IEEE Technology and Society Magazine - September 2016 - 63

source of light will provide an indication as an active
glint in the pupil of the eye to show it if is a live eye, or
a false video eye.
Secondary biometrics can further improve the robustness of biometric systems to defend against potential
spoofing attacks. Daugman has published results of studies involving the order of hundreds of billions of crosscomparisons of iris codes [5] showing that while one iris
code might be duplicated across a large segment of the
population, there is almost no statistical likelihood for a
pair of people to have both iris patterns duplicated. This
does not occur even in the case of identical twins. A similar logic follows if we use a combination of two different
biometrics, e.g., iris and palmprint, or iris and fingerprint.
And using two complimentary methods of liveness detection can also reduce the potential for spoofing. Thus an
analysis of the lips region of the face and a comparison
with extracted word structures from an audio recording
could be used as a liveness measure for speech detection. The combination of the speech itself with the voice
characteristics could be used as a biometric.
So while identity theft via stolen biometrics is feasible,
it is far from easy to implement. The challenges posed
are only likely to grow more sophisticated in the future.

asked to verify themselves on the telephone they are
invariably asked a sequence of questions about their
past life - where they lived, their first car, their first pet,
their best friend at school, mother's maiden name and
of course, their date of birth; this exact same information has been provided to tens, even hundreds of other
companies, services, and websites. Just one of these
entities could have a dishonest employee willing to steal
and sell on such data - how is this less a risk than committing one's biometric data to a modern electronic
device that sits in a jacket pocket most of the day? At
least the user knows the device's location 24/7 until they
manage to lose it, and even then they can get to a network and hit the kill switch.
What if we return to the use of a personal signature as
the legally binding baseline of authentication? How difficult is it to perform a high-resolution scan of your signature from a page and extract the corresponding bitmap
image? And the latest version of Adobe PDF viewer allows
you to import just such a bitmap to facilitate electronic
signing of PDF documents! So which is more secure - an
advanced biometric obtained using liveness detection to
ensure that you were actually holding and operating the
device, or your handwritten signature? I know where I'd
prefer to put my money, literally as well as figuratively.

Supervised Versus
Unsupervised Authentication

Privacy Concerns

This is perhaps the biggest leap with smartphone biometrics - almost all other existing biometric systems
employ a supervised authentication process. There is
always a human overseer who can step in where the process fails. This is a luxury that is not available to the purveyors of consumer products and if the device does not
perform as expected without the need for user supervision, it invariably ends back with the manufacturer.
A continual challenge with consumer systems is that
everything is expected to work and to work consistently,
even in difficult non-standard conditions. Looking back
historically to the mid-2000s when the first pocket-PC
devices appeared with a sweep fingerprint scanner we
can hypothesize that their short timeframe in the market
was probably due to poor reliability of the fingerprint
authentication system. While this was never admitted
publicly, it makes sense. The achievement of Touch IDTM
where others have failed must be acknowledged [6].
The other challenge of unsupervised authentication is,
naturally, that you are not around to detect when the "bad
guys" try to crack your authentication system. This challenge is less tested and there will definitely be a great deal
of discussion and publicity directed here as biometrics
become further embedded into mainstream devices.
But the use of biometrics is not less secure than
many of the "manual" systems used today to secure our
credit cards and bank accounts. When a customer is

This brings us to the topic of biometrics and privacy. If
personal biometrics are to be used as a means of
authentication it becomes critical to consider the use
cases employed. Legacy biometric techniques gather
data in centralized databases, and these "enrolled biometrics" become a permanent record of your identity.
Thus the owner of the data becomes the effective arbiter of your identity.
Clarke [7] has written in detail on this topic. He separates privacy into a number of sub-categories and
emphasizes the need for safeguards, depending on the
particular use of biometric data. These safeguards are
essential if biometric technology is not to fall into illrepute, even in relatively free societies. In more authoritarian societies, he considers that the worst fears
expressed in popular culture may well become reality.
While Jain and Nandakumar [8] focus more on the
maturity of biometric technology, they recognize the
importance of considering privacy in any particular
application of biometrics. More specifically they raise
several key concerns:
■ Who owns the biometric data, the individual or the
service provider?
■ Will the use of biometrics be proportional to the need
for security in a given application? Should a user be
required to provide a fingerprint in order to purchase
a hamburger or access a commercial website?

september 2016

∕

IEEE Technology and Society Magazine

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - September 2016