MétaCan
Menu
Back to cohort

Domain Adaptation in Beam Management using Asymmetric Autoencoder and Transfer Learning

2025· article· en· W4414405983 on OpenAlexaff
Medhat Elsayed, Melike Erol‐Kantarci

Bibliographic record

Venuenot available
Typearticle
Languageen
FieldEngineering
TopicOptical Systems and Laser Technology
Canadian institutionsUniversity of Ottawa
Fundersnot available
KeywordsAutoencoderTransfer of learningLeverage (statistics)Beam searchEncoderBeamformingBeam (structure)Reduction (mathematics)Baseline (sea)Throughput

Abstract

fetched live from OpenAlex

In 5G millimeter wave (mmWave) networks, beam management is crucial for establishing optimal transmitter-receiver beam pairs between the base station (gNB) and user equipment (UE) through sequential adjustment of azimuth and elevation angles and transmit power. Conventional approaches like exhaustive search increase exponentially with growing beam configurations, while predefined codebook-based beamforming vectors yield suboptimal solutions in dynamic networks. Although machine learning (ML) techniques offer potential solutions by learning efficient beam selection policies, the high-dimensional search space makes direct policy learning computationally intensive. Transfer learning (TL) can leverage knowledge from simpler network optimization tasks; however, dimensional mismatches between source and target domains present significant challenges. To address these dimensional mismatches, we present a novel domain adaptation (DA) framework that uses an asymmetric autoencoder combined with Maximum Mean Discrepancy (MMD) loss to align latent representations of source and target domains. This architecture achieves state space reconstruction accuracy exceeding 93% and latent space alignment with total loss below 0.01. Building on this framework, we introduce two transfer learning methods for beam management: a representation transfer learning with fine-tuning (RTLF) approach leveraging the pretrained encoder, and an adaptive source weight transfer (ASWTR) mechanism combining the encoder with selective source domain weight transfer. Compared to baseline training with DRL, both TL methods demonstrate minimum 16% faster convergence and 81% reduction in training episode steps. The proposed TL methods demonstrate comparable beam management performance to baseline DRL in throughput and energy efficiency, while maintaining optimal beam configurations and outperforming the non-DRL method by 50% in energy efficiency and 45% in throughput with 20 UEs.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.002
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Not applicable · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.002
Threshold uncertainty score0.005

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0010.002
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.001
Scholarly communication0.0000.001
Open science0.0010.001
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0010.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.007
GPT teacher head0.205
Teacher spread0.198 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designNot applicable
Domainnot available
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same topicOptical Systems and Laser TechnologyFrench-language works237,207