DPD Inverse Model Updating for Rare Peak Signal Distortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High power amplifiers in wireless base stations exhibit non-linear input-output characteristics, leading to distortion in transmission signals, and existing distortion compensation methods struggle with accuracy when the input signal with maximum value appears infrequently, resulting in incomplete model updates and inaccurate compensation.
Innovation Solution
A distortion compensation circuit that estimates and updates an inverse model based on input and output signals, regardless of the presence of the peak value, and uses multiple threshold values or predicted maximum values to ensure accurate coverage of signal ranges, allowing for continuous and precise distortion correction even when the input signal with maximum value has a low appearance frequency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the inverse model is updated only when an input signal with peak value is detected, then the model accuracy is improved, but the update frequency decreases when communication data amount is small
Solution Approach 1:
The system performs preliminary sampling of input and output signals continuously, preparing data for model estimation even before peak values are detected. This preliminary action ensures that when peak values do occur, the model can be immediately updated with accurate data, resolving the contradiction between maintaining high update frequency and ensuring model accuracy.
Solution Approach 2:
The distortion compensation circuit uses its own operational data (input and output signals from the HPA) to automatically update its inverse model without external intervention. This self-service mechanism ensures continuous model improvement while maintaining high update frequency, as the system continuously monitors its own performance and updates accordingly.
2Productivity
If the inverse model is updated frequently with limited data, then the update frequency is improved, but the model coverage range decreases
Solution Approach 1:
The system continuously samples and stores input-output signal pairs in advance, building up a comprehensive dataset that covers the full signal range. When updating the inverse model, this pre-collected diverse data ensures the model maintains broad coverage even with frequent updates, resolving the contradiction between update frequency and range coverage.
Solution Approach 2:
The sampling and model estimation process operates continuously without interruption, ensuring that the inverse model is constantly refined with new data while maintaining coverage of the entire signal range. This continuous action prevents range reduction that would occur with intermittent updates based on limited data.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A distortion compensation circuit capable of realizing highly accurate distortion compensation by updating a model even under a situation in which the appearance frequency of an input signal having a maximum value is low. A DPD processor 2 includes an inverse model estimation unit 22, which estimates an inverse model for a model expressing input-output characteristics of an HPA 6 based on an input signal S1 to the HPA 6 and an output signal S10 from the HPA 6, a distortion compensation unit 26, which compensates for distortion of the input-output characteristics by adding the inverse model to the input signal S1, and a sampling circuit 20, which samples the signals S2 and S10 in a predetermined time immediately before the sampling and inputs the signals S2 and S10 to the inverse model estimation unit 22. The inverse model estimation unit 22 updates the inverse model based on S2 and S10 input from the sampling circuit 20 regardless of whether the maximum value that the input signal S1 can take is included in a range sampled by the sampling circuit 20.