Modeling Stability and Bitrate of Network-Assisted HTTP Adaptive Streaming Players

Abstract

Abstract-Viewers using HTTP Adaptive Streaming (HAS) without sufficient bandwidth undergo frequent quality switches that hinder their watching experience. This situation, known as instability, is produced when HAS players are unable to accurately estimate the available bandwidth. Moreover, when several players stream over a bottleneck link, their individual adaptation techniques may result in an unfair share of the channel. These are two detrimental issues in HAS technology, which is otherwise very attractive. To overcome them, a group of solutions are proposed in the literature that can be classified as network-assisted HAS. Solving stability and fairness only in the player is difficult, because a player has a limited view of the network. Using information from network devices can help players in making better adaptation decisions. The contribution of this paper is three-fold. First, we describe our implementation in the form of an HTTP proxy server, and show that both stability and fairness are strongly improved. Second, we present an analytical model that allows to compute the number of changes in video quality and the bitrate of a video stream. Third, we validate the accuracy of the model by comparing the modelbased estimations for the number of changes in video quality and for the mean bitrate of a video stream, with results in a real implementation of our HAS assistant. The results show that the model-based results are highly accurate. As such, this model is useful in practice for planning video delivery networks that use in-network HAS assistants, and enables us to analyze the stability and the mean bitrate of HAS streams prior to real deployment.

Enhancing Over-the-Top Video Streaming Quality with DASH Assisting Network Elements

“…Multimedia systems: A proxy server based implementation of a DANE is presented in [9]. HTTP traffic is routed through this proxy server which inspects the traffic to detect DASH flows and their characteristics.…”

A model for evaluating sharing policies for network-assisted HTTP adaptive streaming

“…For example, in previous work we showed that the number of changes in video quality can be reduced while improving the fairness between video streams [16] . However, the sharing policy did not take into account the types of streams or devices, and thus only represented fairness on a bitrate level instead of targeting an equal quality of experience while considering device specific factors.…”

“…These implementations range from traffic shaping at the residential gateway [13] , signaling players from a measurement proxy [14] , using OpenFlow [15] , and our implementation in the form of an HTTP proxy server [16] . Typically, these implementations target networks where the bottleneck link is in the local-or access network, and thus relatively close to the users.…”

“…However, building testbeds to determine the performance of a policy is costly and time consuming. In previous work, we proposed a model that allows to accurately estimate the bitrate of the video streams and the bitrate stability [16] , 1 and as such give an estimation of the quality of experience (QoE) of the viewer. The QoE of the viewer improves when the bitrate of the video increases, while the number of switches in bitrate should be kept low [17,18] .…”

“…However, SAND only specifies the communication between the HAS assistant and the players. This means that it is still up to the HAS assistant to decide how the avail- 1 Reference [16] presents an early version and evaluation of the model. In this paper we extend this model and use it to evaluate different sharing policies.…”