INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS
Milene Serrano and Carlos José Pereira de Lucena
Department of Informatics, Pontifical Catholic University of Rio de Janeiro (PUC-Rio)
Rua Marquês de São Vicente 225, Ed. Padre Leonel Franca 13˚ andar, Gávea, Rio de Janeiro - RJ, Brazil
Keywords: Ubiquitous computing, Mobility, Intentional multi-agent systems, Systematic development, Service
omnipresence, Device heterogeneity, User satisfaction, Adaptability, Context-awareness.
Abstract: Being everywhere, going anywhere and accessing at any time. Ubiquitous computing is the paradigm of
service omnipresence, device heterogeneity, calm technology application and user satisfaction. Therefore,
the success of ubiquitous systems depends on the mobile computing nature. In this paper, we introduce the
application of intentional mobile agents in the systematic development of ubiquitous systems. These agents
are commonly used to perform specific activities in dedicated servers, such as the content adaptability based
on the ubiquitous profiles information. It demands context-awareness, which can be improved by exploring
critical interactions among mobility, smart-spaces and cognitive-based autonomous entities. Finally, we
show how our proposal has been appropriately applied to a ubiquitous system from the e-commerce domain.
1 INTRODUCTION
Ubiquitous computing designs a world in which the
heterogeneous devices are available throughout
different physical environments and the users can
enjoy network services whenever and wherever they
want (e.g. home, office, store and school). However,
the efforts to establish the connection with these
devices, the integration of different physical
environments, the content adaptability and the
context-awareness to better attend the users must be
effectively imperceptible to them. Thus, the
ubiquitous systems pose some challenges that
demand an adequate technological support to
improve the software engineers’ work in the
systematic development of this kind of system. In
this context, the success of ubiquitous systems
depends on the mobile computing nature. Mobile
computing is involved with the mobility issue by
investigating and developing computational
resources to attend it, improve it and manage it.
Autonomous mobile entities can be useful in dealing
with this field by migrating from one smart-space to
another or from one server to another to perform
specific activities (e.g. content adaptability)
according to the ubiquitous context.
In order to contribute to the systematic
development of ubiquitous systems, we suggest the
application of intentional agents by exploring critical
interactions among mobility, smart-spaces and these
collaboration- and cognition-based autonomous
entities as an intentional multi-agent-system.
Multi-agent systems (MAS) focus on the use of
many intelligent agents that interact with each other
in order to achieve specific goals. The interactions
can be either collaborative or selfish. The agent-
based approaches have generated lots of excitement
in recent years, especially in complex contexts, as
they have interesting properties – e.g. autonomy,
flexibility, reactivity and proactivity – and they hold
out promise to be the paradigm for conceptualizing,
designing and implementing the next generation of
software systems. Normally, multi-agent-systems
are based on behavior or intentionality. According to
our experimental research (Serrano et al. 2008)
(Serrano et al. 2009) (Serrano and Lucena 2010a,)
the intentionality abstraction improves the cognition
capacity, the “like-me” recognition (Gordon 2005)
and the goal formation (Dignum and Conte 1997) by
dealing with the user’s wills centered on her/his
beliefs, desires and intentions – the BDI model
(Pokahr et al. 2005) (Georgeff et al. 1998).
Furthermore, intentional agents are capable of
achieving explicit ascription of human mental states
and modeling human practical reasoning.
Ubiquitous computing is the paradigm of service
omnipresence, device heterogeneity, calm
technology application and user satisfaction. In
addition, the success of the ubiquitous systems
depends on the mobile computing nature. Therefore,
147
Serrano M. and Pereira de Lucena C..
INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS.
DOI: 10.5220/0003111201470156
In Proceedings of the 3rd International Conference on Agents and Artificial Intelligence (ICAART-2011), pages 147-156
ISBN: 978-989-8425-41-6
Copyright
c
 2011 SCITEPRESS (Science and Technology Publications, Lda.)

we combine intentional multi-agent systems with the
mobility issue in ever-changing environments.
Furthermore, we also provide a reuse-oriented
building block that supports the systematic
development of ubiquitous systems – independent of
their cognitive domains – by applying intentional
mobile agents. The support was developed as an API
driven by specific protocols in order to improve the
inter-operability and the communication among the
agents. Finally, we show how our intentional-
mobile-agent-based proposal has been appropriately
applied to a ubiquitous system from the e-commerce
domain and compare this support to related work.
