Hybrid Metaheuristic Approach for Scheduling of Aperiodic OS Tasks - Complex System Modelling and Control Through Intelligent Soft Computations - page 826

Another real application issue is that once some task has to miss a deadline, it
should be the least valuable task. Again, many algorithms do not address this fact.

Contribution: Algorithm for Feasibility Testing of Sporadic Task Systems

In the current work, we use as a scheduling policy, the EDF algorithm. After a
recon

w h was applied at run time, the intelligent agent proposes
useful solutions for users by sending sporadic tasks to idle times which are con-
sidered as non-productive times of the processor, by modifying the deadlines of
tasks, the worst case execution time (WCETs), the activation time, by tolerating
some non critical k among m tasks to miss their deadlines or by removing some
tasks according to a prede

guration scenario

ned heuristic.
We suppose that each system n can be automatically and repeatedly recon g-

(0) and after the h th recon

ured. n is initially considered as ʾ guration n turns into

ðÞ
þ

n ðw h Þ , where h first virtual processor to virtually

execute old periodic tasks and VP 2 as the second virtual processor to virtually
execute new sporadic tasks, implementing the system after the hth recon

2@ .Wede ne VP 1 as the

guration

n ðw h Þ , all old tasks from n ðw h 1 Þ are executed by the newly updated scenario w h .In

VP ðw h Þ
1

and the added sporadic tasks are executed by VP ðw h Þ
2

. The proposed intel-

n w h ligent agent is trying to minimize the response time R ðw h Þ of after each recon-

guration scenario w h .

Formalization

n

We assume in this work a system composed of a mixture of n periodic and m

sporadic tasks. The initial processor utilization factor U before any addition sce-
nario of new sporadic tasks to ¼ P i¼1 C i

n ðw h 1 Þ ¼ n is U

T i . An assumed system

n ðw h Þ ¼

fs 1 ; s 2 ; ...; s n g turns after a recon guration scenario w h to

fs 1 ; s 2 ; ...; s n r n þ 1 ; r n þ 2 ; ...; r m g

by considering that m-n new sporadic tasks

n ðw h 1 Þ . After each addition scenario are added to w h , the tasks are logically divided

into two subsets. One contains the so called new sporadic tasks which are added to
the system, and the rest of tasks taken from

n ðw h 1 Þ are considered as old tasks to

form the second subset.

Problem

After any addition scenario w h , the response time can be increased and/or some old/

new tasks can miss their deadlines.
When a recon

guration scenario is automatically applied at run-time, the pro-
posed intelligent agent logically decomposes the physical processor of n ðw h Þ into