# The Keyframes and reference frames identification of encrypted traffic

**Repository Path**: network-intelligent-perception/the-keyframes-and-reference-frames-identification-of-encrypted-traffic

## Basic Information

- **Project Name**: The Keyframes and reference frames identification of encrypted traffic
- **Description**: This project is based on the research work titled "TDS-KRFI: Reference frame identification for live web streaming towards HTTP Flash Video protocol". It aims to efficiently identify keyframes and ref
- **Primary Language**: Python
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 1
- **Forks**: 0
- **Created**: 2023-05-30
- **Last Updated**: 2024-07-17

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Keyframes and Reference Frames Identification in Encrypted Traffic

#### Introduction

This project is based on the research work titled "TDS-KRFI: Reference frame identification for live web streaming towards HTTP Flash Video protocol". It aims to efficiently identify keyframes and reference frames in encrypted real-time traffic, providing research support for improving real-time video Quality of Experience (QoE) in subsequent studies.

#### Software Architecture

Model: Keyframes and reference frames identification model Datahandler: Extracts features from pcap files and determines the positions of keyframes and reference frames in decrypted pcap data using a key file.

#### Modifications Required to Run the Model

1. datahandler.1_Feature_Extract.py: Modify 'pcapfile_dir' to the location of the pcap files in the packetRequireLabel(threshold, time) function. Also, replace 'logfilename.log' in 'command = ["tshark", "-o", "tls.keylog_file:logfilename.log"]' in datahandler.flowcontainer_ipv6.reader.py with the key log file corresponding to the pcap file (refer to Section IV.B of the referenced paper for obtaining the key).
2. datahandler.2_successionDatahandler.py: In the main function, 'dirPath' represents the folder name of the generated txt files from '1_Feature_Extract.py', and 'writedir' represents the folder location of the txt files that conform to the input format of the DGCNN recognition model. Once these modifications are made, the data preprocessing part is completed, and the next step is to use DGCNN for keyframes and reference frames identification.
3. model.main.py: 'dirpath' represents the folder location of the txt files generated by '2_successionDatahandler.py'.

#### Instructions for Use

1. Run datahandler.1_Feature_Extract.py to extract the adjacency matrix of data units, corresponding node features, and labels from the pcap files.
2. Run datahandler.2_successionDatahandler.py to convert the obtained txt files into the required format files (txt) for the recognition model.
3. Run model.main.py to perform training and testing for keyframes and reference frames identification (50 rounds).

#### Recognition Results

The keyframes identification results for 50 rounds in standard-definition (SD) mode can be referred to in model/TDSdata/Databq_results_50_rounds.txt.