Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learned Pulse Shaping Design for PAPR Reduction in DFT-s-OFDM

Fabrizio Carpi, Soheil Rostami, Joonyoung Cho, Siddharth Garg, Elza Erkip, Charlie Jianzhong Zhang

TL;DR

This work proposes a machine learning-based framework to determine the FDSS filter, optimizing a tradeoff between the symbol error rate (SER), the PAPR, and the spectral flatness requirements, and results show that learned FDSS filters lower the PAPR compared to conventional baselines, with minimal SER degradation.

Abstract

High peak-to-average power ratio (PAPR) is one of the main factors limiting cell coverage for cellular systems, especially in the uplink direction. Discrete Fourier transform spread orthogonal frequency-domain multiplexing (DFT-s-OFDM) with spectrally-extended frequency-domain spectrum shaping (FDSS) is one of the efficient techniques deployed to lower the PAPR of the uplink waveforms. In this work, we propose a machine learning-based framework to determine the FDSS filter, optimizing a tradeoff between the symbol error rate (SER), the PAPR, and the spectral flatness requirements. Our end-to-end optimization framework considers multiple important design constraints, including the Nyquist zero-ISI (inter-symbol interference) condition. The numerical results show that learned FDSS filters lower the PAPR compared to conventional baselines, with minimal SER degradation. Tuning the parameters of the optimization also helps us understand the fundamental limitations and characteristics of the FDSS filters for PAPR reduction.

Learned Pulse Shaping Design for PAPR Reduction in DFT-s-OFDM

TL;DR

This work proposes a machine learning-based framework to determine the FDSS filter, optimizing a tradeoff between the symbol error rate (SER), the PAPR, and the spectral flatness requirements, and results show that learned FDSS filters lower the PAPR compared to conventional baselines, with minimal SER degradation.

Abstract

High peak-to-average power ratio (PAPR) is one of the main factors limiting cell coverage for cellular systems, especially in the uplink direction. Discrete Fourier transform spread orthogonal frequency-domain multiplexing (DFT-s-OFDM) with spectrally-extended frequency-domain spectrum shaping (FDSS) is one of the efficient techniques deployed to lower the PAPR of the uplink waveforms. In this work, we propose a machine learning-based framework to determine the FDSS filter, optimizing a tradeoff between the symbol error rate (SER), the PAPR, and the spectral flatness requirements. Our end-to-end optimization framework considers multiple important design constraints, including the Nyquist zero-ISI (inter-symbol interference) condition. The numerical results show that learned FDSS filters lower the PAPR compared to conventional baselines, with minimal SER degradation. Tuning the parameters of the optimization also helps us understand the fundamental limitations and characteristics of the FDSS filters for PAPR reduction.
Paper Structure (19 sections, 2 equations, 6 figures)

This paper contains 19 sections, 2 equations, 6 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Conventional Receiver
  4. Performance Metrics
  5. FDSS Design with Machine Learning
  6. Loss Function
  7. SER component $\mathcal{E}$
  8. PAPR component $\mathcal{P}$
  9. FDSS shape component $\mathcal{S}$
  10. Learned Filter Taps Model
  11. Constrained FDSS Design
  12. Non-flat
  13. Flat
  14. Flat with vestigial sideband
  15. Results
  16. ...and 4 more sections

Figures (6)

  • Figure 1: Block diagram DFT-s-OFDM with SE and FDSS. The SE operation is sketched in the middle.
  • Figure 2: Constrained FDSS designs: (a) non-flat; (b) flat; (c) flat with vestigial sideband (zero-ISI). The vertical dimension represents the filter values in frequency domain, while the horizontal axis (omitted) represents the subcarrier index.
  • Figure 3: Learned FDSS filters for limiting cases when $\text{EBW}=14.2\%$. Resulting performance w.r.t. RRC FDSS:
  • Figure 4: Learned FDSS filters with different constraints when $\text{EBW}=14.2\%$. Only the first half of the subcarriers is shown in this plot. Resulting performance w.r.t. RRC FDSS:
  • Figure 5: CCDF of PAPR for the scenario in Fig. \ref{['fig:FDSS']}.
  • ...and 1 more figures