DEVELOPMENT OF CONTEXT-AWARE APPLICATIONS IN
UBIQUITOUS INFORMATION SYSTEMS
Benselim Mohamed Salah
LabSTIC Laboratory, University of 08 Mai 45, Guelma, Algeria
Seridi-Bouchelaghem Hassina
LabGed Laboratory, University of Badji Mokhtar, Annaba, Algeria
Keywords: Ubiquitous computing, Context of use, Concerns separation, MDA, Context modeling, UML.
Abstract: Nowadays, software engineering is moving towards the development of ubiquitous and distributed
applications. This tendency is constrained by parameters such as mobility and heterogeneity that
characterize the current situation of a user. Each new application will be able to adapt its services with the
change of context of use and satisfy all user's preferences. The aim of this paper is to propose a new
development approach that can take into account the change in context of use during the application
development process. It permits us to develop contextual aspects of a system in a separate way and
independently from the business aspects of this system and from the technological constraints of the chosen
platform. Our proposal, based on the principles of MDA (Model Driven Architecture), is defined by three
steps. First, the separation of contextual aspects by introducing the 3TUP process (3 Track Unified Process)
and the development process as PSI "Ψ". Second, the context modeling using UML (Unified Modeling
Language) and conform to a proposed context metamodel. And thirdly, the integration of the contextual
model in MDA process using model merging operation.
1 INTRODUCTION
In a pervasive (or ubiquitous) environment, the great
challenge of designers will be how to take into
account all continuous changes of user's current
situation. So, ubiquitous information systems are
influenced by several factors and features like
mobility, heterogenity and distribution. Thus, new
application will be able to adapt its services with the
change of context of use and satisfy all user's
preferences. Our study aims to provide an approach
for developing context-aware applications that can
guarantee the contextualization of future
applications. Our proposal uses the principles of
MDA (OMG, 2003), and it is based on the concept
of aspects separation and particularly the separation
of contextual constraints. This separation allows us
to study all contextual information in an independent
way and without remaking the entire development
process because of any changes in the context of
use. Then we will use a graphical approach to
represent the elements of context in order to obtain a
contextual model. This one will be built according to
a proposed contextual metamodel. The contextual
model will be integrated into the process of MDA by
model merging operation. Merging is to combine
business model elements with those of the
contextual model to form a merged model which
will be transformed into a specific model according
to the specifications of a chosen platform.
2 RELATED WORKS
Several studies and researchs have been made in the
field of context and especially for taking into
account of contextual changes that surround the
execution of an application and introducing user's
desired preferences. There are few works that are
oriented towards using the MDA approach to design
adapted applications and obtain personnalized
information. Vale and Hammoudi showed in (Vale,
2008) the importance of MDA in context modeling
and context-aware applications development, and for
223
Mohamed Salah B. and Hassina S..
DEVELOPMENT OF CONTEXT-AWARE APPLICATIONS IN UBIQUITOUS INFORMATION SYSTEMS.
DOI: 10.5220/0003488802230228
In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 223-228
ISBN: 978-989-8425-55-3
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

that they designed the CSOA (Context-aware
Service Oriented architecture) based on ECA
viewpoints (Enterprise Collaboration Architecture).
In the study (Ou, 2006), the authors have designed a
contextual model using ontologies in order to
represent context information, then they proposed a
new architecture MDIA (Model Driven Integration
Architecture) dedicated to the implementation of
context-aware applications. Dey present in (Dey,
2001) a toolbox named "context toolkit" that can
serve as support to adapt applications to context.
Another study presented by Chen (Chen, 2000)
consists to capture various contextual information
that will be treated and modeled (location model,
data structures). The project PUMAS (Peer
Ubiquitous Multi-Agent System) is a framework
based on agents. PUMAS selects the desired
information from one or more sources, and adapts it
with user's characteristics and with the mobile
device features (Chaari, 2005). The project SECAS
(Simple Environment for Context Aware Systems) is
interested in adapting applications to use context
(user's preferences, environment, device used, ...).
SECAS aims to make adapted applications to
different context of use on three components: data,
services and presentation. To ensure such an
adaptation, this project uses Web services (Chaari,
2005).
Our study aims to demonstrate the importance of
separation between three aspects, namely: business
aspects, contextual aspects and technical aspects.
Our goal is to develop and design a new technique
(extended MDA) by isolation of contextual
constraints in a separate branch. This technique is
defined by three steps: contextual aspects separation,
context modeling and integration of the contextual
model.
