prikaz prve stranice dokumenta Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu
No public access
master's thesis
Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. urn:nbn:hr:168:010649
Salamun, Karla
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Salamun, K. (2020). Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:010649

Salamun, Karla. "Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. https://urn.nsk.hr/urn:nbn:hr:168:010649

Salamun, Karla. "Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2020. https://urn.nsk.hr/urn:nbn:hr:168:010649

Salamun, K. (2020). 'Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 07 November 2024, https://urn.nsk.hr/urn:nbn:hr:168:010649

Salamun K. Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2020 [cited 2024 November 07] Available at: https://urn.nsk.hr/urn:nbn:hr:168:010649

K. Salamun, "Primjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2020. Available at: https://urn.nsk.hr/urn:nbn:hr:168:010649

Metadata
TitlePrimjena genetskog programiranja za raspoređivanje zadataka u sustavima za rad u stvarnom vremenu
Title (english)Application of Genetic Programming for Real-Time Task Scheduling
AuthorKarla Salamun
MentorHrvoje Džapo (mentor)
Committee memberHrvoje Džapo (predsjednik povjerenstva)
Committee memberMario Cifrek (član povjerenstva)
Committee memberŽeljka Lučev Vasić (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2020-07-10, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Electrical Engineering
Abstract
U ovom radu prikazane su mogućnosti primjene strojnog učenja za raspoređivanje zadataka u preopterećenim sustavima za rad u stvarnom vremenu. Istraženi su postojeći formalni algoritmi za raspoređivanje u preopterećenim sustavima za rad u stvarnom vremenu, kao i metode za smanjenje opterećenja u tranzijentnim i permanentnim uvjetima preopterećenja. Implementiran je radni okvir za evoluciju heuristika za raspoređivanje koristeći genetsko programiranje. U implementirani radni okvir integrirana je potpora za višekriterijsku optimizaciju (NSGA-II algoritam) i kooperativnu koevoluciju. Rezultati evoluiranih heuristika analizirani su pomoću simulatora izvođenja periodičkih zadataka koji koristi raspoređivanje s istiskivanjem. Sučelje za generiranje skupova zadataka za učenje i testiranje implementirano je koristeći UUniFast algoritam. Rezultati testiranja predstavljene tehnike uspoređeni su s rezultatima algoritma Red Tasks as Late as Possible (RLP) s obzirom na postignutu kvalitetu usluge (QoS). Primjenom evoluirane heuristike na veliki broj sintetski generiranih setova zadataka, kvaliteta usluge u uvjetima preopterećenja povećana je do 50% u odnosu na RLP algoritam. Rezultati implementiranog radnog okvira verificirani su usporedbom s rezultatima postojećeg radnog okvira za evolucijsko računanje. Napravljena je modifikacija operacijskog sustava FreeRTOS za korištenje evoluiranih heuristika za raspoređivanje zadataka.
Abstract (english)
This thesis presents the possibilities of using machine learning approach for task scheduling in overloaded real-time systems. The existing formal algorithms for scheduling in overloaded real-time systems were investigated, as well as the methods for reducing system load in transient and permanent overload conditions. A framework for evolving scheduling heuristics using genetic programming was implemented, with support for multi-objective optimization (NSGA-II algorithm) and cooperative coevolution. The performance of the evolved heuristics is measured by a simulator of periodic tasks execution with preemptive scheduling. The interface for generating task sets for training and testing is implemented using the UUniFast algorithm. The results of testing the considered technique are compared to the results of the Red Tasks as Late as Possible (RLP) algorithm, with respect to Quality of Service (QoS) metric. Applying the evolved heuristic to a large number of synthetically generated task sets increased the QoS achieved in overload conditions up to 50% in comparison with RLP approach. An existing framework for evolutionary computation is applied to the task scheduling problem in order to validate the results achieved by the implemented framework. The FreeRTOS operating system was modified to use the evolved heuristics for task scheduling.
Keywords
Keywords (english)
Languageenglish
URN:NBNurn:nbn:hr:168:010649
Study programmeTitle: Electrical Engineering and Information Technology
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2023-01-13 13:45:17