CSI-based Proximity Estimation: Data-driven and Model-based Approaches

Official code and dataset. Implementations are made on Python with PyTorch.

Instructions

Python Environment Setup

We recommend using Conda for creating an isolated environment dedicated to running our code and ensuring that your own packages won't be disturbed. Installation guidelines can be found here.

After installing Conda, create the environment from our required package list using:

conda env create -f env.yml

Then activate the environment using conda activate csi_proximity and you're ready to start.

Data-Driven

The hyperparameter sweep across variants of the proposed neural network architecture is based on Weights & Biases. To create the sweep, run:

cd data_driven
python NN_sweep.py

which also prints the sweep ID and a link for accessing the sweep results later on. Then run an agent (or multiple) to start training indefinitely. Add -c 1 to the end of the command below to run a single hyperparameter combination.

python agent.py SWEEP_ID --online

Model-Based

Code for implementing the model-based solutions is available upon request.