3 EXTENDED MDA
Future applications have to provide exact results and
personnalized information and they must be
adequate with every variation of user's current
situation. Ubiquitous information systems are
characterized by many features like mobility,
heterogenity and distribution. These constraints
represent the context of user's current situation and
are not introduced explicitly in MDA life cycle.
MDA is based on models and model transformation
and its advantage is the concerns separation (only
two branches : business and technical). But this
approach does not take into account, in clear way,
the continuous changes and variation of context
especially with modern information systems
operating in pervasive environment.
Figure 1: Architecture of extended MDA.
To extend this limit, we propose two solutions
that can introduce contextual information in MDA
life cycle. This information can be represented as
rules or models as the preferred option. So, the first
solution consists in creating "contextual rules" that
represent the contextual elements of the current
situation of a user and then to introduce these
"contextual rules" in a mapping of the PIM-PSM
transformation. The second solution uses a
"contextual model" that represents all contextual
aspects of a system and which will be integrated in
the process of MDA using a model merging
operation. Among these proposed solutions, we
choose the second one to develop a new vision of
MDA architecture. This new approach is defined by
three steps: separation of contextual aspects, context
modeling and integration of the contextual model in
MDA process.
Figure 1 presents an extension of MDA
approach. This new vision aims to distinguish
between the contextual constraints and the two types
of classical constraints (business and technical) in
MDA. Context separation allows us to study all the
environmental changes and user's preferences in an
Contextual
aspects
PDM
CM
MM
PSM
Mer
g
in
g
Transformation
Context
modeling
PIM
Business
aspects
Aspects separation
Technical
aspects
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
224

independent single branch. This will minimize the
risk of remaking all the application development
process because of these changes. Once the
contextual aspects are isolated and separated, we
proceed to model them in order to build a contextual
model (CM) based on separated contextual
information. The contextual model thus obtained is
combined with the business model (PIM: Platform
Independant Model). So, we obtain a merged model
(MM). This one will be transformed into a specific
model (PSM: Platform Specific Model) using the
specifications of a chosen platform stocked in the
Platform Description Model (PDM).
3.1 Contextual Aspects Separation
Why we separate the contextual aspects? (Figure 2)
Indeed to answer this question, we must demonstrate
that each of the contextual elements can not be
handled by the business branch neither by the
technical branch. For example, in ubiquitous
computing, time and space are very important and
they cover the concept of ubiquitous information
systems (anytime, anywhere).
Also, any changes of
the surrounding objects can directly affect the state
of the current situation of a user. On the other hand,
each user may have particular preferences related to
content, presentation or information display.
The concept of "context" has been defined in
(Dey, 2001) as follow: “Context is any information
that can be used to characterize the situation of an
entity. An entity is a person, place, or object that is
considered relevant to the interaction between a
user and an application, including the user and
applications themselves”.
From this definition,
we can extract the basic
elements that constitute the main components of
context of use. These elements are: the user,
location, time and environment. In the field of
ubiquitous computing, these elements have a
common feature such as change and variation.
Ubiquitous computing offers many opportunities
for access to information like random use, in time
and space, of mobile devices by varying kinds of
users including distant persons and nomads. Because
of user mobility, not only the time and location are
constantly changing but also the surrounding objects
(including nearby people) and environmental
factors. So, the user is facing a different situation
that he is in a new context of use and with which he
must be adapted. Modification or change in one
element of the context of use involves the transition
of this context toward another state of the user's
current situation. Because of these changes, we
must not remake the whole application development
process, but only the part that deals with such
constraints.
This brings us to distribute all constraints of an
ubiquitous information system into three distinct
types: business, contextual and technical. Hence the
name "3-Track" which means "three ways" or "three
paths" to introduce the new concept of: 3TUP
process (3 Track Unified Process).
This separation permits us to develop contextual
aspects of a system in a separate way and
independently firstly, from the business aspects of
this system and secondly, from the technological
constraints of the chosen platform. Hence the need
to consolidate all the contextual specifications that
represent the context of use in a new independent
branch named "contextual branch". It is to say that
the development process can be done on three
independent ways. And here we introduce a new
concept of development process as PSI "Ψ" (Figure
3). Any application development by using PSI
process is based on four branches (three parallel
branches and one common branch).
Figure 2: Contextual aspects separation.
