Wi-Fi Sensing Through Few-Shot Multi-Antenna Multi-Receiver CSI Learning

This repo contains code and dataset of the paper "ReWiS: Reliable Wi-Fi Sensing Through Few-Shot Multi-Antenna Multi-Receiver CSI Learning"

Dataset

Collected CSI dataset can be found at https://drive.google.com/drive/folders/11jLRUp7WBp96vtw8Yn5Rz0zballdSsu6?usp=sharing
The folder Formatted_data_frames contains the cleaned, formatted and pre-processed data frames collected from 3 distinct environements, namely A1, A2 and A3. Data folders starts with mXcY where X indicates the number of monitors and Y is the number of antenna used for collecting the data. trained_A1 is used for training the Few-shot model and test_A2 and test_A3 are used for testing. Each folder contains data for 4 activities, i.e., standing, walking, jumping and empty room.
The folder Raw_pcap contains raw .pcap collected CSI data.

Code

For datasets with one monitor use one_mon.py and one_mon_trained_embedding.py.
one_mon_trained_embedding.py utilizes a CNN for embedded training, while one_mon.py utilizes resnet 12 for embedding training.\

For datasets with three monitors use three_mons.py and three_mon_trained_embedding.py.
three_mon_trained_embedding.py utilizes a CNN for embedded training, while three_mons.py utilizes resnet 12 for embedding training.\

Usage

• Download the dataset in folder Formatted_data_frames
• Place datasets in folder few_shot_datasets
• Pick the propoer datafolder for training: m1c1_xxx or m1c4_xxx for one_mon_trained_embedding.py and m3c1_xxx or m3c4_xxx for three_mon_trained_embedding.py.
• Pick the desired testing environment (either A2 or A3).