20.500.12556/DKUM-12185 METODE STATISTIČNEGA MODELIRANJA IZMERJENEGA OMREŽNEGA PROMETA ZA NAMENE SIMULACIJ METHODS FOR THE STATISTICAL MODELING OF MEASURED NETWORK TRAFFIC FOR SIMULATION PURPOSES Doktorska disertacija je s področja modeliranja, vrednotenja in simulacij telekomunikacijskih omrežij. Kot cilj smo si zadali razvoj novih metod modeliranja omrežnega prometa, ki predstavljajo zelo pomemben segment simulacij komunikacijskih omrežij. Pri modeliranju prometa, za simulacijske namene, smo kot izhodišče izbrali v realnih omrežjih izmerjeni promet, katerega statistične lastnosti želimo s predlaganimi modeli čim bolj verno posnemati. V doktorski tezi je izpostavljenih več originalnih prispevkov k znanosti. Prva dva se nanašata na dve predlagani metodi ocenjevanja parametrov statističnega modela podatkovnih virov: metoda s posnemanjem defragmentacije paketov ter metoda s primerjavo histogramov. Znotraj druge metode smo identificirali še en pomemben originalni prispevek, ki ga predstavlja analitični model fragmentacije prometa podatkovnih virov. Kot četrti originalni prispevek smatramo modificiran hi kvadrat test, ki bolje kot nekateri drugi znani testi, ovrednoti odstopanje med modeliranim in izmerjenim prometom. V doktorski tezi predstavljene algoritme in metode smo implementirali v programskem orodju TraffMod. S pomočjo tega orodja, v povezavi s simulatorjem OPNET Modeler, smo opravili obširne raziskave, na osnovi katerih smo potrdil uporabnost razvitih metod ter veljavnost omejitev in pogojev pod katerimi posamezne metode dajo uporabne rezultate. This dissertation covers the areas of modeling, evaluation and simulations of telecommunication networks. The goal of our research is the development of a new network traffic modeling methods, which represent very important segments of communication network simulations. During the traffic modeling we developed methods for statistical descriptions of measured network traffic in the best possible way. More of the original contributions to science are exposed in this dissertation. The first two contributions present two estimation methods for data source statistical model parameters: a method of mimic packet defragmentation, and a method based on histogram comparisons. The second method also makes another contribution to science. This is the developed analytical model for packets´ fragmentation. The fourth original contribution to science presents a modified chi squared test, which in comparison with other known tests, better evaluates discrepancies between measured and modeled network traffic. The represented methods and algorithms in this dissertation were implemented within a developed application, called TraffMod. This application, together with the use of the OPNET simulation tool, was used as a tool during extensive research work, which confirmed the validity of the developed methods, models and algorithms, and those conditions and limitations where these estimation methods would be useful. omrežni promet modeliranje simulacije naključni proces porazdelitev verjetnosti network traffic modeling simulations random process distribution true true false [M. Fras] Slovenski jezik Angleški jezik Doktorska disertacija 2009-10-13 08:27:07 2009-10-14 09:20:30 2022-04-12 07:17:09 2022-04-19 03:15:25 2009 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 [519.22+004.414.23]:621.39(043.3) 13507094 URN:SI:UM:DK:XKXQNFX1 DR_Fras_Matjaz_2009.pdf DR_Fras_Matjaz_2009.pdf 1 8CB2D238D9AC747563977707D3D7AD25 1a259583d89d036f8c22cd4b1c528fb5dfb7b57c86e02424b4914b1772eb8654 56bd05eb-7c03-11eb-bb7a-00155d0001ca 20.500.12556/dkum/82b1fb0b-d3b6-4848-aaaa-41ef9a8e5b4c https://dk.um.si/Dokument.php?lang=slv&id=11394 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0