Sybil Attacks

1
Sybil Attacks

• Sachin Chaudhry

2
Sybil ???

• The subject of the book Sybil, a case study of
  a woman with multiple personality disorder.

3
Definition

• Sybil attack is one in which an attacker
  subverts the reputation system of a peer-to-peer
  network by creating a large number of
  pseudonymous entities, using them to gain a
  disproportionately large influence.

4
Reputation Systems

• A reputation system aggregates the users
  collective experiences in order to allow a user
  to form an opinion about someone with whom he has
  not previously interacted
• Reputation Systems exist in
• - p2p systems
• - ad hoc networks
• - online markets

5
Vulnerability to Sybil Attacks

• Symmetric
• one in which, an identitys reputation depends
  solely on the topology of the trust graph, and
  not the naming or identity of nodes
• Asymmetric
• one in which, there are specifically trusted
  nodes from which all reputation values propagate

Cheng Friedman Sybil-proof reputation
mechanisms. Third Workshop on the Economics of
Peer-to-Peer Systems, 2005.
6
Symmetric Reputation Systems

• Sybil identities to create a copy of the existing
  graph representing trust relationships
• cannot distinguish original nodes from the copies
• not sybil-proof
• Example google PageRank

Cheng Friedman Sybil-proof reputation
mechanisms. Third Workshop on the Economics of
Peer-to-Peer Systems, 2005.
7
Asymmetric Reputation Systems

• each entity separately computes a trust value
  along their unique paths to every other identity
  in the system
• trusted nodes cannot be impersonated, so no gain
  by duplicating the graph
• Sybil-proof

Cheng Friedman Sybil-proof reputation
mechanisms. Third Workshop on the Economics of
Peer-to-Peer Systems, 2005.
8
SybilGuard

• SybilGuard implements basic idea of asymmetric
  reputation system
• This ensures that regardless of the number of
  sybil nodes a malicious user created, an honest
  user will share an edge with a fixed minimum
  number of them
• This protocol enables honest nodes to impart
  trust to a large fraction of the other honest
  nodes

Yu, Kaminsky, Gibbons, Flaxman - SybilGuard
Defending Against Sybil Attacks via Social
Networks SIGCOMM06
9
Salient Features

• Divide all the nodes into groups. And each group
  can have at most w nodes.
• To prevents clustering of sybils in one group,
  any new node is joined randomly to one group in
  the pool of groups.
• Each group has a Verifier V (an honest node).

10
Salient Features (contd.)

• Each honest node constrains its degree d within
  some constant
• When a Suspect node (S) tries to establish an
  edge with an honest node, then, the honest node
  uses random walks.
• Random Walk is traversing the route to the V node
  of the Suspect node.
• The honest node uses d (degree) random walks to
  reach the V-node of S-node.

11
Salient Features (contd.)

• If d/2 (or d/4 or d/8) random walks contain the
  S-node, then only the edge is established
• V-node of the honest node also accepts S as an
  honest node

12

• Thanks