[Bloat] pacing, applied differently than bbr

[Taran Lynn]

On 2/26/20 12:12 PM, Dave Taht wrote:
> As for the actual content of the paper... umm.. code is where?
The code can be found on out site here
https://lambda-11235.github.io/control_theo_network/projects/. Note hat
the code there is the implementation used in the paper. We have since
made significant changes, but will wait to publish the code until we
publish our results.


On 2/26/20 1:00 PM, Jonathan Morton wrote:
> Hmmm.  I see some qualitative similarities to BBR behaviour, but the
> algorithm doesn't seem to be very robust since it seems to improve a
> lot when given approximate a-priori information via cap and collar
> settings.

Current work is focused on improving robustness in this area, and some
of our recent work has already seen some improvements.


On 2/26/20 1:00 PM, Jonathan Morton wrote:
> How does it treat ECN information, or does it set itself Not-ECT?

The algorithm currently only uses delay based feedback. It does not use
loss or ECN information.