[opendtv] Re: China writes its own digital TV standard
- From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
- To: <opendtv@xxxxxxxxxxxxx>
- Date: Wed, 28 Jun 2006 12:45:00 -0400
Doug McDonald wrote:
> If you have stretches of COFDM interleaved with
> stretches of pseudorandom PN sequence, you can
> proceed as follows. It is desirable to have
> double-length PN sequences. What you do is
> calculate the cross-correlation between the known
> PN sequence and the data in the vicinity of the PN
> sequence. This gives you, directly, the signature
> of the multipath. You then keep the largest of the
> multipath echoes, and use those to calculate what
> the received signal would be with just the PN
> sequence as transmitted, with the COFDM part zero.
> You then subtract that from the received signal.
> This gets rid of the major part of the noise
> introduced into the COFDM. All this of course is
> done using Fourier transforms, not traversal
> filters. It's quite elegant.
Good, thanks. This is the way a PN sequence-savvy receiver uses the new
training sequence. My question was how these PN sequences, now stuffed
in the GI of COFDM, would affect older receivers.
> If multipath is static or slowly changing, you can
> average different occurances of the PN seqence and
> get, in the static case, essentially perfect
> removal of the PN sequence interference with the
> COFDM caused by multipath. In the static case, in
> other words, even with severe multipath, the PN
> sequence add NO NOISE to the COFDM data, after
> processing.
>
> Oh yes ... pilot carriers are either not needed or
> of reduced need.
Yes, this is interesting indeed, and shows what's happening here.
Convergence. Multicarrier and single carrier receivers working much the
same way.
The advantage of COFDM is still that it can more easily overcome deep
and narrow notches, although deep and narrow notches are also easily
mitigated with the use of diversity antennas.
As long as we're redesigning RF schemes to improve reception robustness,
here's another proposal for 8-VSB.
I'm stuck on the idea of using the frequent segment sync to assist in
training. These occur every 77.3 or so usec, so very frequently, and
much more frequently than the GI. But they are short, only 4 symbols.
What if the segment training sequences were not all identical, but
instead were a known series of PN symbols. A number of these could be
combined by a receiver, perhaps, to be used as a reasonably good
training sequence. I don't know how many it would take, but even if you
need to accumulate 511 of these symbols, that would still occur more
than twice as often at the existing training segment.
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:
