Content-Transfer-Encoding: 8bit
Message-ID: <65AECDF1F89AD411900400508BFC869F0D75CE@pdw-mail-1.dera.gov.uk>
From: "Talbot Andrew" <ACTALBOT@dera.gov.uk>
To: rsgb_lf_group@blacksheep.org
Subject: LF: RE: Where next?
Date: Tue, 20 Feb 2001 09:27:56 -0000
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Mike's comments still seem to make the assumption that a human brain can
do better than software.  Why ?

It doesn't matter how good your eyes are at perceiving subtle
differences of tone or colour interspersed with static crashes, or ears
at hearing minute changes in a single tone, they can only see / hear
amplitude changes and are throwing away all the information contained in
the phase of the signal - 3dB worth if your interpretation is perfect
and probably a lot more.

Why assume a machine cannot do better - the world is now full of
communication signals travelling over links that would be impossible for
analogue information.  We just need to choose a scheme optimised to the
type of signalling we want to do and write software to implement it in
such a way that the average amateur (who won't build anything of course)
can just plug in and go.

It is very arrogant to assume the human brain can always do better than
a machine just because we feel it ought to.

PSK will do better on LF than SLOWCW  DFCW etc. in similar bandwidths,
but we need to implement it properly.  The US Lowfers are there, and
showed this years ago.  All we need is decent frequency stability in
receiver and transmitter and a preparedness to not insist on relying
blindly on Soundcards to meet all our data comms needs.   

If the link won't support lower and lower bandwidths - the simplest
route to weak signal working - then use coding to achieve this with
wider signals.  And here only machines can do the job.   I can't somehow
visualise a human brain doing the Viterbi error correction algorithm in
real time from a Spectrogram
--------
I understand there was a quote published in QST (or QEX ?) some time ago
from the FCC to the ARRL that "If amateurs did not improve their
technology they would lose spectrum".    Can anyone in the US confirm
(or deny) this and point me at the reference.

Andy  G4JNT


> -----Original Message-----
> From: Mike Dennison [mailto:mike.dennison@rsgb.org.uk]
> Sent: 2001-02-19 16:37
> To: rsgb_lf_group@blacksheep.org
> Subject: LF: Where next?
> 
> 
> 
> > ..................... One such candidate could be the Steve 
> Onley's FDK.  I
> > know that, theoretically, PSK can have an advantage over 
> non-coherent
> > FSK, but this doesn't take into account phase distortion over very
> > long paths.  But I will leave to others to discuss about 
> the pros and
> > cons of the two methods.
> > 73  Alberto   I2PHD
> 
> A factor that is often ignored when comparing PSK08 (for instance) 
> and QRSS is the difference between relying on a machine to 
> interpret the results and using the brain. The spectrogram-type 
> programs do wonderful work but the final 10dB or so is down to the 
> brain deciding what is a valid signal and what is not, then deciding 
> whether it makes sense or not.
> 
> We really would have a winner if we can combine the obvious 
> theoretical advantages of using FM (or phase mod) rather than AM, 
> especially when noise is a significant limiting factor, with the 
> advantages of a display that allows the brain to add some dBs by 
> intelligently interpreting the result. 
> 
> For instance, how about produce a display that is the result of 
> subtracting the signals and noise in a given audio band, with the 
> signals and noise in another band of the same bandwidth. This might 
> provide a means of reducing the effect of wideband noise such as 
> static (but not random noise) on a conventional mono-frequency 
> transmission by subtracting the noise in the 'no-signal' band from 
> the band containing the signal. It might also provide a 'comparitor' 
> effect so that FSK signals can be better detected - for a given time 
> period, if the output of 'channel A' is bigger than the output of 
> 'channel B' then it is a binary '1'; if the other way round 
> it is a binary 
> '0'.The important thing is to resist the temptation to feed 
> this into a 
> machine that tries its best to work out whether it is a valid 
> character 
> or not - this is the bit that the brain does better.
> 
> 
> Mike
> 
> 


-- 
The Information contained in this E-Mail and any subsequent correspondence
is private and is intended solely for the intended recipient(s).
For those other than the recipient any disclosure, copying, distribution, 
or any action taken or omitted to be taken in reliance on such information is
prohibited and may be unlawful.