Optimal Spacing of a Linearly Interpolated Complex-Gain LUT Predistorter

Abstract

Modern spectrally efficient modulation techniques produce radio-frequency signals with varying envelopes. Transmitting these signals through nonlinear amplifiers results in spectral regrowth and degradation of the bit error rate. Lookup table (LUT) predistorters have been widely used to mitigate the effects of nonlinear amplifiers. This paper theoretically proves that the use of linearly interpolated LUT predistorters results in superior rejection of residual nonlinear distortions, as compared with conventional noninterpolated LUT predistorters. The refinement of LUT performance with linear interpolation results in 12-dB additional rejection for every bit of precision, as opposed to 6 dB for noninterpolated LUTs. It is also shown that linearly interpolated LUT predistorters benefit more from optimal spacing than their noninterpolated counterparts. Besides improved mapping fidelity, potential benefits also include reduced gate count and faster convergence of adaptive LUT training algorithms.