Hybrid Digital Predistortion for RF Power Amplifier Memory Effects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital predistortion (DPD) techniques face challenges in achieving accurate linearity and efficiency in radio frequency (RF) power amplifiers, particularly due to increasing sampling rates and nonlinear effects such as charge trapping in GaN transistors.
Innovation Solution
A hybrid DPD approach combining a basis-function-based actuator and a neural network-based actuator, which uses a set of basis functions and neural networks to perform predistortion operations, respectively, and combines their outputs to generate a predistorted signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional DPD techniques are used, then implementation is simpler, but linearity and efficiency improvement is insufficient due to nonlinear effects like charge trapping
Solution Approach 1:
The patent combines two different DPD architectures (basis-function-based and neural network-based actuators) into a hybrid system. The basis-function actuator handles immediate predistortion while the neural network actuator models long-term memory effects like charge trapping, achieving superior linearity by merging complementary approaches rather than using either alone
Solution Approach 2:
The DPD system uses a composite architectural approach, integrating different computational models (basis functions and neural networks) with complementary strengths. This composite structure allows the system to handle both short-term and long-term nonlinear effects, improving manufacturing precision in terms of PA linearity while managing the complexity through modular design
2Reliability
If neural network-based DPD is used, then long-term memory effects are modeled accurately, but computational complexity and processing time increase
Solution Approach 1:
The DPD system segments the predistortion task into two parts: the basis-function actuator handles immediate, computationally simple predistortion, while the neural network actuator separately models long-term memory effects. This segmentation allows each component to be optimized independently, maintaining modeling accuracy while managing computational complexity through division of labor
Solution Approach 2:
The system applies partial action by using the neural network only for specific long-term memory effects rather than attempting to model all nonlinearities. The basis-function actuator handles the majority of predistortion needs, while the neural network provides supplementary correction for memory effects, reducing overall computational complexity while maintaining necessary accuracy
3Measurement precision
If higher sampling rates are used, then signal processing accuracy improves, but processing delay and computational load increase
Solution Approach 1:
The system uses periodic action by updating the neural network actuator parameters at lower rates compared to the basis-function actuator. The basis-function actuator operates at full sampling rates for immediate response, while the neural network parameters are updated periodically based on accumulated data, reducing processing delay while maintaining signal processing accuracy through complementary update strategies
Data Source
AI summary
Systems, devices, and methods related to hybrid basis function, neural network-based digital predistortion (DPD) are provided. An example apparatus for a radio frequency (RF) transceiver includes a digital predistortion (DPD) actuator to receive an input signal associated with a nonlinear component of the RF transceiver and output a predistorted signal. The DPD actuator includes a basis-function-based actuator to perform a first DPD operation using a set of basis functions associated with a first nonlinear characteristic of the nonlinear component. The DPD actuator further includes a neural network-based actuator to perform a second DPD operation using a first neural network associated with a second nonlinear characteristic of the nonlinear component. The predistorted signal is based on a first output signal of the basis-function-based actuator and a second output signal of the neural network-based actuator.