BIOMETRIC AUTHENTICATION DEVICES AND SEMANTIC
WEB SERVICES
An Approach for Multi Modal Fusion Framework
L. Puente Rodríguez
Universidad Carlos III de Madrid, Avda de la Universidad, 30, Leganés, Madrid, Spain
M. J. Poza
Universidad Francisco de Vitoria. Ctra. Pozuelo-Majadahonda Km. 1.800. Pozuelo de Alarcón, Madrid, Spain
J. M. Gómez, B. Ruiz
Universidad Carlos III de Madrid, Avda de la Universidad, 30, Leganés, Madrid, Spain
Keywords: Biometrics, Authentication Devices, Emerging Technologies, Data Integration, Semantic Web Services,
Ontologies.
Abstract: Identity verification is now a days a crucial task for security applications. In the near future organizations
dedicated to store individual biometric information will emerge in order to determine individual identity.
Biometric authentication is currently information intensive. The volume and diversity of new data sources
challenge current database technologies. Biometric identity heterogeneity arises when different data sources
interoperate. New promising application fields such as the Semantic Web and Semantic Web Services can
leverage the potential of biometric identity, even though heterogeneity continues rising. Semantic Web
Services provide a platform to integrate the lattice of biometric identity data widely distributed both across
the Internet and within individual organizations. In this paper, we present a framework for solving biometric
identity heterogeneity based on Semantic Web Services. We use a multimodal fusion recognition scenario
as a test-bed for evaluation.
1 INTRODUCTION
Identity recognition is performed now a days by the
use of traditional techniques such as PINs,
passwords, digital signatures, etc. Biometrics
promise to offer a new alternative, portable, easy to
use, free of memory, loss or theft problems. A global
solution will be based on the creation of specialized
organizations offering authentication services. This
Biometric Accreditation Entities (BAE) will base
their services on previously acquired biometric data.
Biometrics authentication usually refers to the
identification of an individual based on his or her
distinguishing traits. In principle, a biometric
identity is based on the premise that a measurable
physical or behavioural trait is a more reliable
indicator of identity than the traditional systems such
as pairs composed by password and username,
Personal identification numbers (PIN) and the akin.
Particularly, since biometric identity technologies
deal with security and privacy issues, the challenge
for the research community is to attain integrated
solutions that address the entire problems from
sensors and data acquisition to biometric data
analysis and system design.
Presently, the lack of performance of biometric
systems is being alleviated by the use of multiple
biometric indicators for identifying an individual in
order to increase its accuracy when using a
technique called Multimodal Fusion (Kittler, J.
Hatef, R. Matas, J. G., 1998) (Jain R. and J. Quian,
2001). As a result of this, biometric information has
grown exponentially and algorithms for feature
extraction, matching score or decision levels handle
a tremendous amount of data. Furthermore, the
95
Puente Rodríguez L., J. Poza M., M. Gómez J. and Ruiz B. (2008).
BIOMETRIC AUTHENTICATION DEVICES AND SEMANTIC WEB SERVICES - An Approach for Multi Modal Fusion Framework.
In Proceedings of the First International Conference on Biomedical Electronics and Devices, pages 95-100
DOI: 10.5220/0001046300950100
Copyright
c
SciTePress
recent years have provided with a lattice of
duplicated efforts in building test databases such as
face recognition databases (e.g. FERET, PIE or
BANCA) (Bailly-Baillière E., S. Bengio et al., 2003)
and a lack of uniform standards and granted open
access to these databases, as discussed in (Ming, A.
Ma, H., 2007).
Hence, arguably the most critical need in
biometric identity recognition is to overcome
semantic heterogeneity i.e. to identify elements in
the different databases that represent the same or
related biometric identities and to resolve the
differences in database structures or schemas, among
the related elements. Such data integration is
technically difficult for several reasons. First, the
technologies on which different databases are based
may differ and do not interoperate smoothly.
