I was reading this a month or so ago, though the date of the work is from 2007 onwards.
http://www.boegan.it/index.php?id=125
An issue was that they needed some kind of 'scale factor' to correct for the beacon's reported 'distance to casualty' - they originally assumed that rock was twice as attenuating, but found on subsequent tests that it was highly dependent on the nature of the rock, and settled on a figure of 1.6-1.8 . I assume the metric given out is calibrated for avalanche snow.
The modern digital beacons work at 457kHz.
My understanding of the technology is that the transponders themselves are permanently transmitting pulses at low power (and do so for at least a week), and each transponder can then be flicked into locate/receiving mode to triangulate onto another's signal.
We wondered whether we might be able to thus shove a beacon at the extremity of one cave, and then attempt to hone in on the signal from complicated regions of passage in another cave that were believed to be near.
I guess the benefits of this technology compared to DIY beacons (which perhaps use more suitable frequencies + accurate methods for deriving distances) is that the beacons are rugged, small, reliable, power efficient and there's lots of devices out there. Just need to find some Alpinists who wouldn't mind lending a couple for the summer
