ABSTRACT

In the present day world, online Depending on the context on which a biometric

shopping using WAP enabled mobile phone has system works, it can be either classified as an

widely come into use. Credit cards serve as the Identification system or a verification

currency during e-business and e-Shopping. (authentication) system. Identification involves

As technology has advanced in the in establishing a person’s identify whereas in

negative side also hackers and spoofers steal verification involves confirming or denying a

misuse credit card numbers, even though the person’s claiming identity.

network has been made secure.

So, in this paper, we have proposed a
multi-biometric model (integrating voice, MULTIBIOMETRICS:
fingerprint and facial scanning) that can be A multi-biometrics system is obtained by
embedded in a mobile phone, this making e- the integration of multiple individual biometrics
transactions more secure. The model is very cost models. A numbers of models integrating hand
effective as we have tried to use the hardware geometry, keystroke dynamics, face and iris
already present in the phone.This paper uses for recognition system have flooded the markets in
image processing or facial recognization and recent years.
finger print. Here we present a multimodal system
that can be embedded in a mobile phone, which
integrates fingerprint, voice and facial scanning.
BIOMETRICS:
It shuts down the problem of high False
A biometric system is a recognition
Rejection Rate of facial scanners, eliminates the
system, which makes a personal identification by
fooling of fingerprint scanners and overshadows
determining the authenticity of a specific
the disadvantage of voice recognition models.
physiological or behavioral characteristic
possessed by the user. This method of Introduction to Biometrics:
identification is preferred over traditional A wide range of biometric
methods involving passwords and PIN numbers traits can be employed:
for various reasons: • Fingerprints, the most common
• The person to be identified is required mechanism
to be physically present at the point of • Iris
identification. • Retina
• Identification based on biometric • Voiceprint
techniques eliminates the need to • Hand geometry
remember a password or carry an • Palm prints (vein structure analysed
identity. via IR light)
•…
Extract features of the
presented trait and match
the result against a stored
template:
• Process is lossy, matches are
approximate
Two Usage Modes for
Biometrics:
Mode 1: Access control:
• Only this exact person is allowed in
• Primary identifier uniquely identifies
someone
– Personal ID (public value)
– PIN/password (private value)
• Biometric backs up the primary ID
– 1:1-match biometric check weeds out
the majority of impersonators
– Match only this one identified person
and no-one else
Mode 2: Identification
• Inexact match used to find things
– Find one of 3 million people (DHS terrorist
list) from a population of 6 billion
• Real-life analogy: “Was this the person who
robbed you” vs. “Find the person who robbed
you in these 25 shelves of books of mugshots”
• The answer to all your terrorism problems
.NEED FOR BIOMETRICS IN
MOBILE PHONES:
Mobile phones have ceased to be
exclusive status of the high class and, today has
become an indispensable electronic gadget in the
life of many. The main reason for their higher
market penetrations in recent days is their
incredible array of functions at an affordable
cost. Apart from setting remainders and sending
e-mails, they are also used in
• e-business
• SMS messaging
• Chatting
• Telemedicine and teleconferencing
Thus, these phones with wide
roaming facility prove to be a really versatile
device
Nowadays, shopping through the internet
has become very popular and surely, a WAP
enabled mobile phone provides the facilities to
consumers to shop online. Credit cards continue
to be an efficient tool for online money
transactions. But, on the other hand, credit card’s
number can be stolen on its way to its destination
and can be misused by hackers. Thus, e-Business
through a mobile phone becomes insecure.
Also, a report in www.download.com
stated that much anti-fraud Software, like those
provided by ArticSoft and ISC, created a back
door entry and were largely involved in data
spoofing. In addition to this, many user and
companies were prone to the attack of many
viruses and Trojan horses.
With so much of problems faced, the
service provide turned their attention towards
biometrics to prevent data spoofing and to
provide secure e-Transactions
Though security applications that verify a
person's identity based on their physical
attributes, such as fingerprint readers or iris
scanners, have been in use for some time,
biometric security has only recently started to
appear in mobile phones, PDAs and notebook
computers where the need for miniaturization
represents a technological challenge.
So far biometric data has been used to tie the
device to a person to prevent it from being used
illegitimately if lost or stolen. But the IST project
SecurePhone is taking a new approach,
employing physical attributes to enable the user
to digitally sign audio, text or image files,
providing proof of their origin and authenticity.
Although existing communications infrastructure
based on the GSM, GPRS and UMTS mobile
systems provides a secure means of
communication, it lacks any robust method of
user identification. Text, audio and image files
can be sent by anyone to anyone with no
authentication and there are no guarantees the
person you are talking to in a phone
conversation, if you've never met them before, is