Standards for cross-database communication allow
the databases (and their users) to exchange
information. Secondly, the precise naming
conventions for many scientific concepts in fast
developing fields such as biometrics are often
inconsistent, and so mappings are required between
different vocabularies.
Hence, we present in this paper a framework for
solving multimodal fusion oriented biometric
identity data heterogeneity problems, keeping the
structure of databases created with the aim of being
used for identity accreditation and distributed over
the Web. Our approach is based on the breakthrough
of adding semantics to Web Services which perform
a role of entry points for such databases.
Fundamentally, this implies that our framework
enables different biometric identity data to be
discovered, located and accessed since they provide
formal means of leveraging different vocabularies
and terminologies and foster mediation.
The remainder of this paper is organized as
follows. In Section 2, a brief state-of-the-art on the
technologies employed in our research is given.
Section 3 defines some terms we use along this
paper. Section 4 identifies the heterogeneity of data
involved in the biometric identification process.
Section 5 describes the framework for solving
problems using Semantic Web Services. Finally,
conclusions and related work are discussed in
Section 6.
2 STATE-OF-THE-ART
Semantic Web Services and Ontologies are the
cornerstone technologies applied in our research. On
the one hand, data interoperability between different
information sources is achieved by means of
ontologies and their mapping. On the other hand,
Web Services semantically annotated are the
software entities responsible for providing a
normalized interface to disparate functionality and
data sources. In this section, a brief description of
each of these technologies is put forward.
2.1 Ontologies
Although a number of different ontology definitions
can be found currently in literature, in this work we
use Borst’s one (Borst, W.N., 1997): “an ontology is
a formal specification of a shared
conceptualization”, where ‘formal’ refers to the need
of machine-understandable ontologies. This
definition emphasizes the need of agreement in
carrying out a conceptualization. On the other hand,
‘shared’ refers to the type of knowledge contained in
the ontologies, that is, consensual, non-private
knowledge. In this work, this definition of ontology
has been adopted.
Ontologies have become the de-facto standard
knowledge representation technology after the
emergence of the Semantic Web along with
Semantic Web Services and the Semantic Grid. For
all these new research branches, ontologies are the
cornerstone technology. Knowledge in ontologies is
mainly formalized using five kinds of components:
classes, relations, functions, axioms and instances
(Gruber, T. R., 1993). There are several formal
languages used to construct ontologies, that is,
ontology languages, including KIF, OCML and F-
Logic. Along with the Semantic Web, new markup
ontology languages have come out such as SHOE,
DAML+OIL, and the current de facto standard,
OWL (Web Ontology Working Group, 2004).
2.2 Semantic Web Services
Semantic Web Services are a new technology
resulting from the combination of other two
technologies, namely, the Semantic Web and Web
Services. On the one hand, the Semantic Web (SW)
aims at adding semantics to the data published on
the Web (i.e., establish the meaning of the data), so
that machines are able to process these data in a
similar way a human can do (Berners-Lee, T.,
Hendler, J., Lassila, O., 2001). Ontologies are the
backbone technology of the SW as they provide
structured vocabularies that describe the
relationships between different terms, allowing
computers (and humans) to interpret their meaning
flexibly yet unambiguously.
BIODEVICES 2008 - International Conference on Biomedical Electronics and Devices
96
On the other hand, Web Services (WS)
technology extends the Web from a distributed
source of information to a distributed source of
functionality. It is based on a set of standard
protocols, namely, UDDI (Universal Description,
Discovery and Integration), SOAP (Simple Object
Access Protocol), and WSDL (Web Services
Description Language). Therefore, WS provide the
means to develop globally accessible loosely-
coupled applications. However, as the Web grows in
size and diversity, there is, composition and
invocation can be carried out by autonomous
software entitiesan increased need to automate traits
of WS such as discovery, selection, composition and
execution. The problem is that current technology
around UDDI, WSDL, and SOAP provide limited
support for all that (Fensel, D. & Bussler, C., 2002).
