Another interesting question - I’ve noticed that Chinese broadcast stations often transmit the same AM signal from different transmitters on different frequencies. This is quite common now - two, sometimes three frequencies simultaneously broadcasting the same AM signal. They also experience fading. Theoretically, for three stations, the signal could be recovered by discarding one of the three. But more often, it’s just two stations. I think that’s enough, as each station duplicates both the LSB and USB spectra, so with two synchronized stations (with calibrated signal delay between them), we get four copies of the spectrum, which should be enough for high-quality signal recovery.
Is there any established mathematics for such operations?
An even more interesting question - is it possible to recover an AM signal damaged by fading using only the LSB and USB parts of the spectrum for single AM station? This assumes that the fading affects only one part of the spectrum. I think it might be impossible, as two copies of the spectrum may not be enough to reliably recover the information. But could it still be possible for AM signals using their specific properties?