The remainder of this paper is organized in
sections. Section 2 discusses our proposal. Section 3
presents the application of our proposal to a
ubiquitous system from the e-commerce domain as
well as the evaluation process performed to verify
and validate our proposal. Section 4 compares our
proposal to related work. Finally, Section 5
concludes and presents future research.
2 OUR PROPOSAL
We conducted our experimental research –
performed in the Software Engineering laboratories
at the Pontifical Catholic University of Rio de
Janeiro (PUC-Rio) and the University of Toronto
(UofT) over the past three and a half years – by
following a process (Figure 1), which is composed
of the activities presented as follows:
 Investigate (1): it represents the investigation of
the ubiquitous computing state-of-the-art to identify
and define the main ubiquitous concerns as well as
the investigation of traditional and emergent
technologies to deal with these concerns.
 Develop (2): it represents the development of
intentional-MAS-driven support as reuse-oriented
building blocks to guide the systematic development
of ubiquitous systems. Among other contributions,
we offer building blocks to deal with the content
adaptability (Serrano et al. 2008); the dynamic data
storage, retrieving and update (Serrano and Lucena
2010b); the non-functional requirements elicitation,
modeling, and operationalization (Serrano et al.
2010); the data sharing, manipulation and access
based on the user’s privacy policies and preferences,
the device features, the network specifications and
the contract information and supported by specific
capabilities and ubiquitous profiles (Serrano and
Lucena 2010b); and mobility by using intentional
agents contemplated with a specific capability – the
ability of migrating from one agent’s platform
container to another. In this last building block, the
containers represent different smart-spaces, which
can contain different servers/providers, services,
contents, resources, devices and collaborative
software agents.
 Apply (3): it represents the application of our
building blocks support to the systematic
development of ubiquitous systems in different
cognitive domains.
 Evaluate (4): it represents the evaluation of our
building blocks based on the results acquired by
performing tests centered on the user’s satisfaction
and other ubiquitous issues.
 Evolve (5): it represents the evolution of the
applied building blocks – generating their evolved
versions – according to the evaluation’s results.
The focus of this paper is on the description of the
building block to deal with the mobility issue.
2.1 Mobility Building Block
The proposed building block approach is based on
the JADEX framework (Braubach et al. 2004) and
the JADE-LEAP platform (Caire 2003). The
framework offers specific protocols and resources to
develop BDI-model-driven mobile agents, which
represent autonomous entities improved by the
intentionality and the mobility properties. The
platform provides two specific execution modes
(split and standalone) to support the integration of
heterogeneous devices (e.g. MIDP and Personal Java
devices) in the agents’ platform. In order to deal
with MIDP devices, in which the memory and the
processing capacities are limited, the split execution
mode allows sharing resources with another
powerful machine that is connected to the platform
using the network (wire or wireless). The platform
and its cognitive agents automatically control the
sharing of resources, which means this is
imperceptible to the users of limited hand-held
devices. Moreover, in order to deal with powerful
devices (e.g. Personal Java devices), which have
capacity to run the platform’s container, the
standalone execution mode is required.
In our approach, following the JADEX
resources, the migration feature, in which a
platform’s agent can migrate between hosts and the
agent’s state is persisted until there is a desire to
restore it, is based on the Java’s serialization
mechanism. Thus, the serialization of the agents is
supported at runtime, wherever and whenever it is
necessary.
ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence
148

Figure 1: Research evolution process – SADT Notation (Marca and McGowan 1987).
Our main intention is to guide the software
engineers in the systematic development of
ubiquitous systems. Therefore, we developed a
reuse-oriented support set to facilitate the
implementation of intentional mobile agents. We use
the concept of capability. Basically, a capability is
the quality of being able to perform specific
activities or services to achieve specific purposes. In
this context, a mobile agent represents an agent with
a special ability. It has the potential for moving from
one smart-space to another and/or from one server to
another, by maintaining its state. In our platform,
each agent has at least one capability – called “root
capability.” The root capability is given by the
beliefs, desires and intentions. In the JADEX, the
beliefs represent the agent’s notion about the real
world; the desires represent the agent’s goals, which
are based on third parties’ goals (e.g. user’s goals
and organization’s goals); the intentions – called
plans – represent sequences of tasks to achieve the
goals. Beliefs, goals and plans are specified in the
agent’s XML file. To create additional capabilities
for reuse purpose in different agents, we provide
different definition files – one definition file for each
created capability. The focus of this paper is on the
mobility capability.