As a consequence, the principles behind SW
technology were applied to WS leading to what we
know as Semantic Web Services (SWS) technology.
It consists of annotating WS with semantic content
so that service discovery.
The W3C is currently examining various
approaches with the purpose of reaching a standard
for the SWS technology: OWL-S (OWL Web
Ontology Language for Services) (OWL-S W3C
Submission, 2004), WSMO (Web Service Modeling
Ontology) (WSMO W3C Submission, 2005), SWSF
(Semantic Web Services Framework) (SWSF W3C
Submission, 2005), and WSDL-S (Web Service
Semantics) (WSDL-S W3C Submission, 2005). The
two most widespread approaches are OWL-S and
WSMO. OWL-S is an ontology for services that
makes it possible for agents to discover, compose,
invoke, and monitor services with a high degree of
automation. Similarly, WSMO provides a
conceptual framework for semantically describing
all relevant traits of WS in order to facilitate the
automation of discovering, combining and invoking
electronic services over the Web.
3 SOME DEFINITIONS
A trait is defined as any physical, motor or
psychomotor human characteristic capable of being
used in biometric identification.
A user is any person for the system to recognize,
and whose traits are stored somehow in the database.
A donor is every person (user or not) whose trait
is captured, voluntary or involuntary, by a sensor of
the system.
A sample is defined in (Mansfield, J. Wayman, J.L.,
2002) as a biometric measure presented by the donor
which eventually results in an image or signal.
4 IDENTITY HETEROGENEITY
PROBLEMS
A typical biometric system presents a well defined
structure (Mansfield, J. Wayman, J.L., 2002) that
includes two phases: enrolment and testing.
Enrolment faces the creation of a type of model
representing the user in a univocal way, while
testing tries to determine if the donor matches or not
the model.
This two phases share the steps related to sample
acquisition, pre–processing signal and feature
extraction. Enrolment completes its chain with a
model creation, while testing do it with a matching
step.
Figure 1: The process.
Acquisition implies that one or more sensors acquire
one or more samples of certain donor biometric
traits presented to the biometric systems (e.g.
fingerprint, face, iris image). Different kind of
sensors capturing different biometric donor traits
generate different kind of samples.
After capture, samples are pre–processed by
cleaning and normalizing in order to adapt the signal
to further data extraction, which means signal
BIOMETRIC AUTHENTICATION DEVICES AND SEMANTIC WEB SERVICES - An Approach for Multi Modal
Fusion Framework
97
filtering, enhancement, energy detection, image
centring...
The feature extraction module obtains certain
information values supposedly related to the donor
in a univocal way. These values are collected in sets
called “feature vectors”. Different feature extraction
algorithms generate different vectors types.
During enrolment a set of homogeneous vectors
are used to create a new model of the related trait of
the user. While testing the identity claim user model
is compared with another set of vectors obtained
from the current donor.
For mono–modal biometric systems only one
trait is scanned, and of course only one model is
generated for each user.
In our multi–modal biometric fusion approach,
multiple user models are generated: one for each
modality. Every kind of sensor, scanning different
traits, generates different signals. Every one of them
flows through different acquisition–preprocessing–
extraction chains. The testing phase matches the
resulting vectors of every chain with the
correspondent model, obtaining as a result a
confidence level which informs of the probability
that the sample belongs to the user identity claim.
Finally, these results are collected by a decision
module which will decide to validate or reject the
donor.
Different capture, pre-processors or feature
extractor algorithms generate different kind of
models. Then, other kind capture–process–extractor
chain generate feature vectors that should not match
properly the model. That’s why semantic
information should be added to raw data in order to
identify such a variety.
