Website Fingerprinting on Anoymous Network

Team Learning Machine
Fall 2024 Machine Learning Project

Project Overview

This project aims to analyze website fingerprinting on anonymous networks and evaluate the performance of various machine learning models under different scenarios.

Team Introduction

• Yunwoo Choi (2176387) - Open World Multi Class Classification, Presentation Development
• Minkyung Lee (2076298) - Open World Binary Classification, Presentation Development
• Jayeong Seo (2170045) - Open World Multi Class Classification, Presenter
• Jiin Han (2276350) - Open World Binary Classification, Presenter
• YouLee Han (2271065) - Closed World Classfication, Script Writer
• Jaehee Gung (2029003) - Closed World Classfication, Script Writer

Project Description

Website fingerprinting is an attack technique that identifies websites a client has accessed by analyzing the patterns of data flow. The project is divided into the following categories:

1. Closed World Classification: Classifying the 95 monitored websites.

2. Open World

• Binary Classification: Distinguishing whether a user is visiting a monitored website(label 1) or an unmonitored one(label -1).

• Multi-Class Classification: Classify 95 monitored website traces with unique labes(0, 1, 2, ..., 94) against additional unmonitored websites with label -1.

File Overview

• closed_final.ipynb: Contains the closed-world classification experiments and results, focusing on classifying traffic from a predefined set of monitored websites.

• open_binary.ipynb: Focuses on open-world binary classification, distinguishing between monitored and unmonitored websites.

• open_multi.ipynb: Includes experiments for open-world multi-classification, identifying specific monitored websites among a larger set of possible targets.

• ScalingEncoding.ipynb: Details various data scaling and encoding techniques used to improve model performance.

• WFdefense.ipynb: Evaluates different defense mechanisms against website fingerprinting attacks, aiming to mitigate the risks of these types of attacks.

Requirements

To run the provided notebooks, you will need the following dependencies:

Python 3.8+

Jupyter Notebook

NumPy

Pandas

Scikit-learn

Matplotlib

How to Run the Project

1. Clone the repository.

2. Install the dependencies listed in the requirements section using the following command:

   pip install -r requirements.txt

3. Open Jupyter Notebook and load the respective .ipynb file for the experiment you want to run.

4. Follow the instructions within each notebook to run the experiments.

Results

The results of the experiments are detailed within each respective notebook. It includes metrics such as accuracy, precision, recall, and F1-score to evaluate model performance.