Nonlinear pre-emphasis for improving QAM performance over channels with additive and multiplicative noise

This paper presents a new approach for improving performance of communication systems immersed in additive and multiplicative noise. Multiplicative noise manifests itself as phase noise and gain noise. Considering modulations with symbols distributed among several magnitudes, the additive noise presents a uniform distribution over the symbols while multiplicative noise increases with symbol magnitude. As a result symbols with a larger magnitude tend to dominate the symbol error rate of the system. We propose to use invertible nonlinear transformations to distort the constellation prior to modulation and invert the distortion in the receiver prior to decision making. The main effect of the transformation is to decrease the contribution of additive noise in the larger-magnitude symbols while increasing it in the low-magnitude symbols. We demonstrate that as a result the symbol error rate distribution with respect to symbol magnitude unifies and the overall symbol error rate decreases. Analysis is carried out for a second order polynomial distortion. Two optimization criteria are proposed: minimizing the symbol error rate and minimizing the error rate variation with respect to symbol magnitude.