5 THE FRAMEWORK
All over the world we can find heterogeneous
databases as a set of biometric recorded data. In this
section, we present an integrated approach that
address the entire problem of enabling entities
(organizations) to confirm individual authentication,
based on biometric models (traits) stored in different
databases all over the Web. The main requirements
of our framework are as follows:
Provide a platform that allows data matching of
acquired biometric samples against individual
biometric models (traits) stored in certain databases
Provide catalogues of data that allows to
determine a given model location and make these
catalogues available via the Web.
Use of the Web Service technology to make
this access a reality and provide the plumbing
communication technology over the wire.
Use of Semantics to find the most accurate
model source for the biometric testing that is taking
place.
And finally use available Web Services in order
to ask for autentication for acquired samples.
In our framework, we have addressed this
process, taking into account the growing complexity
of having a multimodal biometrics test. For that, we
notice that our work is mostly oriented to multiple
biometric fusion strategies, where multiple biometric
measures are utilized (Kittler, J. Hatef, R. Matas, J.
G., 1998).
A graphical representation of our framework is
depicted in the figure 2.
Figure 2: The framework.
The first part of the aforementioned steps i.e. the
capture of the biometric samples and certain steps of
the signal-process is addressed by the Biometric
Data Acquisition System. The goal of this system is
to encapsulate the biometric measure with a “data
cover” that includes the type of the measure (voice,
image, fingerprint, etc.) and a number of significant
attributes that can describe the measure and the
identity claim.
The second one, the Semantic Service Engine,
takes the measure and its “data cover” and accesses
the different data storages looking for the best fitting
for the cover along the Web. A Semantic Service is
an execution environment for the Semantic Web
Services initiatives described in section 2. Along
with some of the W3C submissions, different tools
have been implemented that bring together the major
Semantic Web Services functionalities in an
integrated framework. One example is WSMX (Web
Services Execution Environment), an execution
BIODEVICES 2008 - International Conference on Biomedical Electronics and Devices
98
environment to perform dynamic discovery,
selection, mediation, invocation and inter-operation
of Semantic Web Services. Another example is IRS
(the Internet Reasoning Service), a Semantic Web
Services framework which allows applications to
semantically describe and execute Web Services.
The IRS system supports the provision of semantic
reasoning services within the context of the
Semantic Web.
In both cases, a Semantic Service Engine would
take the measure and its “data cover” to hook it up
with the Web Service that fits the most and behaves
as an entry point of the aforementioned databases, as
it is show in Figure 1.
Finally, the Decision Module component
produces a typical index called matching score. A
match or no-match decision can be made according
to whether this score exceeds a decision threshold or
not. This implies an attempt to validate or not the
claim of identity, which outcomes a final decision
about the biometric identity.
6 CONCLUSIONS
AND RELATED WORK
Since Biometric technologies are intended to tackle
security and privacy issues, the integration of access
control mechanisms and information security are
also areas of growing interest.
The creation of Biometric Accreditation Entities
will be an alternative in the near future to the current
digital certification organisms.
In this environment the heterogeneity of sample
capture and data process should not become a barrier
for the use of this identification technology.
As the use of Semantic Web Services grows, the
problem for searching, interacting and integrating
relevant services is becoming increasingly a hurdle
for the leverage of existing Semantic Web
technologies which have reached a certain level of
maturity.
In this paper, we have proposed a conceptual
approach for a effective solution in this
heterogeneous environment. It is based on the
application of the Semantic Web Services properties.
It requires to add semantic to the data stored in the
biometric accreditation entities databases, and
provide the adequate services to the SWS servers.
It has already been proposed the idea of using
agents. They can take advantage of the machine-
processable metadata provided by the Semantic Web
Services. In (Hendler, J., 2001), the author points out
how the ontology languages of the Semantic Web
can lead to more powerful agent-based approaches
for using services offered on the Web. A more
practical approach is shown in (Gandon, F. and
Sadeh, N., 2004), where the authors describe an
application where intelligent agents, aided by
context information provided by Semantic Web
Services, assist their users with different sets of
tasks.
