P2P Privacy §
Research §
- Nomos Mix simulations: message dissemination time by queue type - Session 1: Measured message dissemination time using 20~80 number of nodes, for each 5 queue types. Noisy Coin-flipping queue shows the lowest dissemination time when there are less number of real messages flown in the network, but shows the worst, if not. Random Sampling queue or Permuted Coin-flipping queue show around 2x longer dissemination time compared to the Non-mixing queue, but the performance degradation is relatively low, as the number of real messages increase. This is the initial interpretation of the result, which we can’t make a conclusion based on.
- Nomos Mix simulations: message dissemination time by queue type - WIP: Session 2: This experiment is still running. This runs more nodes (100~10000) than the session 1, to see how each queue performs in the larger network. Also preparing next experiments: measuring how long messages stay in the queue.
- Extended the methodology document with more parameters for message dissemination experiments. Found one paper that measures properties of the Monero network which also include the node connectivity levels, an extract of it can be found here.
- Designed another type of experiment, that is devoted to observing the quality of “temporal mixing” for a single node on a single path. The results of which will give us better understanding of the predictability of delaying of each of the queues and guide our decision during the selecting process.
- The results for the first session of simulations give some intuition about the impact of each of the parameters on the dissemination time. However, it’s too early to conclude as the quality of mixing is another part of the equation and it’s our main focus now.
- Started looking at the payment mechanism for the mixnet.
- Analysis of the queueing system in the Nomos Mixnet node: considered the FIFO (First In First Out) attack where two sender nodes send messages through k mix nodes to the receiver node. The latter is controlled by an adversary which is also capable of observing other nodes. For a mix node where a message is removed from its out-queue with probability q, derived an expression for the probability of success of the FIFO attack. This probability can be computed numerically by creating a large population of random variables (wrote a code for the latter). Using this framework, showed that the probability of success of the FIFO attack is a monotonic decreasing function of the number of mix nodes k and is bounded from below by ½. Also showed that having different q parameters for sender and mix nodes can reduce the probability of success of the FIFO attack. Finally, showed that heterogeneous latencies of connections can increase the probability of success of the FIFO attack.
Development §
Data Availability §
Research §
Development §
- PR #686: DA BlobInfo instead of Certificate - Removed Certificate and Attestation definitions, introduced DispersedBlobInfo trait and implementations. Also updated the DA mempool, integration tests and other components to use BlobInfo.
- (WIP) DA Mock network service for CLI app: still in progress preparing various components for the updated CLI App.
- PR #685 - finished implementing/fixing the DA network replication layer.
- (WIP) Started implementing the DA network dissemination layer - finished the validator behaviour and started the executor behaviour.
PPoS/Consensus §
Research §
Development §
Coordination Layer §
Research §
- PR #17: integrate the zone withdrawal logic with CL ptx/note structure.
- PR #13: users could create as many zero-valued outputs as they’d like without affecting the balance of a partial tx, the implications of this was not clear so we decided to ban zero-valued outputs for now until we understand what is going on.
- PR #14: boilerplate for the user atomic transfer ptx.
Development §
Testnet + Insights §
Development §
Research §
Development §
Miscellaneous §