Robust Application Layer Routing for Anonymous Communications

1
Robust Application Layer Routing for Anonymous
Communications
Final presentation

• Shreyas Doshi
• Keita Fujii
• Pablo Diaz Gutierrez
• Michael Moore
• Lorenzo Gil Sanchez

Names sorted alphabetically by last name. All of
the above will be presenting today.
2
Outline

• What it is?
• Why is it important?
• How we carried it out?
• System Functionality
• Architectural Details
• Implementation
• Evaluation and Results
• Demonstration

3
What it is

• A robust approach for providing anonymity
  services on the Internet
• Uses a dynamic, application-layer routing method
• Provides for
• Sender anonymity
• Recipient anonymity
• Sender-recipient unlinkability

4
Why it is important

• Lack of anonymity a road-block to the deployment
  of several services over the Internet
• Like elections, anonymous authoring
• Our approach to providing anonymity can be easily
  deployed on the Internet, with the potential to
  provide support for various different protocols
• Like HTTP, SMTP, etc.

5
Comparison to Related Work

• Our approach differs significantly from previous
  work in that
• It is more robust and
• It can potentially adapt well to varying network
  load
• On account of
• Dynamic path creation
• Redundancy in the form of peer groups

6
Algorithm Details

• Our Project
• Dynamic Path Finding
• Group Formation/Usage
• This Presentation
• Peer/Group Initialization
• Failover during Dynamic Path Finding
• Failover during Onion Routing

7
Peer/group Initialization

• Usage of global multi-cast address to locate
  members of any group
• Ping/Pong process
• Ping request others peers to send their
  information
• To global multi-cast address
• Requesters IP/Port, Requesters public key
  (encrypted by global public key)
• Pong response from others sending their
  information
• To requesters uni-cast IP/Port (encrypted by
  requesters public key)
• Receivers IP, Port, Group address, Group public
  key

8
Peer/Group Initialization

• Optional Pong If requester learned about new
  peers, then send own information to each learned
  peer

9
Peer/Group Initialization

• Optional Pong If requester learned about new
  peers, then send own information to each learned
  peer

10
Finding the Next Peer

• Failure handling during path discovery
• Indicated by Timeout TCP-IP connection failure
  to a particular peer
• Selection of a new peer
• Each peer has list of other peers, next peer
  chosen randomly
• Acquiring new peers
• If none of the peers links work, peer can send
  ping/pong to acquire new peer links

11
Path Discovery Failover

• Forwarding
• Random selection of next hop from list of peers
• Creation of reverse onion at each hop
• Hop counter (HC) forward until zero

!! Host D Unreachable
D
A
C
Y
X
B
12
Path Discovery Failover

• Forwarding
• Random selection of next hop from list of peers
• Creation of reverse onion at each hop
• Hop counter (HC) forward until zero

D
A
C
Y
X
B
13
Grouping Mechanism

• Groups
• Statically defined each peer already belongs to
  a particular group and can not change group
• For each group, group members listen to a
  different group multi-cast address(mechanism to
  locate alternative group members via ping/pong
  process)

14
Grouping Failover

• During onion routing, group addresses were placed
  in onion (not individual peers) and encrypted
  by group public keys

!! Host C4 Unreachable
B1
B1 to forward to member of Group gC
15
Grouping Failover

• During onion routing, group addresses were placed
  in onion (not individual peers) and encrypted
  by group public keys

B1
B1 to forward to member of Group gC
16
System Weaknesses

• Dynamic Path Formation
• Ability of first peer to control entire path
  returned
• Can know entire path and trace messages
• Can set end peer and perform more effective
  intersection attack
• Ability of peers to control what traffic they
  receive
• Can collude such that other peer will only
  receive traffic through collusion
• Grouping
• Even spreading of traffic
• For better anonymity would like traffic to
  aggregate

17
System Architecture

• Communicator
• Discovers peers/groups
• Chooses one peer in the specified group
• Transmits data to another peer
• Cipher
• Encrypts/decrypts data
• Generates public/private keys
• Router
• Chooses next group at path discovery phase
• Forward message to the next group specified in
  Onion Route

18
Message Format

• Message Type
• Path Discovery / Data Delivery
• ID
• Checked to discard duplicate messages
• Onion Route
• Route information through which message should be
  forwarded
• Reverse Onion Route
• Route information used to reply this message
• Hop Count
• Number of peers allowed to be forwarded for path
  discovery
• Body

19
Detailed System Architecture
20
(No Transcript)
21
Communicator (Peer Discovery)

• Multicast Ping message

B
A
C
D
D joined the network
22
Communicator (Message Transfer)

• Incoming Message
• Decrypt message
• Pass to Communicator
• Outgoing Message
• Choose next peer in the specified group
• Encrypt message
• Transmit message

23
Router (Path Discovery)

• If HopCount0
• Send back the discovered path to the sender
• If Hop Countgt0
• Decrement Hop Count
• Add this group to Reverse Onion Route
• Select next group
• Send path discovery message to the group

24
Router (Data Delivery)

• If Onion Route is empty (meaning this is the last
  peer)
• If message contains path information
• Create Onion Route
• Send data with the route
• If message contains data
• Pass data to application
• If not
• Obtain next group from Onion Route
• Add this group to Reverse Onion Route
• Forward message to next group

25
Final Test and Results

• How did we evaluate the system?
• Empirical test
• Theoretical study
• What aspects did we consider?
• Performance
• Robustness/Reliability
• Scalability

26
Time / Message
27
Robustness / Reliability
28
Scalability
29
Demo