* Methods for authentication
When screening large numbers of people, linking identification to real-world identity (that is, authentication) is a tough problem. As readers probably know, there are four basic methods for authentication:
* What you know that others don’t (e.g., passwords).
* What you have that others don’t (e.g., tokens such as keys or smart cards).
* What you do that others can’t (e.g., the way you sign your name or the phrase on a keyboard).
* What you are that others aren’t (e.g., your fingerprints, retinal patterns, iris characteristics, or face).
Passwords don’t work very well for crowds. Tokens are used all the time – consider airline tickets and passports – but in today’s digital scanning and printing world, they are easy to counterfeit (I’ll be looking at new mechanisms for safeguarding passports in another article).
A report last year by Jonathan Krim of the Washington Post pointed out that facial recognition systems using photographs can have serious problems: “… [F]ederal researchers who have tested face-recognition technology say its error rate is unacceptably high – up to 50% if photographs are taken without proper lighting.” https://www.washingtonpost.com/wp-dyn/articles/A43944-2004Aug5.html An American Civil Liberties Union (ACLU) report revealed that in face recognition trials at the Palm Beach Airport in 2002, “…the system failed to match volunteer employees who had been entered into the database fully 503 out of 958 times, or 53% of the time.” https://www.aclu.org/Privacy/Privacy.cfm?ID=10340&c=130
Unlike fingerprint and retinal scans, both facial and ear recognition can be relatively non-intrusive, requiring little interference with or involvement by the subject (no physical contact or unusual procedures such as staring into a lens).
Iris recognition is another biometric technology that has required some cooperation by the subject; however, there have been reports that new technology should permit iris recognition at a distance. Tabassum Zakaria, reporting for the Australian Broadcasting Corporation in 2003 https://abc.net.au/science/news/stories/s982770.htm quoted U.S. CIA officials as working on new biometric systems with a ten-fold improvement in recognition rates.
A July 14, 2005 report by Duncan Graham-Rowe explains that University of Southampton (U.K.) biometrics researcher Mark Nixon is finding that ears may provide excellent features for biometric identification systems https://www.newscientist.com/channel/mech-tech/dn7672. Nixon points out that ears are relatively stable compared with other facial features and do not change with people’s expressions. His initial trials used pictures of 63 people and found 99.2% accuracy – an error rate much lower than for facial recognition systems.
So unfortunately for us in the snow belt, ear muffs may eventually be seen as threats to security.
