Direct Learning Neural Network Digital Predistortion Using Backpropagation Through a Memory Power Amplifier Model

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This work presents a new direct learning for a neural network based digital predistortion linearizer using backpropagation through a memory power amplifier. The learning method is compared with indirect learning among other methods. The proposed learning methods are evaluated on a wideband CMOS power amplifier using a 160 MHz 802.11ax WiFi signal with an 11.5 dB peak-to-average power ratio and a center frequency of 1 GHz at 8.2 dBm average transmit power. The direct learning using memory backpropagation shows superior performance to other proposed learning methods, achieving a correction of 10.8 dB to the error vector magnitude to -39 dB EVM.

Original languageEnglish
Title of host publication2023 IEEE/MTT-S International Microwave Symposium, IMS 2023
Pages791-794
Number of pages4
ISBN (Electronic)9798350347647
DOIs
StatePublished - 2023
Event2023 IEEE/MTT-S International Microwave Symposium, IMS 2023 - San Diego, United States
Duration: 11 Jun 202316 Jun 2023

Publication series

NameIEEE MTT-S International Microwave Symposium Digest
Volume2023-June

Conference

Conference2023 IEEE/MTT-S International Microwave Symposium, IMS 2023
Country/TerritoryUnited States
CitySan Diego
Period11/06/2316/06/23

Keywords

  • digital predistortion
  • direct learning architecture
  • indirect learning architecture
  • neural networks
  • power amplifiers
  • wireless communication

All Science Journal Classification (ASJC) codes

  • Condensed Matter Physics
  • Radiation
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Direct Learning Neural Network Digital Predistortion Using Backpropagation Through a Memory Power Amplifier Model'. Together they form a unique fingerprint.

Cite this