Mjerenja prometa i performansi usluge YouTube na temelju podataka prikupljenih iz mreže i s korisničkog uređaja

master's thesis

• Title (english): Measurements of YouTube traffic and performance based on network and client-side data collection
• Author: Dario Pevec
• Mentor: Lea Skorin-Kapov (mentor)
• Committee member: Lea Skorin-Kapov (predsjednik povjerenstva)
• Committee member: Maja Matijašević (član povjerenstva)
• Committee member: Mirko Sužnjević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2016-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Jedna od najpopularnijih višemedijskih usluga u Internetu jest YouTube. YouTube kao davatelj usluge ima potpunu kontrolu nad dostavom i upravljanjem kvalitetom: vrši odabir poslužitelja s kojeg sadržaj struji te postavlja razinu kvalitete sadržaja koji se isporučuje. Kada YouTube video promet dođe u mrežu operatora krajnjeg korisnika, operator isporukom ne može upravljati na temelju parametara koje klijent razmjenjuje s YouTube poslužiteljima, budući da je promet šifriran. Ovo predstavlja problem i za davatelje usluge i za mrežne operatore koji žele pružiti korisnicima najbolju moguću iskustvenu kvalitetu. Predloženo rješenje je YouQ sustav. YouQ sustav na temelju šifriranog mrežnog prometa i parametara sa aplikacijskog sloja kreira ulaz za algoritme strojnog učenja koji mogu zaključiti razinu iskustvene kvalitete koja se manifestira kod krajnjeg korisnika. Na temelju te informacije, pružatelji mrežne usluge mogu učinkovito raspodijeliti mrežne resurse kako bi krajnjem korisniku omogućili najbolju moguću iskustvenu kvalitetu. Glavni fokus ovoga rada je sustav YouQ i njegove komponente.

Abstract (english)

One of the most popular multimedia service on the Internet is YouTube. YouTube as a service provider has control over content delivery and quality management: it chooses the servers from which videos are streamed and decides which quality levels should be delivered to the end user. When YouTube traffic pass through network infrastructure of a network provider, the network provider does not have control over those packets, because payload is encrypted. Hence, network providers are “left in the dark” when it comes to estimating end user QoE for their customers watching YouTube content. As a proposed solution to this problem, the system YouQ is developed. The YouQ system collects both application-level KPIs and encrypted network traffic features and prepares them as an input for machine learning algorithms. Learning algorithms detect patterns and correlations in data and create a model that can classify video streaming into one of QoE classes, based on network-level data only. Based on that learned model, network operators can efficiently allocate network resources and provide best possible quality to the end users. This work mainly focuses on the YouQ system and its components.

• Keywords: Iskustvena kvaliteta; Ekstrakcija mrežnih značajki; YouTubeov algoritam adaptacije kvalitete; Dinamičko prilagodljivo strujanje sadržaja putem HTTP protokola; Šifrirani mrežni promet
• Keywords (english): Quality of Experience; Network Feature Extraction; YouTube adaptation algorithm; Dynamic Adaptive Streaming over HTTP; Encrypted network traffic
• Language: english
• URN:NBN: urn:nbn:hr:168:848653
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 11:17:09