FastZIP Data
FastZIP Dataset and Code
EnvironmentOpen Data Commons Attribution License v1.0Introduced 2021-06-09
Structure of code/data folders and how to use them
fastzip-code
- Contains codebase to generate results in fastzip-results folder
- Individual notebooks contain comments on their functionality/how to use them
- FastZIP-Resample.ipynb (optional)
- Resamples the collected sensor data to desired sampling rates (eliminates the effect of sampling rate instability/drift)
- Input: fastzip-data/exp-X/raw
- Output: fastzip-data/exp-X/adv and fastzip-data/exp-X/non-adv
- FastZIP-Process.ipynb
- Computes error rates for adversarial and benign devices in different configurations, generates binary fingerprints (see comments inside the notebook)
- Input: fastzip-data/exp-X/adv and fastzip-data/exp-X/non-adv
- Output: fastzip-results/logs and fastzip-results/fps
- FastZIP-Results.ipynb
- Parses and caches results generated by FastZIP-Process.ipynb notebook to be used for plotting and data analysis
- Input: fastzip-results/logs and fastzip-results/fps
- Output: fastzip-results/cache
- The system paths that are used by all notebooks are set in fastzip-code/const/globconst.py --> adjust them before running the code!
- Uses folder fastzip-data as input
- Uses folder fastzip-results as output
- The above notebooks were run using Python 3.6.5 on Ubuntu 18.04.5 LTS bionic (x64)
- The list of python packages with versions installed on the test machine is in fastzip-code/python3-packages.txt
fastzip-data
- Contains sensor data collected from multiple devices in running/stationary cars in various conditions (e.g., in a city, etc.) from three experiments: exp-3, exp-4, and exp-5
- Each exp-X folder has the same structure:
- adv - sensor data collected when two cars drive one after another
- non-adv - sensor when two cars drive the same route but not one after another
- In adv and non-adv the sensor data is resampled (see description of FastZIP-Resample.ipynb )
- raw - sensor data collected in the experiment before resampling or split into adv and non-adv
- Each exp-X folder has the same structure:
fastzip-min_entropy
- Contains input and evaluation results for min-entropy estimation of our generated fingerprints using NIST SP 800 90B tests
- See README inside the folder for more detail
fastzip-results
- Results in json or json.gz formats generated by FastZIP-Process.ipynb and FastZIP-Results.ipynb
- See comments in these notebooks for the type of results stored in folders inside fastzip-results
fpake
- Contains the implementation of the fPAKE protocol as well as input and output for its benchmarking
- See README inside the folder for more detail