Finally, our future work will focus on creating a
complete adapted ontology and to define a standard
for required services on SWS, identifying real–world
scenarios and validating the efficiency of our
approach and to determine its feasibility. This work
is related to existing efforts about ontology merging
and alignment. A future version of our framework
will be orientated towards that direction.
ACKNOWLEDGEMENTS
This work is founded by the Ministry of Science and
Technology of Spain under the PIBES project of the
Spanish Committee of Education & Science
(TEC2006-12365-C02-01).
REFERENCES
Bailly-Baillière E., S. Bengio et al., 2003. “The BANCA
Database and Evaluation Protocol,” in Springer
LNCS-2688, 4th Int. Conf. Audio- and Video-Based
Biometric Person Authentication, AVBPA’03. 2003,
Springer-Verlag.
Berners-Lee, T., Hendler, J., Lassila, O., 2001. The
Semantic Web. Scientific American, May 2001, pp.
34-43.
Borst, W.N., 1997. Construction of Engineering
Ontologies for Knowledge Sharing and Reuse. PhD
Thesis. University of Twente. Enschede, The
Netherlands.
Fensel, D. & Bussler, C., 2002. The Web Service
Modeling Framework WSMF. Electronic Commerce
Research and Applications, 1(2).
Gruber, T. R., 1993. A translation approach to portable
ontology specifications. Knowledge Acquisition Vol.
5:199-220.
Web Ontology Working Group, 2004. OWL Web
Ontology Language Guide.
OWL-S W3C Submission, 2004. OWL Web Ontology
Language for Services. Available at:
http://www.w3.org/Submission/2004/07/
WSMO W3C Submission, 2005. Web Service Modeling
Ontology. Available at: http://www.w3.org/
Submission/2005/06/
BIOMETRIC AUTHENTICATION DEVICES AND SEMANTIC WEB SERVICES - An Approach for Multi Modal
Fusion Framework
99
SWSF W3C Submission, 2005. Semantic Web Service
Framework. Available at: http://www.w3.org/
Submission/2005/07/
WSDL-S W3C Submission, 2005. Web Service
Semantics. Available at: http://www.w3.org/
Submission/2005/10/
Hendler, J., 2001. Agents and the Semantic Web. IEEE
Intelligent Systems, 16(2): 30-37, March/April 2001.
Gómez, J. M., Rico-Almodóvar, M., García-Sánchez, F.,
Martínez-Bejar, R. & Bussler, C., 2004. GODO: Goal-
driven Orchestration for Semantic Web Services.
WSMO Implementation Workshop, September 2004.
Jain, K. Bolle, R. et al. Biometrics, 1999: Personal
Identification in Networked Society. Kulwer
Academic. 1999.
Jain R. and J. Quian, 2001: Information Fusion in
Biometrics. Proc. 3rd International Conference on
Audio and Video Based Person Authentication
(AVBPA) pp. 354-391,Sweden, 2001.
Gibbins, N., Harris, S., Shadbolt, N., 2003. Agent-based
Semantic Web Services. In Proc. of the 12
th
Int. World
Wide Web Conference, May 2003.
Gandon, F. and Sadeh, N., 2004. Semantic Web
Technologies to Reconcile Privacy and Context
Awareness. Web Semantics Journal, 1(3), 2004.
Ming, A. Ma, H., 2007. An Algorithm Tested for the
Biometrics Grid. Proceedings of the Second
International Conference in Grid and Pervasive
Computing (GPC07). Paris, France. 2007.
Mansfield, J. Wayman, J.L., 2002. Best Practices in
Testing and Reporting Performance of Biometric
Devices. National Physics Lab for Mathematics and
Scientific Computing. 2002.
Kittler, J. Hatef, R. Matas, J. G., 1998. On Combining
Classifiers. IEEE Transactions on PAMI, vol. 12
(1998). Pp. 226-339.
BIODEVICES 2008 - International Conference on Biomedical Electronics and Devices
100