 |
|
 |
One of the main goals in developing a middleware is to maintain a high
degree of openness and flexibility, making it applicable in as many
different contexts as possible.
During JADE design and realisation we worked with this concept in mind
and the several different applications that we learned from actual Open
Source members tells us that we reached the goal.
The general peer-to-peer intelligent agents approach permits to model systems that represent environment in which independent autonomous elements have to communicate and collaborate in order to achieve the desired results: and this description applies, at least theoretically, to almost all the situations of the human life.
Best results of this technology are delivered, however, in those cases where an interaction between numerous elements is required, and in which an autonomous and dynamic adaptation to complex relations is needed.
Two general perspectives may be considered in describing circumstances
in which peer-to-peer intelligent agents approach is considerably helpful:
• a single point of view, where a single element has to tackle
the complexity of the environment in which is stuck: the agent that
represents him does the work (or part of) for him, communicating with
the other actors in the field, trying to get the wanted outcomes
• a multiple point of view, where all the elements in the system
try to collaborate as a whole in order to achieve a common result: the
system job is to subdivide tasks and to control individual contribution
optimising resources and communications.
Starting from both JADE real application examples and theoretical P2P intelligent agent studies, it is possible to imagine many different contexts of use of our middleware. We propose hereafter a first general categorisation with no scientific premises but just to help in getting a general understanding of the possible usages of JADE:
• Mobile applications
• Internet applications
• Corporate applications
• Machine-to-Machine applications
The focus in this case is in the support of a person on the move where
the peer can represent a “personal agent” that helps its
owner. Its goal is to facilitate the search and discovery of information
through the interaction with other peers, being both other persons or
“service providers”.
In general JADE agents are extremely suited to act in the context of
Mobile PIM – Personal Information Management: their ability of
autonomous and proactive acting and seamless communications with the
others allows to conceive applications addressed to the every-day life
organisation, like meeting organiser, info search or services negotiation
(i.e. to find the nearest and cheapest parking, or to book the best
restaurant according to user preferences).
Agent intervention can support its owner in concrete needs: several
examples illustrate the strengths of JADE peers acting as assistants
for travellers, being private tourists or mobile workers. Other applications
have been developed in the mobile work or sales support systems, in
which P2P agents duties are dynamic task & info sharing and exchange.
In other cases agent role could be played in the entertainment field:
JADE middleware can be the base for multiparty gaming applications,
in which a real interaction between players is offered, or can be the
building block for an enhanced kind of mobile community where the peer-to-peer
communication allows a more direct and richer relation amongst members.
In the Internet domain several application concepts based on P2P agents
technology have been proposed during the years, as the main concept
has been to beneficiate from the increased connectivity, in terms of
bandwidth and relations amongst people.
JADE-based systems allow end users to deal with the complexity and number
of opportunities, and to exploit the possibility to access remote resources
and services in a seamless way.
Key elements in JADE-based applications designing are direct P2P communication
support, smart info retrieval capabilities and negotiation techniques.
Starting from these basic principles, various sectors have been considered
like for instance e-learning or e-healthcare, and in general all the
context of e-commerce/e-trading have been tackled.
Moreover many of the consideration put forward in the mobile environment
can been easily extended to a fixed consumption, allowing an integrated
organisation of the personal and working life between PC and mobile
devices.
Finally also the entertainment sector has been analysed starting from
the obvious consideration on community services, multiparty gaming and
content sharing applications.
Peer- to-peer intelligent agent paradigm is an evident choice when
the task is enhancing company productivity and efficiency. The point
is to facilitate the collaboration and co-operation amongst systems
and people in order to achieve better results.
Many different examples of the usage of P2P agents technology have been
provided already in order to support company processes: when it comes
the time to share information or coordinate tasks the deployment of
JADE-based system becomes very effective and useful.
Instances have been proposed in “soft contexts”, such as
knowledge management and personnel administration or in general for
the support to decision making processes.
Other proposals have come in “harder” environments, including
companies core processes, like logistic or production: valuable demonstrations
concern systems for factory control, exploiting JADE capabilities in
optimising tasks and coordinating resources.
Interestingly enough some JADE applications exceed even company boundaries
supporting activities amongst different businesses: agents ability in
the negotiation and information retrieval amongst services/resources
provider leads to the construction of JADE-based supply-chain or e-procurement
products.
Machine-to-Machine applications
In those cases where peers represent machine and not humans, JADE features
can be exploited at its best. Autonomy and proactivity of agents play
a decisive role in the management of complicated systems: when the number
of elements and the complexity of the relations raise, the opportunity
of a distributed control simplifies significantly system operations.
Complex algorithms and heavy elaborations, typically concentrated in
a single central point can be subdivided in each peer, increasing overall
efficiency and system performance and reducing the risks connected with
this concentration, such as fault tolerance and scalability issues.
Classical examples in this context are automatic control or traffic
management systems, but the helpfulness of JADE approach can go down
in more depth: several studies are under way in order to extend P2P
agents paradigm to network management, with the concept of peers mapping
onto network equipments and interacting amongst them for resources optimisations
and system control.
Finally it’s interesting to cite the extensive application of
P2P intelligent agents models for simulation uses. The concept of numerous
elements acting independently but together in the environment in which
they are part of, easily applies to several contexts, from different
science sectors (for instance biology, ecology and natural science in
general) to social and economic studies.
The use of theoretical models able to embody trustworthily the environment
under study and perform reliable simulations allows to achieve scientific
results that are practically impossible to get from the observation
of the real world domain.