Deep Convolutional Neural Networks for Massive MIMO Fingerprint-Based Positioning

Joao Vieira, Erik Leitinger, Muris Sarajlic, Xuhong Li, Fredrik Tufvesson

Research output: Contribution to journalArticleResearch

Abstract

This paper provides an initial investigation on the application of convolutional neural networks (CNNs) for fingerprint-based positioning using measured massive MIMO channels. When represented in appropriate domains, massive MIMO channels have a sparse structure which can be efficiently learned by CNNs for positioning purposes. We evaluate the positioning accuracy of state-of-the-art CNNs with channel fingerprints generated from a channel model with a rich clustered structure: the COST 2100 channel model. We find that moderately deep CNNs can achieve fractional-wavelength positioning accuracies, provided that an enough representative data set is available for training.
Original languageEnglish
JournalarXiv.org e-Print archive
Publication statusPublished - 21 Aug 2017

Fingerprint

MIMO systems
Neural networks
Wavelength

Keywords

  • stat.ML
  • cs.IT
  • math.IT

Cite this

Deep Convolutional Neural Networks for Massive MIMO Fingerprint-Based Positioning. / Vieira, Joao; Leitinger, Erik; Sarajlic, Muris; Li, Xuhong; Tufvesson, Fredrik.

In: arXiv.org e-Print archive, 21.08.2017.

Research output: Contribution to journalArticleResearch

@article{b60dd08877064aa88bac46b2048ab128,
title = "Deep Convolutional Neural Networks for Massive MIMO Fingerprint-Based Positioning",
abstract = "This paper provides an initial investigation on the application of convolutional neural networks (CNNs) for fingerprint-based positioning using measured massive MIMO channels. When represented in appropriate domains, massive MIMO channels have a sparse structure which can be efficiently learned by CNNs for positioning purposes. We evaluate the positioning accuracy of state-of-the-art CNNs with channel fingerprints generated from a channel model with a rich clustered structure: the COST 2100 channel model. We find that moderately deep CNNs can achieve fractional-wavelength positioning accuracies, provided that an enough representative data set is available for training.",
keywords = "stat.ML, cs.IT, math.IT",
author = "Joao Vieira and Erik Leitinger and Muris Sarajlic and Xuhong Li and Fredrik Tufvesson",
note = "Accepted in the IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) 2017",
year = "2017",
month = "8",
day = "21",
language = "English",
journal = "arXiv.org e-Print archive",
publisher = "Cornell University Library",

}

TY - JOUR

T1 - Deep Convolutional Neural Networks for Massive MIMO Fingerprint-Based Positioning

AU - Vieira, Joao

AU - Leitinger, Erik

AU - Sarajlic, Muris

AU - Li, Xuhong

AU - Tufvesson, Fredrik

N1 - Accepted in the IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) 2017

PY - 2017/8/21

Y1 - 2017/8/21

N2 - This paper provides an initial investigation on the application of convolutional neural networks (CNNs) for fingerprint-based positioning using measured massive MIMO channels. When represented in appropriate domains, massive MIMO channels have a sparse structure which can be efficiently learned by CNNs for positioning purposes. We evaluate the positioning accuracy of state-of-the-art CNNs with channel fingerprints generated from a channel model with a rich clustered structure: the COST 2100 channel model. We find that moderately deep CNNs can achieve fractional-wavelength positioning accuracies, provided that an enough representative data set is available for training.

AB - This paper provides an initial investigation on the application of convolutional neural networks (CNNs) for fingerprint-based positioning using measured massive MIMO channels. When represented in appropriate domains, massive MIMO channels have a sparse structure which can be efficiently learned by CNNs for positioning purposes. We evaluate the positioning accuracy of state-of-the-art CNNs with channel fingerprints generated from a channel model with a rich clustered structure: the COST 2100 channel model. We find that moderately deep CNNs can achieve fractional-wavelength positioning accuracies, provided that an enough representative data set is available for training.

KW - stat.ML

KW - cs.IT

KW - math.IT

M3 - Article

JO - arXiv.org e-Print archive

JF - arXiv.org e-Print archive

ER -