Skip to content

navigation.md

Latest commit

 

History

History
35 lines (35 loc) · 2.08 KB

navigation.md

File metadata and controls

35 lines (35 loc) · 2.08 KB
Raw

NAVIGATION

  • requirements.txt - required Python modules
  • convert_compile.sh - g++ compiler script for data preparation, outputs to /bin/convertdata
  • docs/ - project docs
    • navigation.md - you are here
    • data_format.md - detailed explanation of our data storage format
    • paper.odt / paper.pdf - final paper in .odt / .pdf format
  • data/ - due to big file sizes (up to 1GB) data is NOT stored on GitHub
    • raw/ - raw data i.e. Wireshark capture files
    • prep/ - prepared data would be here, for detailed format here
      • train/ - training data
      • test/ - test data
  • src/ - actual source code
    • prep/ - data preprocessing code in C++
      • convert/ - main program(s) to prepare the frame data
      • include/ - all header files, constants and reference tables
        • constants.hpp - different constants and aliases used all throughout the project
        • crc32.hpp - generating CRC32 checksum of a vector of bytes, credit to https://github.com/timepp for reference table generation function
        • encoding.hpp - software-level re-implementation of 8b/10b encoding for vectors of bytes
        • encodingtables.hpp - reference tables used in 8b/10b encoding and decoding
        • filesaver.hpp - class for managing file I/O during data preprocessing
        • frame.hpp - various operations on Ethernet II frames - reading, parsing from Wireshark, saving to file etc.
        • transerrors.hpp - introducing single-bit errors into Ethernet II frames
      • src/ - function definitions
    • ml/ - all Python machine learning code
      • datasets/ - Dataset classes, inheriting torch.util.data.Dataset
        • EtherBits.py - frames are expressed as tensors of bits (each as torch.tensor.bool)
        • EtherBytes.py - frames are expressed as tensors of bytes (each as torch.tensor.uint8)
      • util/ - various utilities
      • modules/ - examples of networks used in the projects, for exhaustive list refer to paper
      • trainer.py - simple class that takes and trains torch.Module
      • training.py - code for training and assesment of the model using the Trainer class