[opendtv] Re: Spectrum is too valuable

  • From: Craig Birkmaier <brewmastercraig@xxxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Fri, 06 Nov 2015 08:42:51 -0500



Regards
Craig
On Nov 5, 2015, at 10:02 PM, Manfredi, Albert E <albert.e.manfredi@xxxxxxxxxx>
wrote:


Craig Birkmaier wrote:

Sorry Bert, but Mother Nature disagrees with you. The MPEG-2 TS
streams are a broadcast service delivered to each home passed.
Broadband is a two-way unicast service;

Aaargh!

The downstream part of DOCSIS uses exactly those same 6 MHz channels as
MPEG-2 TS might be using (actually DOCSIS 3.1 widens the channel, but that
doesn't change anything appreciably for this argument).

Your argument is irrelevant today. There is no question that a system could
serve more homes with broadband if they repurposed spectrum now used for analog
or digital TV streams. The question is how much bandwidth is needed in any
neighborhood to replace the broadcast TV service with on demand UDP TV streams?

The upstream signals are combined from multiple homes, in a PON, however
those are usually lower capacity than downstream. Either way, though, any
spectrum taken up by broadcast one-way streams takes away from the broadband
service, within that PON structure.

Don't know why you changed subjects and started talking about upstream
bandwidth. The problem is providing enough downstream bandwidth to replace a
broadcast service with a unicast service. You love to talk about spectral
efficiency, but ignore the simple math.

There is no way the current FOTA broadcast TV service could deliver unicast
said to every home in a market without massive spectral reuse - i.e. tons of
cells serving a small number of homes. Cable is no different. There is a finite
amount of bandwidth available in each PON that must be divided by the number of
homes served by that PON.

Improvements in spectral efficiency help by stuffing more bits into a given
chunk of spectrum; thus we have seen downstream bandwidth improve with each new
upgrade to the DOCSIS service. There has been no comparable improvement in TV
utilization of cable spectrum as they are locked into a static QAM
constellation and MPEG-2 encoding for TV.

So do the math. Here is a link to a paper from Arris comparing the spectral
efficiency of DOCSIS 3.0 vs 3.1.

https://www.arrisi.com/globalassets/resources/white-papers/arris_spectral_efficiency_of_docsis_wp.pdf
For 0 dB operating margin, the analyses for DS (8K FFT) and US (4K FFT) showed
that DOCSIS 3.1 spectral efficiencies can be 8.1996 bps/Hz and 7.8589 bps/Hz,
respectively, which is about 30% and 90% gain over the estimated DS and US
DOCSIS 3.0 spectral efficiencies of 6.33 bps/Hz and 4.15 bps/Hz, respectively.

Let's assume that there are 900 MHz available for downstream service in a
modern 1 GHz HFC system - the rest is upstream and other system overhead.

So the best we can do today with DOCSIS 3.0 is 6.33 X 900,000,000 = 5.697 Gbps

With DOCSIS 3.1 this will increase to a max of 7.379 Gbps. This will in theory,
serve 295 homes with 25 Mbps or about 148 homes with 50 Mbps.

How much does a home need? 25 Mbps can support about 2 HD streams with MPEG-2,
perhaps 3 streams with h.264. This may be sufficient for a home with two
residents, but may not be sufficient for a home with four or more residents.

The FCC now specifies about 25 Mbps as the minimum needed to be classified as
broadband. So with DOCSIS 3.1, in theory about 295 homes can be served with
broadband from a PON. In practical terms it is likely to be 250 or less.

Today PONs serve between 25 and 2000 homes, with the average being 500.

So even if you reclaimed all of the available bandwidth in a 1 GHz HFC system,
you would still need to reduce the number of homes per PON by half to deliver
25 Mbps broadband to each home.

No amount of words will take away from this simple fact, Craig. You can't
fool Mother Nature.

You must be speechless now!

Regards
Craig

Other related posts:

  1. Home
  2. opendtv
  3. Archive
  4. 11-2015