[Lowfer] Time coherent scheduling [Long, Boring]

[Stewart Nelson]

Hi Mike and all,

There are many practical communications applications where the objective
is to detect the presence of an incoming signal, rather than decode any
message content.  The question asked may be serious (has an enemy missile
been launched at us) or mundane (has the button been pushed on my garage
door opener).  Entire books have been written on this subject.  There is
a good one titled simply "Detection of Signals in Noise", but the math
was far too complex for my brain, so I didn't buy it :)

In most such applications, the detector must be heavily biased against
false positives.  You want to be sure that your garage door won't open
because of noise or QRM, even if that means that you must occasionally
push the button more than once to get it to open.  In the missile case,
even though both possible incorrect decisions could be equally disastrous,
the system must still be biased against the positive, because most of the
time, the enemy is not launching a missile.

In our hobby application, this is also true.  Below, I use your assumption
of one minute pulses sent every five minutes.  Suppose your "decision
threshold", programmed or estimated by your eye/brain, is high enough that
there is only one chance in a thousand that an observed pulse was caused
by noise. If beacon captures are attempted for 12 hours each night, then
on average we will "capture" each of the five beacons once a week, when
they are not on the air!  Of course, as you mentioned, it is a great
improvement to demand two consecutive hits to confirm a capture.  But
even then, there is IMO no suitable threshold.  Look at some of the weaker
Lowfer captures.  Many signal pulses fall below the top 1% of noise hits.
But allowing 1% false positives per pulse would still give one or two false
captures per beacon per season, while also missing many real captures.
I believe that when examining QRSS signals, we are using 10 to 20 key down
element times to confirm reception, demanding more for weaker signals.
The result is much more robust detection.

Most propagation peaks last between 20 minutes and 5 hours, so there is
an opportunity to use more pulses, but evaluation with the present ARGO
would be difficult.  A version which divided the screen into five areas,
filling each for one minute sequentially, would make it easy to see signal
level from each beacon over many pulse periods.  IMO, with PEP of one watt,
this would still be a slight downgrade from present QRSS practice, simply
because less energy is being radiated.  It may, however, be a good tradeoff,
since it permits reception while your beacon is in operation.

I would much rather see a system that is a significant improvement.  First,
it is good to advance the state of the art, and I believe that will help to
maintain hobbyist interest.  But I also have a more selfish reason.  I will
be returning to Reno near the end of January, and hope to have time to build
a Lowfer beacon.  Of course, I'd like it to be received regularly by you
all.  My location puts me at a considerable disadvantage, so I need all
the help I can get.  The challenge is to develop a format which does not
demand much modification to existing transmitters, does not need overly
complex software, can work (perhaps at reduced performance) without GPS,
and still offer enough improvement to be worthwhile for most Lowfers to
implement.  Your observation about the slant of WA is a good start.
The less our signals look like unmodulated carriers, the better.

73,

Stewart KK7KA