Central Overview Repository for Using Explicit (Host-to-Network) Flow Measurements for Network Tomography

This is the central overview repository for the source code accompanying our paper Using Explicit (Host-to-Network) Flow Measurements for Network Tomography. It provides the general capabilities for reproducing our evaluation and relies on a patched version of ns-3.36.1.

Publication

This work has been created in the context of the following publication:

• Ike Kunze, Constantin Sander, Alexander Ruhrmann, Niklas Rieken, and Klaus Wehrle: Using Explicit (Host-to-Network) Flow Measurements for Network Tomography. In ANRW '25: Proceedings of the ACM/IRTF Applied Networking Research Workshop 2025

If you use any portion of our work, please consider citing our publication.

@inproceedings{2025_kunze_efm-network-tomography,
  author = {Kunze, Ike and Sander, Constantin and Ruhrmann, Alexander and Rieken, Niklas and Wehrle, Klaus},
  title = {{Using Explicit (Host-to-Network) Flow Measurements for Network Tomography}},
  booktitle = {{Proceedings of the ACM/IRTF Applied Networking Research Workshop 2025}},
  year = {2025},
  doi = {10.1145/3744200.3744769},
  url = {https://www.comsys.rwth-aachen.de/publication/2025/2025_kunze_efm-network-tomography/2025_kunze_efm-network-tomography.pdf},
  publisher = {{ACM/IRTF}},
}

Content

Folder	Purpose
automation	Contains the main simulation orchestration which this repository uses for conducting the experiments
observer-placement-generator	Contains scripts for generating different observer placements
plotting	Contains scripts for plotting the evaluation results
setup-scripts	Contains scripts for the initial setup of the simulation and evaluation environments
raw-result-processing	Contains code that analyzes and processes the raw simulation results and extracts the contained information

Initial Setup

1. Clone this repository into desired path (WITH submodules!)

git clone git@github.com:COMSYS/ns3-efm-study-overview.git --recurse-submodules

2. Initialize ns-3 simulation and evaluation environment

bash ./setup-scripts/setup-all.sh

3. Create python virtual environment and install requirements

python3 -m venv venv && source venv/bin/activate && pip install -r requirements.txt

Experiments

Our experiment pipeline uses different wrappers, which are designed to facilitate conducting large numbers of simulation and analysis runs. For this, we make use of extensive configuration files (see the corresponding README for instructions on how to create the configuration files).

To run simulations based on configuration file Test-Simulation.json:

python3 automation/run-simulations.py -m Test-Simulation.json

To analyze simulation results based on configuration file Test-Analysis.json:

python3 automation/run-analyses.py -m Test-Analysis.json

Results Reproduction

Should you aim to reproduce our results, please follow these steps.

1. Adjust paths in simulation and analysis configurations. These paths define where you intend to store the results of the simulation and the analysis. Please provide absolute paths.

2. Generate and execute auxiliary simulation configuration file.

## Generate an auxiliary simulation configuration that provides additional input for later (more specifically, it explores the paths the ping flows will later take)
python3 automation/simulation/configuration_generator_sim.py -m generation-config-sim-PING-ROUTES.json
python3 automation/run-simulations.py -m PINGROUTE-Sim.json

3. Generate simulation and analysis configurations

## Generate three sets of simulation configurations
python3 automation/simulation/configuration_generator_sim.py -m generation-config-sim-ANRW-5.json
python3 automation/simulation/configuration_generator_sim.py -m generation-config-sim-ANRW-10.json
python3 automation/simulation/configuration_generator_sim.py -m generation-config-sim-ANRW-11.json

## Generate three sets of corresponding analysis configurations
python3 automation/analysis/configuration_generator_analysis.py -m generation-config-analysis-ANRW-5-FlowAndMonitorSelection_PERC.json
python3 automation/analysis/configuration_generator_analysis.py -m generation-config-analysis-ANRW-10-FlowAndMonitorSelection_PERC.json
python3 automation/analysis/configuration_generator_analysis.py -m generation-config-analysis-ANRW-11-FlowAndMonitorSelection_PERC.json

4. Execute simulations

## Run the simulations. Note that this can take a while.
python3 automation/run-simulations.py -m ANRW-5-Sim.json
python3 automation/run-simulations.py -m ANRW-10-Sim.json
python3 automation/run-simulations.py -m ANRW-11-Sim.json

5. Execute analysis as soon as simulations have completed

## Execute the analysis. Note that this can take a while.
python3 automation/run-analyses.py -m ANRW-5-Eval-FlowAndMonitorSelectionPerc.json
python3 automation/run-analyses.py -m ANRW-10-Eval-FlowAndMonitorSelectionPerc.json
python3 automation/run-analyses.py -m ANRW-11-Eval-FlowAndMonitorSelectionPerc.json

6. Adjust paths in the main plot configuration.

{
    "plot_cache_folder": "/REPLACE/WITH/PATH/TO/WHERE_YOU_WANT_TO_STORE_PLOT_CACHE_FILES",
    "plot_output_folder": "/REPLACE/WITH/PATH/TO/WHERE_YOU_WANT_TO_STORE_PLOTS"
}

7. Plot the results

python3 plotting/plot-main.py 

Note that you will likely not be able to exactly replicate our results as the generation of the simulation configuration files entails randomness.