The intentional agent with the proposed mobility
capability is used in our experimental ubiquitous
environments to perform specific tasks in dedicated
servers and to migrate from a smart-space to another
by following the user’s mobility. Figure 2 illustrates
a typical intentional mobile agent way-of-working to
deal with the content adaptability issue.
The adaptation necessity is determined based on
the context, which involves the user’s preferences
and privacy policies, the features of the user’s hand-
held device, the network connection specifications
and the contract rules established between the user
and the service provider. There is an interface agent
that runs inside the user’s device. This agent is
“light,” based on behavior, to avoid problems with
limited devices. The interface agent is responsible to
intermediate the communication between the user
and the agents’ platform, in which the services and
contents are offered. At the moment that the user
requests a service or a content, the interface agent
integrates the user’s device into the platform. There
are two ways to perform this integration. On one
hand, if the device is limited, the integration is
possible by sharing resources with a powerful
machine – connected into the platform – to run the
container using the split execution mode. On the
INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS
149

other hand, if the device is powerful, the integration
is possible by running the container using the
standalone execution mode. When the device is
integrated – it is normally instantaneous – the
interface agent requests the creation of an intentional
agent that represents the user in the domain, called
domain personal agent. The creation depends on the
AMS agent that represents an intentional agent with
the capability of creating new agents into the
platform. The interface agent delegates the user’s
goals to the domain personal agent. The latter agent
tries to achieve the user’s goals centered on the
ubiquitous context using reasoning and learning
techniques (e.g. fuzzy-logic-based techniques),
ontologies and other technological resources.
Only to illustrate, we can consider that the user
wants to access a video file using her/his device.
Analyzing the context, the domain personal agent
observes that the device has specific features (e.g.
screen resolution and specific file formats for video
visualization). Thus, this agent requests the
adaptation of the content for the adaptation manager
agent. There is one adaptation manager agent in each
container. This agent requests the creation of a
specific agent to execute the adaptation in a
dedicated server. It avoids overloading the
adaptation manager agent. Again, the creation
depends on the AMS agent. As the requested agent
must move from one server to another and/or from
one container to another, it is created with a specific
capability – the mobility capability. The created
intentional mobile agent migrates to the dedicated
server to perform specific tasks in collaboration with
the agents that are specialized in the adaptability
issue – in other words, agents that reason mainly
centered on adaptability techniques. The content is
adapted and the domain personal agent receives it.
This agent sends the adapted content to the interface
agent that performs the content visualization to the
user using her/his device. Context awareness,
adaptation techniques and ubiquitous profiles
investigation are used to better satisfy the user’s
necessities. The entire process execution is
imperceptible to the end user, by respecting the calm
technology principle, in which Mark Weiser (Weiser
and Brown 1995) argues that it is necessary to
integrate the users in different smart-spaces, offer
services anywhere at anytime without disturbing or
even distracting her/him.
Furthermore, the infrastructure required to
perform the presented process (e.g. the resources,
the agents, the capabilities, the ontologies, the
ubiquitous profile persistent model and the
protocols) is provided by our mobility building
block as an API. This API is a jar file that can be
incorporated into the project of the ubiquitous
system-to-be and extended to better attend its goals.
The proposed reuse-oriented support is independent
of the cognitive domain. The agents use the FIPA
Coder and Decoder for SL Language (Bellifemine et
al. 2007) based on specific ontologies (Serrano and
Lucena 2010a). Thus, the offered infrastructure
contemplated the systematic development of the
ubiquitous system-to-be with a certain degree of
commonality.
3 APPLICATION&EVALUATION
In order to evaluate our reuse-oriented building
block approach, we applied it to the development of
Figure 2: Intentional mobile agent way-of-working.
ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence
150

an intentional-MAS-driven ubiquitous system in the
e-commerce domain. As follows, we describe the
case study and present the reuse and the evaluation
of our building block approach.
