Our cryptographers came back with new ideas from the conference week last week. Balazs is working on a lead for speeding up RLN, which could bring proof times down to 10ms, a 6x improvement over the current, already optimized implementation. Given the critical role of RLN on mix performance, this would be no small feat. It is based on this paper.
Our LIONESS cipher spec has finally been merged and closed, after a final round of reviews.
Our first tests with DHT queries over mix have been successful. We are now working on integrating discovery to make it operational on the storage network at large.
The reference implementation for the Ed25519 key blinding that we are planning to deploy in our future hidden-services is now available.
We are making solid progress with NAT traversal, and should have a complete implementation covering the full range of options - from port mapped routers to UPnP/PCP/PMP to relay fallbacks - available for testing within a week or two.
Addressed the review comments on LIONESS specification PR, fixed the math rendering issues, and modified the Mix specification to use LIONESS instead of AES for payload encryption.
Discussed and defined the concrete primitives to be used for the LIONESS specification
studied further promising RLN proof speed optimization opportunities, based on this paper. (Mix network bandwidth would be probably bottlenecked by RLN proofs as of now)