README

Description

This C++ code is designed to make predictions using a Tensorflow Keras model. The main functionality of the code involves converting packet data from pcap format to CSV, processing the CSV data, and making predictions using a pre-trained neural network model.

Dependencies

Before running the code, make sure you have the following dependencies installed:

• TensorFlow C++ API
• frugally-deep
• C++17

Usage

1. Packet Conversion and CSV Processing:

   pcapToCsv();

   This function converts a pcap file (dump.pcap) to CSV format using a shell script (convert_pcap_csv.sh). It then processes the CSV data by removing unnecessary columns.

2. Scientific Notation Processing:

   process_csv_scientific();

   This function processes CSV data, specifically handling scientific notation in the CSV cells. It converts scientific notation to its corresponding decimal form.

3. CSV Data Reading and Model Prediction:

   read_and_process_csv();

   This function reads the processed CSV data from finaldump.csv, loads a pre-trained neural network model (model.json), and makes predictions on each line of the CSV data. The predicted results are mapped to human-readable labels using a predefined status map.

Additional Notes

• Status Map:

  The predictions are mapped to human-readable labels using the statusMap defined in the code.

• Invalid Data Handling:

  If invalid or non-numeric values are encountered in the CSV data during processing, a warning is printed, and the corresponding value is set to 0.0f.

• Output:

  The predicted results are printed to the console and stored in the sharedVector vector.

Running the Code

Compile and run the code using a C++ compiler with C++17 support. Make sure to link against the TensorFlow C++ library and include the necessary dependencies.

g++ -std=c++17 -o your_executable_name your_source_code.cpp -ltensorflow
./your_executable_name

Note: Ensure that the paths to the model file and other resources are correctly specified in the code.

License

This code is provided under the MIT License. Feel free to use, modify, and distribute it according to the terms of the license.