Self-attention cell-free MIMO

Thumbnail Image

Files

master_Palshin_Andrei_2025.pdf (1.03 MB)

URL

Journal Title

Journal ISSN

Volume Title

School of Electrical Engineering | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-11-21

Department

Major/Subject

Communications Engineering

Mcode

Degree programme

Master's Programme in Computer, Communication and Information Sciences
Tieto-, tietoliikenne- ja informaatiotekniikan maisteriohjelma
Magisterprogrammet i data-, informations- och kommunikationsteknik

Language

en

Pages

57

Series

Abstract

In modern wireless communication, rising traffic, tighter latency budgets, and increasing network complexity place additional demands on the scalability and responsiveness of resource allocation and control across distributed networks. These demands motivate attention-based deep learning for wireless resource optimization. This thesis explores the use of a transformer-based neural network for power control in cell-free massive MIMO (CFmMIMO) systems as a case demonstration of how attention-based models can be used in wireless system optimization. In this setting, self-attention efficiently captures global dependencies among users by conditioning on large-scale fading to access points and on pilot-allocation information. The model is trained in an unsupervised manner on a high-performance computing cluster. The training data consist of simulated network snapshots represented by the large-scale fading matrix and the pilot-allocation matrix. Experiments indicate per-user spectral efficiency comparable to established optimization methods, while guaranteeing a significantly lower inference time. Results are consistent across the evaluated configurations. These findings support attention-based models as a practical and scalable solution for real-time resource optimization in CFmMIMO.

Description

Supervisor

Vorobyov, Sergiy

Thesis advisor

Kocharlakota, Kameswara

Keywords

machine learning, wireless communications, self-attention, transformer neural networks, cell-free massive MIMO, power control

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202512169311

Collections

[dipl] Sähkötekniikan korkeakoulu / ELEC