Making waves, making static
Suppose you are concerned about the physical detection of your electromagnetic transmission, regardless of code-cracking concerns. You would want to make your signal as much like random noise as possible.
To do this, you would use the formula
signal = (noise power - signal power)^1/2
and then make sure your noise power roughly equals your signal power so there is no detectable signal.
But you must use a deterministic algorithm that is good at approximating noise and that, added to typical noise over time, nearly equals signal power. The decoder computer then uses the algorithm's inverse to subtract out the pseudorandom noise.
A disadvantage is that pseudorandom functions may be revealed by statistical analysis. Also, monitors may be suspicious of the amplitude of the noise and may detect it as coming in bursts, indicating message traffic.
Such a system might prove useful in masking keyboard transmissions. Keyboards, which typically (though not by design) give off recognizable electromagnetic waves for each key, are vulnerable to remote eavesdropping equipment.
I have not checked with any expert to see whether this idea is practical.