jspωiki
Strength of Function for Authenticators - Biometrics

Overview#

Strength of Function for Authenticators - Biometrics (SOFA-B) includes variables for:

Strength of Function for Authenticators - Biometrics attempts to provide a starting point for the overall Strength of Function for Authenticators framework by identifying the ways in which biometric authenticator strength can be measured and evaluated. It focuses on three core concepts: False Match Rate, Presentation Attack Detection Error Rate, and Effort .

Match Rate Interdependence#

False Match Rate (FMR) and False Non-Match Rate (FNMR) are interdependent: adjusting the biometric system to decrease one of these rates will increase the other. When two biometric samples are compared, a biometric system will generate a similarity score. For example, a range of similarity scores may be 1 (for low similarity) to 100 (an exact match). The binary match or non-match decision is made based on where this score falls with respect to a decision threshold:
  • a value above the threshold is considered a match
  • a value below the threshold is considered a non-match.
Setting this threshold is a Risk Assessment decision because it results in errors, as some number of true match scores will fall below the threshold, while some true non-matches will fall above it. Changing the decision threshold to a higher value will decrease false matches, but it will also increase false non-matches. In other words, the False Match Rates and False Non-Match Rates are interdependent: adjusting the biometric system to decrease one of these rates will increase the other.

Strength of Function for Authenticators - Biometrics for Opportunistic Attacks, the statistic that is of interest is the False Match Rates because it is a measure of the degree of difficulty to find an imposter that randomly matches a legitimate user. False Non-Match Rates is a measure of User Experience, signifying how often a legitimate user will be rejected. However, when False Match Rate is measured, it is necessarily measured at an False Non-Match Rates level because the two statistics are interdependent.

More Information#

There might be more information for this subject on one of the following: