Tor fingerprinting using supervised machine learning

Dissertant

al-Mubayyid, Ala al-Din

Thesis advisor

Hadi, Ali
al-Atum, Jalal Umar

Comitee Members

Ujan, Arafat
Umar, Khamis

University

Princess Sumaya University for Technology

Faculty

King Hussein Faculty for Computing Sciences

Department

Department of Computer Sciences

University Country

Jordan

Degree

Master

Degree Date

2014

English Abstract

Tor is the low-latency anonymity tool and short of “ ” one of the prevalent used open source anonymity tools for anonymizing TCP traffic on the Internet used by around 500,000 people every day.

Tor protects ’ k x difficult for an observer to correlate visited websites in the Internet with the real physical-world identity.

Tor accomplished that by ensuring adequate protection of Tor traffic against traffic analysis and feature extraction techniques.

Further, Tor ensures anti-website fingerprinting by implementing different defenses like TLS encryption, padding, and packet relaying.

However, in this thesis, an analysis has been performed against Tor from a local observer in order to bypass Tor protections, the method consists of a feature extraction from a l w k w ’ possible for a local observer to fingerprint top monitored sites on Alexa and Tor traffic can be classified amongst other HTTPS traffic in the network ’ x ent, several supervised machine-learning algorithms have been employed.

The attack assumes a local observer sitting on a local network fingerprinting top 100 site on Alexa, results gave an improvement amongst previous results by achieving an accuracy of 99.6% and 0.01% false positive.

Main Subjects

Information Technology and Computer Science

Topics

• Security measures

No. of Pages

146

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-413618