[opendtv] Re: TI Exec: Open Standards Crucial for Mobile TV

John Shutt wrote:

> The correct terms are 1/2 rate and 1/4 rate, not 1/2
> additional FEC.
>
> http://www.atsc.org/standards/a_53c_amend-1_corr-1.pdf

I am referring to this:

"Enhanced data shall be expanded by the ratio of 1:2 (for
1/2 rate coding) or 1:4 (for 1/4 rate coding) to form
place-holder bits for the Enhanced convolutional coding and
shall be formatted in MPEG-2 packets that shall be
multiplexed with packets from the Main MPEG-2 stream."

Which are then sent to the trellis encoder.

So this says that the further expansion of 1/2 or 1/4 is
overlaid on the standard 2/3 convolutional FEC.

> As you correctly state, at 1/2 rate, for 542 Kbps, you get
> 18.1 Mbps in the main stream.  You removed 1.3 Mbps from
> the main stream to get 542 Kbps in the robust stream.
> lose 2 bits to gain 1 bit.
>
> For 1/4 rate it's even worse.

As I am now in a habit of asking Craig, "Compared with what?"

If you have to create a more robust sub-stream, no matter
whether you do this by adding some convolutional FEC or
whether you do this by reducing the constellation of the
robust stream, or a combination of both, you *will* chew up
bandwidth compared to a single, wide stream, for equal
robustness of the wide stream.

There's no free lunch, John. You can't create, say, 19 Mb/s
or wide stream + 390 Kb/s of robust stream, have them both
going over the same 6 MHz band, with the wide stream no
less robust than before you added the robust stream.

> As I understand it, with DVB-H, as implemented within a 6
> MHz DVB-T stream (as opposed to a dedicated 5 MHz DVB-H
> carrier as Crown Castle is planning in the US), the DVB-H
> data is time sliced within the DVB-T bitesream, and is a
> straight one-for-one bit usage.  The DVB-H data includes
> additional FEC, so for your 542 Kbps you may need to use
> 25% more bitrate for the additional FEC data. So you lose
> 677 Kbps from the main stream for a single DVB-H payload
> of 542 Kbps.

If you only use 25 percent more FEC, you are not creating a
much more robust bitstream, John. But look at your
documentation and see that they mention 1/2 or 2/3 FEC
(presumably total) and 16-QAM or QPSK. Not 64-QAM.

So in fact, you are knocking down the bit rate of the
robust sub-stream by quite a bit, if you want to keep the
main stream as robust as it was before.

> And I hate to point this out, but E-VSB may not even work
> very well.  E-VSB was designed to improve reception, but
> not to increase battery life of a portable device.

You are probably correct, although not on the bit rate
comparison issue. I'd be much more excited to see 8-VSB
receivers with proper use of the training sequence, and
perhaps diversity antennas, be used for mobile applications.
The idea of watching TV on the LCD panel of my 3G cell
phone just doesn't captivate my imagination. Oh well.
Let's see how successful Verizon is with Vcast, then we can
talk about this some more.

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: