Detection Schemes for High Order M-Ary QAM Under Transmit Nonlinearities


Gulgun Z., YILMAZ A. Ö.

IEEE TRANSACTIONS ON COMMUNICATIONS, vol.67, no.7, pp.4825-4834, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 67 Issue: 7
  • Publication Date: 2019
  • Doi Number: 10.1109/tcomm.2019.2905569
  • Journal Name: IEEE TRANSACTIONS ON COMMUNICATIONS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.4825-4834
  • Keywords: Bit interleaved coded modulation (BICM), inter-symbol interference (ISI), error vector magnitude (EVM), mis-matched decoding, nonlinear channels, nonlinearity of power amplifiers, DIGITAL PREDISTORTION, COMPENSATION, DISTORTION, CHANNELS, ACCESS
  • Middle East Technical University Affiliated: Yes

Abstract

Nonlinearities in various stages of a transmitter may hinder and restrict the transmission rate. As observed in many studies, outermost constellation points are usually more adversely affected by these impairments. To observe these effects, we utilize two power amplifier models that have different effects on transmitted signals. The Rapp model considers only amplitude deformation and the resultant in-phase and quadrature errors can be assumed to be independent on the receiver side. Unlike the Rapp model, the Saleh model exerts both amplitude and phase deformations and the phase deformation introduces correlation between the in-phase and quadrature errors according to our observations. In addition to the correlation, the variances of in-phase and quadrature errors may not be equal to each other. In this paper, we propose receivers that consider error variances of each quadrature amplitude modulation (QAM) symbol. We compare the performances of the receivers with those of other receivers that take average error variances into account for decoding. Furthermore, we propose a practical receiver that directly works on digitized observations based on a look-up table that keeps log-likelihood ratios of the quantized regions in order to reduce computational complexity.