Overview

Example applications of the WiSig dataset.

Citation

If you use this code in your research please cite

S. Hanna, S. Karunaratne and D. Cabric, "WiSig: A Large-Scale WiFi Signal Dataset for Receiver and Channel Agnostic RF Fingerprinting," in IEEE Access, vol. 10, pp. 22808-22818, 2022, doi: 10.1109/ACCESS.2022.3154790.

Directory Description

Jupyter notebooks

d001_plot_Full_WiSig_tx_rx_grid.ipynb : plots the number of signals per Tx-Rx pairs using the file data_summary.pkl. Shows how to display the model of Tx and Rx hardware.

d002_analyze_compact_datasets.ipynb: Loads the compact datasets from disk and shows the number of signals per Tx-Rx pairs

d003_ManyRx_nrx.ipynb: Studies the impact of changing receivers on classification accuracy using the non-equalized dataset

d011_ManyRx_nrx_eq.ipynb: Studies the impact of changing receivers on classification accuracy using the equalized dataset

d004_ManySig_nsig.ipynb: Studies the impact of changing the number of training signals on classification accuracy using the non-equalized dataset

d005_ManySig_nsig_eq.ipynb: Studies the impact of changing the number of training signals on classification accuracy using the non-equalized dataset

d006_ManyTx_ntx.ipynb: Studies the impact of changing the number of Tx on classification accuracy using the non-equalized dataset

d007_ManySig_ndays.ipynb: Studies the impact of changing the number of training days using the non-equalized dataset

d008_ManySig_ndays_eq.ipynb: Studies the impact of changing the number of training days using the equalized dataset

d009_ManyTx_localization.ipynb: Plots the average power received at different Rx localization

d010_ManyTx_localization_network.ipynb: Evaluates the performance of WiSig for localization

Python Files

data_utilities.py: Functions to load the dataset and prepare it for classification

PKL files

data_summary.pkl: Contains number of signal per Tx-Rx for the entire datset

IdSig_info.pkl: Contains the google drive links of all files of Full WiSig

orbit_hardware.pkl: Contains a description of the model of WiFi Tx and USRP rx as described in Orbit

Folders

html: Contain an html copy of all ipynb files

py: Contain a python copy of all ipynb files

weights: Contains the trained neural network weights