Community Computing - Advanced Design Approaches to Emerging Software Systems - page 212

geneous networked devices in a physi-
cal space. The major contribution was to
present active spaces as a programmable
environment instead of a collection of in-
dividual and disconnected heterogeneous
devices. In 2004, an extended version,
Super Space, was proposed to manage
and orchestrate groups of Active Spaces.
However, they did not suggest an abstrac-
tion model to conceptualize pervasive
objects constructing the space or coopera-
tive relationships between objects in their
space.

requirements are given, they are concerned about
the way to design and implement necessary agents
to offer the required services. However, in the case
of a pervasive system intending to provide services
using existing agents, we can say that participant
agents are already defined. Accordingly, in such a
case, it is more important to consider how to meet
the system requirements by existing agents rather
than what agents are required. Furthermore, most
multi-agent based development approaches do not
deeply concentrate on cooperation. To achieve our
goal, however, cooperation is the most important
aspect, so we have to consider cooperation in
more detail than others have. In this section, we
briefly introduce previous works on multi-agent
based pervasive systems development.

• PICO (Kumar, 2003; Sung, 2002):

PICO (Pervasive Information Community
Organization) is a middleware framework
for dynamically creating mission-oriented
communities of autonomous pervasive
software objects offering pervasive ser-
vices. In several agent cooperation models,
organization concepts have already been
introduced (Jennings, 2003; Wooldridge,
2002), but PICO has applied this concept
to pervasive domains. In this project, a
community was defined as a pervasive
object consisting of one or more agents
working towards a common goal. In addi-
tion, they introduce community computing
as a framework for collaboration among
agents. Their fundamental concept satis-
fies requirements of pervasive computing,
such as proactive real-time collaborations
for automated and continuous services pro-
vided in a heterogeneous environment.

• Gaia (Wooldridge, 2000; Jennings,

2003): It introduced a methodology for
analysis and design to develop a multi-
agent system. In Gaia, a multi-agent sys-
tem is regarded as a collection of compu-
tational organizations consisting of various
interacting roles and allows an analyst to
go systematically from requirement state-
ments to a design through a process of de-
veloping increasingly detailed models of
the system to be constructed.

• AALADIN (Ferber, 1998): It is a meta-

model of a multi-agent system based on
organizational concepts. It allows for de-
scribing any kind of organization using
only the core concepts of groups, agents,
and roles. In the extended version (Ferber,
2003), the model was improved into the
AGR model (Agent/Group/Role model).
In that model, the dynamic aspect was add-
ed by specifying the creation of a group,
the entering and exiting mechanism of an
agent within a group, and the role acquisi-
tion mechanism.

As the middleware-based infrastructure works,
multi-agent based approaches are frequently used
to develop pervasive systems, because agents'
features such as flexible and autonomous problem
solving behavior and the richness of interactions
guarantee dynamic and intelligent services. Ex-
isting multi-agent approaches are interesting to
study the way to seek out necessary agents to
meet requirements of a pervasive system. When

• BRAIN (Cabri, 2003): It is a framework

for supporting the different phases of the
development of interactions in multi-