Modeling End-User Behavior In Data Networks

dc.contributor.authorLopez Oliveros, Luisen_US
dc.contributor.chairResnick, Sidney Iraen_US
dc.contributor.committeeMemberJarrow, Robert A.en_US
dc.contributor.committeeMemberSamorodnitsky, Gennadyen_US
dc.description.abstractA session is a cluster of packets that represents end-user activity in data networks, e.g. surfing the web, transferring files, streaming media, Internet-calling. We study three problems of sessions. The first problem examines why for real sessions the distribution of the number of bytes or number of packets per unit time traveling through a network node is approximately Gaussian, despite previous theoretical results say it may also be stable L´ vy. The second and third problems study four key session e features: size or number of bytes transmitted, duration, average transfer rate, and initiation time. We focus on marginal distributions of size, duration and rate, dependence structure between the marginals, and distribution of the difference between consecutive initiation times. We group sessions according to peak transfer rate in the second problem, and network application in the third problem. The ultimate goal is to teach computers how to mimic network sessions and understand network end-user behavior.en_US
dc.identifier.otherbibid: 7955495
dc.titleModeling End-User Behavior In Data Networksen_US
dc.typedissertation or thesisen_US Universityen_US of Philosophy D., Statistics


Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
4.2 MB
Adobe Portable Document Format