Re: [Sframe] Inquiry about matrix authentication in SFrame

Richard Barnes <rlb@ipv.sx> Wed, 27 September 2023 13:15 UTC

MIME-Version: 1.0
References: <CAGSk29fZfZp=twfy=ewt9aXA-LC=OK8=H0Jr29EhvVci-QvMaQ@mail.gmail.com>
In-Reply-To: <CAGSk29fZfZp=twfy=ewt9aXA-LC=OK8=H0Jr29EhvVci-QvMaQ@mail.gmail.com>
From: Richard Barnes <rlb@ipv.sx>
Date: Wed, 27 Sep 2023 09:15:38 -0400
Message-ID: <CAL02cgRt9EOAn7tPTa2CthXtAy6HysBWxKft_+dmivESzdOiKA@mail.gmail.com>
To: 酒見由美 <yumi.sakemi@gmo-cybersecurity.com>
Cc: sframe@ietf.org, Satoru Kanno <satoru.kanno@gmo-cybersecurity.com>
Content-Type: multipart/alternative; boundary="000000000000713630060656fd4f"
Archived-At: <https://mailarchive.ietf.org/arch/msg/sframe/XT3WX4t6wbcL4raIaVMQ_JoynbU>
Subject: Re: [Sframe] Inquiry about matrix authentication in SFrame
Precedence: list

Hi Yumi,

Sorry for the delayed reply.  I may have mis-spoken at the side meeting;
"matrix authentication" does not exist.

The authentication that SFrame provides is typical symmetric-key
authentication, of the sort that HMAC or an AEAD function provides.  In a
group setting, that means that a receiver is assured that media came from
the sender **or any other receiver**, since the sender and the receivers
all have the same symmetric key.

Additional authentication properties can be added by virtue of the method
used for distributing SFrame base keys.  For example, in the MLS-based key
distribution scenario described in the document, the senders and receivers
authenticate to each other in MLS, using public-key credentials.  So each
receiver can authenticate identities for the other senders/receivers at the
MLS layer, but at the SFrame layer, a receiver could still spoof packets as
if they were from a sender.

Hope that helps,
--Richard

On Wed, Aug 30, 2023 at 1:31 AM 酒見由美 <yumi.sakemi@gmo-cybersecurity.com>
wrote:

> Dear Richard and members participating in the SFrame side meeting at
> IETF117
>
> I participated in talking about efficient (low-latency) hashing
> algorithms at the SFrame side meeting at IETF117.
> Thank you for allowing me to talk there.
>
> At the side meeting, I talked about it because I expected that
> ultra-low latency cryptography "Areion," which my team is focusing on,
> would help make SFrame more secure.
>
> As a comment at that time, Richard shared information that while
> SFrame does not have per-sender authentication, it has introduced a
> technique called matrix authentication.
> Since then, I have read the SFrame I-D and other documents, but I need
> help finding information about features and mechanisms of matrix
> authentication.
>
> Therefore, I would appreciate it if you could share any papers or
> documents on Matrix Authentication.
>
> Best regards,
> Yumi
>
> --
> Yumi Sakemi, Ph. D.
> GMO Cyber Security by IERAE, Inc.
> yumi.sakemi@gmo-cybersecurity.com
>

[Sframe] Inquiry about matrix authentication in S… 酒見由美
Re: [Sframe] Inquiry about matrix authentication … Richard Barnes