Publication:

Macromodeling of Electrically Long Structures Using a Delay-Aware Rational Szego Kernel

 
cris.virtual.department#PLACEHOLDER_PARENT_METADATA_VALUE#
cris.virtual.department#PLACEHOLDER_PARENT_METADATA_VALUE#
cris.virtual.department#PLACEHOLDER_PARENT_METADATA_VALUE#
cris.virtual.department#PLACEHOLDER_PARENT_METADATA_VALUE#
cris.virtual.department#PLACEHOLDER_PARENT_METADATA_VALUE#
cris.virtual.orcid0000-0002-9524-4205
cris.virtual.orcid0000-0002-9587-6923
cris.virtual.orcid0009-0005-6366-321X
cris.virtual.orcid0000-0001-6600-1792
cris.virtual.orcid0000-0003-2899-4636
cris.virtualsource.department7bac28ac-f3c2-462d-aea4-cc71c4892295
cris.virtualsource.department0a9dec9b-6490-4505-b9c3-a259df58d053
cris.virtualsource.departmente2dd6768-13eb-410a-afd3-3a3729d930c2
cris.virtualsource.department3d0467d5-8f2f-463b-9a89-cd2e89911f08
cris.virtualsource.departmente8043942-f5dc-4e9f-b5ef-85780b08f47a
cris.virtualsource.orcid7bac28ac-f3c2-462d-aea4-cc71c4892295
cris.virtualsource.orcid0a9dec9b-6490-4505-b9c3-a259df58d053
cris.virtualsource.orcide2dd6768-13eb-410a-afd3-3a3729d930c2
cris.virtualsource.orcid3d0467d5-8f2f-463b-9a89-cd2e89911f08
cris.virtualsource.orcide8043942-f5dc-4e9f-b5ef-85780b08f47a
dc.contributor.authorLindemans, Yens
dc.contributor.authorUllrick, Thijs
dc.contributor.authorCouckuyt, Ivo
dc.contributor.authorDeschrijver, Dirk
dc.contributor.authorDhaene, Tom
dc.date.accessioned2026-11-17T10:39:05Z
dc.date.available2026-11-17T10:39:05Z
dc.date.createdwos2026
dc.date.issued2026
dc.description.abstractGaussian process regression has recently gained significant traction for modeling and optimization of microwave system frequency-domain responses, owing to its nonparametric nature, data efficiency, robustness to noise, and inherent uncertainty quantification. These properties are enabled through the complex-valued rational Szegő kernel, which preserves key physical characteristics of frequency responses, including causality, Hermitian symmetry, and stability. However, the conventional Szegó kernel struggles to accurately represent electrically long structures, where propagation delays induce strongly oscillatory behavior. This work proposes a delay-aware rational Szegő (DRS) kernel that explicitly incorporates propagation delays via complex exponential terms while preserving the rational macromodel form. The approach is evaluated on the scattering parameters of a terminated lossy transmission line and benchmarked against the standard Szegő kernel. Results demonstrate that both models achieve high accuracy when trained on dense data, while the DRS kernel retains this accuracy under sparse sampling, highlighting its suitability for data-efficient macromodeling of electrically long structures.
dc.description.wosFundingTextThis work is supported by Flemish Research Foundation (FWO Vlaanderen) under grant G095224N and the Flanders AI Research Program.
dc.identifier.doi10.1109/spi68887.2026.11594726
dc.identifier.isbn979-8-3315-9076-5
dc.identifier.issn2475-9481
dc.identifier.urihttps://imec-publications.be/handle/20.500.12860/60836
dc.language.isoeng
dc.provenance.editstepusergreet.vanhoof@imec.be
dc.publisherIEEE
dc.source.beginpage1
dc.source.conferenceIEEE 30TH WORKSHOP ON SIGNAL AND POWER INTEGRITY, SPI
dc.source.conferencedate2026-06-14
dc.source.conferencelocationTurin, Italy
dc.source.endpage4
dc.source.journal2026 IEEE 30TH WORKSHOP ON SIGNAL AND POWER INTEGRITY, SPI
dc.source.numberofpages4
dc.title

Macromodeling of Electrically Long Structures Using a Delay-Aware Rational Szego Kernel

dc.typeProceedings paper
dspace.entity.typePublication
imec.internal.crawledAt2026-11-11
imec.internal.sourcecrawler
imec.internal.wosCreatedAt2026-11-11
Files

Original bundle

Name:
9111.pdf
Size:
3.37 MB
Format:
Adobe Portable Document Format
Description:
Published
Publication available in collections: