ITIS 60108010 Wireless Network Security

1
ITIS 6010/8010 Wireless Network Security

• Dr. Weichao Wang

2

• Pairwise key establishment in sensor networks
• In many of secure routing mechanisms, we have
  assumed the knowledge of keys
• We will investigate how these keys are
  established among wireless nodes

3

• Key predistribution method (Eschenauer and
  Gligor, CCS02)
• Back in 2002, symmetric encryption is still the
  only choice for sensors
• No trusted third party for key distribution
• Predistribution
• Group key
• Pairwise key
• Each has its problems sensor compromise, add new
  node, forward/backward secrecy

4

• Probabilistic key sharing
• A pool of P keys are generated offline
• Every sensor will randomly get k keys when it is
  deployed
• With a certain probability, any pair of sensors
  may share at least one key
• For those that do not share keys, multihop path
  can be used to establish such a key

5

• Example P 10,000, how may keys should every
  sensor have so that the probability is 0.5? (75)
• Some simple analysis
• Shared key discovery b/w neighbors
• Broadcast key identifiers
• Broadcast plaintext and corresponding ciphertexts

6

• Path key establishment
• Establish keys through shared neighbors
• Revocation
• A compromised sensor only discloses a small part
  of the keys
• Addition of new sensors
• A question can we link the pre-distributed keys
  to node identity or other information?
• What will be the advantages and disadvantages?

7
(No Transcript)
8

• Three extensions (by Chan et al, Oakland)
• q-composite
• Multipath reinforcement
• Random pairwise key
• Evaluation criteria
• Resilience to node capture
• Revocation
• Scale
• Clone attack

9

• q-composite pre-distribution
• A pair of nodes have to share at least q keys to
  establish a secure link b/w them
• When q increases, attackers have to compromise
  more nodes to break a link
• We still need to preserve a certain probability
  that any pair of sensors can establish a key
• How to balance the two factors?

10

• A simple analysis
• What is the probability that a pair of sensors
  have exactly i shared keys?
• The probability that a pair of nodes have at
  least q shared keys is 1- p(0) - p(1) - -
  p(q-1)
• Improvements in resilience to node capture
• q-composite demonstrates better resilience than
  basic scheme when a small group of sensors are
  compromised
• It makes the system more robust since it is more
  difficult to compromise a large group of sensors

11

• Say we have node A and B, neither of them has
  been compromised. What is the probability that
  the link between them is not safe if x nodes have
  been captured?
• For a specific key, the probability that it is
  compromised is?

12
(No Transcript)
13

• Multipath key reinforcement
• Try to establish a link key through multiple
  paths
• It will be great to use with the basic mechanism,
  but not the q-composite
• Now the malicious node has to compromise more
  keys to get the link key

14

• Let us assume that A and B have already find a
  single shared key
• Now A and B will determine a new link key through
  multiple independent paths
• A locates j link-disjoint paths
• Each link in the paths has established a link key
• A generates j random numbers and each random
  number will be sent through a different path
• the final key will be k xor R1 xor R2 xor --- xor
  Rj
• The final key is protected by all j random numbers

15

• The more paths
• The safer is the key
• The more communication overhead
• For each path, the longer is the path
• The higher probability that it is not safe
• 2-hop multipath
• Only go through the shared neighbors

16

• If the probability that an attacker compromises
  one link key is b, then the probability that it
  compromises a k two-hop path key is
• b (2b b2)k
• Why multipath and q-composite should not work
  together?
• q-composite needs a smaller key space

17
(No Transcript)
18

• Random pairwise key scheme
• In the basic scheme, we cannot authenticate the
  node
• The malicious node can conduct Sybil attack
• We need to link the knowledge of a key and the
  identity of the node
• Every key is known by only two nodes
• Every node remembers m keys and the other half
  that knows the key

19

• Can be used to
• Authenticate the other node
• Conduct intruder identification
• Prevent Sybil attacks
• Support node revocation