Detection of Impaired OFDM Waveforms Using Deep Learning Receiver

Jaakko Pihlajasalo; Dani Korpi; Taneli Riihonen; Jukka Talvitie; Mikko A. Uusitalo; Mikko Valkama

doi:10.1109/SPAWC51304.2022.9834021

Detection of Impaired OFDM Waveforms Using Deep Learning Receiver

Jaakko Pihlajasalo
, Dani Korpi
, Taneli Riihonen
, Jukka Talvitie
, Mikko A. Uusitalo
, Mikko Valkama

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

5 Citations (Scopus)

51 Downloads (Pure)

Abstract

With wireless networks evolving towards mmWave and sub-THz frequency bands, hardware impairments such as IQ imbalance, phase noise (PN) and power amplifier (PA) nonlinear distortion are increasingly critical implementation challenges. In this paper, we describe deep learning based physical-layer receiver solution, with neural network layers in both time- and frequency-domain, to efficiently demodulate OFDM signals under coexisting IQ, PN and PA impairments. 5G NR standard-compliant numerical results are provided at 28 GHz band to assess the receiver performance, demonstrating excellent robustness against varying impairment levels when properly trained.

Original language	English
Title of host publication	2022 IEEE 23rd International Workshop on Signal Processing Advances in Wireless Communication, SPAWC 2022
Publisher	IEEE
ISBN (Electronic)	9781665494557
ISBN (Print)	9781665494564
DOIs	https://doi.org/10.1109/SPAWC51304.2022.9834021
Publication status	Published - 2022
Publication type	A4 Article in conference proceedings
Event	IEEE International Workshop on Signal Processing Advances in Wireless Communication - Oulu, Finland Duration: 4 Jul 2022 → 6 Jul 2022

Publication series

Name	SPAWC
ISSN (Print)	1948-3244
ISSN (Electronic)	1948-3252

Conference

Conference	IEEE International Workshop on Signal Processing Advances in Wireless Communication
Country/Territory	Finland
City	Oulu
Period	4/07/22 → 6/07/22

Funding

In this paper, we demonstrated how different ML-based receiver architectures can be trained to deal with various hardware impairments. Our study covered two ML-based receivers, one of which is operating only in the frequency domain (DeepRx), and another one that has processing layers both in time and frequency domains (HybridDeepRx). We provided extensive numerical results in 28 GHz mmWave network context, which indicate that a frequency-domain neural network is well-suited for mitigating the effects of PN, whereas a hybrid time/frequency-domain network is required for dealing with IQ imbalance and nonlinear distortion. The results also indicated that both ML-based receivers outperform conventional baseline receivers in the appropriate impairment ACKNOWLEDGMENT This work was supported in part by Nokia Bell Labs and in part by the Academy of Finland under the grants #319994, #332361, and #338224.

Keywords

5G NR
6G
deep learning
hardware impairments
IQ imbalance
machine learning
mmWave
nonlinear distortion
phase noise
power amplifier
sub-THz

Publication forum classification

Publication forum level 1

ASJC Scopus subject areas

Electrical and Electronic Engineering
Computer Science Applications
Information Systems

Access to Document

10.1109/SPAWC51304.2022.9834021

Detection of Impaired OFDM Waveforms
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Accepted author manuscript, 136 KBLicence: Unspecified

https://urn.fi/URN:NBN:fi:tuni-202210077499Licence: Unspecified

Cite this

@inproceedings{397a8d520a90444b9c37d5fc40dbf676,

title = "Detection of Impaired OFDM Waveforms Using Deep Learning Receiver",

abstract = "With wireless networks evolving towards mmWave and sub-THz frequency bands, hardware impairments such as IQ imbalance, phase noise (PN) and power amplifier (PA) nonlinear distortion are increasingly critical implementation challenges. In this paper, we describe deep learning based physical-layer receiver solution, with neural network layers in both time- and frequency-domain, to efficiently demodulate OFDM signals under coexisting IQ, PN and PA impairments. 5G NR standard-compliant numerical results are provided at 28 GHz band to assess the receiver performance, demonstrating excellent robustness against varying impairment levels when properly trained.",

keywords = "5G NR, 6G, deep learning, hardware impairments, IQ imbalance, machine learning, mmWave, nonlinear distortion, phase noise, power amplifier, sub-THz",

author = "Jaakko Pihlajasalo and Dani Korpi and Taneli Riihonen and Jukka Talvitie and Uusitalo, \{Mikko A.\} and Mikko Valkama",

