Performance of Stochastic Quantizers Employing Non-linear Processing

In the paper, an acquisition setup employing dither is considered. Both linear and non-linear post-processing of the quantized samples are analyzed with the aim of assessing the resolution and accuracy performance of the considered system. Uniformly and Gaussian distributed dither are assumed and results are given that enable the comparison of the two considered post-processing schemes. It is shown that the performance of the non-linear one is often better than that of the linear algorithm, so that the former can be employed to improve the input signal estimation.