[tele.com test Methodology]
>>Steve: I can't say if round 1 or round 2 provided those numbers in the test.
Well than I wouldn't make any conclusions from this test about system performance over long noisy Loops. Trust the T1/E1 committee. They really did their homework.
>I think I'm
missing the point about when a Westell modem uses 7.2 watts and the power sent
down the line from the modem transmitter.
Looks like the tele.com test people did too. Bellcore got it right.
>This wouldn't have shown up in the ~13
different tests by Teledotcom on performance?
It only shows up if you know to look for it.
> Should I concluded that a DMT
modem places less power on the line than a CAP modem?.. until later.
Rupert did a great job on this issue (see below). Bottom line Bill, you can get 7Mbps over a wet noodle if you tweak the transmit power. It doesn't mean anything if it breaks the wire plant. Not measuring the PSD of each modem on each loop is a fundamental flaw of any loop performance test. If the tele.com test does not disclose these numbers than I would place the issue next to the toilet. It could come in handy some day.
Power Spectral Density and Spectral Compatibility
Spectral compatibility may sound like a bureaucratic issue; but it is not. It is of great
importance for any system that is going to be deployed in a real network. Spectral
compatibility defines how much energy a system will put out - this controls to what
extent it will interfere with other systems trying to use other copper pairs in the area.
A badly designed system can quite literally prevent other users from communicating.
This cross-talk will easily affect other copper pairs in the same binder group and can,
in severe cases, affect pairs in another binder. As the network moves towards
deregulation, having clear and universally respected rules becomes even more
important, for with many operators in the same space, accessing different nodes in the
network with a proliferation of different suppliers and technologies the possibility of
such cross-talk is bound to increase. Such interference could easily prevent a company
from supplying a service or connecting to a customer - raising a lot of potential for
conflict and litigation. The only way to prevent such conflict (or to avoid the lawsuits) is
to have strict guidelines that are universally followed.
To prevent this conflict happening, each standard defines a "PSD mast", describing
what power spectral density the system can employ for upstream and downstream
transmissions by frequency. This mask was established to allow ADSL to have
reasonable rates, but to allow compatibility with other services and reasonably easy
deployment. If a system does not obey such masks then an operator risks damaging
other services - perhaps services that people are paying premium rates for (e.g.
guaranteed throughput services using HDSL or Frame Relay).
In fact, as there has been more deployment, these masks have been tightened to ensure
co-existence. For example, there used to be a `high frequency boost' option (26dBm)
to allow for longer reach; that has now been blocked. Or the roll-off above 1.1MHz
has been made more stringent, in order to be compatible with VDSL, which had not
been a concern when the mask was first described. (These changes are not a problem
for existing modems designed to T1.413, which already supported such operations).
This PSD mask was defined by T1E1 for ADSL as a generic technology, and not of
any one line code - i.e. whatever the implementation, it must meet the requirements and
strictures to not cause interference. This was established by all concerned, and was
one of the first contributions formally agreed on in the ad hoc group [8] (this group is
discussing CAP as a technology for ADSL, and must follow the interference guidelines
already for ADSL). Unfortunately, the current CAP document and existing
implementations do not follow this mask, and breach the agreements.
For the upstream, rather than rolling off at 140 kHz CAP continues up to 180 kHz.
That additional 40 kHz overlaps directly with other technologies, especially - and
fatally - with standard DMT ADSL. As a result, CAP ADSL as currently defined
cannot be deployed in the same binder as ANSI standard ADSL. A similar overlap
occurs at the upper end of the downstream band: instead of rolling off at 1.1 MHz this
continues to 1.5 MHz - severely interfering with VDSL systems.
Specifically, T1.413 states the following: for the upstream direction the passband
region is 25-138 kHz. Within this band the power spectral density (PSD) is specified
to be -38dBM/Hz, and then drops so that above 181 kHz, the PSD is required to be
at least 24 dB below -38 dBm/Hz, i.e. at a -62 dBm/Hz level.
The CAP document proposes the following, and breaches the allowed mask
ú An upstream passband region of 55.4-171 kHz.
ú T1.413 which specifies that the passband ends at 138 kHz - not at 171 kHz, meaning
that the CAP upstream violates space reserved for downstream transmissions
ú A power level of -41 dBm/Hz at 181.2 kHz
ú T1.413 where the requirement is for -62 dBm/Hz at frequencies above 181 kHz
ANSI review contributions (8,9,10) quantify for specific loops the degradation caused
by the proposed CAP/QAM system on standard compliant DMT systems. The impact
will be severe:
ú a loss in data rate of hundreds of kilobits
ú a margin reduction of 5 dB
This will sabotage the performance of standard DMT to such an extent it is unlikely it
could be deployed in the same binder. A standards compliant DMT system is `better
behaved', and will not interfere with other technologies (nor with a CAP system).
The 97-134 proposal for CAP/QAM does not meet the spectral compatibility
requirement agreed by the group per T1E1 Ad Hoc/97-11.
Similarly, on the downstream, CAP continues out to 1.5 MHz, instead of 1.1 Mhz. In
other words, CAP breached even the older, more relaxed standards - not to mention
the new "VDSL-friendly" ones that DMT now complies to. This extra bandwidth
places it directly into the VDSL band and causes significant interference and cross-talk
there. It appears (11-12) that a single ADSL system with these properties would
dominate all other noise sources and potentially make VDSL unusable on all but the
very shortest of loops.
Both of these arise from the fact that CAP is less efficient that DMT, requiring more
`space' and signal power to get the same data rate. Further, CAP still specifies the
power boost option, now removed from T1.413. It appears from (5) that rather than
merely being an option, CAP actually requires this boosting if it is to achieve rates >
4.08 Mbps over a full-length CSA loop. This will introduce severe cross-talk,
compromising not only standard ADSL and VDSL, but impacting technologies such as
HDSL or S-HDSL - even T1 in an adjacent binder (13).
As such, it is clear that CAP's cross-talk renders it completely incompatible with a
large number of other solutions, including existing standard ADSL, and VDSL.
Power-boost was removed as an option from T1.413 on the grounds of not causing
cross-talk; it appears to be required for CAP (ironic for a technology promoted as
being lower power). Furthermore, the proposed technology is incompatible with its
own existing specifications for compatibility; where this has been discussed, the
specifications are relaxed.
later |
