Owner: @Youngjoon Lee

Reviewers: 🟢@Marcin Pawlowski

Purposes

• Verify that the actual inbound transmission rate matches the intended persistent transmission rate when all Tiers 1-3 are integrated, even with immediate forwarding.
• Determine appropriate tolerance thresholds for identifying malicious or unhealthy connections.

Related Specifications

• Note about frequencies
• Connection Maintenance
• Frequencies Measurement

Experiment Setups

• The https://github.com/logos-co/nomos-simulations is used to run experiments using the Rust implementation.

  nomos-simulations/simlib/blendnet-sims at mix-conn-maintenance · logos-co/nomos-simulations

• Common Parameters

  | network latency | data msg lottery interval | persistent transmission rate | temporal processor max delay | blend hops | conn monitor time window | | --- | --- | --- | --- | --- | --- | | 40ms | 20s | 1 msg/s | 10s | 2 | 20s |

• Network topology construction

  • Network topology is created at the beginning of each experiment, and does not change.
  • The simulation ensures that each node has exactly peering degree connections with randomly chosen peers.

Results

| network size | peering degree | network diameter | cover traffic epoch / slot duration | # cover msgs generated by network per epoch | # cover msgs generated per node per epoch | min rate | avg rate | max rate | malicious tolerance | unhealthy tolerance | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | 10 | 3 | 2.10 | 20s / 1s | 10 | 1 | 0.20 | 0.75 | 1.30 | 0.30 | 0.80 | | | 6 | 1.29 | | | | 0.50 | 0.90 | 1.25 | 0.25 | 0.50 | | | 9 | 1.05 | | | | 0.55 | 0.95 | 1.35 | 0.35 | 0.45 | | 100 | 3 | 4.19 | 200s / 1s | 100 | 1 | 0.05 | 0.75 | 1.90 | 0.90 | 0.95 | | | 6 | 2.57 | | | | 0.15 | 0.90 | 2.05 | 1.05 | 0.85 | | | 9 | 2.10 | | | | 0.25 | 1.00 | 2.10 | 1.10 | 0.75 | | | 20 | 1.54 | | | | 0.30 | 1.05 | 2.10 | 0.95 | 0.70 | | | 40 | 1.25 | | | | 0.25 | 1.05 | 2.00 | 1.00 | 0.75 | | 1000 | 3 | 6.29 | 2000s / 1s | 1000 | 1 | 0.05 | 0.70 | 1.70 | 0.70 | 0.95 | | | 6 | 3.86 | | | | 0.15 | 0.90 | 2.15 | 1.15 | 0.85 | | | 9 | 3.14 | | | | 0.15 | 0.95 | 1.75 | 0.75 | 0.85 | | | 20 | 2.31 | | | | 0.15 | 1.05 | 1.95 | 0.95 | 0.85 |

Persistent Transmission Rate

• Although the protocol forwards incoming messages immediately to the connected peers, the actual inbound transmission rate does not differ significantly from the intended transmission rate of 1 message per second. This aligns with the expectations set during the design phase.
• It is because the cover traffic parameters have been set to generate cover messages every blend_hops = 2 seconds approximately, as recommended in the Assumptions. Although each node forwards incoming messages immediately, the messages are initially generated at relaxed intervals, resulting in an actual transmission rate that remains close to 1.

Effective Emission Frequency Tolerance

The protocol expects that all nodes adhere to the Effective Emission Frequency ($EEF$). If a node does not comply with it, the node is marked as either malicious or unhealthy. A tolerance parameter $\delta_{EEF}$ is used to accommodate minor variations.