Re: [Secdispatch] EDHOC Summary

[Tero Kivinen <[e-mail]>]

Göran Selander writes:
> [GS] As I mentioned in my recent reply, given the changes you make
> to TLS to make message sizes on par with EDHOC, it is a new protocol
> so the statement about relying on the analysis of TLS is
> questionable. Comparing implementations there are clearly more of
> TLS, but, again, this is a new protocol.

I do not think it is new protocol, it could be done by just applying
static header compression to TLS.

Making static context header compression rules, that will omit for
example most of the TLS framing which are not needed (as they are
static) will allow take normal TLS frame, compress it to much smaller
frame to be sent to other end.

Then on the other end the static context header compression rules will
decompress the frame and will get exactly same full TLS frame that was
in the sender end. From the TLS library point of view the packets are
full TLS 1.3 packets without any modifications. There might be some
constrains what kind of TLS 1.3 frames the compression can do, for
example there might be constraints that there cannot be any generic
extensions, i.e., the TLS library would need to be configured so it
will never even try to use them, the length of nonce might be fixed to
specific length etc.
-- 
[e-mail]