Rob [-_-_-_-_-_- WHY SDMT -_-_-_-_-_-]
Robert Hawley > By the way, who administered the tests that gave rise to your "CAP Leaks!" quip? Was it an independent group, or a group of DMT backers?
I'm guessing that this "demonstration" was orchestrated by the DMT
camp just as surely as other "demonstrations" will be orchestrated
by the CAP camp. There is a lot of money to be made here, so people are willing to tailor the "truth" to further their own agendas.
British Telecom administered the tests of a VDSL/CAP modem that doesn't notch out the amateur radio bands. This was in order to test weather amateur radio bands are affected by VDSL. An audio tape was made of the test, an it was played publicly a number of times.
The tape starts with a Ham operator speaking, the sound very clear. They then turned on the CAP modem, and the Ham operator's signal was completely wiped out. Only static was heard. Then the CAP modem was turned off, and the Ham operator once again came in loud and clear.
This process was then repeated, showing the same results.
The BT experiment/tape plainly illustrates that amateur radios bands need to be avoided by VDSL modems. The CAP proponents continue to play this down, including my discussions with Paradyne at SuperCom 97.
I'm not trying to be the champion of single-carrier VDSL. I'm just
trying to point out that there are proponents on each side. I find
it interesting that people who backed multi-carrier ADSL are now
backing single-carrier VDSL and wanted to bring up the issue. Maybe
they thought DMT VDSL was not feasible, but Amati has now solved
these problems by going half-duplex (ping-ponging)? I have no idea. I was hoping that someone with more knowledge might comment.
I fully admit I'm not a Comm. Engineer, I have an Electronics, Flight Simulation and Computer Engineering background. Though communications is progressing more in my direction every day. I fully admit my academic background does not even come close to the Stanford engineers working for Amati.
Though it does not take a Rocket scientist to understand the fundamental concepts, one only needs a reasonable electronics background. I wish we had participation from some of the Amati or other ADSL company engineers, since this is not the case, I will try to supply what information I can, in order to help others better understand the issues, and technologies at hand.
If however some CAP proponent Engineers were to post, I would tend not
to believe everything I hear. Having found many untruths in the CAP white paper, and what they say to you face to face.
I am always open for discussion with a knowledgeable individual.
No disclaimers this time around.
I will try to answer any questions others might have, and if needed
I could go into more detail than is presented below in some areas.
But I will not fall into a tit for tat debate on the "facts" presented
with anyone who does not have a background in electronics.
I spent ample time ensuring the information provided is accurate.
I have tried to present the following using a few simpler terms and wording, in hopes that most here might be able to follow along. Though I understand it may not work out that way. The information can not be simplified to much, otherwise it loses all credibility.
WHY SDMT for VDSL
One of the fallacies I find over and over in the CAP point of view, is
the comparing CAP and DMT as not being that different, and both
technologies can achieve the same end, optimizing the transmit
bandwidth.
This is true in theory, but is not true in the real world.
In theory you are allowed infinite complexity, but in the real world this is not feasible in many instances, in other instances, it can not be accomplished an efficient, and cost-effective design.
At this point to save me a few thousand words, please see my post Why DMT techstocks.com
or please reference Dr. John Cioffi's Article
internettelephony.com
which explains how DMT achieves it's desired results, if needed.
With that said -- Robert let me try to help you and others understand why CAP cannot compete with DMT in the VDSL arena.
In order for CAP to achieve the same optimization as DMT, CAP/QAM must rely on transmit filters and receivers to remove energy in the bands.
Determining how well the energy is removed is directly proportional to
the number of taps used in the filters, this is limited by the complexity (which equals cost) of the system. Generally speaking, hundreds of taps are required to sufficiently remove energy beyond the
useful band.
Over and over again we have heard how much more complex DMT is in
comparison to CAP. ** Well pay close attention Amatians, do to the
complexity/cost constraints only a fraction of the required number of
taps are used in CAP. For this reason CAP modems may operate on short
distances across copper with few problems, the number of taps are
adequate. But on longer distances CAP modems fail catastrophically. On longer lines CAP/QAM confine the transmitted signal to the useful
frequency band, this allows the reliable signal in the correct part of
the band, to be corrupted by the bad signal in the wrong part of the
band.
Loop lengths are shorter for VDSL, and after the above description, it
might seem logical for one to assume this would be an advantage for
CAP. Sorry to disappoint some of you.
Though the loop lengths are shorter, the frequencies used are much
higher in VDSL. VDSL using the 11 Mhz and above, where ADSL uses 1.1
Mhz. Now what problems may CAP in counter one might ask? The higher in frequency the more the signal is effected by noise!
