Hybrid Metaheuristic Approach for Scheduling of Aperiodic OS Tasks - Complex System Modelling and Control Through Intelligent Soft Computations - page 824

We assume that for all the real-time tasks, we have M automatic recon guration

scenarios ( w 1 ; w 2 ; ...; w M ). We mean in this thesis by an automatic recon gu-

ration scenario M), any operation that adds, removes or also updates

tasks at run-time which corresponds to this automatic recon

w h (h 2 1 ...

guration scenario w h .

Automatic updates of tasks mean modi cations of their temporal parameters e.g.

Periods and/or deadlines, or modi cations of their Worst Case Execution Times.

In our model, a sporadic task is represented by

r w h
i

for each recon guration

scenario w h and it ' s completely characterized by specifying its worst case execution

time C w h
i

, its relative deadline D w h
i

, the deadline tolerance value m w h
i

, and its value

I w h
i

w h . In the following, we assume that the task
class can be derived from the task value. For instance, tasks with value I w h
i

for each recon guration scenario

¼ H can

be considered as HARD and others with value I w h
i

¼

S are considered as SOFT.
For the proofs which provide a necessary and suf cient condition for the

schedulability of task sets at each task activation, we require a notation that iden-
ti es the kth task in the current ordered list at time t. For this, we use the notation

r w h

i ; 1 ; r w h i ; 2 ; ...; r w h

r w h
i ; k

i ; m where

is the task in the kth order for each recon guration

scenario w h .

In our model, we assume that the minimum interarrival time of each sporadic
task is equal to its relative deadline D w h
i

r w h i can be completely
characterized by specifying its worst case execution time C w h
i

, thus a sporadic task

and its relative

deadline D w h
i

. Hence, a sporadic task set will be denoted as follows:

w h
i

C w h
i

D w h
i

n w h 1toM.

In summary and in our work, a sporadic task set will be denoted for each
recon

¼ fr ð ; Þg; i ¼ 1tom ; h ¼

w h as follows:

guration scenario

¼ fr w h
i

C w h
i

D w h
i

m w h
i

I w h
i

n w h

ð ; ; ; Þg; ; :

i ¼ 1tom h ¼ 1toM

Within this framework, different solutions are used for handling sporadic tasks in
an optimal fashion. In particular, the tasks are scheduled based on their deadline,
guaranteed based on C i , D i , m i , I i , and rejected based on I i value.
The other known method is to use response time analysis, which consists of
computing the worst-case response time (WCRT) of all tasks in a system and
ensuring that each tasks WCRT is less than its relative deadline. To avoid these
problems, and to have a feasible system in this thesis work, our proposed tool
RT-Recon guration can be used. For this reason, we present the following rela-

tionships among the parameters de

ned above for each recon guration scenario w h :

d w h
i

a w h
i

D w h
i

¼ þ ð 1 Þ

L w h
i

d w h
i

a w h
i

C w h
i

¼ ð 2 Þ