Introduction

The Nomos Mix simulator:

• implements the Nomos Mix protocol
• runs it during the specified simulation time
• analyzes the bandwidth usages and communication patterns.

Installation

git clone <https://github.com/logos-co/nomos-specs.git> -b mixnet-v2-sim-fin
cd nomos-specs
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Getting Started

Before running simulations with realistic parameters, this section describes how the simulation works and how to configure it, by running it with simple parameters.

Base Settings

This is a setting to produce the base result (i.e. control group) that will be compared to other results later.

simulation:
  # Desired duration of the simulation in seconds
  # Since the simulation uses discrete time steps, the actual duration may be longer or shorter.
  duration_sec: 1000

network:
  # Total number of nodes in the entire network.
  num_nodes: 5
  latency:
    # Maximum network latency between nodes in seconds.
    # A constant latency will be chosen randomly for each connection within the range [0, max_latency_sec].
    max_latency_sec: 0.1
    # Seed for the random number generator used to determine the network latencies.
    seed: 0
  peering:
    # Target number of peers each node can connect to (both inbound and outbound).
    degree: 6

mix:
  # Global constant transmission rate of each connection in messages per second.
  transmission_rate_per_sec: 10
  mix_path:
    # Maximum number of mix nodes to be chosen for a Sphinx packet.
    max_length: 3
    # Seed for the random number generator used to determine the mix path.
    seed: 3

logic:
  sender_lottery:
    # Interval between lottery draws in seconds.
    interval_sec: 1
    # Probability of a node being selected as a sender in each lottery draw.
    probability: 0.000
    # Seed for the random number generator used to determine the lottery winners.
    seed: 10

Discrete Simulation Time

• The simulation uses the discrete time mechanism that doesn’t rely on the real time unit.
  • The simulation processes duration_sec * 1000 discrete time steps, i.e. the time precision is “milliseconds”.
  • If the computations in a single step are not heavy, the step will be done faster than 1 millisecond.
• In other words, the simulation can be finished earlier than duration_sec if computations are not heavy.
  • Of course, it may take longer than duration_sec if computations are heavy.

Network

• Latency
  • A network latency of each connection is chosen randomly within a range [0, network.max_latency_sec].
  • i.e. Each connection will have a different constant latency.
• Peering

  • To build a peer-to-peer network, each node establishes connections with the target number of peers (peering.degree).

    • The target number includes both inbound and outbound connection establishments.

  • In other words, all connections are duplex (bidirectional) in order to simulation the real networking as similar as possible.

    • i.e. Once a connection is established between node A and B, both A and B can send messages through the connection at the same time.

  • For simplicity, nodes are connected as a “ring” topology for now (regardless of peering.degree).

    (node-3) --> node-0  <--> node-1 <--> node-2 <--> node-3 <-- (node-0)
    

    • The simulator will support more complex topology soon.

Mix