Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Machine learning-based Precoder Type Selection in Massive MIMO Networks
Blekinge Institute of Technology, Faculty of Computing, Department of Computer Science.
2025 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Background. Massive MIMO underpins modern 4G/5G networks by boosting spectral efficiency through beamforming. Precoding can follow either a codebook or a reciprocity principle, and the superior choice depends on the instantaneous radio context.

Objective. This thesis investigates whether machine learning can learn the context and select the more efficient precoder type from production-network data.

Methods. Precoder selection is cast as a two-model regression problem. One Random-Forest regressor is trained on transmissions that used codebook precoding, and a second regressor on those that used reciprocity. At run-time each HARQ process is evaluated by both models; the scheme with the higher predicted spectral efficiency is chosen. Performance is estimated offline on historical data.

Results. Trained on 400,000 HARQ processes, the two models achieve low cross-validated and test nRMSE and unbiased residuals. The learned selector overrides the legacy rule in 52.5% of transmissions and increases aggregate efficiency by +6.4 %. On the switched cases the mean uplift is +16.7%; the 95th percentile reaches +44.2%. Feature-importance analysis matches domain intuition.

Conclusions. While most transmissions change little, the selector consistently finds a large subset where an alternative precoder yields substantial gains. The approach is data-driven, interpretable and lightweight enough for real-time 5G deployment, though live trials are advised to validate causal uplift and tune safety margins.

Abstract [sv]

Bakgrund. Massive MIMO är grunden för dagens 4G/5G-nät och höjer den spektrala effektiviteten med hjälp av strålformning. Precoding kan utföras antingen kodboks-baserat eller via reciprocitet; vilken variant som är bäst beror på den aktuella radiokanalen.

Syfte. Arbetet undersöker om maskininlärning kan lära sig detta sammanhang och välja effektivaste precoder med verkliga basstationsloggar som grund.

Metod. Precoder-valet formuleras som ett regressionsproblem med två separata Random-Forest-modeller: en tränad på kodboks-överföringar och en på reciprocitets-överföringar. Vid varje HARQ-process körs båda modellerna; den precoder som ger högst förväntad spektral effektivitet väljs. Prestandan utvärderas offline på historisk data.

Resultat. Modellerna, tränade på cirka 400,000 HARQ-processer, uppvisar låg korsvaliderad och test nRMSE och väntevärdesriktiga residualer. Selektorn ändrar precoder i 52,5% av sändningarna och höjerden totala spektrala effektiviteten med +6,4%.  När byte sker är medelvinsten +16,7%; den 95e percentilen når +44,2%. Variabelvikterna överensstämmer med kända domänfaktorer.

Slutsatser. Även om de flesta sändningar påverkas marginellt identifierar modellen en stor andel där alternativ precoding ger betydande vinster. Metoden är data-driven, tolkbar och tillräckligt lätt för realtidsinförande i 5G-basstationer, men fältförsök rekommenderas för att bekräfta vinsten och justera säkerhetsmarginaler.

Place, publisher, year, edition, pages
2025. , p. 45
Keywords [en]
MIMO, 5g, Machine learning, Artificial intelligence
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:bth-27745OAI: oai:DiVA.org:bth-27745DiVA, id: diva2:1962534
External cooperation
Ericsson
Subject / course
Degree Project in Master of Science in Engineering 30,0 hp
Educational program
DVAMI Master of Science in Engineering: AI and Machine Learning 300 hp
Supervisors
Examiners
Available from: 2025-06-11 Created: 2025-05-31 Last updated: 2025-09-30Bibliographically approved

Open Access in DiVA

fulltext(2117 kB)225 downloads
File information
File name FULLTEXT01.pdfFile size 2117 kBChecksum SHA-512
3408c07ca9fb1eb4a77fbfecaececd93906439468856d108ce9d91852113ee29db199b49891273866b085b9c2dad4b8824337dac8904f4fe4ffc23647b301221
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Iseni, Arlind
By organisation
Department of Computer Science
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 227 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 204 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf