Re: Interpretations, subjectivity & spread-spectrum

From: Hans Moravec <hpm.domain.name.hidden>
Date: Wed, 14 Jul 1999 16:19:45 -0400

The "intelligent life evolved in alternate iterpretations of the sun"
discussion led me to notice an almost perfect analogy in everyday
life.


Spread-spectrum techniques were developed for the military to reduce
the detectability, interceptability and interfereability of radio
transmissions. Now they are in everyday use, for instance in portable
phones.

Instead of putting all the transmission energy into a single carrier
frequency, a spread-spectrum transmitter distributes its energy along
an entire broad band of frequencies. In one kind the frequency hops
rapidly in a pseudo-random pattern. We could make the analogy with
that, but it is more direct with another approach.

Imagine the transmission band thinly sliced into a large number of
narrow frequency bins. A transmitter broadcasts simultaneously at all
these frequencies, but puts a different amount of energy into each bin
according to a specific gain profile. The coefficients in this
profile may be negative for some bins, so that their power goes down
with modulation as it goes up in other bins.

The receiver multiplies the signal it gets in each bin by its own gain
profile and sums the results to recover the modulation. The signal
received in every bin may be below the noise, but the correlated
signal emerges from the uncorrelated noise in the summation process.
Only, of course, if the receiver's profile of coefficients matches the
transmitter's. With an incorrect profile, the receiver sees only
noise.

The number of possible profiles is exponential in the number of
frequency bins.

Many transmitters with sufficiently different profiles can share the
same band. Each new transmitter burdens the others with a little more
background noise (though usually less than the average fluorescent
light or hair dryer).

You can't have exponentially many transmissions at the same time, but
you can probably have more than the number of frequency bins.


On receiving unknown transmissions of this kind, you could search
through the combinatorial gain profile space looking for non-random
modulations. The search might turn up quite a few different
signals.

------------------
So:

These days you can often find several organized signals in RF hiss by
searching an exponential space of linear spread-spectrum transforms.

Is it not just as reasonable to imagine that one might find several
modes of organized activity by searching an exponential space of
linear transforms of an object like the sun?
Received on Wed Jul 14 1999 - 13:21:45 PDT