P2P Privacy §
Research §
- The new mixnet executable spec is being updated in the PR #98 but has not been merged yet. The new spec will be reviewed for changes that need to be updated in the executable spec.
- Updated mixnet simulation is in the branch of nomos-specs, a PR hasn’t been opened yet because additional team discussion is needed to check whether it meets our analysis requirements. A document has been prepared to explain what the simulation can do and how its result can be used. Based on this, we will discuss what needs to be done as the next experiment.
- Updated mixnet gadget document with queuing designs: added a section about queuing mechanisms. There are two flavors, one is time bounded and another is time unbounded. Time unbounded is better from the privacy perspective as an adversary will not know any upper time limit during which a message is going to be propagated. Also presented two versions of the unbounded with discussion about the design properties and their impact on privacy.
- Started working on the level 1 noise generation mechanism, it is sketched in the NomMix design already but requires more refinement. In short, the idea is to introduce another type of noise (cover traffic) that will be limited to some threshold (to protect the network from spam) and its generation will be incentivized to encourage nodes to send it.
- Discussed how to proceed with simulations and analysis of the mixnet. One of the outcomes is a design of a new model that can be used to describe the network which is connection oriented (rather than node oriented) - more details in the point below.
- To model communication in the network, considered two approaches: the “node-centred” and “connection-centred” . The former assumes that only the state of a node (such as sending, receiving, etc.) is observed and the latter assumes that also connection is observed. Also, considered Hamming distance as a measure of distance between systems with LEVEL 0 and LEVEL 0 + LEVEL 2 noises, i.e. the difference between a system which “communicates” and “doesn’t communicate”. This work is summarised in this document.
Development §
Data Availability §
Research §
- WIP: (DA python PoC) Executor large number of persistent connections to remote host test - During the offsite, there were some concerns raised that the executor might not be able to keep a large number of open connections to remote hosts. A test using a python DA Node and Executor was performed, having the executor connecting to multiple instances of DA Nodes on a remote machine. Results need to be formally documented, but simple tests of pushing data over thousands of persistent TCP connections to a remote host succeeded on commodity hardware.
- WIP: (DA python PoC) PoC of Executor dispersal and sampling protocol for direct connections - A simple executor node is in progress - initially it was used for the connections feasibility test, now it’s developed further to work with the DA Node used in the subnet PoC.
- Created the Runnable DA PoC Specification.
- Updated the GitHub repo with current code (Status: Simple python DA nodes and simple python DA Executor implemented; executor can send packets) - WIP PR.
- Tweaked the DA encoder benchmark to cross-check data & column sizes.
Development §
PPoS/Consensus §
Research §
- WIP: fixing a bug that was found in the wealth concentration code and producing and analyzing all simulation results again, based on the fix.
Development §
Coordination Layer §
Research §
- CL PoC plan has been added to Notion.
- Did a few more prover optimizations in this PR.
- Summary of all the CL specs has been created.
- The Caulk paper is being studied. The protocol is understood in general terms. The SemaCaulk definitions have been reviewed. A general description and protocol steps have been added for the related protocol here. Since KZG is used instead of a Merkle tree, proof calculations are faster. Creating a proof of set membership involves precomputation and quick proof generation. Precomputation can be done early and updated efficiently, while proof generation is fast and constant-time.
- Research has been done on which other zkVMs could be tested for the CL. Also talked to external contributors to get a second opinion on the topic. Written the initial notes (not very detailed) about the related protocols here.
Development §
Testnet + Insights §
Development §
- Prepared Nomos Explorer architecture document.
Research §
Development §
Miscellaneous §