Re: satisfied customer

["Andrew C. Esh" <andrewes@cnt.com>]

  W. Curtiss Priest wrote:
  > 
  > Matthew Sparby said, regarding ISDN:
  > >It doesn't do any modulation at all
  > 
  > My communication theory is rusty, but I thought that certain phase shift
  > encoding schemes were inherently more noise immune and better reached
  > toward the theoretical limits of the bandwidth available than simply
  > sending a "digital signal" (a simple rise and fall in voltage on a
  > wire).
  
  Try to stop concentrating on the analog aspects of this. That part of
  the equation falls out when you're talking about ISDN, because of the
  change to full digitial Customer Premises Equipment.
  
  Try to think in terms of distance. The reason POTS is analog from the
  CO to the customer is because A) The end user equipment is cheaper, and
  B) The distances are too far for older high speed digital lines like
  DS1 to work. Besides, the customer is perfectly happy with his 3000Hz
  of analog bandwidth. All he needs it for is phone calls, anyway.
  
  And: The real problem is all that old copper wire that runs from
  the CO to the customer. It can't all be replaced with fiber (not yet,
  anyway). The old copper isn't all that good for running digital.
  
  That was then, this is now.
  
  Now we have cheaper CPE which can be used to digitize a voice call.
  We want to do this because we can also use the topology for digital
  data, which is gaining popularity as people connect to the Internet.
  
  Another thing has changed: The phone company is putting fiber in the
  ground, and is starting to put equipment into what are called
  "Distribution Areas". This means there is digital service from the
  CO to a box buried in your neighborhood. That box can serve 10-20
  houses. Since the box is close, the short range/high speed digital
  signalling techniques work. Those techniques have also improved,
  so a higher digital speed can be supported at a longer range than
  was possible as little as five years ago.
  
  Since we are now in range, and have the cheap CPE which does the
  job of digitizing, we can now begin to run digital from end to end.
  This means all the digital equipment in the CO, and between the COs,
  can be more efficient. They can detect bad data, and correct it. They
  don't have to blindly transmit digitized noises, both intentional and
  unintentional. Almost every bit being carried is effective, since
  it's been carefully checked by the equipment from the end user to the
  end user.
  
  What's wrong with analog:
  
  The problem with the analog world, is that there is no way to check
  the analog signal with 100% assurance that what comes out the other
  end of the line is exactly what went in. There are filtering and
  noise cancellation methods, but they can only be applied in a general
  fashion.
  
  The problem with a 28.8K modem is that it has to live in the imperfect
  analog world. The electrical noise made by your neighbor's garage door
  opening is happily transmitted along the same line as your phone signal.
  At the far end of the line, the equipment digitizes that sound mix, and
  sends it along to the remote modem, which also receives a mix of moises
  from its local line. the result is that much of the data that arrives is
  unusable.
  
  Maybe a simple analogy would help: analog electrical signalling is like
  using a firehose to move water. Digital is like a pipe. Taking piped water
  and spraying it at the CO will get most of the water there, but some will
  miss, and whatever is falling as rain will get added. The CO catches it and
  pumps it through a pipe to the remote CO. It all gets there. Then the remote
  CO sprays it out to your house, and the loss/corruption problem happens
  again. Effectively, you wind up getting drenched. What a digital signalling
  medium being extended to the house is doing is extending the pipe all the way
  to the customer. Not a drop is lost, and nothing is added. 
  
  Let's be digital:
  
  Since a fully digital system can check the data at every point, bit errors
  can be detected and eliminated. The data that arrives at the far end is
  entirely free of errors. You may argue that this is what a 28.8K modem is doing,
  but with ISDN, the CO-to-you portion of the circuit is under agreeable control,
  and is engineered to provide a certain level of digital service. On an analog
  line, the CO equipment is not cooperating in a digital manner, and cannot check
  or correct what it receives. It just sends the noise along to the other end.
  
  What this may mean is that the phone system is about to become a lot more
  efficient, but that efficiency will be eaten up by higher throughput demands.
   
  > Is this true of ISDN that if I took a scope and looked at the
  > PC bus where the digital signal was passing, and then looked on the
  > ISDN line, I would see the same, identical wave form?
  
  Probably not, but it doesn't matter. This is a paradigm shift. Stop
  comparing apples to oranges.
  
  I've attached a logic analyzer ("scope") to an Ethernet LAN captured the
  waveform of a packet being transmitted. I decoded the bits from the
  scheme they used to endcode them. I understand how the analog portion
  of that system works. It works well, but at it's base, it's still analog.
  The point at which that system becomes useful to me is above the digital
  level. I don't listen to the packets, I use my Ethernet card, and my
  computer networking stack (software) to decode them. From my experience,
  I also know that the Ethernet analog waveform is different from RS-232,
  T1, and POTS voice. It's easy to guess that it's different from many other
  forms, as well.
  
  Think of it like a layer cake. At the bottom is the analog copper wire
  (or fiber, or microwave). As long as we are only transporting sounds, we
  can put them directly on the bottom layer of the cake. Also on top of
  that layer, we can encode digital communication. Along with that encoding,
  we can provide other services like error detection, error correction,
  and compression. If we use CPE to digitize sound, then we can still make
  a phone call, but it gets carried on the top layer of the cake, above all
  the digital services. The other services that can be carried on that top 
  layer, such as data, cannot be carried at the bottom layer effectively. 
  They need those services which are supplied by the middle layer, and are
  performed by the equipment in your ISDN "modem", and by all the CO equipment.
  
  -- 
  Andrew C. Esh			mailto:andrew_esh@cnt.com
  http://www.mtn.org/~andrewes - ACE Home Page