3.1 e-Commerce Case Study
The case study involves two Ph.D. students,
heterogeneous devices, different clients and various
service/content providers. The main goal is to offer
services and contents anywhere at any time, by
adapting them according to the context and using
calm technologies. There are various challenges
posed by this case study, such as: context awareness,
content adaptability, distributed environments,
specific ubiquitous profiles need, device
heterogeneity, different levels of user satisfaction,
process complexity invisibility, like-me recognition,
data storage/retrieving/update at runtime and
mobility. We applied different building blocks to
deal with all these concerns. Moreover, the services
and contents are distributed and can be offered by
different servers. These servers can be located into
the main container – the same container that runs the
ubiquitous systems – or into a dedicated container.
The communication among these containers is
supported by intentional mobile agents and the
network connection (wire or wireless). The main
offered contents are media contents (e.g. music,
video, text and image) and file contents (e.g.
executable files, docs and presentations). The
security, the media/file size and the price are some
quality criteria used by the agents to evaluate the
contents and choose the best one to attend the user’s
need. Most of the tasks involved in this ubiquitous
scenario – e.g. content adaptability, agents’ creation,
context analysis, user-environment integration,
content evaluation – are performed at runtime. Thus,
we use intentional agents to cooperatively combine
their capabilities in order to quickly and
appropriately achieve a number of goals, which are
imposed in this e-commerce case study. In terms of
mobility, we solve the problem by instantiating our
intentional-mobile-agents-based building block.
3.2 Our Approach for Reuse
We instantiated the offered APIs to facilitate and
improve the systematic development of the e-
commerce ubiquitous system. Thus, our e-commerce
ubiquitous system was contemplated with specific
capabilities, agents, ontologies, protocols and other
support in different levels – e.g. interface, domain
and persistence. Concentrating our efforts in the
reuse of the building block for mobility issue, Figure
3 illustrates the application of intentional mobile
agents to perform tasks related to the adaptability,
the service omnipresence, the context awareness and
the user-environment integration issues in ever-
changing smart-spaces. First, we integrate the user’s
device to the agents’ platform, more specifically to
the main container, in which the e-commerce
application is running. The integration is based on
the split and the standalone executions modes - see
Figure 3 (Part A). Connected to the platform, the
interface agent interacts with other platform agents
(e.g. AMS agent and domain agent) and also can
access different services (e.g. yellow pages, white
pages, download service and print service). Based on
the user’s request (e.g. video download and file
print), the interface agent requests the creation of a
specific agent to represent the user in the e-
commerce domain. The latter agent’s goals are:
“video must be downloaded” and “file must be
printed.” Thus, it consults the appropriate
knowledge base that contains the agent’s beliefs and
it also analyzes the ubiquitous profiles, which are
stored in a dynamic database. The beliefs represent
what the domain agent knows about the real world.
The ubiquitous profiles contain the user’s profile,
the device profile, the network profile, the contract
profile, the content profile, the service profile and
the smart-space profile. The agent’s knowledge base
and the ubiquitous profiles combined with a fuzzy
logic library, which is centered on the security, price
and download time quality criteria, are some
mechanisms used to provide context-aware services
and contents to the final users in our e-commerce
case. The domain agent uses these mechanisms to
make decisions, personalize the requested service
and/or the content and to better attend the user’s
goals at runtime. It is also important to say that these
mechanisms have been developing based on the
users’ satisfaction quality criterion, by respecting,
for example, the users’ preferences and privacy
policies. If it is necessary to adapt the content, the
domain agent requests the adaptation to the
adaptation manager agent. The manager delegates
this goal to a specific agent – a mobile agent – which
is created at runtime by using the white pages
services (AMS). The mobile agent migrates to a
dedicated server – adaptation dedicated server. It
collaborates with the adaptation agent to achieve the
delegated goal. At that moment, specific adaptation
techniques are applied. In our case study, these
techniques are categorized as:
 adaptation based on resizing, in which the
content is adapted according to the device screen
INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS
151

resolution;
 adaptation based on transcoding, in which the
content is modified from one format to another;
 adaptation based on reduction, in which the
content is adapted by using data compression;
 adaptation based on replacement, in which a
sequence with still frames composes a slide show;
 adaptation based on integration, in which the
content is adapted by using service composition. For
example, a video can be obtained by combining
different image frames with the corresponding
audio.
The content is adapted and the mobile agent
returns to the container in which the adaptation
request was performed. The domain agent receives
the adapted content – in the e-commerce domain, the
adapted video – and sends it to the final user in
her/his hand-held device.
In the print service use, the mobile agent’s
creation is requested by the domain agent and the
AMS agent performs the request. A mobile agent
was necessary when the desired service – e.g. print
service - is offered in a server that is located in
another container of the platform, different from the
container from which the user requested the service.
The mobile agent migrates to the service’s container
and cooperates with other agents placed in this
container to achieve its goals – “file must be
printed.” In this case, as the final user must
physically access the printed file, the service’s
container is determined based on the physical
location of both containers – the user’s container and
the service’s container – by using a GPS. If it is not
possible to perform the service because of a physical
limitation, the domain agent informs the user about
the problem and offers some alternatives: (i) the user
can move herself/himself to another smart-space –
offered/suggested by the agent based on the user’s
location – that offers the service; and (ii) the user
can abort the goal. The latter alternative is not
recommended as it negatively contributes to the
user’s satisfaction. However, it can be the only
possible alternative in critical situations. Every
decision depends on the context, whose analysis is
performed at runtime and adequately supported by
context-aware and location-aware technological sets.
As the reuse-based process involves different
Figure 3: Intentional mobile agent – integration, adaptation and service omnipresence.
ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence
152

support sets to deal with specific ubiquitous
concerns, Figure 3 (Part B) just shows part of this
process. Other important details are presented as
observations: (i) the agents’ communication is based
on specific protocols (e.g. FIPA Request Protocol
(FIPA 2002a)), languages (e.g. FIPA-ACL (FIPA
2002b) and FIPA SL Content Language (FIPA
2002c)) and ontologies (Serrano and Lucena 2010a).
The protocols, the languages and the ontologies
define a set of allowed speech acts and their
associated semantics, by improving the agents’ inter-
operability and facilitating the support’s reuse; (ii)
the data storage, retrieving and update are supported
by our dynamic database infrastructure (Serrano and
Lucena 2010b); (iii) the content adaptability is
guided by our intentional-MAS-driven framework
for content adaptation (Serrano et al. 2008); (iv) the
agent deals with the human-mental states centered
on the BDI model, reasoning techniques and
learning techniques. In this context, we use a fuzzy
logic library (Bigus and Bigus 2001,) which is
instantiated to incorporate different quality criteria
according to the cognitive domain of the ubiquitous-
system-to-be; and (v) we use a unique identifier to
represent each agent at the platform. Moreover, to
avoid problems with the agent’s identification and
the desired service’s analysis, all the agents and the
services are registered into the yellow pages of the
platform (or Directory Facilitator DF (Pokahr et al.
2005) (Braubach et al. 2004)) when they are created
and they are deregistered at the end of their “life.”
The agent’s identifier as well as its registration can
be accessed – considering the determined privacy
policies – by all containers connected to the
platform. Thus, all the servers, providers, services,
contents and devices mentioned in the e-commerce
domain are connected to the platform and have a
virtual location based on their container. It is
important to emphasize that if it is necessary to
obtain the real physical location, we also use
different location-aware technological support, such
as the GPS.
3.3 Evaluating our Approach
In order to evaluate the adequacy of the proposed
support, we performed different tests, such as:
agents’ spent time to achieve the adaptability-related
goals from the user’s request to the solution; the
AMS response time to create a new mobile agent at
runtime; and the user’s satisfaction in relation of the
received services and contents, which were
respectively performed and adapted by intentional
agents (both mobile and not mobile). In order to
facilitate the reader’s evaluation, the tests were
basically performed using heterogeneous devices –
memory and processing capacities; three notebooks
– each of them as the main machine of the container
in which they were virtually located; several content
servers and service providers – distributed in the
network; and the network wireless connection. More
details are presented in Table 1.
Table 1: Performed tests’ conditions.
