[Standards-JIG] Secure Groupchat (was Re: The Great Encryption Debate)

Justin Karneges justin-keyword-jabber.093179 at affinix.com
Tue Aug 16 19:54:00 UTC 2005


On Tuesday 16 August 2005 11:51 am, Ian Paterson wrote:
> > > Three very different e2e cases have been identified:
> > >
> > >    A. One-to-one (including session, offline and object encryption)
> > >    B. One-to-many
> > >    C. Many-to-many
> >
> > You can scratch many-to-many unless you want a bunch
> > of people sharing a resource.
>
> You're right if we're talking about one-off messages (as in the
> one-to-many case). But if we're talking sessions then many-to-many would
> be valid (e.g. a conference room).

Speaking of conference rooms, I was just looking at some SILC documents, and 
one technique they use is to change the session key every time someone enters 
or exits the room.  The idea behind this is to prevent a user from decrypting 
messages prior to entering (if the key didn't change when he entered) or 
decrypting messages after exiting (if the key didn't change when he exited).

One nice thing about the above method is that it means all participants should 
be available at the time of key negotiation (or renegotiation), which may 
open the door to some sort of multi-party diffie-hellman (if such a thing 
exists, I recall it being mentioned in the past).

Ideally, secure groupchat should work over regular groupchat servers, and only 
the participants would have the keys.

Another issue with groupchat is message signing.  With one-to-one sessions it 
is clear who is sending which message (if it wasn't you, then it was the 
other guy).  However, with groupchat it seems that the stanzas will need to 
be public key signed.  I'm not sure yet how this would affect repudiability.

Anyway, just some ideas to start with.

-Justin



More information about the Standards mailing list