"Hank Oredson" writes:
http://pulsar.princeton.edu/~joe/K1JT/

Thanks, that's quite interesting, even though the description is very
incomplete. I wish there was more technical documentation there.

I know that a possibly-similar multi-tone FSK scheme was suggested by
Phil Karn KA9Q some years back, though I don't know whether he ever
implemented it. The goal was to be able to write the modem in
software on a PC and connect to a radio's audio channel using a sound
card. He wanted to operate at RTTY speeds (50 bps?) while I'm
thinking in terms of going much slower.

I'm pretty much a newbie to this stuff but I have the impression that
if you can escape that audio constraint (i.e. if you can directly
digitize your receiver's mixer output and use software along the lines
of GNU Radio for demodulation instead of messing with sound cards),
it's better to use wideband direct sequence SS to minimize
interference potential.