Modified systolic array implementation of QRD-RLS algorithms for solving the generalized Hammerstein models

Abstract

Generalized Hammerstein models can be used to accurately mimic the dynamic nonlinear behaviors of RF power amplifiers or transmitters. A high efficient parameter extraction algorithm for the generalized Hammerstein model is an essential problem for implementing the nonlinear model in a cheap microprocessor. In this paper, a modified systolic array based QRD-RLS algorithm, which is suitable for implementing in a FPGA-based soft-core microprocessor such as Altera NIOS II, is presented to extract the parameters of the generalized Hammerstein models of an RF power amplifier. The validation results based on a LDMOS power amplifier with a 4-carrier 3GPP WCDMA signal reveal that the modified systolic array based QRD-RLS algorithm can accurately identify the model parameters.