DEVELOPMENT OF CONTEXT-AWARE APPLICATIONS IN UBIQUITOUS INFORMATION SYSTEMS
225

Figure 3: Development process as PSI "Ψ".
The three parallel branches are:
- Business Branch: regroups all basic
functionalities of the system,
- Contextual Branch: studies and represents only
the contextual constraints,
- Technical Branch: describes the specific chosen
platform.
All these branches nearly have the same steps
(Requirements, Specifications, Analysis) but
according to each kind of constraints. They are
independent and they can be done simultaneously.
The common branch corresponds to the phase of
realization. This phase begins as soon as the
previous three branches are complete and it consists
on several steps like: Design, implementation,
coding and maintenance.
3.2 Context Modeling
Context modeling approaches are as follows (Strang,
2004) : KeyValue Models, Markup Scheme Models,
Graphical Models, Object Oriented Models, Logic
Based Models and Ontology Based Models. These
approaches are classified by the scheme of data
structures which are used to exchange contextual
information in the respective system.
Before building the contextual model we are
going to present our context metamodel (Figure 4).
Figure 4: The context metamodel.
Our context metamodel is made by using UML
(Unified Modeling Language) notation. The
proposed metamodel allows designers to produce
more consistent and more homogeneous contextual
models. This metamodel is described as follows:
The context of use "CONTEXT"of a user (or a
situation) can be composed by several contextual
elements "ContextElement". Each of these elements
is defined by: the type, the attributes, the constraints,
the relevance and the source (sensor).
The type of a contextual element "ContextType"
is provided by one of the following entities: the user,
the application, the environment or the behavior. So
each contextual element has one of the following
types:
- "UserContext" : when information is related to the
user,
- "ApplicationContext" : if information comes from
the application means,
- "EnvironmentContext" : if information is provided
by the surrounding objects (other than the user and
the application),
- "BehaviorContext" : its provided by all behavioral
iteractions between the three entities (user,
application, environment).
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
226

Figure 5: The contextual model.
The attributes "ContextAttribute" permit to
represent the properties of a contextual element class
and define all values that can be attached to
instances of this class or an interface.
Constraints "ContextConstraint" are defined by a
restriction or condition on an element class and they
are used to model complex and important
information such as domain, table …
The source of the contextual element
"ContextSource" allows to represent the specific
hardware or software used for contextual
information sensing. "ContextSource" assures the
processing of all information captured by the
"Sensor".
"ContextRelevancy" indicates the relevance
quantification of an element. It represents the level
of importance that each contextual element may
have.
Our contextual model (Figure 5) is built using a
graphical approach based on UML and it is conform
to the context metamodel. This one contains main
elements representing the context of use, namely:
location, time, device, user's preferences, and
surrounding objects.
The current situation of a user may be influenced
by several factors and constraints as follows:
- Constraints related to the user himself (Identity,
Profile, Behavior, Preferences).
- Constraints related to the application (Software,
Hardware, Networks).
- Constraints related to the environment (time,
Location, weather, nearby persons, surrounding
objects and available resources).
3.3 Integration of the Contextual
Model
The integration process of the contextual model in
MDA approach is essentially based on merging
models operation. This operation combines the
elements of business model (PIM) with those of
contextual model (CM). Merging operation can be
decomposed into four distinct phases: comparison,
conformance checking, merging and reconciliation
(or restructuring) (Kolovos, 2006). The result of this
operation will be a merged model (MM) which will
be transformed to a specific model (PSM) according
the technical specifications of a chosen platform.
The integration process has five phases: comparison,
conformance checking, merging, restructuring and
transformation.
In comparison phase all correspondences
between equivalent elements of the source models
(PIM+CM) are identified. It is a phase of "matching"
elements in pairs. In EML (Epsilon Merging
Language) (Kolovos, 2008), match-rules are used in
order to compare elemets and to decide if they match
or no. The decision has a boolean type and it will be
used in merging phase. This will eliminate duplicate
elements in the merged model.
In conformance checking phase all identified
elements during the comparison phase (matching)
are tested and verified. They are examined for
conformance with each other. This phase is also
performed using match-rules that decide whether
compared elements are conform or not.
The merging phase is the phase of union and
combination of source elements for creating target
elements according to the two decisions (match and
conform) from previous phases. Here, there are two
possible approaches, and this, according to the result
of matching. Indeed, if the compared elements are
"full correspondent" then they are unified as a single
element in the merged model by using merge-rules.
