[e2e] TCP Loss Differentiation

[Detlef Bosau]

David P. Reed wrote:
> The primary causality I was trying to reflect by that choice of words 
> is that in all systems there are large-scale causality relationships 
> (application layer or
I think, "large scale" is the very point.

When the drop/loss ratio is approximately constant over a certain period 
of time, it does not really matter whether the sender's individual
response to a loss event is correct or not. As long as the sender does 
congestion recovery "adequately often", anything is fine.

                                                                                                                                                         

> user layer, in most cases) that break the fundamental assumption that 
> there are a collection of memoryless processes in the system.  These 
> are really significant in most networks, ignored by most models:
>
> - users behave differently when networks get slow (they go to the cafe 
> or panic and start hitting keys harder and faster).

This exactly reflects my "bus stop scenario". A user may move some feet 
to the left or to the right - and even this may result in a Rayleigh 
fading maximum or a Rayleigh fading minimum. And we did not even talk 
about other causes for fading or interference.

> - wireless network transmissions are a primary source of noise to 
> other transmissions, so any correlations or dependence can get 
> amplified when noise causes retransmission by other nodes, causing 
> more noise, ...  the probability that wireless networks have modes of 
> highly synchronizing "resonances" that correlate rather than 
> decorrelate signals is high - and can be used to increase SNR by 
> techniques like analog network coding (zigzag, for example) if you 
> realize that the phenomenon is not a noise process at all, since it 
> adds no uncertainty.

That's, what's continously ignored by quite a lot of work.

Now, the problem for me is twofold.

First: If an end-to-end loss differentiation in a highly dynamic mobile 
networking environment is not feasible, this would be helpful for the 
little sketch
I submitted recently. (And now, I'm eager to read the reviewer comments.)

Second: I'm still thinking, how an alternative approach can be 
evaluated. All the simulations I know so far assume Markov models, 
Gilbert Markov models,
Jake's model and the like, which exhibit a more or less stationary or 
quasi stationary behaviour, particularly on large time scales.



-- 
Detlef Bosau                          Mail:  detlef.bosau at web.de
Galileistrasse 30                     Web:   http://www.detlef-bosau.de
70565 Stuttgart                       Skype: detlef.bosau
Mobile: +49 172 681 9937

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3351 bytes
Desc: S/MIME Cryptographic Signature
Url : http://mailman.postel.org/pipermail/end2end-interest/attachments/20090314/cd90631e/smime.bin