Neural Network-based scheme for PAPR reduction in OFDM Systems

Diego Reinoso-Chisaguano, Felipe Grijalva, Martha Cecilia Paredes Paredes, Jorge Carvajal-Rodriguez

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

Abstract

This paper proposes a neural network-based scheme for Peak-to-Average Power Ratio (PAPR) reduction which also replaces the Inverse Fast Fourier Transform (IFFT) block of an Orthogonal Frequency Division Multiplexing (OFDM) transmitter. The scheme is composed by one neural network per subcarrier, which are implemented only in the transmitter. The training inputs of each neural network are frequency-domain OFDM symbols and the outputs are time-domain PAPR reduced OFDM symbols obtained using a Branch-and-Bound Constellation Extension (BBCE) scheme. The results show that our scheme achieves a PAPR reduction and Bit Error Rate (BER) similar to constellation shaping techniques but with reduced complexity.

Original languageEnglish
Title of host publication2019 IEEE 4th Ecuador Technical Chapters Meeting, ETCM 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728137643
DOIs
StatePublished - Nov 2019
Externally publishedYes
Event4th IEEE Ecuador Technical Chapters Meeting, ETCM 2019 - Guayaquil, Ecuador
Duration: 13 Nov 201915 Nov 2019

Publication series

Name2019 IEEE 4th Ecuador Technical Chapters Meeting, ETCM 2019

Conference

Conference4th IEEE Ecuador Technical Chapters Meeting, ETCM 2019
Country/TerritoryEcuador
CityGuayaquil
Period13/11/1915/11/19

Keywords

  • BBCE
  • Neural Networks
  • OFDM
  • PAPR

Fingerprint

Dive into the research topics of 'Neural Network-based scheme for PAPR reduction in OFDM Systems'. Together they form a unique fingerprint.

Cite this