And if these elements are considered "not
correspondent" then they are propaged distincly in
the merged model by using transform-rules.
The target model, obtained after the merging
phase, may contain inconsistencies or anomalies.
These ones must be cleaned or fixed in the
restructuring phase to make a valid model.
The last step in the integration process of
contextual model is to transform the merged model
to a specific model according to the technical
specifications of the chosen platform.
DEVELOPMENT OF CONTEXT-AWARE APPLICATIONS IN UBIQUITOUS INFORMATION SYSTEMS
227

4 CONCLUSIONS
In this study, we presented a new vision of MDA
approach which is able to introduce the context of
use in its life cycle and, consequently, produce
context-aware applications in pervasive environment
characterized by ubiquitous information systems.
Our proposal goes through three steps. The first is to
separate contextual aspects of a system from other
ones (business and technical). This separation is
based on a new introduced concept, namely the
3TUP process. The second step aims to represent
contextual information as a model. To construct the
contextual model, we used a graphical approach
based on UML and we proposed a context
metamodel which can be used by other researchers.
The final step proposes a method to integrate the
contextual model in the development process of
MDA approach, and this, with merging models. At
the end we have introduced a new idea for creating
an extended UML or an UML profile especially for
the domain of context of use.
As perspective we try to present an UML
extension for representing the context of use. An
extension of UML can be described by an
extensisbility mechanism that should contain:
Stereotypes, tagged values, constraints and interface.
Our future goal consists on creating a stereotype for
representing the relevancy of a contextual element.
This stereotype will be described by several tagged
values (destined for the measurement of relevancy)
and some constraints (destined to specify the
conditions of using). This extension is based on
UML notation and it permits obtaining a specific
graphic representation of a contextual situation. We
have to extend UML so that it permits us to
represent all of the conceptual elements that can
influence the current situation of a user. Each
contextual element should be able to be represented
by thus extension of UML. As perspectives, we
expect to continue this study by constructing a
contextual mapping. This will include new
transformation rules to describe the context of use
and that will guide the transition from PIM
(including context) to PSM. Our vision extends to
the implementation of information systems that will
be "contextually parameterized ". Applications must
include variables representing all possible contextual
information that will be introduced by constraints
and profiles of user's current situation.
REFERENCES
Chaari, T., Laforest, F., Flory, A., 2005. «Adaptation des
applications au contexte en utilisant les services web».
Proceedings of the 2nd French-speaking conference
on mobility and ubiquity computing, ACM
International Conference Proceeding Series, Vol. 120
pp 111-118, Grenoble, France.
Chen, G., Kotz, D., 2000. «A survey of context-aware
mobile computing research». Dartmouth Computer
Science. Technical Report TR2000-381.
Dey, A.K., Abowd, G.D., Salber, D., 2001. «A conceptual
framework and a toolkit for supporting the rapid
prototyping of context-aware applications». Human-
Computer Interaction vol.16, 2001, pp 97-166.
Kolovos, D.S, Paige, R.F., Polack, F.A.C., 2006.
«Merging Models with the Epsilon Merging Language
(EML) ». In Proc. ACM/IEEE 9th International
Conference on MDE Languages and Systems
(Models/UML 2006), Italy.
Kolovos, D.S, Paig,e R.F., Rose, L.M., Polack, F.A.C.,
2008. «Epsilon». Epsilon Book From Department of
Computer Science, University of York, UK.
OMG., 2003. «MDA Guide V1.0.1». www.omg.org
OMG ., 2007. «Unified Modeling Language (OMG
UML), Superstructure, V2.1.2».
www.omg.org/spec/UML/2.1.2/Superstructure/PDF
Ou, S., Georgalas, N., Azmoodeh, M., Yang, K., Sun, X.,
2006. «A Model Driven Integration Architecture for
Ontology-Based Context Modelling and CAA
Development» . Rensink and Warmer (Eds.):
ECMDA-FA 2006, LNCS 4066, pp. 188 – 197.
Strang, T., Linnhoff-Popien, C., 2004. «A context
modelling survey». The 6th international conference
on ubiquitous computing, workshop on advanced
context modeling, reasoning and management as part
of ubicomp 2004, , pp 33-40.
Vale, S., Hammoudi, S., 2008. «Towards Context
Independence in Distributed context-aware
applications by the Model Driven Approach». ACM
SIPE'08, Sorrento, Italy.
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
228