Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Beam selection using Machine Learning in Massive MIMO systems

Persson, Simon LU (2023) EITM01 20231
Department of Electrical and Information Technology
Abstract
In mobile communications the demand for increased capacity and speeds is a
constant. 5G NR is the latest generation which is now seeing widespread adoption. Crucial technologies like Massive MIMO and beamforming are enabled by
the use of large antenna arrays. While these arrays allow a 5G network to handle
larger amounts of data at higher speeds they do have drawbacks. Specifically,
the power requirements associated with using many large arrays are sizable.
Antenna selection is a potential solution to this problem, allowing the antenna array to use only a subset of elements at off-peak times. Selecting the
most favorable beams for transmission and reception is an important step towards achieving this goal. The aim of this thesis is... (More)
In mobile communications the demand for increased capacity and speeds is a
constant. 5G NR is the latest generation which is now seeing widespread adoption. Crucial technologies like Massive MIMO and beamforming are enabled by
the use of large antenna arrays. While these arrays allow a 5G network to handle
larger amounts of data at higher speeds they do have drawbacks. Specifically,
the power requirements associated with using many large arrays are sizable.
Antenna selection is a potential solution to this problem, allowing the antenna array to use only a subset of elements at off-peak times. Selecting the
most favorable beams for transmission and reception is an important step towards achieving this goal. The aim of this thesis is therefore to apply machine
learning techniques to this problem in order to predict the locations of optimal
beams.
Throughout the testing process several machine learning models and approaches to solving the problem were explored. In each step of the process the
prediction accuracy was evaluated and improvements were made. Several of
the models displayed the ability to make accurate predictions, which will aid in
solving the antenna selection problem.
The highest accuracy was achieved when predicting one of two predefined
beam regions based on Precoding Matrix Indicator-values using a Deep Neural
Network model. This test yielded a prediction accuracy of 87.4%. (Less)
Popular Abstract
The fifth generation of mobile communications brings great improvements to
speed and capacity compared to its predecessors. While this may be a significant
advantage of 5G there are drawbacks like increased power consumption. Antenna
selection looks to be a promising solution to this problem.
A crucial part of any 5G system is the use of large antenna arrays consisting
of a large number of smaller antenna elements. These elements work together
and allow the capacity across a radio link to be multiplied.
Beamforming is an important technique enabled by the use of antenna arrays. The phase and amplitude of a signal can be modified at each antenna
element in such a way that constructive interference is created and aimed in a
specific... (More)
The fifth generation of mobile communications brings great improvements to
speed and capacity compared to its predecessors. While this may be a significant
advantage of 5G there are drawbacks like increased power consumption. Antenna
selection looks to be a promising solution to this problem.
A crucial part of any 5G system is the use of large antenna arrays consisting
of a large number of smaller antenna elements. These elements work together
and allow the capacity across a radio link to be multiplied.
Beamforming is an important technique enabled by the use of antenna arrays. The phase and amplitude of a signal can be modified at each antenna
element in such a way that constructive interference is created and aimed in a
specific direction, also known as a beam. Using all antenna elements for this
process allows transmissions to be focused while reducing interference for received signals. The downside of this is high power consumption. For this reason
this project works towards finding a way of using only a subset of the available
antennas. An important step towards achieving this goal is beam selection, or
selecting the best beams for a given signal.
This project has investigated ways of selecting the most suitable beams using
a variety of machine learning techniques. In a set of received beams the suitability of each one can be determined based on amplitude, meaning the higher
the power the better the beam. The aim of this project is therefore to predict
which beams will have the highest power based on a variety of input variables.
Several approaches to the problem and machine learning models were tested
over the course of the thesis. The most promising results were achieved when
dividing the set of beams into predefined regions and predicting which has the
highest power beams. Making these predictions based on so PMI data, which
is used by the base station to enable multiple data streams, yielded the highest
accuracy.
In the future this work will have to be expanded upon to enable antenna
selection, however having a reliable way of selecting beams is a crucial step towards achieving this goal. (Less)
Please use this url to cite or link to this publication:
author
Persson, Simon LU
supervisor
organization
course
EITM01 20231
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Massive MIMO, Beamforming, Machine Learning, 5G, NR, Antenna selection
report number
LU/LTH-EIT 2024-962
language
English
id
9142403
date added to LUP
2024-02-15 15:46:01
date last changed
2024-02-15 15:46:01
@misc{9142403,
  abstract     = {{In mobile communications the demand for increased capacity and speeds is a
constant. 5G NR is the latest generation which is now seeing widespread adoption. Crucial technologies like Massive MIMO and beamforming are enabled by
the use of large antenna arrays. While these arrays allow a 5G network to handle
larger amounts of data at higher speeds they do have drawbacks. Specifically,
the power requirements associated with using many large arrays are sizable.
Antenna selection is a potential solution to this problem, allowing the antenna array to use only a subset of elements at off-peak times. Selecting the
most favorable beams for transmission and reception is an important step towards achieving this goal. The aim of this thesis is therefore to apply machine
learning techniques to this problem in order to predict the locations of optimal
beams.
Throughout the testing process several machine learning models and approaches to solving the problem were explored. In each step of the process the
prediction accuracy was evaluated and improvements were made. Several of
the models displayed the ability to make accurate predictions, which will aid in
solving the antenna selection problem.
The highest accuracy was achieved when predicting one of two predefined
beam regions based on Precoding Matrix Indicator-values using a Deep Neural
Network model. This test yielded a prediction accuracy of 87.4%.}},
  author       = {{Persson, Simon}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Beam selection using Machine Learning in Massive MIMO systems}},
  year         = {{2023}},
}