[LINK] RFI: ADLS2+ Performance

[TKoltai]

> -----Original Message-----
> From: link-bounces at mailman.anu.edu.au 
> [mailto:link-bounces at mailman.anu.edu.au] On Behalf Of Craig Sanders
> Sent: Tuesday, 10 July 2012 8:23 AM
> To: link at mailman.anu.edu.au
> Subject: Re: [LINK] RFI: ADLS2+ Performance
> 
> 
> On Thu, Jul 05, 2012 at 08:42:54PM +1000, TKoltai wrote:
> > If in June 2011 they had 640,000 DSL customers and in July they had 
> > only 500K ports, that's a very respectable 20% floating 
> customer base.
> > 
> > In other words, a most reasonable Signal to Noise ratio.
> 
> until i read this, i thought that you're a kook but you do 
> have a good knowledge of the ISP industry and a reasonable 
> grasp of the technology behind it.
> 
> but what you're saying there makes no sense. what you claim 
> is SNR is actually customer (or device) contention, the 
> number of consumers competing for a given limited resource. 
> in this case, users:ports - 640K:500K isn't bad at all, 
> many/most users don't leave their computers connected all the 
> time even with "always on" ADSL.
> 
> it is not, however, anything like or at all related to 
> signal-to-noise ratio.
> 
> SNR refers to the strength of the SIGNAL (the useful 
> information being transmitted as seen by the receiver) versus 
> the level of background noise. i.e. it's a measure of the 
> overall quality of the connection path
> - whether that's wires or wireless.
> 
> To greatly simplify: a high SNR is "good". low SNR is "bad". 
> too much noise can reduce (or completely overwhelm) the 
> information carrying capacity of the channel.
> 
http://en.wikipedia.org/wiki/Signal-to-noise_ratio

An excellent reference source which was not written for the average Fred
Flinstone.

The term "Signal to Noise Ratio" can and is used for many different
purposes.
I believe that Roger meant it as per your Wikipedia Ref.
I meant it (partly) as - lookit all these spurious distractions. Too
many advertisements and not enough "content".

Sometimes my wit is overly subtle. Which makes me either witless or a
kook.
Feel free to choose the most appropriate label.

However, very occasionally, behind the wit is actual science.

The frequencies generated by the different connection speeds, local loop
tail resistance, constant rx/tx frequency changes, MDF crosstalk and
backplane traffic does indeed impede and affect the SNR on a DSLAM on a
nanosecond basis.
Modern DSLAMS attempt to "filter" out this noise... By calculating the
optimum "aggregate" frequency. With the addition of local modem "fine
tuning" available to consumers. However in reality, just as in the first
reference [1] (wireless related but essentially the nub of any digital
connection schema), data signalling is affected by the spectrum
efficiency within the "acceptable noise" permitted environment of
copper.

i.e.: we could increase signal and decrease noise considerably by
replacing all copper tails with silver or 18k gold... (And removing all
twists [twist tie and looping terminations] and bends from all
subterranean copper tails. - this removing resistance loops procedure is
often called "Sweeping the copper").

Other factors include even a 1 degree alteration in soil or pit
temperatures which can send the Noise up and the Signal down (by
expanding the copper due to heat) requiring more "Signal" from the DSLAM
which then lowers the aggregate filtering to all users on that DSLAM.
Resulting of course in lowered "Signal" and increased "Noise".
 
References:
[1] http://www.cnl.fi/mvarela/pdf/varela-mesaqin05.pdf
[2] And US Patent http://www.patents.com/us-6904082.html attempts to
addresses some of the issues.

Sorry, don't have time today to find fifty references but in conclusion,
all digital signalling for communications depends on frequencies being
transmitted along a carrier wave. (Copper, WiFi, Fibre...).  As such
they can be affected (distorted) by anything that emits frequencies -
e.g. a starter motor in a car on the street, a power transformer
switching from peak to off-peak, the ham radio guy down the street, an
FM radio station in a rain storm... A Nebula formation from 3.99
billion years ago, 4 billion miles from earth.

My knowledge of XDSL matters commenced in 1984 or 85 with a 19
kilometres 9.6/19.2K "swept" copper connection between Craig Crescent
coconut Grove and Stokes Hill Power Station in Darwin.

My problem then was the same as today, I didn't know it couldn't be
done... So we just did it.


Regards,


TomK