Recent Changes and Advances in Tor Tran, Tuan Tu Technische - - PowerPoint PPT Presentation
Chair of Network Architectures and Services Department of Informatics Technical University of Munich Recent Changes and Advances in Tor Tran, Tuan Tu Technische Universitt Mnchen Seminar Future Internet SS2018 Departments of Informatics
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Tran, Tuan Tu Technische Universität München Seminar Future Internet SS2018 Departments of Informatics München, 06. April 2018
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
But there is hope: The Tor Network
Intro joke do not take seriously
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Tor network: Basic structure and function
Threats/Attack taxonomy
Countermeasures Some unsolved issues of the Tor network
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client/ Onion proxy Server Onion router Onion router Onion router TLS Connection TLS Connection
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client/ Onion proxy Server Entry node Middle node Exits node
The data is transported in cells through the circuit
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Entry node n1
Create cell: E(g^x1, n1 onion key)
n1 choose secret y1 Created cell: g^y1; H((g^x1)^y1) Compute: (g^x1)^y1 Compute: (g^y1)^x1 Common key k1: (g^y1)^x1 = (g^x1)^y1 = g^(x1*y2)
(E (object, key) = Encrypted)
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Entry node n1
Middle node n2
E(Extend cell: E(g^x2, o.key n2), k1 ) Repackage Extend Cell in a Create cell Created cell: g^y2; H((g^x2)^y2) Repackage Created Cell Extended cell Only client and n2 now knows common key k2
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Entry node n1
Exit node n3
Server
Pay Load
K3 K2 K1
Pay Load
K3 K2
….
Pay Load
Encrypted with k1
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Entry node n1
Exit node n3
Server ….
Pay Load
K3
Pay Load
K3 K2 K1
Pay Load
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server Onion router Onion router Onion router
Directory Server
Descriptor: Information about the node like the public identity key, addresses
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server Onion router Onion router Onion router Directory Server Information about the network like addresses of the nodes
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server Onion router Introduction Point Onion router
Distributed Hash table Distributed Hash table
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server Onion router Introduction Point Onion router
Distributed Hash table Distributed Hash table
Descriptor: Introduction point , public key
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server Onion router Introduction Point Onion router
Distributed Hash table Distributed Hash table
Out of Band : .onion address of Service retrieve descriptor
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server rendezvous point Introduction Point Onion router
Distributed Hash table Distributed Hash table Connect to introduction point Choose rendezvous point
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Server rendezvous point Introduction Point Onion router
Distributed Hash table Distributed Hash table
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Different type of attacks against the Tor network:
− Website Fingerprinting − Keyword Fingerprinting
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Entry node Tor Adversary TLS encrypted connection
Analysing traffic pattern, like number of packets send/receive
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Image 1
Image file of webpage send in a specific number of packets
Html file … Webpage
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Tor Adversary Server Packets Samples Analysing
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
(according to Analysis of Fingerprinting Techniques for Tor Hidden Service ) Do not scale well to increasing number of websites With the growing numbers of hidden services, attack may be less effective
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Search engine like Google are an essential part of Internet Search queries also contain a lot of information about us Adversary maintains list of monitored phrases (keywords) Finding the Tor connections carry the search result traces The goal of this attack is to learn about the search queries of a user
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client
Adversary
TLS encrypted connection
Search Engine
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client Tor Adversary Search Engine Monitored Keyword List Analysing
Cat picture Cat video
Searching cat video
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
(according to Fingerprinting Keywords in Search Queries over Tor) Precision decreased with the increase of possible unwanted search queries Incremental search (JavaScript enabled) carries more information for the adversary Keyword fingerprinting attacks in all his variants showed reasonable result and resemble a new threat against Tor
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Conventional attacks often only assume:
New attack :
Possible adversary: Autonomous Systems (ISP,…)
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
In Proceedings of the 24th USENIX Security Symposium August 2015
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Prevent clients to use Tor is to block access to Tor network entirely Client
Tor Address Blocking Directory Server
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Alternatively the adversary can block any Tor traffic regardless of source or destination Possible through easy distinguishable Tor certificate used for the TLS connection
Client ….
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Defences against Traffic analysis
Measure against hindrance of censorship circumvention
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Defense implementation on the application-layer of the OSI-model ALPaCA aims to morph the objects of webpages to hide it traffic pattern For example: add in object like HTML file random data as comment
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
In order to know, how the object must be morph:
− collect sample data from other hidden service website
− use input parameter from the service provider/administrator and data from the page itself
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Target Original Webpage Morphed
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Input:
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Procedure :
(target size) with size(o) < t
remaining target sizes in T
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client 1 Client 2 Client 1
Tor
Directory with pre-morphed Webpage
Every client get the same page differently morphed
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Obscure the order and timing of request the client sends Adds a random delay to every request.
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Application layer defense does not require a change of the source code of the tor protocol The padding is added directly to the objects of the webservice
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Client
Tor
Bridge BridgDB/Bridge authority
Address Blocking
Descriptor: Address, Port number… Descriptor via E-mail
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Pluggable Transports transform the Tor traffic to look like other traffic Example:
Azure cloud), and from there to the actual bridge
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Public bridges often do not offer Pluggable Transports support and some bridges offer out dated transports Common use of default bridges hard coded in the tor browser bundle
Chair of Network Architectures and Services Department of Informatics Technical University of Munich
Technologies ,2017(2), April 2017.
Symposium, August 2004.
Proceedings of the Network and Distributed Security Symposium - NDSS ’17. Internet Society, February 2017.
October 2017.
In Proceedings of the Workshop on Privacy in the Electronic Society (WPES 2017), November 2017.
Proceedings of the 24th USENIX Security Symposium, August 2015
Accessed: 16.03.2018.
15.03.2018.