This repository holds the code for the paper "FlowLens: Enabling Efficient Flow Classification for ML-based Network Security Applications". If you end up using our code for your experiments, please cite our work as follows:

@inproceedings{protozoa,
  title={FlowLens: Enabling Efficient Flow Classification for ML-based Network Security Applications},
  author={Barradas, Diogo and Santos, Nuno and Rodrigues, Lu{\'i}s and Signorello, Salvatore and Ramos, Fernando M. V. and Madeira, Andr{\'e}},
  booktitle={Proceedings of the 28th Network and Distributed System Security Symposium},
  year={2021},
  address={San Diego, CA, USA},
}

##Dependencies and Data

• Install WEKA
• Run pip install -r requirements.txt

• Please check the README.md in each specific security task folder

• The Security Tasks Evaluation folder includes the code we used for evaluating different ML-based security tasks when using FlowLens. The code applies different combinations of our quantization and truncation approaches and allows for checking FlowLens flow markers trade-offs between accuracy and memory footprint

• The Flow Marker Accumulator folder includes an adaptation of the P4₁₆ code we used for implementing FlowLens' flow marker accumulator in a Barefoot Tofino switch. Due to NDA concerns, we make public this adapted version of our code that can be run on the P4's BMV2 behavioral model.

Todo: Provide a full end-to-end dummy example of FlowLens running in BMV2 - e.g. on P4's tutorial VM.