User-Driven Adaptive CSI Feedback With Ordered Vector Quantization

🚀 Welcome to the repository for the paper "User-Driven Adaptive CSI Feedback With Ordered Vector Quantization"! This repository contains the code and resources needed to reproduce the key results from our study. For full details, please refer to the paper.

Table of Contents

• Dataset Preparation
• Usage
• Results
• Citation
• Acknowledgements

Dataset Preparation

This project uses the COST2100 "outdoor" dataset for training and evaluation.

To download and prepare the dataset, run the following commands after cloning the repository:

curl -L -o COST2100_dataset.zip "https://www.dropbox.com/scl/fo/tqhriijik2p76j7kfp9jl/h?rlkey=4r1zvjpv4lh5h4fpt7lbpus8c&e=2&st=pmf7duk6&dl=1"

unzip COST2100_dataset.zip -d COST2100_dataset

rm -f COST2100_dataset.zip

Usage

This repository demonstrates the proposed Ordered Vector Quantization (OVQ) scheme using TransNet and CRNet as base models.

Pretraining

To reproduce the pretraining results, run the following commands:

CRNet

python3 main_cost.py -d out -m crnet -r 4 -b 10 -e 4

TransNet

python3 main_cost.py -d out -m transnet -r 2 -b 10 -e 8

Fine-tuning

To fine-tune the pretrained models and obtain final results, use:

CRNet

python3 main_cost.py -d out -m crnet -r 4 -b 10 -e 4 -ft

TransNet

python3 main_cost.py -d out -m transnet -r 2 -b 10 -e 8 -ft

Results

• The checkpoints folder contains the "best_model" checkpoints for both the pretraining and fine-tuning phases.
• The tables folder contains .csv files that can be used directly for your plots.

✨ You can explore the final results and interactive tables on Weights & Biases.

Citation

📚 If you find our work helpful in your research, we’d be happy if you cite us!

@article{rizzello2023user,
  author={Rizzello, Valentina and Nerini, Matteo and Joham, Michael and Clerckx, Bruno and Utschick, Wolfgang},
  journal={IEEE Wireless Communications Letters}, 
  title={User-Driven Adaptive CSI Feedback With Ordered Vector Quantization}, 
  year={2023},
  volume={12},
  number={11},
  pages={1956-1960},
  doi={10.1109/LWC.2023.3301992}
}

Acknowledgements

1. COST2100 dataset
   C.-K. Wen, W.-T. Shih, and S. Jin, "Deep Learning for Massive MIMO CSI Feedback," in IEEE Wireless Communications Letters, vol. 7, no. 5, pp. 748–751, Oct. 2018.

2. TransNet base model
   Y. Cui, A. Guo, and C. Song, "TransNet: Full Attention Network for CSI Feedback in FDD Massive MIMO System," in IEEE Wireless Communications Letters, vol. 11, no. 5, pp. 903–907, May 2022.

3. CRNet base model
   Z. Lu, J. Wang, and J. Song, "Multi-resolution CSI Feedback with Deep Learning in Massive MIMO System," ICC 2020 - IEEE International Conference on Communications, Dublin, Ireland, 2020, pp. 1–6.