Sionna CIR Dataset Generator

Generates paired Channel Impulse Response (CIR) datasets using Sionna ray tracing on BostonTwin 3D city models. Each sample contains baseline and ablated scenes (with randomly removed buildings) for ISAC research.

Quick Start

1. Download BostonTwin dataset (required once):

cd boston_twin/
python3 download_data.py

2. Configure generation in gen_path_dataset.py:

NUM_SAMPLES = 10000  # Number of samples to generate
AREA_RADIUS = 500    # Radius of the area to cover (in meters)
RESOLUTION = 1       # Resolution of the coverage map
OUTPUT_DIR = "out/"  # Directory to save the output
PLOT = True          # Enable/disable plotting

3. Run the Docker Container

Build and run the Docker container to generate the dataset:

docker compose up --build [-d]

Update the Subtree

Pull Latest Changes

Pull the latest changes from the boston_twin repository into your subtree:

git subtree pull --prefix=boston_twin/ git@github.com:wineslab/boston_twin.git antenna_mgmt --squash

Push Local Changes

Push your local changes from the subtree back to the repository:

git subtree push --prefix=boston_twin/ git@github.com:wineslab/boston_twin.git antenna_mgmt