Browse Prior Art Database

Personal Identification Systems Performance Improvement Algorithm

IP.com Disclosure Number: IPCOM000052726D
Original Publication Date: 1981-Jul-01
Included in the Prior Art Database: 2005-Feb-11
Document File: 3 page(s) / 50K

Publishing Venue

IBM

Related People

Grossman, RS: AUTHOR [+3]

Abstract

This algorithm increases the effectiveness of typical Personal Identification Verification (PIV) systems. PIV systems utilize various identification tests; for example, an instrumented pen may be used to measure accelerations, timings, pen lifts, etc., which occur during the writing of a signature, and these can be compared to a previously recorded profile. The failure rate of a typical PIV system is measured by the rate at which the system both incorrectly rejects individuals with valid identities and accepts individuals with false identities.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Personal Identification Systems Performance Improvement Algorithm

This algorithm increases the effectiveness of typical Personal Identification Verification (PIV) systems. PIV systems utilize various identification tests; for example, an instrumented pen may be used to measure accelerations, timings, pen lifts, etc., which occur during the writing of a signature, and these can be compared to a previously recorded profile. The failure rate of a typical PIV system is measured by the rate at which the system both incorrectly rejects individuals with valid identities and accepts individuals with false identities.

Typically, in PIV systems, a score is generated for each trial and compared to a PASS/FAIL threshold. If the score for a trial exceeds the threshold, the individual's identity is verified and access to the controlled resource is granted. If the score does not exceed the threshold, access is denied and additional trials may be given. Variations in the system and people essentially require that multiple trials be given to achieve an acceptable rate of false rejection. Multiple trials, however, increase the false acceptance rate as well as increasing the computational load on the system.

This algorithm provides the means to reduce the false acceptance rate. This improvement is accomplished in three ways:

The first feature of this algorithm is the implementation of an active rejection strategy. This algorithm does not give all users the same number of trials, and will actively reject users who do not perform well enough.

The second feature is the use of the individual's history in this session to determine the action to be taken. The algorithm does not treat each trial as in independent event but rather as one of a sequence of events. In this way, sustained performance (either good or bad) against the thresholds is taken into account and the system responds appropriately.

The third feature is the use of thresholds variably biased by statistically chosen multipliers. Some systems use a GO/NO-GO threshold. If the user's score does not equal or exceed the threshold, regardless of how close it may be over many trials, the user is still rejected. In contrast, in this algorithm, users who are close to passing, as measured by the variable thresholds, get more trials. No user is accepted, however, unless he does in fact meet his computed threshold.

The algorithm below shows a typical implementation for a signature verification application. This description of the algorithm processing is keyed to the algorithm flow chart. BLOCK ACTION, EXPLANATION 10 The algorithm is invoked by the start of a verification session. 12 The user identifies himself via employee number, etc., signs his name, and a score is generated by the corre...