SNR Based Automatic Rate Control for Optimized Video Streaming over 802.11 by CLS

A fundamental limit of indirect statistics-based feedback is that it classifies link conditions as
either “good” or “bad.” This binary information provides some notion about the direction in which
to adapt the rate setting, but does not suffice to select the appropriate rate at once. This leads
to a slow step-by-step accommodation to large changes in conditions, and introduces the risk of
oscillation in stable conditions. A better approach is to use direct measurements of the link
conditions.The SNR is directly related to the bit error rate (BER) in the link, and hence to the
FER. Consequently, the SNR is linked to the packet delay and jitter, and throughput, and has the
potential to provide rich feedback for automatic rate control. Knowing the current SNR and the
throughput vs SNR curves for each rate setting (e.g., Fig. 2) would solve the rate selection
problem instantly.

Figure 2. Throughput vs. SNR for some 802.11  modulation schemes

Despite the advantages, SNR-based rate control has not been applied in practice yet because of the
following three problems:

1) In reality, for certain link conditions the relation between the optimal rate and SNR is highly
variable. This is due to the imperfectness of the models describing the radio channel.

2) It is not trivial to obtain a reliable estimate of the SNR of a link. Many radio interfaces only
provide an uncalibrated signal strength indication (SSI).

3) The rate controller, which is on the sending side, in fact needs the SNR observed at the
receiving side.

Most work on using SNR information for automatic rate control is based on simulation and does not
consider the practical difficulties of obtaining good SNR estimates. It concentrates on the way in
which the noisy and drifting SNR can be used to determine the correct rate setting to address the
issue of how to communicate back SNR values, but their rate selection algorithm still relies on a
straight SNR threshold technique. Another approach is discussed in, where the assumption is made
that the channel is symmetric, meaning that the SNR observed at either station is very similar for
any given point in time. This assumption allows to use the SNR of the last ACK frame as an indicator
of the SNR at the other side, and to use it for selecting the rate of the next data frame to be sent.

Advertisements

, , , , , , , , , , , ,

  1. Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: