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

From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
Date: Fri, 21 Jan 2011 19:21:52 -0600
Mark Aitken wrote:

> A complimentary paper that amplifies some areas of discussion...
>
http://www.tvantenna.tv/papers/SFN%20Analysis%20V1.pdf

Thanks for the light reading, Mark.

The risk of articles like this one is that people think there are easy answers. 
Just use COFDM. But it's not the case. SFNs are problematic no matter what. 
They're useful to save on spectrum, and they MAY be useful to make reception 
easier in certain locations (if they are designed for that purpose), but they 
will not save on transmitting towers. You always lose "coverage" with SFNs, 
compared to single towers or towers in a MFN.

A couple of minor points. Some of the better ATSC receivers are supposed to be 
able to achieve recpetion even if the pilot has been obliterated entirely. So 
the signal looks like honest to goodness SSB. IIRC, Broadcom did this first, 
way back when receivers were still pretty awful. (One step at time.)

Dynamic echo is bad news for 8-VSB, but the CRC tests did show that the Gemini 
receiver (of a half decade ago!) was considerably better than even the 5th gen 
Zenith. And I think Limberg/Patel/McDonald have ideas to improve on that.

Quoting from the beginning,

"The echo environment of SFN is entirely different. Because the height of SFN 
towers is expected not exceed 300 feet, the echo environment in the covered 
area would be characterized as Rayleigh. This means that in urban environments 
with short towers and without line-of-sight to the receive antenna, the 
interfering signals, not just their echoes, would be severe over the entire 
coverage area. In a Rayleigh environment, well known in mobile communication 
systems using short towers, the echo power is essentially equal to the dominant 
signal power and the directions of arrival, signals or echoes, would be random. 
In such environment, directional receive antennas cannot, in general, 
discriminate against the reception of significant and undesired echo or signal 
power."

This sounds very credible to me. But again, how many 50 KW towers did Qualcomm 
have to install, to cover mainly four urban markets and a thin line between 
them? Answer: 30. COFDM has similar problems, obviously not as severe, but it's 
no panacea. Lots of interference zones created by the multiple towers.

Concerning:

"The Threshold SNR in a distortion-free channel (i.e. no Equalization Noise) is 
15dB. The thermal noise floor is -106.2dBm + NF. If the NF is 7dB, the thermal 
noise floor is -99.2dBm. Therefore, the minimum signal power in a laboratory 
setting is -99.2dBm + 15dB = -84dBm. This level does not account for the noise 
that would be generated by the attenuation and processing of undesired signals 
and echoes."

CNR loss is common to both 8-VSB and COFDM, when echoes exist, and more loss 
exists in Rayleigh channels than in Ricean, also for COFDM. Let's not forget, 
for example, that the difficult Brazil E profile, three paths shifted by 1 
usec, of equal amplitude, which this paper also describes as being very bad, 
required 25 dB C/N by the Zenith 5th gen, but only 19.9 dB C/N by the Gemini.

COFDM was only ever mentioned very briefly, in passing, way at the end. Would 
be nice to see an equally well researched treatise on COFDM SFNs, instead of 
the usual generalities like the one quoted below:

"For these reasons SFN for ATSC-8VSB can only be successful in a shielded 
terrain such that the transmitters are isolated from each other and no longer 
require synchronization. OFDM-based SFN could be successful in a flat terrain 
because in that system the multiplicity of echoes and active signals, by 
design, enhance the dominant signal instead of degrading it."

Not so fast with those generalities, please!

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

Other related posts:
