Submitted an MRL issue highlighting areas the decoy selection algo could improve, and included 4 ways to improve them. Feedback welcome :)

.

Hey everyone,

I've recently completed my master thesis on comparing various scalable ring signatures / ZKPs (Triptych, Lelantus, RingCT 3.0). During this, I think I have discovered a new way to even further improve upon Triptych.

With this, I'm able to bring down the signature size of Triptych from 3 log(N) to just 2 log(N).

Security-wise, I believe I have successfully reduced my construction to discrete log so it should still be on par with Triptych in that regard (no dual-target dlog or anything like that).

In terms of verification performance, it should produce no noticeable overhead (although it may mess with batch verification, I haven't really looked into that yet).

I'll probably try to publish this new construction as a separate paper at some point, but in case anyone's interested in already having a look at it I've copied out all the relevant parts of my thesis and uploaded them here: github.com/nkraetzschmar/triptych-p…/releases/download/v0.1.0/paper.pdf

Would this be of any use to Monero? I'd love any feedback or maybe for someone to point out if I overlooked any glaring flaws in my security proof 😅

nwk: Congrats on your thesis :)

Thanks 😊

I will add it to my reading list

nwk: I notice the paper does not have an email address. Do you have a preferred contact method?

UkoeHB: Either just message me on here, or if you want I can pm you my email (don't want to post contact details publicly 😅)

Nice! Looking forward to reading this nwk

Thanks sarang 😊