Digital Predistortion LUT Updating for Noisy Nonlinear Amplifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital predistortion methods for power amplifiers face issues with noise in output measurements leading to biased convergence and the inability to commute nonlinear filters, affecting the accuracy of predistortion.
Innovation Solution
The method involves time-aligning complex input and output signals, performing array division, and using curve-fitting algorithms to update real and imaginary lookup tables, and iteratively combining these tables to improve the digital predistortion circuit's performance, specifically by computing and updating parameters of non-linear filter blocks in both direct and indirect learning architectures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If indirect learning method is used to reduce computational complexity, then device complexity is reduced, but measurement precision deteriorates due to noise in output measurements leading to biased convergence
Solution Approach 1:
The patent inverts the traditional indirect learning approach by using direct learning architecture where the predistorter parameters are directly optimized to minimize the error between the actual and desired output. This inversion allows the system to directly address the convergence accuracy issue while maintaining computational efficiency through iterative optimization algorithms.
Solution Approach 2:
The patent implements a feedback mechanism where the actual output of the power amplifier is continuously monitored and used to update the predistorter parameters. This feedback loop enables the system to compensate for noise and measurement errors by iteratively adjusting the parameters to minimize the difference between desired and actual output, thereby improving convergence accuracy.
2Device complexity
If indirect learning method is used, then computational complexity is reduced, but manufacturing precision deteriorates because nonlinear filters cannot be commuted
Solution Approach 1:
The patent inverts the traditional approach by directly optimizing the predistorter parameters through direct learning, bypassing the need for post-inverse filtering. This inversion eliminates the commutation issue entirely, as the system directly computes the predistortion parameters without requiring the nonlinear filters to be commuted, thereby improving predistortion accuracy.
Solution Approach 2:
The patent extracts the essential predistortion function from the complex indirect learning architecture, focusing directly on optimizing the predistorter parameters without the intermediate post-inverse filtering step. This extraction simplifies the system while improving accuracy by eliminating the source of the commutation problem.
3Use of energy by moving object
If power amplifier is operated in non-linear region near saturation, then power efficiency is improved, but adjacent channel leakage ratio becomes unacceptable due to spectrum expansion
Solution Approach 1:
The patent applies preliminary anti-action by pre-distorting the input signal to counteract the non-linear effects of the power amplifier before amplification occurs. The predistorter introduces an inverse non-linearity that cancels out the amplifier's non-linearity, preventing spectrum expansion and adjacent channel leakage while allowing the amplifier to operate efficiently in its non-linear region.
Solution Approach 2:
The patent converts the harmful non-linear effects of the power amplifier into a benefit by using the same non-linearity in reverse. The predistorter deliberately introduces non-linear distortion that is the inverse of the amplifier's non-linearity, so that when the signal passes through the amplifier, the two non-linearities cancel each other out, resulting in linear output with high efficiency.
Data Source
AI summary
A method used in a transmitting device comprising a non-linear amplifier and a digital predistortion (DPD) circuit. The method updates real and imaginary look-up tables used by the DPD. The method comprises: i) time-aligning a complex input signal, A, and a complex output signal, E. Signal A is a scan from 0 to a maximum value comprising N discrete points and signal E also comprises N discrete points. The method comprises: ii) performing an array division of A/E=(Yr, Yi), where Yr and Yi are the real and imaginary components, respectively; and iii) computing a real curve and an imaginary curve using curve-fitting algorithms that best fit the data with coordinates (|A|, Yr) and (|A|, Yi). The method stores the real curve in a real lookup table (LUT) and the imaginary curve in an imaginary lookup table (LUT). The method iteratively updates the real LUT and the imaginary LUT.