How to deploy:

#This will call the Sionna + AMC+ Dashboard+ NS3+ metric_server
make deploy-amc-benchmarks 

#On tmux: set-up the mouse:
(ESC then Ctrl+B then : (two points))
set -g mouse on

Change amc config:

To change AMC config in the dashboard, you can:

• Kill the amc.py (top right pane)
• run the following command

#To change the AMC config:
python src/amc/amc.py [OPTIONS]

Possible options:

• to disable gan: "--no-gan"
• To use table: "--method table"
• To use RL (default) : "--method rl"
• Call for help :"--help" ex:

#Calling the amc without the gan and with the table:
python src/amc/amc.py --no-gan --method table

General setup

1. Create a local environment file from the template: cp example.env .env

After copying, open .env and set the path to your virtual environment directory so scripts can find your venv (for example, VENV=.venv or VENVNS3=$HOME/venv/ns3).

Prerequisites: this project expects tmux and docker to be installed on your system. On Debian/Ubuntu you can install them with:

sudo apt update
sudo apt install -y tmux docker.io

On Fedora/CentOS derivatives use:

sudo dnf install -y tmux docker
sudo systemctl enable --now docker

Additional prerequisites

• Python: this project targets Python 3.12. Create and activate a venv:

python3.12 -m venv .venv
source .venv/bin/activate

• Python dependencies: install from requirements.txt:

python -m pip install -r requirements.txt

• System build tools (required to build ns-3 and native extensions). Example for Debian/Ubuntu:

sudo apt update
sudo apt install -y build-essential cmake pkg-config git wget lsof nmap python3.12-venv python3.12-dev

• Docker permissions: Docker commands may require sudo or adding your user to the docker group:

sudo usermod -aG docker $USER
# then log out and back in for the group change to take effect

• CUDA / GPU (optional): If you plan to run models on GPU, ensure matching NVIDIA drivers and CUDA toolkit are installed and install a compatible PyTorch build.

• Optional helpers: python-dotenv is optional but the project will auto-discover a .env file if available. For ns-3 dev the README already recommends cppyy and pygraphviz.

2. Create and activate a Python 3.12 virtual environment for local development.
3. Install the Python dependencies from requirements.txt.
4. Review the available local tasks with make help.
5. Start the services you need with the matching make target, such as make run-amc or make run-dashboard-metrics.

Notes

SSH commmand to run the amc on the server and benchmarks locally

ssh -R 5001:localhost:5001 username@server

ns3 Build+Install

This was made to work properly with the pre-built python ns3 wrapper

create a workspace

cd
mkdir amc_ws
cd amc_ws

venv creation

Creating a basic virtual environement for python 3.12

cd ~/amc_ws
mkdir venv
python3.12 -m venv  --system-site-packages venv/ns3 

##recommended
pip install cppyy pygraphviz
pip install ns3
pip install sionna-no-rt
### if a package breaks others, we recomend doing:
pip install <package-name> --force-reinstall
###or even maybe:
pip install <package-name> --force-reinstall --no-cache
###example (after we broke tensorflow version):
pip install sionna-no-rt --force-reinstall --no-cache 

custom PATH creation

###inside .profile or .bash_profile
export VENVNS3=$HOME/venv/ns3

# Pour que le compilateur trouve les fichiers .hpp
export CPATH=$VENVNS3/include:$CPATH

# Pour que l'éditeur de liens trouve les libs au moment de la compilation
export LIBRARY_PATH=$VENVNS3/lib:$VENVNS3/lib64:$LIBRARY_PATH

# Pour que le système trouve les libs au moment de l'exécution
export LD_LIBRARY_PATH=$VENVNS3/lib:$VENVNS3/lib64:$LD_LIBRARY_PATH

export PIP_SITE=$VENVNS3/lib/python3.12/site-packages

installing JSON in PATH

Installing JSON.hpp as default lib

cd ~/amc_ws
git clone https://github.com/nlohmann/json.git
cd json
mkdir build && cd build
cmake .. -DCMAKE_INSTALL_PREFIX=$VENVNS3
make -j$(nproc)
make install

custom configured and built ns3

cd ~/amc_ws
git clone https://github.com/kohler/click
cd click/
./configure --disable-linuxmodule --enable-nsclick --enable-wifi --prefix=$VENVNS3
make -j
make install

cd ~/amc_ws
wget https://www.nsnam.org/releases/ns-allinone-3.44.tar.bz2

tar -xvjf ns-allinone-3.44.tar.bz2
cd ns-allinone-3.44

#in .profile or .bash_profile
export NS3SRC=$HOME/ns-allinone-3.44/ns-3.44/

cd $NS3SRC/contrib
#you should check right nr version compatible with ns3 on https://cttc-lena.gitlab.io/nr/html/index.html#autotoc_md153

git cl one https://gitlab.com/cttc-lena/nr.git -b  5g-lena-v4.0.y

cd $NS3SRC

./ns3 configure --build-profile=debug --enable-examples --enable-tests --disable-werror \
--enable-python-bindings --enable-build-version --prefix=$VENVNS3/ \
-- -DNS3_BINDINGS_INSTALL_DIR="${PIP_SITE}"

TODO: Find-out why the installation doesnt work when using custom built python bidings

SUGESTIONS: Python bidings could make programming diferent ns3 scenarios easier

ns-3.44:

This gives a hint that new demos all respect the UMi-Street Canyon scenario (need further investigation)

Changes from NR-v2.6 to v3.0

This release contains the upgrade of the supported ns-3 release, i.e., upgrade
from ns-3.40 to ns-3.41.

Changed behavior:
cttc-nr-demo example's parameters are updated to be according 3GPP TR 38.901 UMi-Street Canyon. Also,
the logged values in the NR tutorial are updated accordingly with the updated results.