Main Devices
Basic Specification
Notebook 1
2.53GHz Intel® Core 2 Duo P9500 Processor
3GB of RAM
320GB Serial ATA Hard Drive
Windows Vista Business
Notebook 2
1.83GHz Intel® Centrino™ Duo T2400
Processor
2GB of RAM
120GB Hard Drive
Windows XP
Notebook 3
1.6GHz Intel® Centrino™ M Processor 730
1GB of RAM
100GB 4200 Hard Drive
Windows XP
Network
Basic Specification
Skyline 1
Skyline 2
Skyline 3
Wireless Connectivity
54mbps of Speed
Heterogeneous Device
Some Examples
Mobile and Fixed
Limited and Powerful
Devices
Simple Cell-phone
Smartphone
Palm
Notebook
Desktop
Figure 4 illustrates the agents’ spent time to
adapt images from the user’s request to the solution.
For each test we considered one image adaptation.
The original resolution of the images varies from
1024 x 768 to 266 x 335 pixels. Moreover, the most
common adaptation techniques can be categorized as
the combination of resizing to adapt the resolution
with transcoding to change the media format. We
performed various tests in this field by using a log
agent to precisely determine the time. However, we
are only considering the worst 12 results in order to
facilitate the visualization. The spent time varies
from 0.219 seconds to 0.297 seconds.
Figure 5 shows the AMS response time to create
a new mobile agent at runtime. Again, we illustrated
the 12 worst results to facilitate the visualization.
The response time varies from 0.196 seconds to
0.232 seconds. It means that it is almost
instantaneous with the creation process.
INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS
153

Figure 4: Adaptability issue evaluation.
Figure 5: Mobile agent’s creation evaluation.
Figure 6 presents the users’ satisfaction. We used
a scale that varies from excellent (minimum of 90%
of satisfaction) to unacceptable (no satisfaction), in
which we have five intermediary alternatives: very
good (minimum of 80% of satisfaction), good
(minimum of 70% of satisfaction), regular
(minimum of 60% of satisfaction), poor (minimum
of 40% of satisfaction) and very poor (minimum of
30% of satisfaction).
Figure 6: User’s satisfaction issue evaluation.
We illustrated the 10 worst results to facilitate
the visualization. The satisfaction varies from good
to excellent. It helped us to conclude that we are on
the right direction and that the proposed building
blocks set represented a suitable support in the
development of the e-commerce ubiquitous system
mainly centered on the user’s satisfaction issue.
4 RELATED WORK
There is interesting work that proposes support for
the mobility issue in ubiquitous and pervasive
contexts. Some of them are:
In (Senart et al.2006,) Senart et al. describe the
application of context-based reasoning to support
mobility in ubiquitous systems. They use sentient
objects as mobile intelligent entities to extract,
interpret and manipulate context information in
order to drive their behavior in ad-hoc mobile
networks. The sentient objects communication is
supported by a middleware centered on the event
abstraction. The authors applied their proposal to an
application from the Intelligent Transportation
System.
In (Satoh 2002,) Satoh presents a framework
based on location-tracking systems to navigate
mobile agents to stationary or mobile computers.
Each mobile agent is a collection of Java objects and
equipped with an identifier - called TaggedAgents.
The author presents some practical applications (e.g.
follow-me applications and a user-navigation
system.) Moreover, he describes some important
design principles, which are used to develop the
proposed framework, such as: autonomy, scalability,
extensibility and personalization.
In (Sousa and Garlan 2002,) Sousa and Garlan
describe an architectural framework - AURA - that
supports two important concerns in ubiquitous
systems development: mobility and adaptation. First,
it allows a user to preserve her/his work when
moving among different environments. Second, it
allows performing adaptations in a particular
environment at runtime. Various elements compose
the proposed framework, such as: task manager,
service suppliers, context observer and environment
manager.
The previous work and others found in the
literature related to mobility and adaptation issues in
ubiquitous computing do not use the intentionality
abstraction to design and implement the autonomous
entities that support the development of ubiquitous
systems.
