An information theoretical analysis of self-demodulating waveform channels

This paper studies nonlinear waveform channels which show a self-demodulating characteristic from an information theoretic point of view. These channels contain in the most simple case a square-law nonlinearity followed by a lowpass filter. In this paper, the use of orthonormal series expansions based on sets of sampling functions is discussed for both the input and the output waveform. Based on this, an exact discrete-time representation of a self-demodulating waveform channel is stated. The information theoretical analysis is then discussed based on the statistics of the waveform samples. Finally, bounds on the capacity of the self-demodulating waveform channel are proposed. Furthermore, achievable rates obtained by modulation techniques that are robust for the communication over corresponding channels are presented.