Multistage Digital Predistortion for Frequency-Selective Linearization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital predistortion (DPD) systems for power amplifiers (PAs) are complex and inefficient, failing to meet stringent linearization requirements across different frequency bands, leading to excessive computational complexity and power consumption, especially in applications like Sub-Band Full Duplex (SBFD) and Millimeter Wave (mmW) operations.
Innovation Solution
A multistage digital predistortion (DPD) approach is implemented, where the signal is processed in stages with varying sampling rates and complexity, with earlier stages covering broader bandwidths at higher rates and later stages focusing on specific sub-bands with more stringent requirements, allowing for reduced complexity and targeted linearization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional digital predistortion (DPD) systems are used to meet stringent linearization requirements across different frequency bands, then linearization performance is improved, but computational complexity and power consumption increase excessively
Solution Approach 1:
The patent divides the frequency spectrum into multiple sub-bands and processes each sub-band separately through parallel DPD filters. This segmentation allows the system to meet stringent linearization requirements for specific bands without applying full computational complexity across the entire spectrum, thereby reducing overall computational burden while maintaining precision where needed.
Solution Approach 2:
The patent applies different DPD filter characteristics and computational resources to different frequency sub-bands based on their specific linearization requirements. Bands with stricter OoBE requirements receive more aggressive predistortion processing, while other bands use simpler processing, optimizing the balance between linearization performance and computational complexity locally rather than uniformly across all bands.
2Device complexity
If frequency selective linearization is applied to specific sub-bands with reduced sampling rates, then computational complexity is reduced, but linearization performance may be compromised
Solution Approach 1:
The patent dynamically adjusts the sampling rate and filter order for each sub-band based on its specific bandwidth and linearization requirements. This dynamic adaptation allows the system to use lower sampling rates for bands with less stringent requirements, reducing computational complexity, while maintaining higher sampling rates and more aggressive filtering for bands requiring superior linearization performance.
Solution Approach 2:
The patent changes key processing parameters including sampling rate, filter order, and predistortion coefficients for different sub-bands. By optimizing these parameters individually for each band's characteristics and requirements, the system achieves computational efficiency in bands where full performance is unnecessary while preserving linearization accuracy in bands where it is critical.
3Use of energy by moving object
If multistage DPD approach is implemented with varying sampling rates, then power consumption is reduced, but system design complexity increases
Solution Approach 1:
The patent implements a multistage DPD architecture where the signal processing is divided into multiple parallel stages, each handling specific sub-bands at optimized sampling rates. This segmentation enables power-efficient processing by avoiding uniform high-rate processing across all bands, while the modular stage-based structure manages design complexity through systematic organization of processing functions.
Solution Approach 2:
The patent creates a universal multistage DPD framework that can handle multiple sub-bands with different requirements using a common architectural structure. This multi-functional design reuses processing blocks and algorithms across different bands and stages, reducing overall system design complexity despite the increased functionality required to handle varying sampling rates and band-specific requirements.
Data Source
AI summary
In various embodiments of the present disclosure, provided is a system to perform reduced complexity frequency selective linearization. In an embodiment, multistage digital predistortion (DPD) can linearize different parts of the spectrum differently. The first stage captures the signal across the entire linearization bandwidth where the least stringent linearization requirement and general unwanted emission requirements are targeted. Second stage pre-distorts a down-sampled signal where second least stringent requirements across the linearization bandwidth are aimed for. This “peel-off” process continues until the final stage which pre-distorts the portion of the spectrum with the most stringent linearization requirements. Consequently, earlier stages run at higher rates to linearize wider spectrum while later stages run at lower rates to selectively further linearize portions of spectrum that have already passed through earlier stages.