Segmented Digital Predistortion for High-PAPR RF Linearization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital predistortion techniques suffer performance degradation when used with complex modulation schemes characterized by high peak to average power ratios (PAPR) and wide signal bandwidth, especially under dynamic conditions such as LTE-TDD mode.
Innovation Solution
The implementation of segmented digital predistortion (DPD) techniques, where input sample blocks are classified based on time slice and dynamic range segments, allowing for specific DPD model selections and predistortion coefficient application to improve power amplifier linearization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional digital predistortion techniques are used with complex modulation schemes, then power amplifier linearization can be improved, but performance degrades significantly under high PAPR and wide signal bandwidth conditions
Solution Approach 1:
The input signal is divided into multiple segments based on signal characteristics (PAPR levels, bandwidth conditions, modulation types). Each segment is processed by a dedicated DPD model optimized for its specific characteristics, preventing performance degradation that occurs when a single DPD model tries to handle all signal conditions
Solution Approach 2:
The system dynamically selects and switches between different DPD models based on real-time signal conditions. The DPD apparatus monitors signal parameters and adapts the predistortion approach accordingly, transitioning between models with different complexity levels to maintain optimal performance under varying operational conditions
2Device complexity
If a single DPD model is used for all signal conditions, then device complexity is reduced, but DPD performance suffers under dynamic conditions such as LTE-TDD mode
Solution Approach 1:
The DPD apparatus is designed with multiple DPD models that can handle different signal conditions, making the system universal rather than specialized for a single condition. The system includes DPD models for various modulation schemes, PAPR levels, and bandwidth conditions, allowing it to function effectively across diverse operational scenarios
Solution Approach 2:
The system employs dynamic model selection that adapts to changing signal conditions in real-time. Switching mechanisms enable transitions between different DPD models based on current operational parameters, ensuring optimal performance during dynamic conditions like LTE-TDD mode while managing complexity through intelligent selection rather than simultaneous operation of all models
3Manufacturing precision
If backoff levels are increased to meet linearity conditions, then spectrum emission mask and adjacent channel power specifications are satisfied, but power amplifier efficiency decreases
Solution Approach 1:
Different DPD models with varying precision levels are applied to different signal segments based on their specific characteristics. For signals that require strict linearity (high PAPR, wide bandwidth), more complex DPD models are used to achieve better linearization with smaller backoff, while simpler models handle signals with more relaxed requirements, optimizing the trade-off between linearity and efficiency locally for each signal condition
Data Source
Figure 1
Figure 2
Figure 3
AI summary
In an RF transmitter, a digital predistortion circuit receives a sequence of input sample blocks, and performs a digital predistortion process to produce a predistorted output signal. The digital predistortion process includes selecting a set of predistortion coefficients for an input sample block from a plurality of different sets of predistortion coefficients. Each of the plurality of different sets of predistortion coefficients is associated with a different combination of one of a plurality of time slices within a radio frame and one of a plurality of power ranges. The selected set of predistortion coefficients is associated with a time slice within which the input sample block is positioned and a power range calculated for the input sample block based on block power statistics of the sample block. The process also includes applying the selected set of predistortion coefficients to the input sample block to produce the predistorted output signal.