Re: [Secdispatch] Power measurements

[Carsten Bormann <[e-mail]>]

On Mar 6, 2019, at 13:21, Hannes Tschofenig <[e-mail]> wrote:
> 
> Hi all, 
>  
> Yesterday a few measurements of 3GPP & LoRAWAN power measurements were presented. I asked whether the provided data are estimates based on some model or whether they are the results of some measurements.

Imagine the estimates were made on the back of an envelope, informed by some measurements but not by measuring exactly what was postulated.
I understand we are all curious (I certainly am), but for the purposes of secdispatch:
How would it help to know the detailed technique that was employed?

The design technique that led to EDHOC could be termed ALARA (as low as reasonably achievable): If there is an opportunity to reduce the cost, we are happy if that can be used.
Of course, we need to clarify what is “reasonable” (e.g., losing some security was “reasonable” for TLS 1.3 0RTT because it was believed that property can be explained).
Data from the various deployment environments can help us decide whether we are there yet.
But that decision is a bit like the influence of patents: Every one of us has to make it; there won’t be consensus about the criteria, and trying to generate a consensus document about that would be so ludricous that it would be obvious why people want that.

It is good that work on making (D)TLS more compact continues.  It started with DTLS 1.3, and once that is done, we will immediately be able to take advantage of that for CoAP.  There is also the ATLAS work.  Work on what I’ll call cutlass (“cTLS”) is just starting, and I’m looking forward to an even more concise version of (D)TLS.  But it is not a given that what is “reasonable” for EDHOC will turn out “reasonable” for the TLS brand, so I’m not holding by breath.  And I don’t have to, because EDHOC is there.  

Grüße, Carsten