Message-ID: <566B6B8A.3020505@abelian.org>
Date: Sat, 12 Dec 2015 00:34:18 +0000
From: Paul Nicholson <vlf0403@abelian.org>
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130106 Thunderbird/17.0.2
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <566B354C.4060003@virginbroadband.com.au> <Pine.LNX.4.64.1512112040130.30040@plato.ucs.mun.ca>
In-Reply-To: <Pine.LNX.4.64.1512112040130.30040@plato.ucs.mun.ca>
Subject: Re: LF: VO1NA - carrier 0900 - 1200 UTC
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Status: O


Well done Joe, epic 3-day transmission and good data.

The full plot

  http://abelian.org/vlf/tmp/151212a.png

The diurnal doesn't repeat very well from day to day but
diurnals often don't.   But it appears mostly diurnal as
opposed to oscillator drift and it looks like the long term
stability of the tx is very good.

For comparison here is 24kHz NAA over the same time period

  http://abelian.org/vlf/tmp/151212b.png

There's enough data here to draw some tentative conclusion.

NAA diurnal is much more repeatable and its dark path phase
varies by no more than 50 degrees.  At 8kHz signal phase is
even steadier on long paths.  But at LF the variation is much
more, roughly in proportion to the frequency as we might expect.

If this is generally the case at LF and it probably is, it
doesn't bode well for the sort of ultra narrow band coherent
BPSK that works so well at VLF where we use message durations
of up to several hours.  We're just not seeing that kind of
window here.

When signals are so weak that the reference phase must be found
by trial and error, it's quite easy to guess the phase when
it's likely to be steady or nearly so - not many trials are
needed. But guessing becomes impractical if the phase wanders
about, too many phase patterns must be tried [*].

It looks from these measurements that the narrow band signaling
that works well at VLF will not work at LF.   The longest
duration we could use is probably around 60 to 90 minutes.

On the other hand, LF signals are much stronger than VLF signals
and the noise floor is lower so wide band high speed BPSK
might do well and phase guessing works fine when the message
duration is not too long.  If the message is rapid enough,
the phase is going to be approximately steady whatever the
propagation is doing.

Maybe the boundaries to push above VLF are going to be with
high speed messages, perhaps at higher frequencies because
137 isn't wide enough.

Operation would still be around 0dB Eb/N0, aiming for the
channel capacity, but at a much higher rates and with stronger
signals.

So, 1mS symbols, anyone?

[*] Not just impractical: increasing the number of phase trials
reduces the overall coding gain because it increases the chance
of a false detection.

--
Paul Nicholson
--