There are some advantages to developing
intentional-agents (Gordon 2005) (Dignum and
Conte 1997) (Bigus and Bigus 2001) (Yu 1997,)
such as the BDI-based agents presented on our
reuse-oriented building blocks for mobility and
ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence
154

adaptability issue. Considering that the new goals'
formation is a fundamental feature of autonomous
entities, “existing formal theories of agents are
found essentially inadequate to account for the
formation of new goals and intentions of the agent”
(Dignum and Conte 1997). In addition, the agent’s
cognition capacity and the rationale significantly
increase using the distributed intentionality (Yu
1997) as a goal-orientation-centered approach. The
agents based their decisions on reasoning and
learning techniques (e.g. the BDI model and the
JADEX agents’ engine – Figure 7) and the user’s
satisfaction, being aware in relation to the ubiquitous
context. In this scenario, the context awareness is
centered on the agents’ beliefs, desires and
intentions as interpretation of the human-mental
states. Moreover, some common problems are
avoided by using BDI-based agents, for example: it
is really simple to deal with the agents’ adaptability
according to different ever-changing environments,
by dynamically updating the agents’ knowledge
bases, their beliefs, their goals’ formation and their
sequence of tasks to achieve the desired goals. In the
proposed support set a fuzzy logic library improves
the agents’ reasoning engine (Figure 7).
Furthermore, the intentionality abstraction is closer
to the human-reasoning representation than the
object and/or the behavior abstractions.
Strengthening our argumentation, we can define the
world “intention” as the state of one's mind at the
time one carries out an action.
5 FINAL CONSIDERATIONS
Multiple devices with multiple features, from multi-
ple vendors with heterogeneous technological
capabilities, equipped with various communication
technologies and distributed in different smart-
spaces. This is the typical scenario in ubiquitous
contexts, in which the mobility, the adaptability, the
context-awareness, the integration need and other
issues are essential. Moreover, not disturbing the
users or even distracting them is desirable.
Therefore, the use of calm technologies is
appropriate. Contributing to this field, we propose
different reuse-oriented building blocks to support
the systematic development of ubiquitous systems.
In this paper, we concentrate our efforts on the
presentation of the building block to support the
mobility issue in ubiquitous systems. We evaluate
our approach by applying it to the development of an
e-commerce ubiquitous system, by focusing on
adaptability, mobility, context-awareness,
integration, device heterogeneity and user
satisfaction issues. Finally, we compare our
intentional-agent-based proposal to existing
approaches, centered on sentient objects and
behavioral agents, by describing some advantages
through working with intentionality: (i) agents with
powerful cognition capacity to appropriately deal
with the human-mental states; (ii) easy agents’
adaptability according to the ever-changing context;
and (iii) adequate reasoning engine based on the
agents’ beliefs, desires and intentions and a fuzzy
logic library, in which the fuzzy sets and the fuzzy
variables represent quality criteria (e.g. security,
price and download time) used to better achieve the
user’s goals.
As future work, we intend to improve our reuse-
oriented building blocks set by following the
technological evolution. Recently, we have been in-
Figure 7: JADEX agents’ engine improved by our approach.
INTENTIONAL MOBILE AGENTS IN UBIQUITOUS SYSTEMS
155

vestigating other alternative solutions related to the
agents’ reasoning and learning support (Letier 2002)
(Riedmiller and Merke 2002) (Wiewiora et al.2003).
Our main goal in this field is a comparative study
that shows the advantages and disadvantages of each
alternative according to specific issues: human-
practical reasoning, agent cognition capacity, goals
formation support and human-mental states
interpretation/representation.
REFERENCES
Bellifemine, F.; Caire, G.; Grenwood, D., 2007.
“Developing Multi-Agent Systems with JADE. Wiley
Series in Agent Technology,” John Wiley & Sons Ltd,
ISBN-13: 978-0-470-05747-6, 286 pages.
Bigus, J. P.; Bigus, J., 2001. “Constructing Intelligent
Agents Using Java: Professional Developer's Guide,”
2nd Edition. ISBN 10:047139601X, ISBN
13:9780471396017, John Wiley & Sons.
Braubach, L.; Pokahr, A.; Lamersdorf, W., 2004. “Jadex:
A Short Overview,” In Net. ObjectDays’04:
AgentExpo, pp. 195-207, September.
Caire, G., 2003. “LEAP User Guide,” TILAB, December.
Dignum, F.; Conte, R., 1997. “Intentional Agents and
Goal Formation,” Proceedings of the 4th International
Workshop on Intelligent Agents IV, ISBN:3-540-
64162-9, Vol. 1365, pp. 231-243.
FIPA ORG. 2002a. FIPA Request Interaction Protocol
Specification. Document no. SC00026H.
FIPA ORG. 2002b. FIPA ACL Message Structure
Specification. Document no. SC00061G.
FIPA ORG. 2002c. FIPA SL Content Language
Specification. Document no. SC00008I.
