DLoc Network Architecture Codes

This repository contains the PyTorch implementation of DLoc from Deep Learning based Wireless Localization for Indoor Navigation.

The datasets (features) required to run these codes can be downloaded from the WILD website. You can also download the raw channels from the WILD webpage to run your own algorithms on them.

Requirements

To install requirements:

pip install -r requirements.txt

Note:

The requirements have been tested on Ubuntu 18.04 Docker Image with PyTorch version 1.4.0

Training and Evlautaion

To train the model(s) in the paper and evaluate them, run this command:

python train_and_test.py

The file automatically imports the parameters from params.py.

The parameters and their descriptions can be found in the comments of the example implementation of the params.py file.

To recreate the results from the paper refer to the README of the params_storage folder.

MATLAB codes to transform the raw CSI channels opensource at WILD can be accessed at CSI-to-Features