EUVIMEDEUVIMED entdecken
Crossref

A Fading Memory Pruning Strategy for Polynomial Models of Broadband Power Amplifiers

Chenrui Xu, Jianfeng Zhai, Zhuang Xu, Zhiqiang Yu, Jianyi Zhou, Nianzu Zhang, Chao Yu, Zhang-Cheng Hao

IEEE Transactions on Microwave Theory and Techniques · 2026-08 · 74 (8) · 6851-6859

DOI: 10.1109/tmtt.2026.3694022 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Microwave Theory and Techniques, 2026-08. Verfasst von Chenrui Xu und 7 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Gefördert von: Scientific Research Innovation Capability Support Project for Young Faculty, National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As the signal bandwidth increases in the wireless communication system, power amplifiers (PAs) exhibit strong nonlinear characteristics, and the scale and complexity of behavioral models for PA digital predistortion (DPD) are also increasing sharply. Traditional data search-based pruning algorithms, such as the orthogonal matching pursuit (OMP) and the doubly OMP (DOMP), have significantly limited practicality due to issues such as excessive search time and their continued reliance on generating a complete basis function space. Based on the memory effects of PAs and the decay characteristics of transmission signals, a grounded recommendation for the selection of memory depth in predistortion models and a low-complexity fading memory pruning (FMP) strategy for the polynomial model is proposed in this article. In addition to serving as a direct pruning scheme to eliminate less important terms, the FMP strategy can also act as a valuable a priori knowledge to precompress the candidate space for the OMP or DOMP method. The DPD tests were conducted on two high-efficiency PAs excited by a 100- or 400-MHz 5G new radio (NR) signal. The experimental results demonstrate that when the FMP strategy is applied as a stand-alone pruning method, it can greatly reduce the number of coefficients and real multiplications required in the coefficient extraction stage while maintaining considerable model accuracy. Furthermore, when utilized as a priori knowledge to narrow the basis function space for search-based algorithms, it not only achieves excellent linearization performance but also effectively enhances the search efficiency, saving over 43% (for OMP) and 89% (for DOMP) of real multiplication resources.

Förderung

  • Scientific Research Innovation Capability Support Project for Young Faculty
  • National Natural Science Foundation of China

Zitieren

APA

Xu, C., Zhai, J., Xu, Z., Yu, Z., Zhou, J., Zhang, N., Yu, C., & Hao, Z. (2026). A Fading Memory Pruning Strategy for Polynomial Models of Broadband Power Amplifiers. IEEE Transactions on Microwave Theory and Techniques, 74(8), 6851-6859. https://doi.org/10.1109/tmtt.2026.3694022

Vancouver

Xu C, Zhai J, Xu Z, Yu Z, Zhou J, Zhang N, et al. A Fading Memory Pruning Strategy for Polynomial Models of Broadband Power Amplifiers. IEEE Transactions on Microwave Theory and Techniques. 2026;74(8):6851-6859. doi:10.1109/tmtt.2026.3694022

BibTeX
@article{Xu2026,
  author = {Xu, Chenrui and Zhai, Jianfeng and Xu, Zhuang and Yu, Zhiqiang and Zhou, Jianyi and Zhang, Nianzu and Yu, Chao and Hao, Zhang-Cheng},
  title = {A Fading Memory Pruning Strategy for Polynomial Models of Broadband Power Amplifiers},
  journal = {IEEE Transactions on Microwave Theory and Techniques},
  year = {2026},
  volume = {74},
  number = {8},
  pages = {6851--6859},
  doi = {10.1109/tmtt.2026.3694022}
}
Kostenlos starten →