Neural Network Predistortion for Dynamic Power Amplifier Nonlinearity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional predistortion models and power amplifier models are deficient in nonlinear representation capability and suffer from poor predistortion correction due to lagging responses to dynamic changes in the power amplifier system caused by separated processing with neural networks.
Innovation Solution
A predistortion system incorporating a predistortion multiplier, complex neural network, and radio frequency power amplifier output feedback circuit, where training is performed with direct connection to the power amplifier, enabling simultaneous learning and predistortion correction, utilizing real-time power normalization units and complex error back propagation algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If separated processing is used when applying neural networks to predistortion systems, then the system structure is simplified and easier to implement, but the response to dynamic changes of the power amplifier system becomes lagging and predistortion correction performance deteriorates
Solution Approach 1:
The patent merges the training process and predistortion correction process into a single integrated operation. The neural network is directly connected to the power amplifier system during training, allowing the system to learn and adapt to dynamic changes in real-time while performing predistortion correction, thereby eliminating the lagging response caused by separated processing
2Device complexity
If conventional predistortion models are used, then the system complexity is reduced, but the nonlinear representation capability is inherently deficient
Solution Approach 1:
The patent replaces conventional mathematical predistortion models with a neural network-based model. The neural network's distributed parallel structure and non-linear activation functions provide superior nonlinear representation capability compared to traditional polynomial or memory polynomial models, while the network's adaptive learning capability allows it to automatically capture complex nonlinear relationships in the power amplifier system
3Reliability
If neural networks are applied with direct connection to power amplifier for training, then the nonlinear representation capability and adaptability are enhanced, but the system complexity and processing requirements increase
Solution Approach 1:
The neural network performs self-learning and self-adjustment by directly processing feedback from the power amplifier system during training. The network automatically adjusts its weights and biases through backpropagation to optimize predistortion correction performance, reducing the need for manual system configuration and complex external training equipment
Data Source
AI summary
Disclosed are a predistortion method and system, a device, and a non-transitory computer-readable storage medium. The predistortion method is applicable to a predistortion system which may include a predistortion multiplier, a complex neural network, and a radio frequency power amplifier output feedback circuit. The method may include: inputting a training complex vector to the predistortion system to obtain a complex scalar corresponding to the training complex vector, which is output by the predistortion system; training the predistortion system based on the training complex vector and the complex scalar until a generalization error vector magnitude and a generalization adjacent channel leakage ratio corresponding to the predistortion system meet set requirements; and inputting a service complex vector to the trained predistortion system to obtain a predistortion corrected complex scalar.