Georgeff, M.; Pell, B.; Pollack, M.; Tambe, M.;
Wooldridge, M., 1998. “The Belief-Desire-Intention
Model of Agency,” 5th International Workshop on
Intelligent Agents: Agent Theories, Architectures, and
Languages (ATAL98), ISBN:3-540-65713-4, Vol.
1555, pp. 1-10.
Gordon, R. M., 2005. “Intentional Agents Like Myself.
Perspectives on Imitation: From Mirror Neurons to
Memes,” Hurley & Chater, N., MIT Press, Chapter 15.
Letier, E., 2002. “Reasoning about Agents in Goal-
Oriented Requirements Engineering,” Thesis
Dissertation, UCL-FSA/INGI - Département
d'ingénierie informatique.
Marca, D. A.; McGowan, C. L., 1987. “SADT: structured
analysis and design technique,” McGraw-Hill,
ISBN:0-07-040235-3, 392 pages, USA.
Pokahr, A.; Braubach, L.; Lamersdorf, W., 2005. “Jadex:
A BDI Reasoning Engine,” Multiagent Systems,
Artificial Societies, and Simulated Organizations, Vol.
15, pp. 149-174.
Riedmiller, M.; Merke, A., 2002. “Using Machine
Learning Techniques in Complex Multi-Agent
Domains,” Adaptivity and Learning, Springer.
Satoh, I., 2002. “Physical mobility and logical mobility in
ubiquitous computing environments,” Lecture notes in
computer science, International conference on mobile
agents No. 6, Vol. 2535, pp. 186–201, Spain.
Senart, A.; Bouroche, M.; O'Connor, N.; Hughes, B.;
Singh, K.; Cahill, V., 2006. “Supporting Mobility
using Context-Based Reasoning,” System Support for
Ubiquitous Computing Workshop, 7 pages,
September.
Serrano, M.; Lucena, C. J. P. de, 2010a. “Applying FIPA
Standards Ontological Support Intentional-MAS-
Oriented Ubiquitous System,” 12th International
Conference on Enterprise Information Systems
(ICEIS'10), Vol. SAIC, pp. 114-121, Funchal,
Madeira, Portugal, June.
Serrano, M.; Lucena, C. J. P. de, 2010b. “Agent-Oriented
Dynamic Database for Intentional Development of
Ubiquitous Systems,” PUC-Rio, Monograph on
Computer Science, No. 07/10, ISSN 0103-9741, 24
pages, May.
Serrano, M.; Lucena, C. J. P. de; Mylopoulos, J.; Yu, E.,
2010. “Experiences with Model Reuse: Non-
Functional Requirements Catalogues for Ubiquitous
Systems,” PUC-Rio, Monograph on Computer
Science, No. 08/10, ISSN 0103-9741, 20 pages, May.
Serrano, Milene; Serrano, Maurício; Lucena, C. J. P. de,
2008. “Framework for Content Adaptation in
Ubiquitous Computing Centered on Agents
Intentionality and Collaborative MAS,” Fourth
Workshop on Software Engineering for Agent-
oriented Systems (SEAS’08/SBES’08), pp. 1-12.
Serrano, Milene; Serrano, Maurício; Lucena, C. J. P. de,
2009. “Ubiquitous Software Development Driven by
Agents' Intentionality,” 11th International Conference
on Enterprise Information Systems (ICEIS'09), Vol.
SAIC, pp. 25-34, Milan, Italy, May.
Sousa, J. P.; Garlan, D., 2002. “Aura: an Architectural
Framework for User Mobility in Ubiquitous
Computing Environments,” 3rd Working IEEE
Conference on Software Architecture, pp. 29-43,
Montreal.
Weiser, M.; Brown, J. S., 1995. “Designing Calm
Technology,” Xerox Parc, 6 pages, December.
Wiewiora, E.; Cottrell, G.; Elkan, C., 2003. “Principled
Methods for Advising Reinforcement Learning
Agents,” Twentieth International Conference on
Machine Learning (ICML-2003), Washington, DC.
Yu, E., 1997. “Towards Modelling and Reasoning
Support for Early-Phase Requirements Engineering,”
3rd International Symposium on RE, pp. 226-235,
January.
ICAART 2011 - 3rd International Conference on Agents and Artificial Intelligence
156