note = "Funding Information: In this paper, we demonstrated how different ML-based receiver architectures can be trained to deal with various hardware impairments. Our study covered two ML-based receivers, one of which is operating only in the frequency domain (DeepRx), and another one that has processing layers both in time and frequency domains (HybridDeepRx). We provided extensive numerical results in 28 GHz mmWave network context, which indicate that a frequency-domain neural network is well-suited for mitigating the effects of PN, whereas a hybrid time/frequency-domain network is required for dealing with IQ imbalance and nonlinear distortion. The results also indicated that both ML-based receivers outperform conventional baseline receivers in the appropriate impairment ACKNOWLEDGMENT This work was supported in part by Nokia Bell Labs and in part by the Academy of Finland under the grants \#319994, \#332361, and \#338224. Publisher Copyright: {\textcopyright} 2022 IEEE. JUFOID=88648; IEEE International Workshop on Signal Processing Advances in Wireless Communication ; Conference date: 04-07-2022 Through 06-07-2022",

year = "2022",

doi = "10.1109/SPAWC51304.2022.9834021",

language = "English",

isbn = "9781665494564",

series = "SPAWC",

publisher = "IEEE",

booktitle = "2022 IEEE 23rd International Workshop on Signal Processing Advances in Wireless Communication, SPAWC 2022",

address = "United States",

}

Pihlajasalo, J, Korpi, D, Riihonen, T , Talvitie, J, Uusitalo, MA & Valkama, M 2022, Detection of Impaired OFDM Waveforms Using Deep Learning Receiver. in 2022 IEEE 23rd International Workshop on Signal Processing Advances in Wireless Communication, SPAWC 2022. SPAWC, IEEE, IEEE International Workshop on Signal Processing Advances in Wireless Communication, Oulu, Finland, 4/07/22. https://doi.org/10.1109/SPAWC51304.2022.9834021

Detection of Impaired OFDM Waveforms Using Deep Learning Receiver. / Pihlajasalo, Jaakko; Korpi, Dani; Riihonen, Taneli et al.
2022 IEEE 23rd International Workshop on Signal Processing Advances in Wireless Communication, SPAWC 2022. IEEE, 2022. (SPAWC).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Detection of Impaired OFDM Waveforms Using Deep Learning Receiver

AU - Pihlajasalo, Jaakko

AU - Korpi, Dani

AU - Riihonen, Taneli

AU - Talvitie, Jukka

AU - Uusitalo, Mikko A.

AU - Valkama, Mikko

N1 - Funding Information: In this paper, we demonstrated how different ML-based receiver architectures can be trained to deal with various hardware impairments. Our study covered two ML-based receivers, one of which is operating only in the frequency domain (DeepRx), and another one that has processing layers both in time and frequency domains (HybridDeepRx). We provided extensive numerical results in 28 GHz mmWave network context, which indicate that a frequency-domain neural network is well-suited for mitigating the effects of PN, whereas a hybrid time/frequency-domain network is required for dealing with IQ imbalance and nonlinear distortion. The results also indicated that both ML-based receivers outperform conventional baseline receivers in the appropriate impairment ACKNOWLEDGMENT This work was supported in part by Nokia Bell Labs and in part by the Academy of Finland under the grants #319994, #332361, and #338224. Publisher Copyright: © 2022 IEEE. JUFOID=88648

PY - 2022

Y1 - 2022

N2 - With wireless networks evolving towards mmWave and sub-THz frequency bands, hardware impairments such as IQ imbalance, phase noise (PN) and power amplifier (PA) nonlinear distortion are increasingly critical implementation challenges. In this paper, we describe deep learning based physical-layer receiver solution, with neural network layers in both time- and frequency-domain, to efficiently demodulate OFDM signals under coexisting IQ, PN and PA impairments. 5G NR standard-compliant numerical results are provided at 28 GHz band to assess the receiver performance, demonstrating excellent robustness against varying impairment levels when properly trained.

AB - With wireless networks evolving towards mmWave and sub-THz frequency bands, hardware impairments such as IQ imbalance, phase noise (PN) and power amplifier (PA) nonlinear distortion are increasingly critical implementation challenges. In this paper, we describe deep learning based physical-layer receiver solution, with neural network layers in both time- and frequency-domain, to efficiently demodulate OFDM signals under coexisting IQ, PN and PA impairments. 5G NR standard-compliant numerical results are provided at 28 GHz band to assess the receiver performance, demonstrating excellent robustness against varying impairment levels when properly trained.

KW - 5G NR

KW - 6G

KW - deep learning

KW - hardware impairments

KW - IQ imbalance

KW - machine learning

KW - mmWave

KW - nonlinear distortion

KW - phase noise

KW - power amplifier

KW - sub-THz

U2 - 10.1109/SPAWC51304.2022.9834021

DO - 10.1109/SPAWC51304.2022.9834021

M3 - Conference contribution

AN - SCOPUS:85136025859

SN - 9781665494564

T3 - SPAWC

BT - 2022 IEEE 23rd International Workshop on Signal Processing Advances in Wireless Communication, SPAWC 2022

PB - IEEE

T2 - IEEE International Workshop on Signal Processing Advances in Wireless Communication

Y2 - 4 July 2022 through 6 July 2022

ER -

Detection of Impaired OFDM Waveforms Using Deep Learning Receiver

Abstract

Publication series

Conference

Funding

Keywords

Publication forum classification

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this