UW CSEP 590 Term paper

1
UW CSEP 590 Term paper

• Biometric Authentication
• Shankar Raghavan

2
Definition and Advantages

• Physical or behavioral characteristics
• Much longer, random than a traditional password
• Always there with the person
• Unique to a person

3
Common biometric identifiers
Biometric Identifier Distinctiveness Permanence Performance Acceptability
Fingerprint High Medium High Medium
Hand Medium Medium Medium Medium
Iris High High High Low
Retinal High High High Low
Voice Low Low Low High
Signature Low Low Low High
FRR/FAR are measures for accuracy
4
Authentication mechanism, threats

• If template is compromised, is the biometric
  identifier lost for ever?
• Is there any good method for Tye 1 or type 2
  attacks?

5
Fingerprint minutiae
6
Attacks and Defences

• Dummy finger (Type 1)
• Gummy finger Matsumoto (Type 1)
• Hill climbing attack (Type 4)
• Liveness detection (Type 1)
• WSQ data hiding (Type 2)
• Advantages of iris/retinal scannings
• Image based challenge response systems

7
Fuzziness in biometrics

• n of N attributes match
• Application of Shamirs secret sharing
• Identity based encryption using bilinear maps
• Uses different polynomials for each user
• Generates a private key for every attribute user
  has, this is distinct and not shared with another
  user.
• Interpolates polynomial in an exponent

8
Fuzzy biometrics (backup)

• Fuzzy commitment
• x and x will both decode to a similar value
• Not able to handle rotational/translational
  aspects of order invariances in an image very
  well.
• Fuzzy vault
• Maps a set that hold the key to a secret onto a
  polynomial p ie each value in the set is an x
  coordinate for a point evaluated by p.
• Adds some noise or chaff so that the encrypted
  set becomes p(x),

9
Questions?