Date: Sat, 10 Jan 2015 22:44:00 +0100
From: Pieter-Tjerk de Boer <ptdeboer@cs.utwente.nl>
To: rsgb_lf_group@blacksheep.org
Message-ID: <20150110214400.GO29958@cs.utwente.nl>
Mail-Followup-To: rsgb_lf_group@blacksheep.org
References: <20150108220755.GA20377@cs.utwente.nl>
 <54AF0F8C.8030106@virginbroadband.com.au>
 <54B03993.4090605@abelian.org>
MIME-Version: 1.0
Content-Disposition: inline
In-Reply-To: <54B03993.4090605@abelian.org>
User-Agent: Mutt/1.5.21 (2010-09-15)
Subject: Re: LF: Eb/N0 values for amateur modes
Content-Type: text/plain; charset=us-ascii
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Status: O

On Fri, Jan 09, 2015 at 08:26:59PM +0000, Paul Nicholson wrote:

> CW by ear scores better than I would have guessed.

Indeed, I was surprised to see the number for the three CW cases
match so nicely.
However, precisely in the CW case, the results should be taken with
grains of salt worth several dB.
The aural CW results are based on W2RS' paper about the ZRO tests, and
there is a considerable difference between different operators, so I had
to make a somewhat arbitrary choice.
W2RS' paper can be found here:
http://web.archive.org/web/20050207235207/http://www.n1bug.net/tech/w2rs/humanear.html

> CRC and synchronising bits shouldn't be counted as payload
> information so some of the table entries need higher Eb/N0.

You're of course right in the practical case of comparing complete modes
as given. However, when comparing the underlying modulation and coding
schemes, it's less clear-cut.

In the case of Opera's CRC bits, I have assumed the CRC is only used as
a last check, so usually will be correct. In that case, one could argue
those bits have also been transfered almost always correctly and could
thus have been used for data.
O.t.o.h., if the CRC is used to select among multiple likely outcomes
of the decoding of the (outer) code, then of course those bits are coding
overhead and not payload.
Which of the two is truely the case in Opera, I don't know.

As for the synchronising bits in WSPR and WSJT: I would argue those could
be left out (or at least be reduced considerably) at the expense of
having much harder time searching at the receiving side, or needing to
know the precise frequency and time through some other means. The energy
in the sync bits doesn't contribute to the actual decoding, once the
signal has been found.

Regards,
  Pieter-Tjerk, PA3FWM