Hybrid Metaheuristic Approach for Scheduling of Aperiodic OS Tasks - Complex System Modelling and Control Through Intelligent Soft Computations - page 844

task is considered as a new one at t time units. By considering a feasible System Sys
before the application of the recon guration scenario, each task of n old is feasible,

e.g. the execution of each instance is finished before the corresponding deadline.

4.2.4 Experimental Analysis and Discussion

In this section, in order to check the suggested con gurations of tasks allowing the

system s

behavior on several test sets in order to rate the performance of the polling server
and the background server in our hybrid scenario.

' s feasibility and the response time minimization, we simulate the agent '

Simulation

We have conducted several test sets in order to rate the performance of the polling
server and the background server in our hybrid scenario. We have set up a real-time
recon

guration that allows us to randomly generate
task sets, schedule them according to the proposed hybrid method, and displays the
schedules for visual control. Our test rows have been on each 1,000 randomly
generated task sets, while the number of tasks is significantly higher. We have
scheduled task sets with the polling server and the proposed hybrid method.

guration tool named RT-Recon

Discussion

In each of these examples, many aperiodic requests occur at any moment of the
time. The response time performance of only polling service or only background
service for the aperiodic requests is poor. Since background service occurs when the
resource is idle, with the polling server, the response time performance for the
aperiodic requests is better than both single background service and single polling
service for all requests. For these examples, a polling server is created with an
execution time of 1 time unit and a period of 5 time units. Also note that since any
aperiodic request only needs half of the polling server s capacity, the remaining half

is discarded because no other aperiodic tasks are pending. Thus, these examples
demonstrate how polling and background can provide an improvement in aperiodic
response time performance over background service or polling one and are always
able to provide immediate service for aperiodic requests. The proposed hybrid
scheduling algorithm may thus be appropriate for many applications.
The other most important thing is the complexity which is decreased in our
approach to O(nlog(n)) because the proposed algorithm is recursive, and the Ear-
liest Deadline First algorithm also, would be maintaining all tasks that are ready for
execution in a queue. Any freshly arriving task would be inserted at the end of
queue. Each task insertion will be achieved in O(1) or constant time, but task

'