L. Gan and E. Abd-Elrady (Austria)
Memory polynomial systems, NFxLMS, prediction error method, predistortion, recursive identification.
Digital Predistortion of memory polynomial systems is considered in this paper. The predistorter is first modeled as a memory polynomial system and its coefficients are estimated based on the Direct Learning Architecture (DLA) approach. The Nonlinear Filtered-x Least Mean Squares (NFxLMS) algorithm is derived assuming that the memory polynomial system has been identified. The Indirect Learn ing Architecture (ILA) approach can be used to estimate the coefficients of the predistorter using the Recursive Prediction Error Method (RPEM) algorithm if the nonlinear physical system has not been identified. The predistorter can also be constructed using linear and nonlinear FIR filters as suggested in [1]. The simulation results show that both the memory polynomial predistorter and the predistorter of [1] can effectively compensate the nonlinear distortion and reduce the spectral regrowth.
Important Links:
Go Back
