Improve the behavior when the BWE times out and when we have too little data to determine the incom…

Improve the behavior when the BWE times out and when we have too little data to determine the incoming bitrate.

This is done by changing the RateStatistics so that it resets its window when the accumulator is empty. It also keeps a dynamic window, so that the rates computed before a full window worth of data has been received will be computed over a smaller window. This means that the rate will be closer to the true rate, but with a higher variance.

BUG=webrtc:5773
R=perkj@webrtc.org, sprang@webrtc.org

Committed: https://crrev.com/fb8fc5391e097dc4c2f89b3a38724f5fa109d8f9
Cr-Commit-Position: refs/heads/master@{#12470}

[stefan-webrtc, 2016-04-21 14:08:17]

stefan@webrtc.org changed reviewers:
+ perkj@webrtc.org, sprang@webrtc.org

[stefan-webrtc, 2016-04-21 14:10:52]

sprang for the main review since he has written most of the RateStatistics
class.

perkj for base approval, feel free to review as detailed as you wish.

[stefan-webrtc, 2016-04-21 14:10:52]

sprang for the main review since he has written most of the RateStatistics
class.

perkj for base approval, feel free to review as detailed as you wish.

[perkj_webrtc, 2016-04-22 05:24:58]

Can you please add your self and sprang as owners of the files you modify. You
can have per file ownership in base.

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics_u...
File webrtc/base/rate_statistics_unittest.cc (right):

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics_u...
webrtc/base/rate_statistics_unittest.cc:60: uint32_t kExpectedBitrate = 8000000;
nit: const uint32_t

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics_u...
webrtc/base/rate_statistics_unittest.cc:108: uint32_t kExpectedBitrate =
8000000;
dito

[sprang_webrtc, 2016-04-22 10:36:58]

lgtm with nits

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics.cc
File webrtc/base/rate_statistics.cc (right):

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics.c...
webrtc/base/rate_statistics.cc:45: assert(now_offset < num_buckets_);
dcheck

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics_u...
File webrtc/base/rate_statistics_unittest.cc (right):

https://codereview.webrtc.org/1908893003/diff/1/webrtc/base/rate_statistics_u...
webrtc/base/rate_statistics_unittest.cc:123: now_ms += 501;
Can we add a named constant for the 500ms window? Looks like there are other
values that could be inferred rather than explicitly stated as well?

[stefan-webrtc, 2016-04-22 11:07:09]

Added owners.

[stefan-webrtc, 2016-04-22 11:07:22]

perkj, ptal

[stefan-webrtc, 2016-04-22 11:08:09]

The CQ bit was checked by stefan@webrtc.org to run a CQ dry run

[commit-bot: I haz the power, 2016-04-22 11:08:17]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1908893003/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1908893003/40001

[perkj_webrtc, 2016-04-22 11:29:26]

lgtm

[commit-bot: I haz the power, 2016-04-22 13:08:51]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-22 13:08:52]

Dry run: Try jobs failed on following builders:
  android_arm64_rel on tryserver.webrtc (JOB_TIMED_OUT, no build URL)

[stefan-webrtc, 2016-04-22 13:48:41]

Description was changed from

==========
Improve the behavior when the BWE times out and when we have too little data to
determine the incoming bitrate.

This is done by changing the RateStatistics so that it resets its window when
the accumulator is empty. It also keeps a dynamic window, so that the rates
computed before a full window worth of data has been received will be computed
over a smaller window. This means that the rate will be closer to the true rate,
but with a higher variance.

BUG=webrtc:5773
==========

to

==========
Improve the behavior when the BWE times out and when we have too little data to
determine the incoming bitrate.

This is done by changing the RateStatistics so that it resets its window when
the accumulator is empty. It also keeps a dynamic window, so that the rates
computed before a full window worth of data has been received will be computed
over a smaller window. This means that the rate will be closer to the true rate,
but with a higher variance.

BUG=webrtc:5773
R=perkj@webrtc.org, sprang@webrtc.org

Committed:
https://chromium.googlesource.com/external/webrtc/+/fb8fc5391e097dc4c2f89b3a3...
==========

[stefan-webrtc, 2016-04-22 13:48:43]

Committed patchset #3 (id:40001) manually as
fb8fc5391e097dc4c2f89b3a38724f5fa109d8f9 (presubmit successful).

[commit-bot: I haz the power, 2016-04-22 14:05:50]

Description was changed from

==========
Improve the behavior when the BWE times out and when we have too little data to
determine the incoming bitrate.

This is done by changing the RateStatistics so that it resets its window when
the accumulator is empty. It also keeps a dynamic window, so that the rates
computed before a full window worth of data has been received will be computed
over a smaller window. This means that the rate will be closer to the true rate,
but with a higher variance.

BUG=webrtc:5773
R=perkj@webrtc.org, sprang@webrtc.org

Committed:
https://chromium.googlesource.com/external/webrtc/+/fb8fc5391e097dc4c2f89b3a3...
==========

to

==========
Improve the behavior when the BWE times out and when we have too little data to
determine the incoming bitrate.

This is done by changing the RateStatistics so that it resets its window when
the accumulator is empty. It also keeps a dynamic window, so that the rates
computed before a full window worth of data has been received will be computed
over a smaller window. This means that the rate will be closer to the true rate,
but with a higher variance.

BUG=webrtc:5773
R=perkj@webrtc.org, sprang@webrtc.org

Committed: https://crrev.com/fb8fc5391e097dc4c2f89b3a38724f5fa109d8f9
Cr-Commit-Position: refs/heads/master@{#12470}
==========

[commit-bot: I haz the power, 2016-04-22 14:05:51]

Patchset 3 (id:??) landed as
https://crrev.com/fb8fc5391e097dc4c2f89b3a38724f5fa109d8f9
Cr-Commit-Position: refs/heads/master@{#12470}