Re-architecting Traffic Analysis with Neural Network Interface Cards

This repository contains the code for the paper

G. Siracusano, S. Galea, D. Sanvito, M. Malekzadeh, G. Antichi, P. Costa, H. Haddadi, R. Bifulco - Re-architecting Traffic Analysis with Neural Network Interface Cards [USENIX NSDI 2022]

The structure of the directories is the following:

• dt_rf_bnn: Jupyter notebooks comparing Decision Tree (DT), Random Forest (RF) and Binary Neural Network (BNN) models on Security and IoT datasets
• NNtoP4: BNN to P4 compiler

Additional README.md files are provided in the sub-directories above.

Reference

If you use N3IC for your research, please cite the following paper:

@inproceedings{siracusano2022n3ic,
  title={Re-architecting Traffic Analysis with Neural Network Interface Cards},
  author={Siracusano, Giuseppe and Galea, Salvator and Sanvito, Davide and Malekzadeh, Mohammad and Antichi, Gianni and Costa, Paolo and Haddadi, Hamed and Bifulco, Roberto},
  booktitle={19th USENIX Symposium on Networked Systems Design and Implementation (NSDI 22)},
  year={2022},
  address = {Renton, WA},
  url = {https://www.usenix.org/conference/nsdi22/presentation/siracusano},
  publisher = {USENIX Association},
  month = apr,
}