Overview

This is a PyTorch implementation of BCsiNet inference. The key results in paper Binary Neural Network Aided CSI Feedback in Massive MIMO System can be reproduced.

Requirements

The following requirements need to be installed.

• Python >= 3.7
• PyTorch >= 1.6

Project Preparation

A. Data Preparation

The channel state information (CSI) matrix is generated from COST2100 model and setting can be found in our paper. On the other hand, Chao-Kai Wen provides a pre-processed COST2100 dataset, which we adopt in BCsiNet training and inference. You can download it from Google Drive or Baidu Netdisk.

B. Checkpoints Downloading

The checkpoints of our proposed BCsiNet can be downloaded from Baidu Netdisk (passwd: cism) or Google Drive

C. Project Tree Arrangement

We recommend you to arrange the project tree as follows.

home
├── BCsiNet  # The cloned BCsiNet repository
│   ├── dataset
│   ├── models
│   ├── utils
│   ├── main.py
├── COST2100  # COST2100 dataset downloaded following section A
│   ├── DATA_Htestin.mat
│   ├── ...
├── Experiments
│   ├── checkpoints  # The checkpoints folder downloaded following section B
│   │     ├── a2
│   │     ├── b3
│   │     ├── ...
│   ├── run.sh  # The bash script
...

Key Results Reproduction

The key results reported in Table IV of the paper are presented as follows.

Compression Ratio	Methods	Scenario	NMSE	Params	Checkpoints Path
1/4	BCsiNet-A2	indoor	-17.25dB	33K	a2/in01/model.pth
1/4	BCsiNet-A2	outdoor	-8.35dB	33K	a2/out01/model.pth
1/4	BCsiNet-B3	indoor	-20.31dB	33K	b3/in01/model.pth
1/4	BCsiNet-B3	outdoor	-9.77dB	33K	b3/out01/model.pth
1/8	BCsiNet-A2	indoor	-12.38dB	17K	a2/in02/model.pth
1/8	BCsiNet-A2	outdoor	-6.26dB	17K	a2/out02/model.pth
1/8	BCsiNet-B3	indoor	-12.77dB	17K	b3/in02/model.pth
1/8	BCsiNet-B3	outdoor	-6.86dB	17K	b3/out02/model.pth
1/16	BCsiNet-A2	indoor	-8.99dB	8K	a2/in03/model.pth
1/16	BCsiNet-A2	outdoor	-4.17dB	8K	a2/out03/model.pth
1/16	BCsiNet-B3	indoor	-10.71dB	8K	b3/in03/model.pth
1/16	BCsiNet-B3	outdoor	-4.52dB	8K	b3/out03/model.pth
1/32	BCsiNet-A2	indoor	-6.79dB	4K	a2/in04/model.pth
1/32	BCsiNet-A2	outdoor	-2.69dB	4K	a2/out04/model.pth
1/32	BCsiNet-B3	indoor	-7.93dB	4K	b3/in04/model.pth
1/32	BCsiNet-B3	outdoor	-2.74dB	4K	b3/out04/model.pth

In order to reproduce the aforementioned key results, you need to download the given dataset and checkpoints. Moreover, you should arrange your project tree as instructed. An example of Experiments/run.sh can be found as follows.

python /home/BCsiNet/main.py \
  --data-dir '/home/COST2100' \
  --scenario 'in' \
  --pretrained './checkpoints/a2/in01/model.pth' \
  --batch-size 200 \
  --workers 0 \
  --reduction 4 \
  --encoder-head A \
  --num-refinenet 2 \
  --cpu \
  2>&1 | tee log.out

Note that the checkpoint must match exactly with the reduction, encoder_head and num_refinenet.

Acknowledgment

This repository is modified from the CRNet open source code. Please refer to it for more information.

Thank Chao-Kai Wen and Shi Jin group again for providing the pre-processed COST2100 dataset, you can find their related work named CsiNet in Github-Python_CsiNet.