Long road ahead for biometrics, developer says
Connecting state and local government leaders
A gathering of open-source software developers in Washington heard yesterday that niche biometrics applications are not yet mature or scalable enough for widespread use.<br>
Taylor Boone, who has spent five years developing middleware for biometric applications, is a fan of the technology.
'I believe biometrics have great promise,' said the former chief technology officer of BNX Systems Corp. of Vienna, Va. 'The big application I see coming is for fingerprints on smart cards.'
But Boone yesterday told a gathering of open-source software developers in Washington that niche biometrics applications are not yet mature or scalable enough for widespread use.
'For the most part, this technology is barely above the toy stage,' Boone said at the Secure Trusted Operating System Consortium symposium in Washington. Dependable enterprise applications are three to five years away, he predicted.
Agencies are intensely interested in biometric measurements for authenticating people for physical as well as logical access. The most common measurements such as fingerprints, iris scans and facial recognition all have their advantages and disadvantages, Boone said.
The FBI's Automated Fingerprint Information System uses fingerprints, the best-known and most successfully automated biometric. Despite a criminal stigma attached to prints, Boone thinks it is the most likely technology to reach widespread deployment soon. It is 'head and shoulders above the rest' and easily used with smart cards to authenticate users, he said.
A more troubled biometric technology is facial recognition, which has been used in public places to identify people without their knowledge.
'To my mind, that is an invasion of privacy,' Boone said.
So far, pilots for facial recognition have failed, 'but a lot of research is going into it, which is scary to me,' Boone said. 'Eventually there is going to be enough capacity to filter out the noise and get matches.'
Regardless of the technology used, biometrics faces common challenges. The accuracy of any biometric system depends on the integrity of the initial enrollment and the security of the underlying network and databases.
Nor is biometrics an exact science. The templates that identify a user are not perfectly reproducible, so only approximate matches are possible. The degree of assurance for a match depends on the system requirements. Strict requirements can result in too many false rejections, and low requirements raise the chance of false acceptances.
And for all biometric technologies, 'some segment of the population cannot use it,' Boone said. 'You'd better have multiple ways to authenticate users.'