Information Technology Reference
In-Depth Information
preference relation defines preference between individual pieces of information, so we
can use preference relations between members of
Π
1
and
Π
2
to define
g
(
p
)
.
Suppose that
t
(
q
1
)
(
q
2
)
, i.e., that we prefer to execute task
t
(
q
1
)
to executing
t
(
q
2
)
g
(
p
)
the task
t
. We can define
as a function of the Techne preference relation,
def
=
i.e.,
, that is, from the information that the preference relation already
includes. Namely, in this example it is appropriate to say that, if
t
g
(
p
)
f
(
)
,then
Π
1
g
(
p
)
Π
2
. Since we have only
Π
1
and
Π
2
, it is enough to know that
Π
1
g
(
p
)
Π
2
to know everything we need to define the relegation relation.
Namely, the relegation relation
(
q
1
)
(
q
2
)
t
(
(
p
)
,Π
1
,Π
2
,
g
(
p
)
)
tells us that, if we cannot sat-
g
isfy
g
through
Π
2
.
Finally, we define the
influence relation
.Notethatitissimplehere,sincewehaveno
numerical values, so we cannot speak about influence as correlation. We can only say
that some information influences some other information if the absence of the former
makes it impossible for us to satisfy the latter.
(
p
)
through
Π
1
then we will relegate to
Π
2
, i.e., satisfy
g
(
p
)
Definition 14.
Influence relation:
An influence relation is a binary relation from
ψ
∈
L
SAS
to
φ
∈L
SAS
, iff either:
1.
∃
Π
⊆
Δ
∪
C
s.t.
Π
|
cτ
φ
and
Π
\
ψ
|
cτ
φ
,or
2.
∀
Π
⊆
Δ
∪
C
s.t.
Π
|
cτ
φ
and
Π
\
ψ
|
cτ
φ
.
wi
−→
In the first case above, we say that
ψ
weakly influences
φ
, denoted
ψ
φ
.Inthe
si
−→
second case above, we say that
ψ
strongly influences
φ
, denoted
ψ
φ
.
si
−→
wi
−→
Remark 4.
If
ψ
φ
, then we have no way to satisfy
φ
if
ψ
is not satisfied. If
ψ
φ
,
then some ways of satisfying
φ
cannot be used to do so if
ψ
is not satisfied.
4.1
Runtime Requirements Adaptation Problem Illustration
We now revisit the above definitions and use scenarios from travel exemplar case study
to illustrate how SAS, instantiating CARE and running on user's mobile phone, resolves
the “runtime requirements adaptation problem” at runtime.
For example, user arrives at the airport to avail her flight from Italy to Canada via
Paris for a business meeting. While at the airport after the boarding, user want to connect
to the Internet using her mobile phone to check emails and flight details before checking
in for the plane. Moreover, user wants to be informed about any flight delay.
Taking the above example, we now present SAS adaptation sequence at runtime in
case of change in context
C
along the time
T
t
1
, ....t
n
as shown in the Fig.2. Let
CS
be a set of candidate solution, thereby determining the runtime requirements adaptation
problem as a combination of instances of the tasks
T
∗
)
=
and domain assumptions
K
∗
such that
G
∗
,
Q
∗
and
S
M
are satisfied. In case of changes in the context
C
C
1
, ....C
n
overtime for which
CS
needs to be re-evaluated by the system and
R
is required to be
used or identified in a given context
C
to realize
CS
. By re-evaluation we mean that
system at runtime exploits its monitored information, evaluate all the possible alterna-
tive
CS
or search for new ones (i.e. exploiting available services) that can satisfy the
runtime adaptation problem in response to changes in the
C
therefore adapting to the
=
