Content-Transfer-Encoding: 8bit
Message-ID: <5.1.0.14.0.20011119112735.00a76080@gemini.herts.ac.uk>
Date: Mon, 19 Nov 2001 12:48:25 +0000
To: rsgb_lf_group@blacksheep.org
From: "James Moritz" <j.r.moritz@herts.ac.uk>
Subject: LF: Long-Baseline Phased Arrays on LF
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dear LF Group,

Well, it works for radio astronomers and early warning radars, so why not 
on LF? Here are a few comments:

The basic idea seems to be to coherently add together receiver outputs from 
different locations; the signal amplitudes would add directly, whereas the 
non-correlated noise would only increase as the square root of the number 
of stations, so you would get 3dB more SNR for 2 receivers, 6dB for 4, etc. 
By doing this, one would also effectively obtain a strong directional pattern

The implicit assumption is that the noise at the different locations will 
come from different sources and so not be correlated, wheras the signal 
comes from the same source and will be. However, under reasonably good 
receiving conditions, most of the noise seems to come from atmospheric QRN, 
usually at great distances. So the receivers will probably all be hearing 
the same static crashes, and the noise at the different locations will be 
correlated to a significant degree. A lot of the local, artificial noise is 
connected in one way or another with the mains, which since it is all 
connected to the same national grid, will presumably be correlated to some 
extent too. So I suspect there will not be a great gain in SNR.

You could argue that distant sources of  noise could be rejected by 
synthesising a directional pattern with nulls in appropriate places, 
provided there was not a thunderstorm in the same direction as the desired 
signal. The nulls are effectively produced by cancelling the noise signals 
arriving at different receivers from the offending direction against each 
other. As well as needing accurate relative phase information to do this, 
accurate relative amplitude information is also required, needing some 
means of performing real-time gain calibration on all the receivers and 
antennas in the array. My recent experience of performing about 300 field 
strength measurements showed that a standard deviation of about 1dB in 
amplitude, with individual readings up to +/- 3dB out, occurred due to 
environmental factors, even though the same receiver and antenna was used 
at all locations. Although variations at different locations seemed to be 
fairly constant over a period of a couple of weeks, it is difficult to know 
what long-term variations might happen.

On HF, at various times diversity reception seems to have been popular. 
These systems basically had a number of ganged RXs with separate antennas, 
and summed the demodulated outputs so that whichever receiver was getting 
the strongest signal contributed the most to the overall output. Obviously 
not coherent in any way, but seems to have been quite effective against 
selective fading. Fading is also a problem on LF, especially for the very 
slow transmission modes, so might this cruder approach also work for LF? It 
would depend on fades occuring at different times at different receiving 
locations. I seem to remember we found this was the case during the group 
monitoring experiment we tried on the CFH signal last winter ( see 
http://www.qsl.net/on7yd/t181200.htm). If so it could be done for QRSS by 
simply superimposing spectrograms obtained with the same settings received 
at different locations; it would be a bit harder with something like Wolf, 
requiring accurate timing info to combine a number of recordings.

Cheers, Jim Moritz
73 de M0BMU