Skip to Main content Skip to Navigation
Theses

QoE-based Content Distribution Network architecture

Abstract : Nowadays, with the rapid growth of high quality network applications, the Quality of Service mechanisms, which are used to guarantee required levels of performance to data flows in network, are not enough anymore to satisfy end-users, mostly in the context of Content-Oriented Networks (CON) that is used to redesign the current Internet architecture to accommodate content-oriented applications and services. We propose in this thesis a new system that introduces the Quality of Experience (QoE), a new paradigm that represents the user’s perception, in an adaptive control/command network routing chain. Our proposed system has been applied to a Content Distribution Network (CDN) architecture. We focused on two main functions of a CDN, the routing and the server selection method, in order to improve the QoE level at end-user side. For the routing method, an algorithm named QQAR (QLearning QoE-based Adaptive Routing) has been proposed. This algorithm uses the technique of Reinforcement Learning based on Markov models and take into account as a metric the QoE perceived by the end user. Regarding the server selection method, we proposed a server selection algorithm that is motivated by the Multi-Armed Bandit problem. The two proposed algorithms are validated and compared to traditional approaches using a simulation model and a real platform
Complete list of metadatas

Cited literature [195 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02886838
Contributor : Abes Star :  Contact
Submitted on : Wednesday, July 1, 2020 - 7:13:11 PM
Last modification on : Thursday, July 2, 2020 - 3:41:30 AM
Long-term archiving on: : Wednesday, September 23, 2020 - 5:36:18 PM

File

TH2012PEST1108.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02886838, version 1

Collections

Citation

Hai Anh Tran. QoE-based Content Distribution Network architecture. Other [cs.OH]. Université Paris-Est, 2012. English. ⟨NNT : 2012PEST1108⟩. ⟨tel-02886838⟩

Share

Metrics

Record views

67

Files downloads

38