Dynamic nonlinear modeling system of broadband radio frequency power amplifier

Abstract

The invention discloses a dynamic nonlinear modeling system of a broadband radio frequency power amplifier. The system comprises a non-memory effect processing unit, a coefficient decomposition memoryeffect processing unit and an adder, wherein the non-memory effect processing unit is represented by adopting a static nonlinear non-memory polynomial form, and the coefficient decomposition memory effect processing unit is realized by adopting a coefficient decomposition memory polynomial model, a coefficient decomposition generalized memory polynomial model or a coefficient decomposition improved generalized memory polynomial model. The system has the advantages that on the basis of ensuring high modeling precision and low complexity, the number of model coefficients can be reduced, the amount of calculation resources consumed during training is reduced, and hardware implementation is facilitated.

Description

technical field [0001] The invention relates to a dynamic nonlinear modeling system, in particular to a dynamic nonlinear modeling system of a wide-bandwidth radio frequency power amplifier. Background technique [0002] In recent years, dynamic nonlinear modeling systems (ie, dynamic nonlinear models) of wideband RF power amplifiers have been extensively studied. As the basis of dynamic nonlinear model, Volterra series model has high modeling accuracy, but there is a problem of high complexity. In order to overcome the problem of high complexity of the Volterra series model, the researchers put forward some simple dynamic nonlinear models by serial-parallel combination of the Volterra series model, such as the traditional memory polynomial nonlinear model, the improved Hammerstein nonlinear model and Generalized memory polynomial nonlinear models, etc. Although the complexity of these nonlinear models is lower than that of the Volterra series model, in order to ensure the...

AI Technical Summary

Problems solved by technology

Although the complexity of these nonlinear models is lower than that of the Volterra series model, in order to ensure the modeling accuracy of these nonlinear models, the nonlinear order and memory depth will be relatively high, and with the nonlinear order and memory depth As the value increases, the number of model coefficients will increase significantly, resulting in a significant increase in the computing resources consumed during training, and there are problems such as difficulty in hardware implementation.

Images