[pim] Re: [MBONED]Q: "open" group multicast sender authentication...

There are two that Jake and I were involved with:

[AMBI](https://datatracker.ietf.org/doc/draft-ietf-mboned-ambi/) uses
asymmetric signatures to authenticate manifests of cryptographic packet
hashes that establish authenticity via chaining to a signed manifest, but
with much less than one signature per packet overhead. This comes at the
cost of handwave latency (i.e., sufficient client ingest buffering)
proportional to the reduction in overhead (since you can't sign something
in the future) and requires sufficient redundancy or timely unicast
recovery to avoid a missed deadline when a frame wants data from a packet
whose authenticity has not yet been established.

[ALTA](https://datatracker.ietf.org/doc/html/draft-krose-mboned-alta) was
something we put together based on a paper from 2001 that described an
algorithm for authenticating a stream of packets in the presence of random
loss, without separate explicit manifests, all while minimizing overhead
and handwave latency relative to a manifest-based solution.

In the fullness of time, ALTA feels like an optimization that might be
worth pursuing once the rest of the technology stack has proven itself.
It's a little complicated to build into the first standardized version of a
solution in this space.

Kyle


On Wed, Oct 29, 2025 at 5:50 AM Toerless Eckert <[email protected]> wrote:

> Hi folks
>
> I am trying to remember if we ever wrote down a mechanism to
> cryptograpically
> authenticate the sender of IP multicast packets - including replay attack
> prevention! - without having to rely on shared keys like in GDOI and hence
> the need to trust a group of receivers.
>
> I am pretty sure that there was at some time an interesting
> cryptographically
> new proposal for this - a few years ago - but i don't remember enough
> keywords
> to find it (which WG, what was the name). Even the AI tools are thoroughly
> useless ;-))
>
> And of course, simple authenticating packets with time-stamps
> signing with certificate is well-known... and well-known expensive...
>
> Cheers
>     Toerless
>
> _______________________________________________
> MBONED mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>

_______________________________________________
pim mailing list -- [email protected]
To unsubscribe send an email to [email protected]