Scott in Aztlan wrote:

Suppose you were participating in a search for a hidden radio transmitter, only
instead of the usual radio receiver with a directional antenna you have a
receiver with an omnidirectional antenna and a GPS receiver. As you wander
around, you collect positions from the GPS and signal strength values at those
positions from the radio; your goal is to crunch these data points into an
estimate of the transmitter's location.

Clearly you could use trilateration with any 3 of the data points and get an
estimate, but how would you make use of the fact that you have an arbitrary
number of data points?

If I understand clearly you are interested in locating an RF source which
has nothing to do with GPS by using RF receivers and nondirectional antennae.
Trilateration is only of value if one can establish accurate timing distances,
synchronized between all receivers.

If you used directional RF antennae, then, in conjunction with receiver positions
determined by GPS receivers, you could simply triangulate to estimate the RF
source position.

Isn't there some algorithm which, the more data points it
is given, the better its estimate gets (similar to the way you can average a
series of position readings taken from a stationary GPS receiver to compensate
for the effects of SA)?

Of course SA was permanently discontinued more than three year ago.

Interagency GPS Executive Board (IGEB) - Special Statement
http://www.ostp.gov/NSTC/html/pdd6.html
http://www.igeb.gov/sa.shtml

Presidential Policy & PRESIDENTIAL DECISION DIRECTIVE NSTC-6
http://gps.faa.gov/gpsbasics/PresPolicy-text.htm
http://www.peterson.af.mil/GPS_Suppo...ts/gps_pdd.htm

Selective Availability
http://gps.faa.gov/gpsbasics/SA-text.htm

Joint Program Office
http://gps.faa.gov/gpsbasics/JPO-text.htm