[opendtv] Re: Seeing Ghosts on a Single Frequency Network

Al Limberg wrote:

The crux of the problem is how the 8-VSB receiver combines the
signals from a pair of 8-VSB transmitters using identical (or
complementary) modulation.  The SFN proponents originally
contemplated combining the signals at the antenna of a single
8-VSB receiver.  This can result in as much as a 3 db
improvement in SNR of the combined received signals if they be
in phase.  However, combining the signals at the antenna of a
single 8-VSB receiver is also apt to cause catastrophic loss of
reception should the received signals be in anti-phase.

Yes, but that also holds true for any other modulation, doesn't it? The trick has to be to avoid areas where the signals are 180 out of phase *and* of the same signal strength. Even with multicarrier systems, you can only wipe out so many carriers and still achieve reception. COFDM, used with 2/3 FEC, won't survive loss of anything more than 1/3 of the carriers. A 0 dB signal 180 out of phase will kill it too!

I thought that Charles Rhodes' graphs showed cases of separate signals, 10 usec and 0.5 dB apart, at a single receive antenna. In principle these could have come from two different towers. And yet you could see that an equalizer with enough taps could salvage those low spots (and did, in four of the receivers tested). The deepest nulls were still about 20 dB above the noise floor, it looks like.

If the 8-VSB signals from a pair of 8-VSB transmitters using
identical baseband modulating signal are to be constructively
combined irrespective of the siting of the 8-VSB receiver, the
8-VSB signals must be kept separate from each other until the
conversion to baseband where both signals have in effect been
de-rotated to a common phase.  The two baseband signals can
then be combined, either in a critically sampled digital regime
or in an over-sampled digital regime that is quasi-analog in
nature.

Very interesting twist on the concept. Strictly can't be called SFN anymore. It requires two receivers and cuts spectral efficiency in half, unfortunately, but it would be good even for legacy receivers. They could pick and choose the best signal source. Then again, if you use something like the 1/4 GI that some of the Euro SFNs use, you cut spectral efficiency too, by 1/4 in that case. SFNs don't come for free.

The other, two towers time sharing on the same RF channel idea, I don't see how that can work with legacy receivers? But I agree, it amounts to the same thing for spectral efficiency. Especially looks similar if you consider two or more broadcasters sharing these RF channels.

The proponents of SFNs for 8-VSB signals seek to transmit
respective 8-VSB RF signals resulting from identical baseband
modulating signals from a pair of 8-VSB transmitters that
transmit with 100% duty cycles using the same RF channel
allocation.  It appears that the proponents of SFNs for 8-VSB
signals now contemplate keeping the respective 8-VSB RF
signals separate from each other by directional antennas or
by different antenna polarization.

Again, very interesting. I didn't know about this new initiative. I agree with you on the polarization concept not being good for mobile devices, or for that matter, for antennas in a cramped fireplace like mine are, but I'll keep an eye out for this new non-SFN SFN idea.

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.

--