[e2e] TCP fairness

[Krause, Matthias]

Dear community,

I'm afraid I'm stuck with some artifacts I have to be able to explain and maybe even control.
I appreciate every small hint in a good direction either towards a paper or to the right existing reference.

While experimenting with video streaming over TCP in equilibrium situations, I found several artifacts.
The environment was as follows. 3 video streaming servers (Windows XP) based on TCP were streaming to 3 receivers
in a  local network over a bottleneck node.
Variations on the this setup were {wireless | wired}, {receiver Linux| receiver WinXP}, distance from
base station (in wireless), number of  network nodes in between, inclusion/exclusion of a token bucket filter
node for bandwidth limitation.

In all the experiments, I permutated the receivers, and run each permutation for 20 to 50 times.
The streamed video sequence was 6 Mbit/s and 560 seconds long.

The artifacts I achieved were:


*         In sum, the bitrates achieved were not fair (as in 'close to equal')

*         The bit rate ratios stayed stable most of the time.

*         Sometimes ratios switch, turning the best performer into the worst performer.

I wonder if these phenomena are known and studied or not even worth the effort thinking of it.

Thank you very much for your time and consideration and I'm looking forward to hearing from you.

Yours sincerely,

Matthias Krause
Philips Research


________________________________
The information contained in this message may be confidential and legally protected under applicable law. The message is intended solely for the addressee(s). If you are not the intended recipient, you are hereby notified that any use, forwarding, dissemination, or reproduction of this message is strictly prohibited and may be unlawful. If you are not the intended recipient, please contact the sender by return e-mail and destroy all copies of the original message.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.postel.org/pipermail/end2end-interest/attachments/20090731/a74ea38a/attachment.html