Neural Network Power Amplifier Linearization With Reduced DPD Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The calculation process for Digital Pre-Distortion (DPD) in power amplifiers is lengthy and complex, requiring significant computing resources due to the need to calculate the inverse function of nonlinear components, complicating the linearization operation.
Innovation Solution
A neural network model is used to generate an intermediate signal, which is amplified by a power amplifier and fed back into an inverse model of an ideal power amplifier to calculate the error, simplifying the linearization process by treating nonlinear components as interference and updating weights based on a loss value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If Digital Pre-Distortion (DPD) is used to perform linearization of power amplifiers, then power-added efficiency is improved, but calculation complexity and computing resources required increase significantly
Solution Approach 1:
The patent replaces the traditional mathematical inverse function calculation system with a neural network model system. Instead of computationally intensive inverse function calculations, the neural network learns the inverse relationship through training, substituting complex arithmetic operations with neural network inference that is more efficient to execute.
Solution Approach 2:
The neural network model is trained in advance to learn the inverse function relationship of the power amplifier. This preliminary training phase captures the complex nonlinear characteristics, so that during actual operation, the pre-trained network can quickly perform linearization without real-time complex calculations.
2Manufacturing precision
If the inverse function of nonlinear components is calculated during DPD process, then linearization is achieved, but processing time increases
Solution Approach 1:
The patent substitutes the time-consuming inverse function calculation process with a neural network model that has already learned the inverse relationship during training. The neural network performs rapid inference to achieve the same linearization effect without real-time complex mathematical computations.
Solution Approach 2:
The complex inverse function relationship is pre-computed and stored in the neural network model during the training phase. This allows the system to bypass time-consuming calculations during actual operation by directly applying the pre-learned inverse transformation through neural network forward propagation.
3Manufacturing precision
If traditional DPD methods are used to address nonlinear components, then linearization is achieved, but computational resources required are huge
Solution Approach 1:
The patent replaces the traditional computational approach based on mathematical inverse functions with a neural network-based approach. The neural network model, once trained, requires significantly fewer computing resources for inference compared to real-time inverse function calculations, reducing the power and computational resource requirements while maintaining linearization accuracy.
Data Source
AI summary
A linearization method for a power amplifier includes the following steps. An input signal and an error associated with the input signal are received. The input signal and the error are input into a neural network model. A first intermediate signal is generated using the neural network model. The first intermediate signal is input into a power amplifier, so that the power amplifier outputs a first output signal. The first output signal is fed back into an inverse model of an ideal power amplifier, so that the inverse model outputs a second output signal. The difference between the first intermediate signal and the second output signal is calculated to obtain the error.