To: rsgb_lf_group@blacksheep.org
References: <57237FB9.2060507@posteo.de>
 <8D3705FBF0E442D-628-9A0506@webmail-vm001.sysops.aol.com>
 <5723C259.3090800@abelian.org> <5723C76E.70002@posteo.de>
 <5724968F.1030904@abelian.org> <E1awVQw-0005Gs-4N@post.thorcom.com>
From: Paul Nicholson <vlf0403@abelian.org>
Message-ID: <5724DE39.4090806@abelian.org>
Date: Sat, 30 Apr 2016 16:32:57 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101
 Thunderbird/38.6.0
MIME-Version: 1.0
In-Reply-To: <E1awVQw-0005Gs-4N@post.thorcom.com>
Subject: Re: AW: VLF: EbNaut transmission on 8270.1 Hz again and again
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Status: O


Markus wrote:

 > you could create a stronger carrier component by slightly
 > unbalancing the PSK phases. For example, switching from -80 to
 > +80 degrees would create a sin(10°)=-15.3 dB in-phase carrier,
 > while weakening the modulated quadrature component by only
 > cos(10°)=-0.133 dB.

Well worth a try and can be easily tested in simulation.
The phase search costs some coding gain because of the increased
opportunity for a false decode to overcome the real decode.

The squared signal provides a starting point for the phase
search, averaged over the whole message.  With the right amount
of vestigial carrier it might be beneficial to measure the
carrier phase in four pieces instead of one, to independently
set each quarter of the phase search.   If that reduced the
number of phase trials there might even be an overall slight
gain.

And the signal would be detectable.  Raises some possibilities.

 From an earlier post

 > flatten out the noise level before decoding (i.e. reduce the gain
 > during high noise background) ...  evaluate out-of-band energy.

I've tried normalising the symbol amplitudes before the
decoder and that always reduces performance.   But what you're
suggesting here is a bit different - to look at the out-of-
band noise and apply an AGC.    I wonder if that would help the
mode when the band noise changes through the message duration?
It would be easy to test using some stock recordings from DF6NM,
DK7FC, W4DEX, and others.

Well, you've given me some work to do here Markus, and I've been
busy trialing different phase search patterns, eg 45 instead of 30,
to see if there's a net coding gain to be found.

--
Paul Nicholson
--