Re: Re: issue 357: Channel Binding Definition
From: Yoshihiro Ohba (yohbatari.toshiba.com)
Date: Tue, 2 May 2006 15:42:47 -0700 (PDT)
On Tue, May 02, 2006 at 03:02:21PM -0700, Narayanan, Vidya wrote:
> I don't think the channel binding information needs to be the same for
> all parties. For e.g., consider a case where the server has a view of
> the SSIDs for which a given MSK is valid. The peer at a given moment may
> only see a subset of the SSIDs - as it moves though, it needs to know
> that a new SSID it sees may or may not use that MSK for TSK derivation. 
> 
> For this reason, actually mixing of the keys with channel binding data
> doesn't quite make sense to me yet. This implies that the peer has a
> complete view of the channel binding data at a given time, which does
> not seem to make practical sense in all models of deployment. 

Since I don't think the complete channel binding data is so large, I don't 
see a practical issue here.  What is deployment scenario in your mind?

> 
> It may make sense for the peer have the entire channel binding data from
> the server - as it moves, it can compare to see if what it sees is a
> subset of that data it received from the server or not. 

I don't think only comparing subset of that data with the server does
not work for Channel Binding verification, as an attacker can still
fool around the rest of data that is not part of the subset.

> Alternately, the
> peer just needs confirmation from the server that what it sees is valid.

This requires channel bindind data to be sent from the peer to server, 
while the server pre-configures the channel bindind data.  This is 
redundant information exchange to me.

Yoshihiro Ohba



> 
> 
> Vidya
> 
> > -----Original Message-----
> > From: Bernard Aboba [mailto:bernard_aboba [at] hotmail.com] 
> > Sent: Tuesday, May 02, 2006 12:23 PM
> > To: Narayanan, Vidya; eap [at] frascone.com
> > Subject: RE: [eap] Re: issue 357: Channel Binding Definition
> > 
> > That makes sense.  Do we also need to indicate that the 
> > channel bindings need to be the same for all parties?  For example:
> > 
> > "Channel Binding
> > 
> > A secure mechanism for ensuring the synchronization and 
> > correctness of channel properties (such as endpoint 
> > identifiers) provided to the EAP peer, authenticator and server."
> > 
> > >Minor clarification:
> > >
> > >"Channel Binding
> > >
> > >A *secure* mechanism for ensuring the correctness of channel 
> > properties 
> > >(such as endpoint identifiers) provided to the EAP peer, 
> > authenticator 
> > >and server. "
> > >
> > >The word secure is to imply that if this data is in fact 
> > sent as a blob 
> > >between the peer and server, it must be integrity protected.
> > >
> > >Vidya
> > 
> > 
> > 
> _________________________________________________________________
> To unsubscribe or modify your subscription options, please visit:
> http://lists.frascone.com/mailman/listinfo/eap
> 
> Arhives: http://lists.frascone.com/pipermail/eap
> 
> 

Results generated by Tiger Technologies using MHonArc.