Estimation of in‐service quality of experience for peer‐to‐peer live video streaming systems using a user‐centric and context‐aware approach based on Bayesian networks

Abstract

In this paper, a modified and improved quality of experience (QoE) estimation approach has been presented on the basis of the user‐behavior modelling and context‐aware modelling. We have implemented this proposed approach for peer‐to‐peer live video streaming systems because of the dynamically adaptive and distributed nature of these systems. The problem of estimating QoE is formulated as a Bayesian network model using user‐behavior and context‐related variables as nodes. The relationship between these nodes are established to find out estimated QoE, which is based on the user‐behavior data and context data, and dynamically adapt to any change in their values. This relationship is further explored to provide the basis for more user‐centric, context‐aware and dynamically adaptive services and network systems as this in‐service QoE is important for network optimization and service provisioning. Copyright © 2013 John Wiley & Sons, Ltd.

A hybrid anycast network architecture and intelligent server selection strategy for multimedia services

“…The problems with route‐by‐name anycast are listed as follows: (i) server selection is solely based on routing metrics, so that it lacks flexibility in server selection; (ii) anycast addresses or name prefixes have to be published into the entire routing system, which will inflate routing tables and significantly threaten routing scalability ; (iii) in name‐based routing architectures such as DONA, NDN and SCAFFOLD, flexible‐length names degrade routing efficiency ; and (iv) it is difficult to support flow affinity , that is, to direct sequential requests within a session to the same server. This is indeed a basic feature for stateful multimedia services .…”