eap Mailing List: RE: Strawman -10/EMSK deletion requirement?

[Glen Zorn (gwz) (gwz]

Jari Arkko <mailto:jari.arkko [at] piuha.net> supposedly scribbled:

> Glen Zorn (gwz) wrote:
> 
>> 
>>>> 
>>>> 
>>> The problem is that since EMSK is the root, its compromise will lead
>>> to the compromise of all derived keys. This also holds locally for
>>> the AMSK_root_0 approach. That is, if AMSK_root_0 is compromised
>>> then any keys derived from that root are compromised (but not keys
>>> in 
>>> other apps).
>>> 
>>> 
>> 
>> I guess I'm just confused, then (not a first!).  I thought that the
>> major problem was that it is basically impossible to no which (if
>> any) _applications_ a person would choose to use during a session. 
>> If that is accurate, I can't really see how the AMSK_root_0 approach
>> solves the problem.  If that's not the problem, what is?    
>> 
>> 
> Oh. Maybe I was confused. AMSK_root_0 does not solve that problem. It
> solves the problem that if your application requires multiple keys
> (as in fast handoff to AP2, AP3, etc) then you still only need one
> AMSK from the EMSK for the application. That AMSK_root_0 can then be
> used to generate all the keys that the application in question needs.

OK, I guess I just assumed that the "M" in "AMSK" implied that usage.

> 
> --Jari

Hope this helps,

~gwz

Why is it that most of the world's problems can't be solved by simply
  listening to John Coltrane? -- Henry Gabriel