OK, I'm a bit worried about bringing up this topic given that my search
of r.r.a.a. archives has turned up the fact that this has been a very
contentious topic in the past... but

GENETIC ANTENNA OPTIMIZATION

There, I got it out.

I'm interested in the possibility of using this kind of search of the
space of possible antennas in those situations that we all end up in,
you've put the antenna too close to the tower, you're shoehorning it
into the attic, or you really really need to mount it on the bumper and
not on the roof.

Has anyone written any software that could do this? I guess what I'm
imagining is that you input the location and extent of fixed
conductors, the tower, the car (yes i know it's hard), the wiring in
the walls, whatever, and then you let the computer chew on what
particular weird zig-zag of metal will give you the pattern you *want*
rather than living with the pattern of an antenna designed without the
additional conductors in place.

For attic antennas, you could have a figure of merit based on some
combination of minimizing loss and minimizing the currents induced in
the fixed conductors. Keep that RF off the house wiring, and the noise
off the antenna, while still radiating most of your power?

I guess there was some prior discussion of use of genetic optimization
processes to design amateur antennas being of dubious legality because
of patents... if this is the case, forget I even posted it.

I could imagine this being of great use to the restricted-space amateur
if the code to do it would be tractable and a sensible interface were
present.

Roy, you wanna write an EZNEC plugin that takes wires #n to #m and
fiddles with them until the far field is within some tolerance of a
desired pattern? I bet it would sell...

It's a thought. I'd be interested in what other people think... the
word "genetic" could even be dropped if it is a sticking point since I
don't really care what particular approach is taken to push the design
toward some goal.

73,
Dan
N3OX
www.n3ox.net

wrote:
It's a thought. I'd be interested in what other people think... the
word "genetic" could even be dropped if it is a sticking point since I
don't really care what particular approach is taken to push the design
toward some goal.

The genetic algorithms are recursive. Maybe "recursive" would
be a better term for something that is not alive.
--
73, Cecil http://www.qsl.net/w5dxp

In article .com,
wrote:

I'm interested in the possibility of using this kind of search of the
space of possible antennas in those situations that we all end up in,
you've put the antenna too close to the tower, you're shoehorning it
into the attic, or you really really need to mount it on the bumper and
not on the roof.

Has anyone written any software that could do this? I guess what I'm
imagining is that you input the location and extent of fixed
conductors, the tower, the car (yes i know it's hard), the wiring in
the walls, whatever, and then you let the computer chew on what
particular weird zig-zag of metal will give you the pattern you *want*
rather than living with the pattern of an antenna designed without the
additional conductors in place.

There are already some general-purpose genetic-optimization packages
freely available in languages such as Perl.

It ought to be possible to write a set of wrappers which uses such a
package to manage the "genome" of an antenna-design space being
searched, convert each individual genome to a NEC configuration, run
NEC2 or a similar package, and extract the results.

For attic antennas, you could have a figure of merit based on some
combination of minimizing loss and minimizing the currents induced in
the fixed conductors. Keep that RF off the house wiring, and the noise
off the antenna, while still radiating most of your power?

I'm by no means an expert in this field, but based on some reading I
did a month or so ago I have a feeling that a single figure-of-merit
is not necessarily the right way to go. The search process can end up
being extremely (and perhaps excessively) sensitive to small changes
in the weighting factors.

Rather, most people these days seem to approach such problems from the
point of view of Pareto optimality. That is, there are a number of
figures of merit, which evaluate or rate different (and usually
independent) objectives. A possible solution is Pareto-optimal if
any change you make in it results in at least one figure of merit
getting worse.

Optimization system based on Pareto optimality maintain a population
of solutions (many or most of which are Pareto-optimal) and then
generate new solutions from that set (possibly by "genetic" means).

You end up with a bunch of workable Pareto-optimal solutions, and then
pick from among them at the end of the process - possibly by applying
a final formula to merge the various figures-of-merit for each
solution into a final "how good is it?" figure.

Deferring this final merging until the end of the process seems to
help keep the search system from prematurely glomming onto what turns
out to be a local maximum in the goodness function.

It'd also be a good idea to have some way of evaluating how "brittle"
a possible solution is. If even a slight change in one of the
parameters causes a previously-optimal solution to become very
non-optimal, then it's probably not a good choice for use in an
amateur installation.

You might want to look for clusters of Pareto-optimal solutions, and
then pick one from near the center of the cluster as being both
"representative", and also "robust"... small changes in the parameters
wouldn't take you outside of the general area of the cluster.

I guess there was some prior discussion of use of genetic optimization
processes to design amateur antennas being of dubious legality because
of patents... if this is the case, forget I even posted it.

That's certainly an issue. Someone wanting to pursue this might be
well advised to do some research into older texts, find out what the
state-of-the-art was a couple of decades ago, and base any
implementation solely on techniques which were in the public domain
back then or which were covered by patents which have since lapsed.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!

On 25 May 2006 15:13:30 -0700, "
wrote:

GENETIC ANTENNA OPTIMIZATION

There, I got it out.

Hi Dan,

Hear the one about the constipated antenna designer?
He worked it out with a smith chart.

GA algorithms in this field are used by those with too much money, or
too much time on their hands, or both.

Experience generally trumps the quality of these computational hungry
products. You can search NASA for its recent GA generated antennas
(offered as a new thread some several months ago). They make nice
modern art designs, but it is arguable the effort was worth the
outcome.

73's
Richard Clark, KB7QHC