Optimum Performance of Nonlinearly Distorted Signals with General Distributions.

Nonlinearities are one of the most common impairments that take place in communication systems. Conventionally, nonlinear distortion is regarded as an additional noise component that degrades the performance, leading to irreducible error floors. However, it was shown that the optimum performance of zero-mean Guassian-like signals (such as the ones associated with regular orthogonal frequency division multiplexing (OFDM) modulations) submitted to severe nonlinear distortion effects can even be better than the one with linear transmitters.In this paper, we consider signals with non-centered Gaussian distributions with strong nonlinear distortion effects. We study the optimum performance, showing that the performances gains observed for centered Gaussian signals can be extended to other distributions. These remarkable theoretical results are confirmed by simulations, leading to a different look on nonlinear effects. ¹