Envelope-to: dave@picks.force9.co.uk
Delivery-date: Thu, 26 May 2005 14:52:16 +0100
From: "Mike Underhill" <mike@underhill.co.uk>
To: rsgb_lf_group@blacksheep.org
Date: Thu, 26 May 2005 14:51:55 +0100
Message-ID: <006d01c561fa$14650450$6405a8c0@MJUSonyLaptop>
MIME-Version: 1.0
In-Reply-To: <000b01c56128$4a65d160$6507a8c0@Main>
Importance: Normal
Subject: LF: RE: Re: Reception with multiple antennas
Content-Type: text/plain; charset=windows-1250; format=flowed
Content-Transfer-Encoding: 8bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

Dear Alan/Mike/Group

I think that reception with multiple antennas is a very interesting
suggestion/proposal. Here are some 'bookwork' theoretical points which could
be taken into consideration?

Optimum (non-coherent) combining of a number of signals without knowledge of
the (relative) carrier phase of each can be done by weighting each, before
combining, by their own signal to noise ratio (SNR).  For four received
signals, only if the SNR of each of the signals are equal will an SNR
improvement of 6dB be achievable (over the best signal).

If carrier phase is known for each signal, then three independent nulls may
be placed on (point-) sources of noise. The 'beam forming' can be done in a
computer on stored records of the signals not in real time.  The carrier
phase is best stored at baseband by down-converting each signal to I and Q
signals and recording these. The time accuracy of the recordings should be
better than 1/100th of the shortest symbol length for 40db nulls.  Obviously
for WOLF this is much shorter than for QRSS. 

After one or up to three simultaneous nulls have been found for the noise
sources the combining weights required for identifying these will also give
their directions of arrival.  Depending on the strength of these noise
sources the SNR can then sometimes be increased considerably more than the
6db for four receiving sites. 

If the sites are less tan about half a wavelength apart an antenna gain of
greater that 6dB but less than 12dB may also be achieved. For antennas
greater than about one wavelength apart the gain cannot be usefully
increased over the 6dB combining limit.

If the signal is say 20dB above noise at each  of a pair of antennas (after
the longest possible time integration) then the angle or direction of
arrival can be determined to an accuracy of about 1/10th of a/lambda where
'a' is the distance between the two antenna as seen from the direction of
arrival of the signals. For antennas that are several wavelengths part,
greater angular accuracy is achieved at the cost of 'ambiguities or grating
lobes' that have to be resolved by other means. But fortunately the
approximate angles and directions of arrival can be guessed for LF
operation, and a long baseline interferometer can provide 'vernier'
information to great accuracy.  A pair of stations along the direction of
arrival will give the elevation angle of arrival; a broadside pair will give
the direction of arrival. 

The integration time required to obtain sufficient SNR for a good
measurement or better detection of the information on the signal can of
course be improved (as suggested) by tracking carrier phase variations
throughout each received symbol. In this way optimum combining of all
received signals may be achieved and the signal processing gain will be
maximised.   

Finding the time delays of multipath signals is in principle also possible
but not well suited to the low bandwidths of LF signals.  WOLF has got to be
better for this. With multiple antennas it is possible to compute a main
beam directed to optimise one of the paths with nulls on the other paths.
This should be done on the stored records for each of the identifiable
multi-paths. Then one computes the relative carrier phases of the optimised
signals and from these the relative path delays may be computed, but with an
ambiguity of 1/136kHz or 7.35 microseconds.  Again this ambiguity has to be
resolved by other means. The necessary information is held in the relative
start (or finish) times of each symbol transmitted. These are much easier to
establish with WOLF than with QRSS. 

In summary I am suggesting the exchange of signal records in the form if I
and Q pairs. This will require the appropriate demodulators for each
'netted' receive station.  

I hope this is useful and may stimulate further thought and action. 

Mike G3LHZ.


