eap Mailing List: Re: Re: EAP key binding discussion

[Jari Arkko (jari.arkko]

Rafa Marin Lopez wrote:

> After I read this, I think it seems to be similar to PANA when PAA is 
> not colocated in the EP for example in the wireless LAN 
> model(http://www.ietf.org/internet-drafts/draft-ietf-pana-framework-03.txt). 
>
>         In fact PaC/EAP peer can derive one PMK per each EP/AP 
> controlled by PAA.  PAA can derive the same keys that would send to 
> different EPs controlled by this PAA (how to derive PMKs per each 
> EP/AP derived from AAA-key that PAA receives  from AAA is ongoing 
> work). So PAA would be like your LKDC. However PAA is also acting as 
> NAS that is a difference what you propose.
>
> On the other hand , when a EAP peer moves to another EP/AP controlled 
> by another LKDC then we are moving the problem to allow a fast handoff 
> between LKDCs. In the case of PANA , it is being treated in 
> http://www.ietf.org/internet-drafts/draft-bournelle-pana-ctp-02.txt 
> and http://www.ietf.org/internet-drafts/draft-ietf-pana-mobopts-00.txt
Moving the key controller node higher in the hierarchy does help a lot.
Both in your example above as well as in various L2 concentrator designs.

Regarding the derivation of "PMKs" per EP, isn't that already specified
in Section 5 of draft-ietf-pana-ipsec-05.txt -- the keys are different 
per EP
address and session ID.

--Jari