Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing

Furudoi, Tomoharu; Takahashi, Takumi; Ibi, Shinsuke; Ochiai, Hideki

doi:https://doi.org/10.1109/WCNC61545.2025.10978405

Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing

Furudoi, Tomoharu; Takahashi, Takumi; Ibi, Shinsuke et al.

IEEE Wireless Communications and Networking Conference, WCNC, 2025

Number of Access:33（2025-09-30 09:44 Counts）

Identifier to cite or link to this item: https://hdl.handle.net/11094/102538

Link to primary information

File	Format	Terms of use	Size	Views	Date.Available	Description	information
IEEEWirelessCommunNetworkingConfWCNC	pdf	None	640 KB	14	2025.08.21

Item Information

Output File Export EndNote Basic Export Mendeley

Title	Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing
Creator	CreatorName Furudoi, Tomoharu
	NRID 1000040844204 1000040844204 researchmap permalink takumi_takahashi takumi_takahashi Scopus Author ID 57193615052 57193615052 CreatorName Takahashi, Takumi
	NRID 1000010448087 1000010448087 CreatorName Ibi, Shinsuke
	NRID 1000020334576 1000020334576 researchmap permalink hideki_ochiai hideki_ochiai CreatorName Ochiai, Hideki
Subject	data-driven tuning
	deep unfolding
	Generalized approximate message passing
	MIMO signal detection
Description	Furudoi T., Takahashi T., Ibi S., et al. Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing. IEEE Wireless Communications and Networking Conference, WCNC , (2025); https://doi.org/10.1109/WCNC61545.2025.10978405.
Abstract	Generalized approximate message passing (GAMP) achieves near-optimal performance in detecting spatially multiplexed massive multiple-input multiple-output (MIMO) signals with significantly reduced computational complexity under independent and identically distributed (i.i.d.) measurements. However, in spatially correlated MIMO channels where the ideal assumption of large-scale uncorrelated observation does not hold, the detection capability severely deteriorates. This performance degradation can be compensated for by optimizing GAMP with embedded learnable parameters via deep unfolding (DU) techniques, i.e., data-driven tuning; however, the learning process becomes quite unstable. To address this issue, we propose a novel method that achieves stable learning by incorporating a monotonic increase constraint on the reliability of propagated messages by learning the differential (incremental) values of the learnable parameters between two consecutive iterations. The efficacy of the proposed method is confirmed through numerical results in terms of loss trajectory in the learning process and bit error rate (BER) of massive MIMO detection.
Publisher	Institute of Electrical and Electronics Engineers Inc.
Source Title	IEEE Wireless Communications and Networking Conference, WCNC
Date of Issued	2025-05-09
Language	English
Handle URL	https://hdl.handle.net/11094/102538
PISSN	15253511
eISSN	15582612
Relation.isVersionOf	DOI (Publisher Version) https://doi.org/10.1109/WCNC61545.2025.10978405
Access Rights	open access
CopyRight	© 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
oaire:version	Accepted Manuscript
Category	会議発表論文 Conference Paper

Resource Type	conference paper
Local Type	Conference Paper
DCMI Type	Text
DCTERMS.bibliographicCitation	IEEE Wireless Communications and Networking Conference, WCNC. 2025
DC.title	Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing
DC.creator	Furudoi, Tomoharu
	Takahashi, Takumi
	Ibi, Shinsuke
	Ochiai, Hideki
DC.publisher	Institute of Electrical and Electronics Engineers Inc.
DC.language' scheme='DCTERMS.RFC1766	English
DC.type' scheme='DCTERMS.DCMIType	Text
DCTERMS.issued' scheme='DCTERMS.W3CDTF	2025-05-09
DC.identifier	https://doi.org/10.1109/WCNC61545.2025.10978405
DC.subject	data-driven tuning
	deep unfolding
	Generalized approximate message passing
	MIMO signal detection
DC.description	Furudoi T., Takahashi T., Ibi S., et al. Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing. IEEE Wireless Communications and Networking Conference, WCNC , (2025); https://doi.org/10.1109/WCNC61545.2025.10978405.
DCTERMS.abstract	Generalized approximate message passing (GAMP) achieves near-optimal performance in detecting spatially multiplexed massive multiple-input multiple-output (MIMO) signals with significantly reduced computational complexity under independent and identically distributed (i.i.d.) measurements. However, in spatially correlated MIMO channels where the ideal assumption of large-scale uncorrelated observation does not hold, the detection capability severely deteriorates. This performance degradation can be compensated for by optimizing GAMP with embedded learnable parameters via deep unfolding (DU) techniques, i.e., data-driven tuning; however, the learning process becomes quite unstable. To address this issue, we propose a novel method that achieves stable learning by incorporating a monotonic increase constraint on the reliability of propagated messages by learning the differential (incremental) values of the learnable parameters between two consecutive iterations. The efficacy of the proposed method is confirmed through numerical results in terms of loss trajectory in the learning process and bit error rate (BER) of massive MIMO detection.
DC.rights	© 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
DC.format	102538
citation_title	Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing
citation_author	Furudoi, Tomoharu
	Takahashi, Takumi
	Ibi, Shinsuke
	Ochiai, Hideki
citation_publisher	Institute of Electrical and Electronics Engineers Inc.
citation_language	English
citation_date	2025-05-09
citation_journal_title	IEEE Wireless Communications and Networking Conference, WCNC
citation_doi	https://doi.org/10.1109/WCNC61545.2025.10978405
citation_public_url	https://hdl.handle.net/11094/102538

Learning Stabilization in Deep Unfolding of Generalized Approximate Message Passing

Link to primary information

Item Information

Show details