Stuart W. Card <swc@uc1.ucsu.edu> wrote:

Trying to transmit meaningful communications at high data rates over water pipes would be a TREMENDOUS technical challenge, requiring a vast amount of signal processing to compensate for multipath reflections, impedance mismatches, thermal noise, pickup of an almost variety of interference soures...

Well, I'm sure that multipath reflections would cause wave interference which would garbel whatever was modulated on that signal, however, the carrier wave itself would still be detectable. Suppose you were to broadcast a short burst of a signal at 250,000 hz into the water system. You would get lots of echos, noise, etc, but it would still be detectable as a 250,000 hz signal. It might not be possible to tell exactly what that signal was, but you could still detect the frequency; you could tell that, in fact, a 250,000 hz signal had been sent. Now you could determine the minimum interval of time necessary between bursts for the receiver to be able to identify them as seperate bursts. Having done this, you could send 250,000 hz bursts at regular interals, and then accompany some with 500,000 hz bursts (one bits) and leave the other pulses without an accompanying transmission (zero bits). I'm not really sure what frequencies would work best, that would probably have to be determined experimentally.