Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO
This article addresses the problem of symbol detection in time-varying Massive Multiple-Input Multiple-Output (M-MIMO) systems. While conventional detection techniques either exhibit subpar performance or impose excessive computational burdens in such systems, learning-based methods which have shown...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Open Journal of Vehicular Technology |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10551475/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832582284868321280 |
|---|---|
| author | Toluwaleke Olutayo Benoit Champagne |
| author_facet | Toluwaleke Olutayo Benoit Champagne |
| author_sort | Toluwaleke Olutayo |
| collection | DOAJ |
| description | This article addresses the problem of symbol detection in time-varying Massive Multiple-Input Multiple-Output (M-MIMO) systems. While conventional detection techniques either exhibit subpar performance or impose excessive computational burdens in such systems, learning-based methods which have shown great potential in stationary scenarios, struggle to adapt to non-stationary conditions. To address these challenges, we introduce innovative extensions to the Learned Conjugate Gradient Network (LcgNet) M-MIMO detector. Firstly, we expound Preconditioned LcgNet (PrLcgNet), which incorporates a preconditioner during training to enhance the uplink M-MIMO detector's filter matrix. This modification enables the detector to achieve faster convergence with fewer layers compared to the original approach. Secondly, we introduce an adaptation of PrLcgNet referred to as Dynamic Conjugate Gradient Network (DyCoGNet), specifically designed for time-varying environments. DyCoGNet leverages self-supervised learning with Forward Error Correction (FEC), enabling autonomous adaptation without the need for explicit labeled data. It also employs meta-learning, facilitating rapid adaptation to unforeseen channel conditions. Our simulation results demonstrate that in stationary scenarios, PrLcgNet achieves faster convergence than LCgNet, which can be leveraged to reduce system complexity or improve Symbol Error Rate (SER) performance. Furthermore, in non-stationary scenarios, DyCoGNet exhibits rapid and efficient adaptation, achieving significant SER performance gains compared to baseline cases without meta-learning and a recent benchmark using self-supervised learning. |
| format | Article |
| id | doaj-art-12a28babbe3640f185855085c432aae4 |
| institution | Kabale University |
| issn | 2644-1330 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Open Journal of Vehicular Technology |
| spelling | doaj-art-12a28babbe3640f185855085c432aae42025-01-30T00:04:29ZengIEEEIEEE Open Journal of Vehicular Technology2644-13302024-01-01579280610.1109/OJVT.2024.341083410551475Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMOToluwaleke Olutayo0https://orcid.org/0009-0002-9755-540XBenoit Champagne1https://orcid.org/0000-0002-0022-6072Department of Electrical and Computer Engineering, McGill University, Montreal, QC, CanadaDepartment of Electrical and Computer Engineering, McGill University, Montreal, QC, CanadaThis article addresses the problem of symbol detection in time-varying Massive Multiple-Input Multiple-Output (M-MIMO) systems. While conventional detection techniques either exhibit subpar performance or impose excessive computational burdens in such systems, learning-based methods which have shown great potential in stationary scenarios, struggle to adapt to non-stationary conditions. To address these challenges, we introduce innovative extensions to the Learned Conjugate Gradient Network (LcgNet) M-MIMO detector. Firstly, we expound Preconditioned LcgNet (PrLcgNet), which incorporates a preconditioner during training to enhance the uplink M-MIMO detector's filter matrix. This modification enables the detector to achieve faster convergence with fewer layers compared to the original approach. Secondly, we introduce an adaptation of PrLcgNet referred to as Dynamic Conjugate Gradient Network (DyCoGNet), specifically designed for time-varying environments. DyCoGNet leverages self-supervised learning with Forward Error Correction (FEC), enabling autonomous adaptation without the need for explicit labeled data. It also employs meta-learning, facilitating rapid adaptation to unforeseen channel conditions. Our simulation results demonstrate that in stationary scenarios, PrLcgNet achieves faster convergence than LCgNet, which can be leveraged to reduce system complexity or improve Symbol Error Rate (SER) performance. Furthermore, in non-stationary scenarios, DyCoGNet exhibits rapid and efficient adaptation, achieving significant SER performance gains compared to baseline cases without meta-learning and a recent benchmark using self-supervised learning.https://ieeexplore.ieee.org/document/10551475/Massive MIMO (M-MIMO)symbol detectionconjugate gradientmodel-based learningnon-stationary channelsonline adaptation |
| spellingShingle | Toluwaleke Olutayo Benoit Champagne Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO IEEE Open Journal of Vehicular Technology Massive MIMO (M-MIMO) symbol detection conjugate gradient model-based learning non-stationary channels online adaptation |
| title | Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO |
| title_full | Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO |
| title_fullStr | Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO |
| title_full_unstemmed | Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO |
| title_short | Dynamic Conjugate Gradient Unfolding for Symbol Detection in Time-Varying Massive MIMO |
| title_sort | dynamic conjugate gradient unfolding for symbol detection in time varying massive mimo |
| topic | Massive MIMO (M-MIMO) symbol detection conjugate gradient model-based learning non-stationary channels online adaptation |
| url | https://ieeexplore.ieee.org/document/10551475/ |
| work_keys_str_mv | AT toluwalekeolutayo dynamicconjugategradientunfoldingforsymboldetectionintimevaryingmassivemimo AT benoitchampagne dynamicconjugategradientunfoldingforsymboldetectionintimevaryingmassivemimo |