really who they claim to be.
The upshot is that data exchanged over mobile
devices is of limited use for legally binding
transactions even though mobile devices, given
their ubiquity, would be a prime candidate for
carrying out e-commerce (or m-commerce),
managing business processes such as signing
contracts or even in securing the exchange of
data in e-healthcare and e-government systems.
A digitally signed and authenticated voice
recording during a telephone conversation
would, for example, give the speaker's words
legal value.
"The aim is to enable users to exchange
information that can't be disputed afterward.
That could be a voice recording that is
authenticated to eliminate any doubt about who
the speaker is, what they actually said and prove
that it has not been manipulated," Ricci explains.
"To achieve that it is necessary to digitally sign
the data and to ensure that only the legitimate
user can perform the signing."
The system developed by the SecurePhone
project partners consists of two main elements.
The first, an authentication module, uses
biometric security applications to verify the
user's identity. That in turn gives them access to
the second module which digitally signs the data
using a Public Key Infrastructure (PKI).
The system, which is designed primarily for
PDA-phones but could also be used in new
generation smart phones and WiFi-enabled
PDAs, offers three methods of biometric
identification. One employs the digital cameras
that have become commonplace in mobile
devices along with a face recognition application
to identify the user based on their facial features.
Another uses voice recognition software – also
detecting any asynchrony between speech and lip
movements - and the third verifies the
handwritten signature of the user on the device's
touch screen. The three methods are used in
combination to enhance the overall levels of
security and reliability, and most importantly
they require no hardware additions to mobile
devices.
FACE RECOGNITION:
Facial recognition is considered to be one of the
most tedious among all scans. Further, difficulty
in acquisition of face and cost of equipments
make it more complex.
 However, some WAP enabled
phones like CX 400K and LG-SD1000
manufactured by LG electronics, have built in
camera that can acquire images and can be
transmitted over internet. This it is sent to the
credit card company to verify the face received
matches with the face in their database. If it
matches, the goods are sent, else the order is
rejected.
Face recognition uses mainly the following
techniques:
• Facial geometry: uses geometrical
characteristics of the face. May use
several cameras to get better
accuracy (2D, 3D...)
• Skin pattern recognition (Visual Skin
Print)
• Facial thermogram: uses an infrared
camera to map the face temperatures
• Smile: recognition of the wrinkle
changes when smiling
Facial geometry:
Many different methods based on geometrical
characteristics of the face have been developed
such as "local feature analysis", "Eigen face or
Principal Component Analysis", ...
Skin pattern recognition:
Visual Skin Print relies on standard hardware -
most web-cams and higher resolution mass-
market video cameras, connected to a PC, will
work. Visual Skin Print™ is based on a simple
yet powerful idea: using the details of the skin
for authentication.
Facial thermogram:
Facial thermogram requires an (expensive)
infrared camera to detect the facial heat patterns
that are unique to every human being.
Technology Recognition Systems worked on that
subject in 1996-1999. Now disappeared.
"Face Recognition in Hyperspectral Images" is
an article describing a variant using several
wavelentghs.
Smile recognition:
The Stony Brook university system relies on
probing the characteristic pattern of muscles
beneath the skin of the face.
Guan takes two snaps of a person in quick
succession, asking subjects to smile for the
camera. He then uses a computer to analyse how
the skin around the subject's mouth moves
between the two images. The software does this
by tracking changes in the position of tiny
wrinkles in the skin, each just a fraction of a
millimetre wide.
The data is used to produce an image of the face
overlaid with tiny arrows that indicate how
different areas of skin move during a smile. This
movement is controlled by the pattern of
muscles under the skin, and is not affected by
the presence of make-up or the size of the
subject's smile. The system is sensitive enough
to produce such a map from muscle twitches
even when people are trying to keep their
expression unchanged
Dynamic facial features:
In A Dynamic Approach to Face Recognition
paper, a new method for face recognition is
proposed, which is based on dynamic instead of
static facial features. They track the motion of
certain features on the face during a facial
expression and obtain a vector field that
characterizes the deformation of the face.
In the training process, two high-resolution
images of an individual, one with a neutral
expression and the other with a facial
expression, like a subtle smile, are taken to
obtain the deformation field.
VOICE RECOGNITION:
The speaker-specific characteristics of speech
are due to difference in physiological and
behavioral aspects of the speech production
system in humans. The main physiological
aspect of the human speech production system is
the vocal tract shape. The vocal tract modifies
the spectral content of an acoustic wave as it
passes through it, thereby producing speech.
Therefore, it is common in speaker verification
systems to make use of features derived only
from the vocal tract.
