Code for the article "Analysis of information dissemination through Direct Communication in a moving crowd"

Contents of this repository

This repository contains

• the source code of the CrowNet simulator. It is included as submodule.
• the source code and the result data that we use in the forward propagation and in the sensitivity analyses.

CrowNet

CrowNet is developed by the members of the research project Improving the efficiency of traffic infrastructures by robust internetworking (roVer) at the Hochschule Muenchen University of Applied Sciences.

See https://www.hm.edu/en/research/projects/project_details/wischhof_koester/rover.en.html for information about the roVer project.

Currently, CrowNet is embedded as git-submodule rover-main. We plan to change the repository name from rover-main to CrowNet in 2021.

Result data

The results data files are stored in uq/.. next to the scripts that are used to generate the data.

Scripts for uncertainty quantification

The scripts used for uncertainty can be found under uq/...

System requirements (hardware)

A system with >=250GB RAM and >=80 cores is required, because the simulations take ~6 days.

Init and update submodules

Init the submodules if necessary Use

git submodule init
git submodule submodule --recursive
cd rover-main
git submodule init
git submodule submodule --recursive
cd ..

To produce the results presented in the MDPI article, use

./checkout_mdpi_state

Set up the simulation model

Use

./install_rover
./pull_images.sh

to install the simulation model.

Run the simulations

We use the python package suq-controller to run the simulations and to collect the results. Install the virtual Python environment

cd uq
./install_venv

Activate the virtual environment

source .venv/bin/activate

Start the forward propagation

python3 forward_propagation_1/forward_propagation.py

If the simulation fails, restart the script.

Analyse the results

After the simulation has finished, we analyse the results. We compute the statistics of the resulting empirical distribution.

python3 forward_propagation_1_analyse/statistics.py

We analyse why information dissemination sometimes failes

python3 forward_propagation_2/forward_propagation_2.py
python3 forward_propagation_2_analyse/export_data.py

We use sensitivity analysis to quantify the influence of the parameters

python3 forward_propagation_1_analyse/sensitivity_analysis.py

We use a kriging model and repeat the sensitivity analysis

python3 forward_propagation_1_analyse/sensitivity_analysis_stochastic.py