The copper infrastructure in place is not shielded, (of course I mean
the twisted-pair telephone lines and not shielded coaxial cable) and
leakage occurs "into" unshielded copper, this commonly known as "ingress" and leakage "out of" copper, is known as egress. Let's start with ingress. VDSL modems must be able to operate in with the presence of over-the-air radio frequency interference (RFI) entering the copper bundles and smaller lines running into homes and offices. I'm sure many are familiar with a Die-Pole antenna (looks like a T) that most have connected to their stereo FM Receivers.
Sure increases your signal strength right. Those radio signals are
picked up just fine on that short 6ft little piece of copper. Now try
to imagine what a huge antenna that line running from the pole to your
house makes, or the lines running along miles of telephone poles to your neighborhood. So VDSL must be tolerant of in-the air transmissions (RFI) - see Ref. in Why DMT).
In AM/FM radio the noise is at a fairly constant rate. In the case of
Ham Radio, the signal is not always there, only when the operator keys
the mike. Before moving on to egress, lets talk about another form of
ingress noise that effects VDSL. Anyone have an Air Conditioner,
Refrigerator, Dishwasher, washer/dryer, or a hair-dryer in the house.
Each time one of these devices cycles on or off it causes noise spikes that can easily effect the signal of VDSL modem. Even a simple light switch can have an effect. This is known as Impulse Noise.
DMT being a multiple carrier having 256 channels, the effect of impulse noise is dispersed across the entire 256 channels, the effect is so minor on each channel it does not cause any interference to the signal. (this is the most simplest way I can explain how it works)
There is no added complexity with the DMT method of handling the above described impulse noise.
CAP/QAM is a "single carrier" signal which cannot handle impulse noise as simply, it must use old Error correction coding, and interleaving of the transmitted signals. Once again this is the simplest way to explain the way it works not necessarily the best, I am writing to the audience, not T1E1.4, nor would I attempt to.
CAP handles RFI by using transmit filters, and receiver equalizer
filters, and must create notches in the bands to be avoided. Again the
complexity of incorporating enough of these filters to remove the noise would be cost prohibitive, when we take into account that many amateur radio bands overlap the VDSL spectrum, is it any wonder that the CAP proponents avoid discussing, this issue, in fact they are promoting the fact that Amateur bands are not of concern. Well I'm here to tell you they are, and even more so in Europe where amateur radio dwarfs the U.S. numbers.
Egress
For CAP's problem with Egress or leaking from copper lines, see the
above paragraph about British Telecom's test. It does not make for
happy campers in Europe, nor in the U.S.
How does DMT cope with Egress, short and sweet, they turn off the
channels that have an affect the Amateur Radio and Short-wave bands. See References above for a more in-depth description.
Maybe they thought DMT VDSL was not feasible, but Amati has now
solved these problems by going half-duplex (ping-ponging)? I have no idea. I was hoping that someone with more knowledge might comment.
Robert, I will continue hopefully answering your question.
As I mentioned in my previous post ping-pong DMT for VDSL is known as
SDMT. The "S" meaning synchronized. SDMT is Low Complexity
Here, there is only one Fast Fourier transform (FFT) per modem. ( FFT - see Ref. in Why DMT).
This one FFT is shared by the transmitter and the receiver of the modem, as only one of the two are on at a time. This is less complex than a modem, which has a separate hardware for both the transmit and receive functions, as is the case with CAP/QAM. It should also be noted that since SDMT has only one transmit and receive band, it only needs one analog filter. SDMT's power consumption is "very" low. ( I do not know exactly, at this time)
Another excellent feature of SDMT is that the same VDSL modem is able to be configured for either Asymmetric or Symmetric transmission.
Once again the CAP modem is at a disadvantage, it must change the
bandwidths, of the downstream and upstream bands in order to change
symmetries.
I must say to those who say that CAP is the better choice for VDSL, they have little imagination, and even less patience to use their engineering background to it's fullest potential, when trying to arrive at the solution for the RFI and other interference's on VDSL transmissions.
I thought about providing a summary, but then thought if any one needs one, it would be easy to create one as the key points stand out on their own.
Perhaps someone may provide one.
Amati's Engineers are undoubtedly some of the best in the world, though in DSL, there is no question in my mind, they are the best. It would appear that some of the largest DSP manufactures in the world share my convictions. In time I feel confident that we will see that praise, stated in much more flattering verse, in a forum far superior to this one. Editors note: This statement was written last week before TI spoke at the Montgomery Conference. Thanks Peter Piper. I'm sure more will follow.
In closing I would like to attempt to bring home the points covered
above, in a much simpler terms, this is for those who find the above confusing.
Since many of you may be audiophiles, or at least enjoy the sound of a
decent stereo system.
Pull out an old 45rpm record or an old Album. Turn the volume way up, and play it. Now don't those Clicks and Pops make you cringe, it's almost sounds as if they could be damaging to your speakers.
This is your DATA on CAP, any questions?
Now put on a CD, I suggest Life in The Fast Lane, by DMT.
JW@KSC |