The microphone in the mobile phone
captures the speech. Then, using cepstral
analysis, an utterance may be represented as a
sequence of feature vectors. Utterances, spoken
by the same person but at difference times, result
in similar yet a different sequence of features
vectors. So, the irrespective of the mood of the
consumer, his transaction is accepted or rejected.
The following algorithm may be used in voice
verification.
Voice Recognition Biometrics, continues to
pioneer methods of speach verification that can
effectively operate without the benefit of an
external server or hard drive. The entire software
program and templates can be operated and
stored in the protected device itself. Optionally,
the voice templates used for validation can be
loaded at verification time from a smart-card or
stored in external RAM due to their small size.
The greatest benefit of our method is that user
security can be self-contained within a device.
Cell Phones, ATM cards, Automobiles and Door
Access are just some of the applications that are
protected by our Voice Protection Technology.
Other speaker verification technologies may rely
on costly database management of many users’
templates, and in some cases risk sending the
biometric data or even the voice stream itself
over phone lines, air waves, or Internet lines.
Our Voice Protection Technology, ensures that
the biometric data stays safe as the spoken
phrase cannot be reconstructed from the "key"
created by the user. For Internet applications, the
user's "voice prints" can be contained in a single,
encrypted IP packet as it is less than 1K in size.
This is your best assurance, that we can deliver
biometric security technology to the market at a
lower cost than anyone else in the industry, with
no re-occurring maintenance costs such as
database management or complicated user
training. Once our Voice Protection technology
is built into a product it will continue to function
independently for the life of the product.
How Voice Recognition Works:
The first step in voice recognition is for an
individual to produce an actual voice sample.
Voice production is a facet of life in which we
take for granted every day, and the actual
process is complicated. The production of sound
originates at the vocal cords. In between the
vocal cords is a gap. When we attempt to
communicate, the muscles which control the
vocal cords contract. As a result, the gap
narrows, and as we exhale, this breathe passes
through the gap, which creates sound. The
unique patterns of an individual's voice is then
produced by the vocal tract. The vocal tract
consists of the laryngeal pharynx, oral pharynx,
oral cavity, nasal pharynx, and the nasal cavity
(1). It is these unique patterns created by the
vocal tract which is used by voice recognition
systems. Even though people may sound alike to
the human ear, everybody, to some degree, has a
different or unique annunciation in their speech.
To ensure a good quality voice sample, the
individual usually recites some sort of text,
which can either be a verbal phrase or a series of
numbers. The individual usually has to repeat
this a number of times. The most common
devices used to capture an individual's voice
samples are computer microphones, cell (mobile)
phones, and the land line based telephones. As a
result, a key advantage of voice recognition is
that it can leverage existing telephony
technology, with minimal disruption to an
entity's business processes. In terms of noise
disruption, computer microphones and cell
phones create the most, and land line based
telephones create the least. There are also other
factors which can affect the quality of voice
samples other than the noise disruptions created
by telephony devices. For example, factors such
as mispronounced verbal phrases, different
media used for enrollment and verification
(using a land line telephone for the enrollment
process, but then using a cell phone for the
verification process), as well as the emotional
and physical conditions of the individual.
Finally, the voice samples are converted from an
analog format to a digital format for processing.
The next steps are unique feature extraction and
creation of the template. The extraction
algorithms look for unique patterns in the
individual's voice samples. To create the
template, a model of the voice is created. In
voice recognition systems, stochastic models,
particularly Hidden Markov models, have been
utilized. With this type of modeling, statistical
profiles are created by comparing various voice
samples to determine any repeating patterns.
The final step is verification of the individual. At
this stage, the live voice sample submitted for
verification is compared to the statistical profiles
created, and a probability score is then computed
which describes the likelihood that the individual
is who he or she claims to be.
FINGERPRINT ACQUISITION:
Finger based scanning is one
of the oldest methods used for verification.
Fingerprints, unique and immunable for all are
made of series of ridges and furrows on the
surface of the finger. These ridges and furrows
determine the uniqueness of the fingerprints.
Apart from these, minute points (i.e. local ridge
characteristics that occur at either a ridge
bifurcation or a ridge ending also play role in
fool-proofing this biometric technique.
To reduce the search time and
the computational complexity, fingerprint
classification is undertaken and thus fingerprints
are classified as whorl, right loop, left loop, arch,
and arch. Recently researchers and scientists
achieved a great feat by improving the
fingerprint classification to 94%.
