[opendtv] Bob likes COFDM

  • From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
  • To: "OpenDTV (E-mail)" <opendtv@xxxxxxxxxxxxx>
  • Date: Sat, 19 Mar 2005 18:18:45 -0500

Bob Miller wrote:

> Was this formula developed with the characteristics
> of COFDM propagation and reflection in mind? The
> more buildings the better from our experience.

Interesting point. What you're saying is that
reflections can help, if they're strong. True. There
are cases where reflections will add constructively
to perhaps give you more localized signal than even
free space propagation would provide at that distance.
But of course, beyond where that hot zone is, the
signal will be very weak now. Weaker than free space
propagation. That's where buildings hurt.

> I like COFDM, it bounces around, doesn't need a lot
> of power but allows you to work with short and
> multiple broadcast antennas and not interfere with
> the next station over. When you get to the wide open
> spaces you can have taller sticks and add power.
> Very versatile.

Yes, but moving beyond evangelism, the same can be
said for any system that uses equalizers or rake
filters. That's what these devices do for their
respective modulation types. They just need to be
good enough to get the job done.

Follow this. If you have a train of symbols
transmitted with no spacing between them, for
maximum spectral efficiency, what happens with
multipath distortion? What happens is that the
symbols now spread out over one another. Parts of
a given symbol are delayed more than other parts,
so you get so-called inter-symbol interference
(ISI). Also parts of the symbol are distorted to
be taller or shorter than the original, due to
constructive or destructive interference with the
symbols from multiple paths. Bad news. The symbols
can't be deciphered.

COFDM cleverly addresses this problem by creating
very slow symbols, and sticking a gap between each
symbol. As long as the spreading out of symbols
does not exceed that pre-determined gap, you're
good to go. Also COFDM cleverly creates many tiny
subbands to transmit these slow symbols in
parallel, and transmits redundant data among them,
so that deep notches in the spectrum that clobber
a few of these carriers can just be ignored.

What do equalizers do instead? Equalizers look at
the incoming symbol train and analyze how the
distortion happened to create this symbol overlap,
across the 6 MHz spectrum (in our TV example).
Then, with multiple hundreds of knobs, the equalizer
compensates for the delay that caused the ISI by
delaying other parts of the symbol, allowing the
delayed parts to catch up. This action also works to
restore the notches across the spectrum, because
after all, it was delayed components of the symbol
that created those notches. As long as nulls aren't
complete, the symbol should be restorable. Or at
least, a close enough facsimile to allow for demod.

So what happens downstream of the equalizer? You
once again have a train of symbols with no spacing
between them. Good deal. You've not had to give
up spectral efficiency, and you've survived the
multipath distortion.

As I've already described some time ago, when these
equalizers are applied to COFDM, e.g. as
STMicroelectronics has done in their latest COFDM
demod, the gap between symbols and the multiple
active carriers used in COFDM are no longer
necessary for successful signal demod.

So what you like about COFDM is true about any RF
modulation scheme that adequately addresses
multipath distortion. Until recently, the question
was whether such equalizers were just pipe dreams.
Now we know they exist, and we also know that in
time these equalizers can get even better. Because
we have Moore's law on our side.

Whatever it is today that keeps *good* 8-VSB
receivers from store shelves would equally keep
good COFDM receivers from shelves. You can always
screw up a good design. That's easy. And that's
what you experienced with the receiver you tested
recently.

Bert
 
 
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at 
FreeLists.org 

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: