Direct Learning Control for Stable Power Amplifier Predistortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Direct Learning Algorithm (DLA) for power amplifiers in wireless communication systems faces instability due to numerical error accumulation, leading to divergence and increased inter-modulation products, which violate 3GPP specifications and hinder efficient operation in compressed regions.
Innovation Solution
The solution involves introducing an error scaling factor and monitoring error gradients to detect divergence, with adaptive parameter adjustment and restart mechanisms, along with the use of an auto-correlation matrix with added noise to stabilize the algorithm and prevent divergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If continuous adaptation of DPD algorithm parameters is performed to maintain linearization performance, then the system can track power amplifier changes and maintain compliance, but numerical errors accumulate causing algorithm instability and divergence
Solution Approach 1:
The patent implements periodic restoration of the DLA algorithm by detecting divergence through error gradient monitoring and resetting parameters to predetermined values when divergence is detected. This periodic intervention prevents continuous numerical error accumulation while maintaining overall system compliance, directly resolving the contradiction between continuous adaptation and stability.
Solution Approach 2:
The patent employs feedback mechanisms by continuously monitoring the error gradient and comparing it against thresholds to detect algorithm divergence. This feedback loop enables the system to identify when numerical errors are accumulating and trigger restoration actions, balancing continuous adaptation with stability maintenance.
2Use of energy by moving object
If the power amplifier operates in compressed regions to reduce power consumption, then energy efficiency improves, but inter-modulation products increase violating 3GPP specifications
Solution Approach 1:
The patent applies preliminary action by implementing digital pre-distortion that anticipates and compensates for non-linear effects before they occur. The pre-distorter modifies the input signal to counteract the power amplifier's non-linear behavior, allowing operation in compressed regions while preventing excessive inter-modulation product generation.
Solution Approach 2:
The patent employs preliminary anti-action by introducing a pre-distorter that applies the opposite of the expected non-linear distortion. This counter-distortion approach cancels out the harmful inter-modulation products generated by compressed region operation, enabling energy-efficient operation while maintaining spectral compliance.
3Device complexity
If a fixed number of iterations is used in the iterative inversion process to obtain the inverse model, then hardware complexity is reduced, but convergence accuracy may be insufficient
Solution Approach 1:
The patent applies partial action by using a fixed, limited number of iterations in the iterative inversion process rather than continuing until full convergence. This approach accepts partial convergence that is sufficient for practical purposes while avoiding the excessive computational complexity and hardware requirements that would result from pursuing complete convergence.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The present invention addresses method, apparatus and computer program product for controlling a Direct Learning Algorithm. Thereby, a power amplifier operating in a non- linear state is controlled. A signal to be amplified is input to a pre-distorter provided for compensating for non-linearity of the power amplifier. The pre-distorted output signal is forwarded from the pre-distorter to the power amplifier. Parameters of the pre-distorter are adapted in plural steps based on an error between a linearized signal output from the power amplifier and the signal to be amplified using an adaptive direct learning algorithm. It is detected whether the error diverges; and adapting of the parameters of the pre-distorter is stopped when it is determined that the error is diverging.