[opendtv] Re: Ed Williams 1999 Sinclair Baltimore Report
- From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
- To: "OpenDTV (E-mail)" <opendtv@xxxxxxxxxxxxx>
- Date: Sat, 26 Mar 2005 19:29:07 -0500
Terry Harvey wrote:
> My interpretation is that the gross data rate was 24.88Mb/s
> for COFDM in a slightly greater bandwidth than 8VSB, such
> that it would not fit in a six MegaHertz channel. The net
> data rate was 18.66Mb/s. (The 1/8 guard interval I expect
> eats up available bandwidth).
>
> 8VSB gross data rate is approx 31 Mb/s in a narrower channel
> and the net data rate is 19.39 Mb/s. This means more FEC has
> been applied to 8VSB transmission.
>
> As the pre-adaptive equalization processes improve, does
> this not prove that 8VSB will have in future greater data
> rate vs bandwidth efficiency for the same robustness of
> reception?
Terry, the simple answer to your question is yes. The
questions that keep lurking, though, are when and at what
cost.
In a 6 MHz channel, 64-QAM, 1/8 GI, and 3/4 FEC in COFDM
provides 18.662 Mb/s. Because this was 2K rather than 8K,
the 1/8 GI only provided 37.3 usec of echo tolerance. To
be as close as possible to 8-VSB with 2/3 FEC, the COFDM
mode is 64-QAM, 1/16 GI, and 3/4 FEC (19.76 Mb/s).
So indeed, you're giving up some FEC to allow for the room
taken up by the guard interval.
But in 1999, this COFDM signal was competing against 8-VSB
receivers that could stand 15 usec of echo tolerance, and
only lagging echo at that. So "only" +/- 37.3 usec of 2K
COFDM at 1/8 GI was still way better than 8-VSB. Even the
greater FEC of 8-VSB wasn't enough to compensate for the
relative lack of echo immunity in the 8-VSB receivers.
The issue is equalizers. How good will they be, by when,
and at what cost. As equalization improves, the tradeoffs
become more interesting for 8-VSB or other single-carrier
schemes. (These same issues were debated in developing
VDSL standards, by the way.)
The 1/16 GI of 8K COFDM in a 6 MHz (19.76 Mb/s) band
allows for echo tolerance of 74.67 usec, leading or lagging.
The 5th gen LG is capable of 50 usec leading or lagging. So
not quite there yet. But as that echo tolerance playing
field gets leveled, then the FEC differences can be
compared more directly.
The spectral efficiency price you pay for a GI in COFDM is
buying you the ability to use cheap equalizers. As effective
equalizers are developed and their cost drops, they can also
be used in COFDM. If they are, then in principle the GI can
be cut back to the minimum 1/32 (ignoring the legacy).
At this point, the comparison would be 8-VSB with 2/3 FEC
providing 19.39 Mb/s vs. COFDM with 2/3 FEC and 1/32 GI
providing 18.096 Mb/s, for equal robustness.
The other part that we didn't mention is the active pilots
used in the two schemes. COFDM transmits its pseudo-random
training sequences on 177 continual pilots (8K mode) and
some more scattered pilots, or carriers with boosted power
levels, vs a single pilot in 8-VSB. Once again, equalizers
are the equalizer here too. The better the equalizer, the
less robust reception depends on pilots, the less power
(C/N margin) is needed for equal robustness.
The clever design of COFDM obviously worked better than the
ATSC had thought it would. "Who knew?" Now it's up to
equalizer design to do the catching up.
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:
