Re: [Bloat] a flood of Bufferbloat-related papers

[Lawrence Stewart <lstewart@room52.net>]

Hi David,

On 10/11/11 21:27, David Täht wrote:
> I sat down on my vacation last week thinking I would write up a review
> of progress since the bufferbloat effort began back in January, 2011. In

I think we should exercise a bit of caution in relation to saying things 
like "the bufferbloat effort started in Jan 2011" - the pre-2011 
literature alone (although not using the term "bufferbloat" 
specifically) extensively covers the issues, symptoms and a myriad of 
solutions for the problem.

> particular, I was interested in discovering to what extent we'd made the
> cross-over not just into other OSes besides Linux (e.g - BSD, windows)
> but into academia.

FWIW, I'm a FreeBSD kernel developer and a PhD student working on 
transport layer congestion control. I've been involved in doing a lot of 
experimental work related to bufferbloat (we refer to it as "collateral 
damage" in our papers) using FreeBSD and Linux.

> In the future I would certainly appreciate the authors of bufferbloat
> related/referencing papers to mention them on this mailing list *as*
> they are published!

Here are a few pointers to relevant papers done by people at the 
research centre where I'm studying:

"A rough comparison of NewReno, CUBIC, Vegas and ‘CAIA Delay Gradient’ 
TCP (v0.1)": http://caia.swin.edu.au/reports/110729A/CAIA-TR-110729A.pdf

"Revisiting TCP Congestion Control using Delay Gradients": 
http://www.springerlink.com/content/mq50134631115076/

"Multimedia-unfriendly TCP Congestion Control and Home Gateway Queue 
Management": http://dx.doi.org/10.1145/1943552.1943558

"Improved coexistence and loss tolerance for delay based TCP congestion 
control": http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5735714&tag=1

"Collateral Damage: The Impact of Optimised TCP Variants on Real-Time 
Traffic Latency in Consumer Broadband Environments": 
http://dx.doi.org/10.1007/978-3-642-01399-7_31


Off the top of my head, you should also check out the work done by Nick 
McKeown's group (although their focus has been on core routers): 
http://yuba.stanford.edu/buffersizing/

There's plenty of other relevant work that I can't think of specifically 
right now, but it's out there.

Cheers,
Lawrence