tonellolab/Iterative-Intelligent-Sampling

[ICT Express (2024)] Official Pytorch implementation of "Recurrent DQN for Radio Fingerprinting with Constrained Measurements Collection"

Iterative-Intelligent-Sampling

Nicola Novello and Andrea M. Tonello

Official repository of the paper "Recurrent DQN for Radio Fingerprinting with Constrained Measurements Collection" published at ICT Express (2024).

---

How to run the code

The folder where the code is located must contain a folder NetsImages comprising the following folders:

• Datasets containing the RSSI dataframe obtained after interpolation of the dataset in https://github.com/beaugunderson/wifi-heatmap and the initial training datasets randomly extracted from the same dataset
• DRQN containing the saved DRQN
• FloorImages containing the map layout
• OptimalPaths containing the images of the saved optimal paths traveled by the agent
• RegModels containing the saved PPNNs
• scalers containing the scaler used to standardize the dataset
• accuraciesComparison containing the saved test accuracies

The file main.py runs the experiments, while main_functions.py, classes.py, and utils.py comprise the classes and functions needed to run the main script.

---

References and Acknowledgments

If you use your code for your research, please cite our paper:

@article{novello2024recurrent,
  title={Recurrent DQN for radio fingerprinting with constrained measurements collection},
  author={Novello, Nicola and Tonello, Andrea M},
  journal={ICT Express},
  year={2024},
  publisher={Elsevier}
}

The implementation is based on / inspired by:

• https://github.com/beaugunderson/wifi-heatmap
• https://mlpeschl.com/post/tiny_adrqn/
• https://github.com/patrickloeber/snake-ai-pytorch

---

Contact

nicola.novello@aau.at