It's DeepPacketAttack project for classification over encrypted traffic.

Installation and preparation:

1. Download and install cuda 11.7 if you want to use GPU: https://developer.nvidia.com/cuda-11-7-0-download-archive?target_os=Windows&target_arch=x86_64&target_version=10&target_type=exe_local

2. Add %CUDA_PATH% env.var.: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7

3. Update %PATH% env.var.: %PATH% + C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\bin + C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7\libnvvp

4. Downdload and install miniconda: https://docs.anaconda.com/free/miniconda/

conda env create -f env_linux_cuda117.yaml

python -m pip uninstall torch

python -m pip cache purge

python -m pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117

python

import torch

torch.cuda.is_available()

exit()

pip install click jupyterlab matplotlib datasets pandas plotly pyspark pytorch-lightning scapy[complete]==2.5.0rc1 scikit-learn seaborn tensorboard

Unexpected dependencies:

Check guide https://kontext.tech/article/377/latest-hadoop-321-installation-on-windows-10-step-by-step-guide (not everything needed, but something is necessary)

hadoop-3.2.1: https://github.com/apache/hadoop/tree/release-3.2.1-RC0?ysclid=lsdd29z0h3791672047 and https://github.com/cdarlint/winutils/tree/master/hadoop-3.2.1/bin

maven-3.9.6: https://maven.apache.org/download.cgi

CLEAR DATASET PIPELINE

python preprocessing.py -s clear_pcaps -t clear_processed_data

python create_train_test_set.py -s clear_processed_data -t train_test_data

python train_cnn.py -p train_test_data/application_classification/train.parquet -r train_test_data/traffic_classification/train.parquet -a model/application_classification.cnn.model -t model/traffic_classification.cnn.model -v both

python eval_cnn.py --ct app --gpu True

python eval_cnn.py --ct traffic --gpu True

---

Check my another project with traffic protection via obfuscation: Put obfuscated pcap files in /obfuscated_pcaps directory and run:

OBFUSCATED DATASET PIPELINE

python preprocessing.py -s obfuscated_pcaps -t obfuscated_processed_data

python create_test_set.py -s obfuscated_processed_data -t only_test_data

python eval_cnn.py --actdp only_test_data/app_classification/test.parquet --ct app --gpu True

python eval_cnn.py --tctdp only_test_data/traffic_classification/test.parquet --ct traffic --gpu True