In today’s world, fingerprint
matching techniques can be placed into two
categories: minutiae-based and correlation based.
In minutiae based technique, the minutiae points
are found and their relative placement are
mapped on the finger whereas in correlation
based technique, the fingerprint acquired from
the person is checked for certain points
previously stored in the database. If both
matches, the person is given authentication, else
he is denied permission.
Transaction scanner
embedded above display screen
The scanner here is a transparent layer
above the screen. The scanner consists of arrays
of capacitors of the size of 0.03µ m. capacitors
with such a small size can be manufactured with
MEMS technology. When the consumer places
his thumb on the scanner, the points at which his
fingerprint touches the screen get discharged
whereas others remain charged. Thus the
fingerprint is scanned and is then sent for further
process
Fingerprint Matching:
Among all the biometric techniques, fingerprint-
based identification is the oldest method which
has been successfully used in numerous
applications. Everyone is known to have unique,
immutable fingerprints. A fingerprint is made of
a series of ridges and furrows on the surface of
the finger. The uniqueness of a fingerprint can be
determined by the pattern of ridges and furrows
as well as the minutiae points. Minutiae points
are local ridge characteristics that occur at either
a ridge bifurcation or a ridge ending.
Fingerprint matching techniques can be placed
into two categories: minutae-based and
correlation based. Minutiae-based techniques
first find minutiae points and then map their
relative placement on the finger. However, there
are some difficulties when using this approach. It
is difficult to extract the minutiae points
accurately when the fingerprint is of low quality.
Also this method does not take into account the
global pattern of ridges and furrows. The
correlation-based method is able to overcome
some of the difficulties of the minutiae-based
approach. However, it has some of its own
shortcomings. Correlation-based techniques
require the precise location of a registration point
and are affected by image translation and
rotation.
Fingerprint matching based on minutiae has
problems in matching different sized
(unregistered) minutiae patterns. Local ridge
structures can not be completely characterized by
minutiae. We are trying an alternate
representation of fingerprints which will capture
more local information and yield a fixed length
code for the fingerprint. The matching will then
hopefully become a relatively simple task of
calculating the Euclidean distance will between
the two codes.
Algorithms are being developed which are more
robust to noise in fingerprint images and deliver
increased accuracy in real-time. A commercial
fingerprint-based authentication system requires
a very low False Reject Rate (FAR) for a given
False Accept Rate (FAR). This is very difficult
to achieve with any one technique. We are
investigating methods to pool evidence from
various matching techniques to increase the
overall accuracy of the system. In a real
application, the sensor, the acquisition system
and the variation in performance of the system
over time is very critical. We are also field
testing our system on a limited number of users
to evaluate the system performance over a period
of time.
Fingerprint Classification:
Large volumes of fingerprints are collected and
stored everyday in a wide range of applications
including forensics, access control, and driver
license registration. An automatic recognition of
people based on fingerprints requires that the
input fingerprint be matched with a large number
of fingerprints in a database (FBI database
contains approximately 70 million fingerprints!).
To reduce the search time and computational
complexity, it is desirable to classify these
fingerprints in an accurate and consistent manner
so that the input fingerprint is required to be
matched only with a subset of the fingerprints in
the database.
Fingerprint classification is a technique to
assign a fingerprint into one of the several pre-
specified types already established in the
literature which can provide an indexing
mechanism. Fingerprint classification can be
viewed as a coarse level matching of the
fingerprints. An input fingerprint is first matched
at a coarse level to one of the pre-specified types
and then, at a finer level, it is compared to the
subset of the database containing that type of
fingerprints only. We have developed an
algorithm to classify fingerprints into five
classes, namely, whorl, right loop, left loop,
arch, and tented arch. The algorithm separates
the number of ridges present in four directions (0
degree, 45 degree, 90 degree, and 135 degree) by
filtering the central part of a fingerprint with a
bank of Gabor filters. This information is
quantized to generate a FingerCode which is
used for classification. Our classification is
based on a two-stage classifier which uses a K-
nearest neighbor classifier in the first stage and a
set of neural networks in the second stage. The
classifier is tested on 4,000 images in the NIST-4
database. For the five-class problem,
classification accuracy of 90% is achieved. For Practical Problems with
the four-class problem (arch and tented arch
Biometrics:
combined into one class), we are able to achieve
a classification accuracy of 94.8%. By
Biometric systems have never
incorporating a reject option, the classification
had to withstand serious
accuracy can be increased to 96% for the five-
class classification and to 97.8% for the four-
attack:

class classification when 30.8% of the images • Smart cards took 15 years of

are rejected. criminals walking all over them before

vendors started taking security
Fingerprint Image Enhancement: seriously.
Fingerprint scanners work
A critical step in automatic fingerprint matching poorly with the elderly, manual
is to automatically and reliably extract minutiae
workers, children:
from the input fingerprint images. However, the
• Children haven’t developed strong
performance of a minutiae extraction algorithm
fingerprints yet.
relies heavily on the quality of the input
• Manual workers and the elderly don’t
fingerprint images. In order to ensure that the
have strong fingerprints left.
performance of an automatic fingerprint
• German passport enrolment system
identification/verification system will be robust
ran into problems with people as “old”
with respect to the quality of the fingerprint
as 40 or 50.
images, it is essential to incorporate a fingerprint
– 10% of senior citizens can’t be
enhancement algorithm in the minutiae
reliably enrolled.
extraction module. We have developed a fast
fingerprint enhancement algorithm, which can
Established wisdom:
adaptively improve the clarity of ridge and 3-4% of the population (goats)
furrow structures of input fingerprint images have unstable biometric traits that
based on the estimated local ridge orientation can’t be identified by
and frequency. We have evaluated the sensors
performance of the image enhancement
• In practice it’s often much, much
algorithm using the goodness index of the
higher
extracted minutiae and the accuracy of an online
• Up to 50% reject rate, see later
fingerprint verification system. Experimental
slides
results show that incorporating the enhancement

algorithms improves both the goodness index Fingerprint readers have

and the verification accuracy. problems with outdoor use
E.g: in winter cold

Attack:
Train the system to accept less and
less reliable images
• Has happened (inadvertently) in real-
world deployments as sensors were
subject to wear and tear.
• System would accept anything
(elbow, nose) as a valid print
People forget which finger
they enrolled with and try
each one in turn
• Alternatively, a failure to verify the
chosen print would lead to them trying
all other fingers just in case
• (Happens to a lesser extent with
passwords as well)
FARs for mass-market
fingerprint readers are
already wound sky-high to
avoid consumer acceptance
problems:
• Need to wind the FAR up to the point where
FRR = 0.The readers will have a fairly broad
tolerance on the basis that products that stop
people using their own cars, computers or
whatever because their fingers are a bit sweaty
won’t turn out to be very popular.
UK passport service required
a biometric-compatible
photo instead of the
standard one:
• 80,000 of the first set of photos
(600K) were rejected for not meeting
the requirements.
• Computer-based systems are vastly
easier to confuse than humans.
UK Passport Service
Enrolment Trial:
• Enrolled 9,000 travellers to
evaluate the enrolment
experience.
• 10% of users couldn’t enrol
using the iris-recognition
system
• 30% of able bodied users
couldn’t have their facial
biometrics verified
50% of disabled users
couldn’t have their facial
biometrics verified (!!)
• Particularly problematic due to laws
like ADA that forbid discrimination
against the disabled.
“Fix” was probably to wind
the sensitivity down so that
almost anything would pass
• This is the universal solution for any
failures in systems that employ
approximate matching..
German passport enrolment 10-year life was that that
requirements stipulate the was about the maximum
use throughput of the
of the least unreliable trait processing system
sample if no reliable one can • Now, vastly more complex passports
have to be rolled over in half the time
be obtained
There doesn’t seem to be any minimum
acceptable quality measure for trait sampling
Failure rates were
• If all else fails, click the “No Hand” button on
the fingerprint significantly higher for
software people with darker skin
If the fingerprint isn’t readable, which isn’t and/or eyes:
uncommon, the passport records “No Hand” • Dark skin/dark eyes absorb more
[“Keine Hand”]. There is no provision for light
“Fingerprint not readable” • Features don’t stand out as much
This is an unfortunate ‘solution’
These are all features of the
ethnic group that these
Verification time was a full systems are targeting:
minute across a variety of • The targets are the ones least likely
systems: to be correctly processed!
• Current verification time for intra-EU
travel is < 5s CONCLUSION:
• Order-of-magnitude slowdown for Thus, this mobile multi-biometrics
traveller verification can be embedded in mobile phone. Phone is cost
effective since no special hardware in required
and is highly secured. Thus, this mobile phone
Enrolment has similar
becomes a reality will provide more e-Business
problems: and e-Transactions.
• German biometric enrolment system .
was advertised as taking 2½ minutes,
in practice takes > 10 minutes

One of the reasons why pre-

biometric passports had a