Getting started

Here you find the code described in [borges2024caviar] (see below), which allows to generate the results in this paper.

*Currently tested only on Ubuntu 22.04 and Python 3.9.16

Pre-requisites

Auxiliary linux packages

First update (if necessary)

sudo apt update

cURL

sudo apt install curl

Unzip

sudo apt install unzip

Setting up the NATS server

Go to https://github.com/nats-io/nats-server/tags , download the latest .deb release and install it.

Installing

1) Clone the project repository

Using SSH:

git clone [email protected]:lasseufpa/caviar.git

Using HTTP:

git clone https://github.com/lasseufpa/caviar.git

2) Set up the 3D scenario

Go to the folder where the 3D executable will be stored:

cd caviar/3d

Download the executable

curl https://nextcloud.lasseufpa.org/s/zdNNfM2YCmfrHsi/download/central_park.zip --output central_park.zip

Unzip the file

unzip central_park.zip

Set it as an executable

chmod +x ./central_park/LinuxNoEditor/central_park/Binaries/Linux/central_park-Linux-DebugGame

Now, go back to the project root

cd ..

3) Install the requirements

We use python 3.9.16*

(Optional): Create and activate a virtual environment with conda

conda create --name caviar python=3.9.16

conda activate caviar

First we should install the package below

pip install msgpack-rpc-python==0.4.1

After this, we install the rest of the requirements

pip install -r requirements.txt

Executing a simulation

In the project root folder, run:

python3 simulate.py

This will start a simulation of the flights defined in caviar/examples/airsimTools/waypoints/trajectories/

To correctly abort a simulation, in the terminal press:

ctrl+C

Configuring the simulation

The configuration parameters used in the simulation are stored in caviar/caviar_config.py

More documentation on the configuration soon

Troubleshoot

On the first run, the drone was teleported to the street, but did not started to fly

Sometimes this error can happen due to the download of YOLO weights. For this case, just exit the current simulation with ctrl+c and try it again.

Got the error: "Can't find libdevice directory ${CUDA_DIR}/nvvm/libdevice. This may result in compilation or runtime failures, if the program we try to run uses routines from libdevice."

This is due to not being able to file the CUDA directory. For this you can execute the following, to add it to your environment variables:

For bash

echo -e "export CUDA_DIR=\"$(whereis cuda | cut -d ' ' -f 2)\"\nexport XLA_FLAGS=--xla_gpu_cuda_data_dir=\"\${CUDA_DIR}\"" >> ~/.bashrc

For zsh

echo -e "export CUDA_DIR=\"$(whereis cuda | cut -d ' ' -f 2)\"\nexport XLA_FLAGS=--xla_gpu_cuda_data_dir=\"\${CUDA_DIR}\"" >> ~/.zshrc

Citation

If you benefit from this work, please cite on your publications using:

@ARTICLE{borges2024caviar,
  author={Borges, João and Bastos, Felipe and Correa, Ilan and Batista, Pedro and Klautau, Aldebaro},
  journal={IEEE Internet of Things Journal},
  title={{CAVIAR: Co-Simulation of 6G Communications, 3-D Scenarios, and AI for Digital Twins}},
  year={2024},
  volume={11},
  number={19},
  pages={31287-31300},
  doi={10.1109/JIOT.2024.3418675}}