MMWave Repack

This is a tool for reconstruction MIMO frames based on the recordings from each radar device on the Cascased MMWCAS-RF-EVM module.

After a successful recording, separate recording files are created for each of the four radar devices as follow:

├── master_0000_data.bin
├── master_0000_idx.bin
├── slave1_0000_data.bin
├── slave1_0000_idx.bin
├── slave2_0000_data.bin
├── slave2_0000_idx.bin
├── slave3_0000_data.bin
└── slave3_0000_idx.bin

This tool thus merges the recording from the radar devices to and split them into MIMO frames. The order of the data in each MIMO frame respects the antenna layout on the MMWCAS-RF-EVM board. Hence, the chirp recordings are saved in MIMO frames for each radar device in the following order

[DEV 4] [DEV 1] [DEV 3] [DEV 2]

With

[DEV 1]: Master
[DEV 2]: Slave 3
[DEV 3]: Slave 2
[DEV 4]: Slave 1

NOTE: In this tool,

• it's assumed that all the RX antenna are enabled as well as all the TX antenna. For custom layout, the script must be updated.
• it's also assumed that recordings are in complex format. As so, for each samples, it's assumed that I and Q signals has been recorded.

Environment setup

# Clone the repository
git clone <repo-url>
cd mmwave-repack

# Set up virtual environment and enable it
python -m venv venv
source ./venv/bin/activate

# Install requirements
python -m pip install -r requirements.txt

Usage

A quick overview of the currently supported CLI options are presented below.

usage: repack.py [-h] [-v] [-ns NUMBER_SAMPLES] [-nc NUMBER_CHIRPS] [-o OUTPUT_DIR]
                 [-i INPUT_DIR]

MMWAVECAS-RF-EVM board recordings post-processing routine. Repack the recordings into
MIMO frames

options:
  -h, --help            show this help message and exit
  -v, --version         Print software version and information about the dataset.
  -ns NUMBER_SAMPLES, --number-samples NUMBER_SAMPLES
                        Number of samples per chirp
  -nc NUMBER_CHIRPS, --number-chirps NUMBER_CHIRPS
                        Number of chirps per frame
  -o OUTPUT_DIR, --output-dir OUTPUT_DIR
                        Output directory for storing the mimo frames
  -i INPUT_DIR, --input-dir INPUT_DIR
                        Input directory containing the recordings

The default values of these options are:

-ns 256 -nc 16 -o output

Examples

1. Only providing the recording directory

python repack.py -i <path-to-recording-folder>

This means that default options would be used for any other option. A folder named output will automatically be created within the recording folder to hold the generated MIMO frames.

2. Defining input and output directories

python repack.py -i <path-to-recording-folder> -o <path-to-output-directory>

If the output directory does not already exist, it will be created. Other options would keep their default values and MIMO frames will be saved in the specified output directory.

3. Defining the number of samples per chirp or chirp loops per frame

python repack.py -ns <nb-samples-per-chrip> -nc <nb-chrip-loop> -i <path-to-recording-folder>

Reading and parsing MIMO frames

After merging and repacking the data, MIMO frames are generated and saved as a sequence of int16 integers in binary files (.bin). Such files can be read as shown the code snippet below:

import numpy as np

ntx: int = 12       # Number of TX antenna enalbled (3 TX antenna per radar device)
nrx: int = 16       # Number of RX antenna enabled (4 RX antenna per radar device)
nchirp: int = 16    # Number of chrip loops
ns: int = 256       # Number of samples per chirp

frame = np.fromfile("frame1.bin", dtype=np.int16, count=-1)

# "2" axis count for the I and Q signals recorded
frame = frame.reshape(ntx, nrx, nchirp, ns, 2)
# Frame fully reconstructed
frame = frame[:, :, :, :, 0] + 1j * frame[:, :, :, :, 1]