Information Technology Reference
In-Depth Information
work, that an invocation of each aperiodic task will follow the Poisson arrival
model. Our new proposed contribution is described in the following paragraph.
Contribution
In our proposed approach we are interested in automatic recon
gurations of
real-time embedded control systems that should meet deadlines de
ned by
user requirements. Automatic recon
gurations are applied by Intelligent Agents
Gharsellaoui et al. (
2012
). These systems are implemented by sets of tasks that we
assume independent, periodic, sporadic and aperiodic.
The goal of our original approach applied to real-time system recon
guration and
scheduling is to construct systems that are guaranteed to meet all hard deadlines and
that minimize the response time for all soft deadlines. We de
ne an agent-based
architecture that checks the system
'
s evolution and de
nes useful solutions for users
when deadlines are not satis
guration scenario. We assume also
that the invocation of each aperiodic task will follow the Poisson arrival model. The
application domain of the Poisson law was limited for a long time to that of the rare
events as the suicides of children, the arrivals of boats in a bearing or the accidents
due to the kicks of horse in the armies. But since a few decades its scope consid-
erably widened. At present, we use it in telecommunications a lot (to count the
number of communications in an interval of given time), the statistical quality
control, the description of certain phenomena bound to the radioactive destruction
(the destruction of the radioactive pits(cores) following, besides, an exponential law
of noted parameter so average), the biology, the meteorology, etc. Two cases of
suggestions are possible to be provided by our intelligent agent: modi
ed after any recon
cation of
worst case execution times of tasks or modi
cation of their deadlines. The users
should choose one of these solutions to re-obtain the system
s feasibility and
to minimize the response time of the soft aperiodic tasks. We developed a tool
RT-Recon
'
guration and tested it in order to support the agent
'
s services. We will
well formalize this approach in the following subsection.
Formalization
We now formally describe our proposed concept on which our work is based. We
de
gurable real-time embedded systems
that should classically meet different deadlines de
ne an agent-based architecture for recon
ned in user requirements. The
agent controls the system
s evolution and provides solutions for users when
deadlines are violated after any recon
'
guration scenario. In our contribution for the
aperiodic real-time tasks, we will restrict to only one recon
guration scenario
(M = 1).
Let Sys be a set of n
1
real-time tasks composed of n
i
1
tasks
i
of type i, n
1
tasks
j
s
s
j
1
; s
j
2
; ...;
of type j and n
k
k
i
1
; s
i
2
; ...; s
i
n
1
; s
tasks
s
of type k;
i.e., Sys ΒΌ
fsi1;
Search WWH ::
Custom Search