Attention Refined Unrolling for Sparse Micro-Doppler Reconstruction

This repository contains the implementation of the experimental results described in: R. Mazzieri, J. Pegoraro and M. Rossi, "Attention-Refined Unrolling for Sparse Sequential micro-Doppler Reconstruction," in IEEE Journal of Selected Topics in Signal Processing, doi: 10.1109/JSTSP.2024.3408041. (available here).

All the experiments are implemented in Python version 3.8.10.

If you find our work and implementation useful for your research, please, cite us using the following BibTeX:

@article{10543012,
  author={Mazzieri, Riccardo and Pegoraro, Jacopo and Rossi, Michele},
  journal={IEEE Journal of Selected Topics in Signal Processing}, 
  title={Attention-Refined Unrolling for Sparse Sequential micro-Doppler Reconstruction}, 
  year={2024},
  volume={},
  number={},
  pages={1-16},
  keywords={Stars;Sensors;Channel estimation;Compressed sensing;Spectrogram;Real-time systems;Human activity recognition;Joint Communication and Sensing;Micro-Doppler signatures;Sparse Reconstruction;Algorithm Unrolling;Attention;gHuman Activity Recognition},
  doi={10.1109/JSTSP.2024.3408041}}

@article{pegoraro2023disc,
  title={DISC: a dataset for integrated sensing and communication in mmWave systems},
  author={Pegoraro, Jacopo and Lacruz, Jesus Omar and Rossi, Michele and Widmer, Joerg},
  journal={arXiv preprint arXiv:2306.09469},
  year={2023}
}

1. Setup

Set up a fresh Python environment using

python3 -m venv STAR_env

Then install the required packages by running:

pip install -r requirements.txt

2. Main Results (Section VI-C and VI-D)

STAR was tested on the DISC dataset, publicly available here, in particular on the uniform_7subj.zip subset.

Our results can be reproduced by training STAR from scratch. To do so:

1. Extract the dataset in the folder data/raw_data/
2. Configure the ablation_runs.csv file, by specifying possibly multiple configurations of the hyperparameters.
3. Train all the models specified in the .csv file by executing the following script:

python3 train.py

By default the ablation_runs.csv file contains all the ablation runs reported in the final paper. These can be modified to implement further experiments.

3. Results in novel environment (Sec VI-E)

This folder contains the additional set of data (to be added to the DISC dataset in the short term) To reproduce the results of the new environment, first train the final model as in step 1, and then execute:

python3 STAR_test_room.py