ICCRG meeting, IETF 98, Chicago, IL, USA
Monday 27th, 9:00-11:30, Zurich E/F


20 (est. 15 +5) min Hannu Flinck: Throughput Guidance 
draft-flinck-mobile-throughput-guidance-04

20 (est.) min Tommy Pauly: ICCRG + TAPS: On deploying new algorithms

20 (est. 15 + 5) min Lin Han: Problem Statement: Transport Support for Augmented and Virtual Reality Applications *

35 (est. 25 + 10) min Koen De Schepper: TCP-Prague ideas and experiments

35 (est. 20 + 15) min Neil Cardwell and Yuchung Cheng: An Update on BBR Congestion Control **

20 min: Praveen Balasubramanian: Reflections on Congestion Control


*
AR/VR applications need both low latency and high throughput. We shall see that the ratio
of peak to mean bit-rate makes it extremely challenging to hit both targets, even as
a niche managed service. The greater challenge is how to hit both targets routinely -
for the mass-market. This either needs both low latency and highly variable bandwidth
to be feasible as part of the regular unmanaged Internet service, or it needs simpler
managed QoS services that would be available routinely for everyone.

We look forward, quantifying the possible scale of the problem in about
5-10 years (2022-2027). The problem for the transport community can be reduced by
innovations at the application layer in coding, compression, and prediction. Also,
it would help if access networks were designed for a higher degree of multiplexing
between users. This first draft sizes the worst-case problem for the transport community
(which seems near-impossibly hard). The aim is to use the first draft to draw out experts
in application-layer and access-network design to quantify how much of the problem
they can absorb. This will help us improve our estimate of the size of the residual
transport layer problem in future draft(s).


**
An update on the BBR congestion control algorithm, including
experiences with its deployment at Google and YouTube.