Environmental Engineering Reference
In-Depth Information
Behind these different terms and research areas lie three key properties:
•
Nomadic
•
Embedded
•
Invisible
In effect, this may lead to the creation of a single system with (poten-
tially) billions of networked information devices. All of these next generation
paradigms, in one form or another, will require an autonomic-self-managing-
infrastructure to provide the successful reality of this envisaged level of invis-
ibility and mobility.
Currently, and for the foreseeable future, the majority of users access com-
puting through personal devices. Personal Computing offers unique challenges
for self-management in itself due to its multidevice, multisituation, and mul-
tiuser nature. Personal autonomic computing is much less about achieving
optimum performance or exploiting redundancy (as in AC) and more about
simplifying use of the equipment and the associated services [
10
,
11
]. Thus, it
is particularly relevant to deal with the move toward a nomadic, embedded,
and invisible computing future [
152
,
157
].
8.3 Research and Technology Transfer Issues
The challenge of autonomic computing requires more than the re-engineering
of today's systems. Autonomic computing also requires new ideas, new in-
sights, and new approaches.
Some of the key issues that will need to be addressed are:
Trust:
Even if the autonomic community manages to “get the technology
right,” the trust of the user will be an issue in terms of the user's handing
over control to the system. AI and autonomous agent domains have suf-
fered from this problem. For instance, neural networks (due to concerns
over the “black-box” approach) and a number of AI techniques (due to
their inherent uncertainty) are often not adopted. Rule-based systems,
even with all their disadvantages, often win adoption, since the user can
trace and understand (and thus implicitly trust) them [
153
]. Note that
even within autonomic computing and autonomic communications, the
bulk of the literature assumes rules will be used instead of other, less
brittle and more adaptable stochastic AI approaches.
Economics:
New models of reward will need to be designed. Autonomy and
Autonomicity may derive another self-
property: selfishness. For instance,
why would an AE perform an operation, e.g., relay information, for an-
other AE that was outside its organization and did not affect or benefit
from it? In particular, if it was operating within a mobile (battery pow-
ered) environment and to do so, incurred personal cost, performing the
operation for the outside unit could shorten its useful life or make it nec-
essary to recharge earlier.
∗
