Adaptive Media Streaming - Scalable Continuous Media Streaming Systems - page 141

8

Adaptive Media Streaming

In this chapter we consider networks where only best-effort service is supported. This type
of network, including the Internet, does not guarantee the delivery of data from the media
server to the media client. Moreover, even if the data are successfully transported to the
client, the time it takes can vary. The media server has no control over the amount of
bandwidth available and can only use whatever is available at the time. Obviously, it will
be very challenging to stream high-quality media over such best-effort networks and at the
same time maintain playback continuity.
As the network provides no guarantees whatsoever, the only alternative is to adapt the me-
dia stream to the network bandwidth available. This is made possible by using layered-video
codec or media transcoders as discussed in Chapter 2. In this chapter, we develop an adap-
tation algorithm to adjust the output bit-rate of multi-layer-encoded media or transcoders
according to the measured available network bandwidth and the estimated client buffer
occupancy. With this adaptation algorithm, the media server can then adjust the bit-rate of
the media stream dynamically (e.g., reducing bit-rate when available bandwidth drops) so
that playback continuity at the client can be improved.

8.1 Introduction
The lack of end-to-end quality-of-service (QoS) support in today's Internet has caused signifi-
cant difficulties for the deployment of media streaming services such as video broadcasting
and video-on-demand. In particular, when the network becomes congested, significant packet
losses will arise, leading to corrupted or even dropped video frames.
Given QoS support is unlikely to be widely available in the near future, researchers have
resorted to another approach to tackle this problem. Specifically, a number of pioneering
researchers have investigated algorithms to adapt the video bit-rate to the network bandwidth
available [1-6]. For example, when the network becomes congested, the sender will reduce the
bit-rate of the encoded video to alleviate the congestion. Clearly, reducing the bit-rate will also
degrade the visual quality. Nevertheless, reducing the video bit-rate in a controlled manner at