RE: Multi-Factor Authentication Concern

["Uber Wannabe" <nucleargeekdown () gmail com>]

Just thought I'd throw this in, too...

Just because the standard definition of each word in "Multi-factor
authentication" could be understood as multiple people and multiple factors,
it is the combination of the words that generates its unique meaning.

If we were to go by standard definitions, then...

Door - noun

1.      a movable, usually solid, barrier for opening and closing an
entranceway, cupboard, cabinet, or the like, commonly turning on hinges or
sliding in grooves.
2.      a doorway: to go through the door.
3.      the building, house, etc., to which a door belongs: My friend lives
two doors down the street.
4.      any means of approach, admittance, or access: the doors to learning.
5.      any gateway marking an entrance or exit from one place or state to
another: at heaven's door.

(Source:  www.dictionary.com)


...a house would be a door.  (See #3)


Which shows why logic should not oppose fact.


-- N/A


-----Original Message-----
From: listbounce () securityfocus com [mailto:listbounce () securityfocus com] On
Behalf Of Justin Ross
Sent: Thursday, August 16, 2007 11:37 AM
To: Tep, Tom M. (CDC/CCHP/NCCDPHP); security-basics () securityfocus com
Subject: RE: Multi-Factor Authentication Concern

I agree. Neither "Bob" nor Chris are wholly incorrect, nor wholly
correct. It's semantics, and the definition is in and of itself wholly
subjective to the requirements, the people implementing it, or it's use.

I also agree that generally speaking, when the INFOSEC community talks
about multi-factor authentication they are talking about a single person
- I think that is a far cry from saying "it ALWAYS refers to".

For example here is one of the arguments made:
Take a look at the wikipedia article again. At the end, it contains
this:
"The U.S. Government's National Information Assurance Glossary defines
strong authentication as:
Layered authentication approach relying on two or more authenticators to
establish the identity of an originator or receiver of information. " 

The originator or receiver of the information, could be a database, it
could be a person, it could be a role (for example, the on duty NOC
manager), multi-factor authentication does not have to be a single
person. That was my point and actually validates "Bob's" argument to
some degree.

As I also mentioned, multiple people/computers using different
verification methodologies of a signature on a document, would also by
definition of the words (from a standard dictionary) qualify as
"Multi-factor authentication of a signed document", which is also not a
single individual person. 

While from a security perspective you would never really want to
authenticate based on a role, it is possible, and it's possible to have
it tied to different forms of authentication. Thereby, making it
multi-factor authentication of something other then a individual user. A
VPN peer, can use a shared secret ("something you know" - like a pin),
an IP address ("who you are" like a call-back number on RAS dial-up or a
thumbprint), a digital certificate ("something you have" like a debit
card) which would therefore count as multiple-factor authentication of a
VPN peer, which is a device not a person. 
  
While I did enjoy reading the responses, including the one that which
relates the responder's experience, I have to say if there is one thing
that I have learned in my extensive experience in INFOSEC is that rarely
(if ever) is anything black or white. 


> > So I've scoured the net and books for something that describes
multi-factor authentication as requiring that all factors identify the
same person. So far, I can't find anything. > >
I do believe multiple-factor authentication means (even in the
dictionary sense of the wording): multiple methods of authenticating a
single "entity" (person, government, company, device, et al.); it does
not necessarily have to be a (same or single) person.

I stand by that belief, even though I am clearly outnumbered hah ;)

Regards,

Justin.Ross
Security Engineer



-----Original Message-----
From: listbounce () securityfocus com [mailto:listbounce () securityfocus com]
On Behalf Of Tep, Tom M. (CDC/CCHP/NCCDPHP)
Sent: Wednesday, August 15, 2007 6:23 AM
To: security-basics () securityfocus com
Subject: RE: Multi-Factor Authentication Concern


Based from everyone responses, neither Bob nor Chris are incorrect in
their understanding.  It depends on the company security policy.  I
believe what Bob is referring to is the Limited Access Privilege in
Physical Security Policy. It requires multiple parties' involvement in
order to grant a person access to a secure room.  On the other hand,
Chris is talking about the multi-factor authentication in system login
which implemented a little differently and require three important
things in Authentication:

 1.  Something you know (i.e Password)
 2.  Something you have (id badge or cryptographic key)  3.  Something
you are (a voice print or other biometric)

It DEPENDS!!!!

Hope I haven't confused anyone.

`tom 


 -----Original Message-----
From: Mike Lococo [mailto:mike.lococo () nyu edu]
Sent: Tuesday, August 14, 2007 2:59 PM
To: security-basics () securityfocus com
Subject: Re: Multi-Factor Authentication Concern

> I looked at all of the suggested links, including the Wikipedia 
> article, and I cannot find anything that explicitly states that the 
> factors in a multi-factor authentication system must all be from the 
> same person.

Because authentication is, by definition, the process of verifying an
asserted identity (that statement is easy to find references for,
including the wikipedia article on authentication).  An access control
system must authenticate _each_ identity separately, even when several
identities are involved in a single transaction and even if the process
is streamlined to 'feel' as though it's a single action.  As you're
thinking and speaking about this, remember the difference between
identification, authentication, and authorization.

1) Identification:  Your identity is your username in the system.  You
may have to say it, or type it, or it may be inferred from a retinal
scan or whatever.  As a basic access control principle, every individual
must have an identity.  Anytime you're accepting credentials from more
than one individual, you are _by_definition_ performing more than one
authentication.

2) Authentication:  An identity is authenticated via password, or
voiceprint, or token, or whatever.  If only one type is required, it's
single factor.  If more than one type is required, it's multi-factor.
If more than one type is available (you have a token and a password),
but either is sufficient (you can log in with your password even if you
lost the token), it's still single factor... you just have options.

3) Authorization:  Once you are authenticated, you may or may not be
_authorized_ to access the resource you're interested in.  If a system
requires more than one user to authenticate in order authorize an
action, it implements split-authentication or split-authorization (often
referred to in the context of passwords/pins as split-knowledge).  Each
identity is still authenticated individually, but more than one is
required before any are authorized.

You're talking about multi-factor authentication.  Your friend is
talking about split-knowledge/authentication/authorization.  No
authoritative source on IDM or access-control is going to talk about
whether multi-factor authentication involves multiple identities because
it's well-established that all authentication schemes have as their
basic goal the verification of a single asserted identity.
Authorization schemes exist that require multiple identities to be
involved in a single transaction (nukes and expensive safe-deposit boxes
work this way), but each is always authenticated individually.

Thanks,
Mike Lococo