Re: Wztching for surface traffic (was: Submarine Movie? or TV?)

[John Brownlee <jonnie@chronic.lpl.arizona.edu>]

HA! Finally, something I know about surfaces from the murky deep of
the psubs list!

	The operative constraints on any such device is what do you want it
to do, and how automated would you like it to be? A directional hydrophone
on a stepper motor-controlled scan platform to listen to manually is one
thing, but a passive sonar 'radar screen,' is quite another. It often takes
more effort to take the man out of the loop than to simply do a small task
yourself, so you have to be pretty careful in asking yourself what you're
after. This is so obvious, but as anyone who does this sort of work for a
living will tell you, automating things as a funny way of becoming a
religious experience. Phil's treatment of man in vs. man out of the water
is an excellent example of this lesson.

	With regard to automation, this sort of thing -IS- possible. A lot
of it comes down to how smart the frequency analysis has to be in order to
correlate the rate of change of the power spectrum with an incoming or
outgoing boat. I would think the only cases you would care about would be
incoming and outgoing in close proximity, which means the Doppler shift you
are looking for should be noticeable to spectrum analysis. Also (and any of
you subbers who are also IFR pilots will immediately recognize this fact)
you need to consider depth as a constraint on the system, as Doppler gives
you only slant-velocity information, so you are less likely to detect the
presence of Joe Drunk Weekend Skipper at depth than near the surface. A lot
of the basic physics is on your side, but the technology is non-trivial.
	I can tell you from experience that this sort of software (at least
that was how we did it at the Loonie Lab) is rather sophistocated and you
need to characterize the hell out of the system. I used a relatively simple
neural net to train for known types of input data that had been through
Fourier analysis and eventually got a pretty good is there-isn't there
descriminator. That would be (from my experience) the easiest way to do
this automatically , as an azimuthal 'threat light' system with a
constantly rotating directional transducer. I am sure it would be possible
to build up this sort of system to look for more interesting things too.

	Probably cheaper to bribe a friend with gas/food/beer/greenbacks to
stand watch and shoo away the ooglers. Interesting problem, though. Brings
back memories from my geophysics days.

							John

John Brownlee
Lunar and Planetary Lab
University of Arizona
jonnie @ lpl . arizona . edu