800-63V1_0_2.pdf
2 Comments
Mary Dunker
Aug 06, 2010Mary pointed out the different requirements for recording ID-proofing documents to the CIC Silver group. On August 6, 2010, she was informed that this item will be brought to the attention of the InCommon Silver Technical Advisory Committee, to be included in a package of proposed changes to InCommon Silver profile.
Greg Kroll
Nov 02, 2010On behalf of Mary Dunker & Randy Marchany:
----Original Message----
From: Dunker, Mary
Sent: Wednesday, October 20, 2010 2:47 PM
To: Marchany, Randolph; Kroll, Greg; Marchany, Randolph
Cc: Stewart, Jeb
Subject: RE: Soft PDC and LoA 3
Thanks for your responses, Randy.
Randy Marchany said:
> Are you trying to keep the users from dumbing down their passwords by enforcing LoA 2 criteria?
Mary Dunker responded:
Yes, that would be desirable. I don't think we have the technical ability to enforce password requirements on the soft PDCs. We can always make recommendations, but the question was regarding policy. We can ask Internal Audit.
Randy Marchany said:
> We'll be
> glad to be in a future meeting with the group . Let us know when it'll be.
Mary Dunker responded:
Our standard schedule is every Thursday from 4:00-5:00, and these soft PDC meetings are showing up on your Exchange calendar when I view it. If we have a meeting where we would like some particular input from the Security Office, I'll try to remind Greg to send you an additional e-mail regarding availability.
Greg, an item for tomorrow's agenda (in addition to Kim's report on certificate usability.)
Thanks,
Mary
----------------------------------------
Mary Dunker
Director, Secure Enterprise Technology Initiatives
Virginia Tech Information Technology
1700 Pratt Drive
Blacksburg, VA 24060
540-231-9327
dunker@vt.edu
-----------------------------------------
> ----Original Message----
> From: randy.marchany@gmail.com randy.marchany@gmail.com On Behalf
> Of randy marchany
> Sent: Wednesday, October 20, 2010 1:20 PM
> To: Dunker, Mary; Stewart, Jeb; Kroll, Greg; Marchany, Randolph
> Subject: Re: Soft PDC and LoA 3
>
Mary Dunker asked:
> If the VTCA issues a soft PDC at LoA 3; i.e., in-person with all the
> appropriate identity-proofing steps, we can require a strong password on
> the P12 key store at initial issuance, but we cannot technically prevent
> the user from changing the password to something less strong or removing
> the password altogether. We could have a policy that required a strong
> password on the soft PDC.
>
Randy Marchany responded:
> What is the application that will use a soft PDC at LoA 3? Is there a list
> of business processes that will require LoA3?
>
> Since you can't technically prevent someone from "dumbing down" their
> password, any policy on this issue is unenforceable. I think creating a
> recommendation doc is appropriate.
>
Mary Dunker asked:
> NIST 800-63, section 8.2.3 Level 3, includes a "soft cryptographic
> token" stored on a general-purpose computer as LoA 3 if either of the
> following is true:
>
> 1) a password or biometric is required to activate the key. The
> password would need to meet NIST Loa 2 requirements.
>
> 2) the claimant is required to use a password as well as the key in
> an authenticated protocol with the verifier. In this case the password must
> meet NIST LoA 2 requirements, which includes password strength described in
> 8.2.2.4 and maybe 8.2.2.3.
>
Randy Marchany responded:
> Entropy is the measure of the difficulty in guessing or determining a
> password or a key. I assume you're talking about the requirements in
> Appendix A in 800-63.
>
Mary Dunker asked:
> Question: If we required the user to maintain a password (by policy)
> on the key store that met the NIST LoA 2 criteria, could we meet LoA 3
> under item #1 above? I realize this may be a question for Internal Audit,
> but I would appreciate your input.
>
Randy Marchany responded:
> I think you can although I always assume that the lowest common denominator
> rule applies. Are you trying to keep the users from dumbing down their
> passwords by enforcing LoA 2 criteria?
>
Mary Dunker asked:
> Question: If we use the soft PDC in conjunction with PID, HOKIES, or
> Oracle password, as described in item 2, do these VT passwords meet the
> NIST criteria for password strength?
>
Randy Marchany responded:
> http://computing.vt.edu/accounts_and_access/pickinggoodpasswords.html has
> the strength rules and they are close to the NIST criteria. I think Table
> A-1 shows that with our password strength rules and "dictionary" check,
> we're in the acceptable range.
>
Mary Dunker asked:
> Please let me know your thoughts. Also, it would be great if you or
> another technically-savvy person from the Security Office could attend a
> meeting to discuss these questions. Please let us know if Greg should put
> that discussion on the agenda for this Thursday, October 21, or for a
> subsequent meeting.
>
Randy Marchany responded:
> We'll be at the VASCAN conference Thursday/Friday of this week. We'll be
> glad to be in a future meeting with the group . Let us know when it'll be.
>
> -r.