Cyphal 1.0 will be GA soon (we’re just cleaning up some loose ends on the Beta and then it’s v1 forever) so I’ve been thinking a bit about V1.x in general and secure communications specifically. I’m not ready to post a large thread on this but I wanted to enter this paper from Carnegie Mellon into the discourse to give people time to read it and think about various schemes for authenticated Cyphal exchanges. This is different from encrypted messaging, of course, but for control systems encryption is typically less important that authentication since tampering is the key attack vector.
The paper proposes an interesting solution that I think could be expanded, refined, and adapted for Cyphal.