 -----Original Message-----
> From: owner-rsgb_lf_group@blacksheep.org [mailto:owner-
> rsgb_lf_group@blacksheep.org] On Behalf Of Alan Melia
> Sent: 25 May 2005 13:48
> To: rsgb_lf_group@blacksheep.org
> Subject: LF: Re: Reception with multiple antennas
> 
> Hi all this is an interesting problem, and I have done some crude
> calculations to try and understand the effects and some causes of fading.
> Often after a geomag. storm there can be multipath which leads to dips in
> excess of 20dB in the received signal. I suspect that the multiple aerial
> would not be a help in these conditions where the phase is often changing
> at
> around 8 degrees per minute.
> 
> As I understand it, two sites receiving the same signal and correcting for
> phase difference would lead to a 6dB increase in the required signal. If
> the
> noise is assumed to be non-coherent combining the singals should lead to
> just a 3dB increase in noise so one should see an increse in S/N of 3dB
> for
> every doubling of the aerial size.
> 
> Now to answer Mike's other query, previous experience on fading suggest
> that
> there is a "footprint" for the incoming signal around 50kms in diameter
> where stations receive with similar fading patterns. Although the phases
> will be different at different locations and will change as the
> ionospheric
> "refection" layer moves, all the stations inside the footprint will see
> sensibly the same changes in phase. ( the parts of the ionosphere "used"
> by
> each station will be sensibly moving up and down the same amount) This is
> a
> phase difference that can be optimised for all the stations initially for
> a
> given target path, then the phase difference should stay sensibly constant
> over practical usable periods. We have noticed a lag in the fading of QRSS
> stations at different locations, most dramartically with John WD2XES and
> Warren WD2XGJ (who are about 50 miles apart) over the winter when there is
> often a 20 minute difference on the time that they fade down, but this can
> remain almost constant for several hours (and several fade cycles).
> 
> This suggests there may be the need for a "calibration" phase, but this is
> not too different from the situation with WOLF, though it may be difficult
> to optimise the initial phase when you dont know what you are looking for.
> This suggests to me that it may be necessary for the tranmitting station
> to
> transmit a "training" signal, in moden parlance, but I suppose this is
> just
> what the interleaved code does in WOLF. One can certainly see a
> phase"locking" in WOLF well before the variable signal becone decodable.
> 
> I find this a fascinating possibility as it might allow me to probe the
> ionosphere in a more detailed way. I had been hoping to be able to monitor
> phase changes during darkness to define how the path lengths changed and
> try
> to understand the interaction between absorption and cancellation in
> determining the signal levels at extreme distances.  Remote reception of
> Loran signals was a possibility but I am not sufficiently expert in
> software
> to write the necessary code, and we did discuss the possible use of DCF39
> Wolf DL4YHF some years ago, but there are different problems there.
> 
> What is interesting is that if you cant "hear" a signal you dont know how
> much to improve the equipment by, so you are handicapped. We used CFH in
> the
> early days to estimate the possibilities of Transatlantic reception at 1w
> ERP. Since those days LF horizons have expanded immensly
> 
> Cheers de Alan G3NYK
> 
> ----- Original Message -----
> From: "Mike Dennison" <mike.dennison@ntlworld.com>
> To: <rsgb_lf_group@blacksheep.org>
> Sent: 25 May 2005 12:34
> Subject: LF: Reception with multiple antennas
> 
> 
> > I have been following this interesting discussion, but I have some
> > perhaps naive observations which could do with some expert answers.
> >
> > Combining the output from two stations will give double the noise and
> > double the signal - so no advantage there. I presume the fix is to
> > make sure the two outputs are in phase, thus producing an improvement
> > as the signals will be coherent but the noise will not.
> >
> > Plainly, there will be phase differences between the two receiving
> > stations because of the different path lengths, and I suppose there
> > would need to be compensation for this at the combining station.
> >
> > I can see this being useful for point to point ground-wave links,
> > where the relative phase between the two received signals can be
> > predicted and is constant. But how can it work when the incoming
> > signal is from an unpredictable direction, and the path length (and
> > hence phase) of a sky-wave signal is varying all of the time?
> >
> > At best you will have diversity reception (as described by Mal), but
> > how do you keep two sky-wave signals in synch?
> >
> > Mike, G3XDV
> > ==========

-- 
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.322 / Virus Database: 266.11.17 - Release Date: 25/05/2005