[ec429]

I strongly suspect that this will prove to be limited by the RF path; it relies (as far as I can tell) on coherent properties of the wavefront. This means it will break as soon as the signal passes through heterogeneous material (such as a building), or as soon as reflections produce multi-path interference.

And you certainly wouldn't be able to use it at HF - imagine what the ionosphere will do to your carefully constructed wave!

In general, using more parameters of the wave reduces your resilience to noise; the usual approach of extracting only amplitude, frequency and (perhaps) phase is a summation operator that smooths out a lot of interference. Conceptually, this is like how QPSK needs a higher SNR than BPSK does - you're using more parameters, so you're reducing the 'distance' between things you want to distinguish, so you're increasing the chance that a given amount of noise will produce errors.