circular convolution

glen herrmannsfeldt skrev:
> Rune Allnor wrote:
>
> > glen herrmannsfeldt skrev:
> (snip)
>
> >>There is a (fictional) book, "The Ninth Day of Creation", which
> >>describes a top secret sonar system.  In a quick description, it
> >>sends out a whole spectrum of signals and analyzes the result coming
> >>back.
>
> > The author is not a sonar guy, then. The problem with sonars,
> > compared to radars, is the frequency-dependent noise and
> > losses. Broad-band signals are all but impossible at all
> > but the shorter ranges, simply because dispersion and
> > frequency-dependent losses see to that only narrow-band
> > parts of the signal can be detected. It's all about the water
> > as transmission medium.
>
> I believe like other authors, it was done after interviewing
> appropriate people in the field.

I hope so. The problem that there are several types of
experts:

- The operators who actually use the kits. These guys usually
do not have the technical expertise to understand how the kit
they use, works. They know how top operate it and what
to search for in the data.

- The technical maintenace people. They know the nuts'n bolts
of the gear and the physical parameters of the signals. But they
usually don't know hoe the kit is used.

- The theoreticians. They know how set up a data processing
scheme for the operatives, but they usually don't know how
the kit is used during operations or how the gear looks like.

No type of expert are likely to understand the full chain from
system design, measurement and final analysis. And of course,
everything is filtered through the author, inserting his or her
misunderstandings and obfuscations everywhere.

>  As well as I remember now, the
> transmitted signal is fairly high power, to get above the noise.
> That also means that everyone else knows where you are.  (I believe
> one then leaves the scene as soon as possible.)

One surely ought to...

> As I remember,
> it was more like a frequency sweep, which would have other time
> dependent effects, but maybe better than a broadband source.

There are plenty of signalling schemes around. All of which
work (or not) at the grace of the oceanography.

> >>Not only can it detect the position of a mystery submarine,
> >>but also enough to identify the country that owns it.  (Presumably
> >>by analyzing known submarines.)
>
> > Well, there is lots of myth and illiteracy regarding sonars.
>
> > The (fictional!) sonar system onboard the US sub in Tom Clancy's
> > "The Hunt for Red October" is capable of doing something similar.
> > In the book, the sonar computer analyzes the first data ever
> > recorded of the Red October, and comes up with the technical
> > details, like type of submarine. In another sequence, where the Red
> > October slips away from the US sub, the sonar computer detects
> > something that it classifies as some geological activity.
>
> > Neither type of automated decision-making are possible in the
> > real world.
>
> In this case, there was an expert on the system, one who was part of
> the development, on the sub at the time.

As consultant for Clancy?!? You have to be kidding!

He did a very poor job, then. Or maybe not. To put it this way: If I
were in the navy and asked to consult on a book on navy sonar
systems, I would obfuscate the details about what the systems
do and how things work, as much as possible.

If you read the book carefully (or see the movie), you will note that
the sonar operative, when explaining why they didn't detect "Red
October" slip away, says something like "The computer system
was originally designed for seismic applications. When it finds
something in the data it doesn't recognize, it classifies it as magma."

Note how carefully that's crafted: A navy guy can say that "I don't
understand how this works, but I don't know seismic software."
A seismic guy can say the same from the other direction.

Of course, all of this is as pure BS as you will ever find it. Neither
seismic or sonar data analysis is done that way. The problem is
that "Red October" backfired, and now everybody believe that's
how things are done, or at least ought to be done.

Which, of course, adds nothing to making any actual progress
in sonar technology. 

Rune