It's fun with software radio to take a long recording of a chunk of spectrum and do the fourier transform and plot it. There will be clear lines indicating car

It's fun with software radio to take a long recording of a chunk of spectrum and do the fourier transform and plot it. There will be clear lines indicating carriers, even ones too faint to be heard over the static. Software can do the same trick for decoding a modulation (at the expense of latency in decoding). Modern tech allows a further neat hack here. If you have a very stable local clock (which any modern digital device does) you can correct for frequency errors in the transmission. Once the clocks are synced, you know exactly when the next symbol should be. And you know how much the transmitter frequency as actually received, is varying from what it should be. And since a convolutional coding with a pseudorandom number might be used, you also know what the next symbol will be (only a small number are valid in any state). With all that together, in effect, the decoder knows when and where the signal will be, and at least part of what it will look like, before it receives it. So it can be compared with what is actually received, making it possible to pick it up well below the noise floor. Just add some error correcting codes for robustness. Amateurs have successfully communicated data between California and New Zealand on longwave using only 1 watt of radiated power at the transmitter. (But the baudrate for that is measured in hours per bit. And the receiving antenna was considerable.)