Dual Feature Indexed Quadratic Polynomial-Based Piecewise Behavioral Model for Digital Predistortion of RF Power Amplifiers

DOI number:10.1109/TBC.2024.3434625

Affiliation of Author(s):中国科学技术大学信息学院

Teaching and Research Group:电子工程与信息科学系

Journal:IEEE Transactions on Broadcasting

Funded by:国家自然科学基金 NSFC 62371436

Key Words:Behavioral modeling, basis function screening, digital predistortion (DPD), piecewise model, power amplifiers (PAs), running complexity, segmentation rules.

Abstract:This paper proposes a dual feature indexed quadratic polynomial-based piecewise (DIQP) behavioral modeling technique for digital predistortion (DPD) of RF transmitters. The proposed DIQP model is used to find the most suitable DPD model by performing a dual feature classification on the optimized submodels with a reuse-based function screening algorithm. The optimized submodel is adapted from the previous instantaneous sample indexed magnitude-selective affine (I-MSA) function-based model by transforming the original single linear term into a quadratic term with stronger fitting ability. This key improvement not only enhances the flexibility of the model but also boosts its fitting capability. The segmentation rule of the piecewise model has evolved from a simple threshold segmentation to a dual feature segmentation based on threshold and clustering segments. This reconstruction provides the model with enhanced feature-building capabilities. Additionally, the corresponding hybrid basis function screening (HBFS) algorithm and running complexity identification algorithm based on basis function reuse are proposed. The ingenious design of this reuse-based function screening algorithm not only enhances running efficiency but also ensures the overall performance of the model. The experimental part uses two different power amplifiers (PAs) for behavioral modeling and linearization tests. And the results of the experiments prove that the screened DIQP model is able to achieve the linearization performance-complexity trade-off excellently.

Co-author:Renlong Han,Chengye Jiang,Guichen Yang,Qianqian Zhang,Junsen Wang

First Author:Hao Chang (昌昊)

Indexed by:Journal paper

Correspondence Author:Falin Liu

Document Code:10.1109/TBC.2024.3434625

Discipline:Engineering

Document Type:J

Volume:70

Issue:4

Page Number:1302-1315

Translation or Not:no

Date of Publication:2024-12-12

Included Journals:SCI、EI