Re: [T2TRG] dtls resume

[Daniel Lux <[e-mail]>]

Thanks, I was searching for that document.
He is however if I read it right, suggesting to change the DTLS record
layer header and not the nonce used for the cipher.
Making a specific cipher for constrained devices, 
where the nonce is reduced to 6 or even only 5 bytes (requiring re-
negotiation when roll over is approaching),
would not require to change the DTLS record layer header.
I don't know however what is easier to push through, changing the
header or introducing a new cipher?
/daniel
On Thu, 2016-11-03 at 20:09 +0100, Hannes Tschofenig wrote:
> Section 4 of
> https://tools.ietf.org/html/draft-fossati-tls-iot-optimizations-00
> 
> Ciao
> Hannes
> 
> 
> On 11/03/2016 09:39 AM, Daniel Lux wrote:
> > 
> > During our meeting in Ludwigsuburg at the implementers workshop we
> > discussed the DTLS resume problem.
> > This problem occurs when a constrained device behind a NAT
> > communicates
> > with a server and the NAT changes the
> > UDP source port or the IP address of the sending constrained
> > device,
> > while the server uses the source address and port number
> > to identify the DTLS connection.
> > 
> > One other solution than adding a parameter to the DTLS records
> > would be
> > to construct the IV/NONCE in a way that
> > allows the server to identify the session from that information.
> > This would mean that we would need to specify a new cipher suite,
> > but
> > otherwise it might be an elegant solution.
> > Has anybody suggested this kind of solution to the resume problem?
> > 
> > Kind regards
> > 
> >   Daniel
-- 
Daniel Lux
Tlf: ([phone]
Seluxit
Hjulmagervej 32B
9000 Aalborg
Denmark

 	 We are currently working on:
* New gateway version
* ENCOURAGE EU project
* Mobile Phone program