5 WHY USE EMBEDDED
TECHNOLOGY?
When our system has been installed in that real-life
environment, we have interviewed operators in order
to find out their degree of satisfaction. One of the
improvements that they have proposed has been that
the IDOLO system is like a black box for them. That
is a system which is easy to maintain and configure
without needing any experience with computers (just
by pressing a button). That have made us think about
the possibility of developing the system by using
embedded technology.
Embedded systems shouldn’t be thought of as
PCs with general characteristics. Although, more
and more, they could be replaced in some
application fields thanks to the development of
microprocessors and microcontrollers.
They could be considered as hard – soft
applications which are more interesting in the
immediate future of the Information and
Communications Technology sector. The main idea
is to use systems which are practically invisible and
made up of microprocessors and software in very
small systems. These systems allow us to obtain
information and accomplish its processing
everywhere, in a quick and easy way (ubiquitous
computation). If we are able to create a network
composed of these embedded systems, which are
almost invisible, and we provide them with
communication and action abilities, we can create a
spatially distributed hardware structure that is
available at all times. It makes user anticipate
(pervasive computation) (Marwedel, 2003) (Sutter,
2003). All of this can be accomplished in a more
secure and reliable way and with a quick and easy
maintenance with an economic costs and electric
consumption lower than those of traditional PCs.
By introducing these systems in some
hierarchical schemes of “surveillance” distributed in
substations or transformation centers, we can
accomplish an automatic surveillance which allows
preventive maintenance operations. In this way we
can anticipate possible anomalies that could produce
blackouts, which are quite problematic for the end
users and, as a result, have significant economic
impact for the electric utilities.
6 WHY USE MULTI-AGENT
TECHNOLOGY?
When we are talking about automatic surveillance
we re referring to not only proactive but also
reactive surveillance. This implies that the elements
which carry out the surveillance have human
capacities and are able to make decisions and act by
themselves. This human behavior has to be
programmed, and we are working to achieve this by
using multi-agent system (MAS) technology.
MAS are systems composed of multiple
interacting computing elements, known as agents.
Agents are computer systems with two important
capabilities (Wooldridge, 2002):
• They are capable of autonomous actions and
can decide by themselves.
• They are capable of interacting with another
agents in a social way (cooperation,
coordination, negotiation…)
So, first we are modeling the different
interactions among the different kinds of agents.
Each defined agent is responsible for the
surveillance in a specific way because each of them
has a fixed intelligent level. Some of them have less
processing capacity, so they can carry out a basic
analysis of captured situations, whereas, others have
more processing capacity, so they can accomplish an
advanced analysis.
What kind of information is going to be
analyzed? We are working with several types of
sensors (senses in Figure 3) which capture image
(visible and infrared), smoke and volumetric
measurements. Depending on the kind of data, the
analysis is carried out in a different way. After
modeling the interactions, we are going to design the
MAS. We have a main goal in mind to cover the
maximum amount of elements in the electrical
facility environment (surveillance targets) with the
minimum agents and at the same time by keeping
agents cooperating. Autonomy and learning
capacities are very important for our system and
both of them are great advantages of MAS.
7 ARCHITECTURE FOR THE
NEW SYSTEM
We use two levels of surveillance. In the lowest
hierarchical level there is a network composed by
guard agents which don’t need to have an excessive
processing capacity. These vigilant systems are
always collecting information (through sensors) and
they have to control their environment by generating
alarms or corrective actions (actuators) if
appropriated. They have to interact with their
environment in a quick and easy way, processing in
real time. They have certain mobility in a limited
area.
ICEIS 2008 - International Conference on Enterprise Information Systems
394