Advanced

User Equipment Characterization using Machine Learning

Fakhoury, Vanessa LU and Zhou, Xin LU (2019) EITM02 20191
Department of Electrical and Information Technology
Abstract
With the ever increasing demand for higher data rates and reliability, efficient management of cellular networks remains a challenge. Among other technologies, fifth generation systems are expected to tackle this challenge using large electronically controllable antenna arrays operating in the time division duplex mode. In such systems, spatial beamforming is implemented with accurate channel estimates at the cellular base station (BS). As a consequence of the high channel dimensions, large amounts of data are being collected at the BS which can be further utilized in order to optimize resource allocation, and to observe trends to facilitate more efficient beamforming to user equipments (UEs). To this end, machine learning methods play a... (More)
With the ever increasing demand for higher data rates and reliability, efficient management of cellular networks remains a challenge. Among other technologies, fifth generation systems are expected to tackle this challenge using large electronically controllable antenna arrays operating in the time division duplex mode. In such systems, spatial beamforming is implemented with accurate channel estimates at the cellular base station (BS). As a consequence of the high channel dimensions, large amounts of data are being collected at the BS which can be further utilized in order to optimize resource allocation, and to observe trends to facilitate more efficient beamforming to user equipments (UEs). To this end, machine learning methods play a vital role to identify useful patterns in data. In this thesis, four machine learning models have been built to categorize whether an UE is moving at a given velocity, or remaining stationary. In particular, binary neural network, multiclass neural network, support vector machine and logistic regression techniques are implemented and analyzed. The uplink sounding reference signal (SRS) channel estimates values are used as input data to the machine learning techniques. The SRS data are generated from a lab simulator at Ericsson AB, Lund. A binary neural network is first built to classify if the UE is moving or remaining stationary. Furthermore, the multiclass neural network is extended to classify movement of the UE at different speeds of 30 km/h or 100 km/h. Further to this, a support vector machine and logistic regression are implemented to compare performance and computational complexity of such approaches, relative to a binary neural network. The obtained results show that the binary neural network has the highest classification accuracy (98%) compared to the support vector machine (95%) and logistic regression (93.8%). For binary classification in this thesis, a larger amount of samples are input into the neural network therefore achieving the highest accuracy. Additionally, the multiclass neural network showed an accuracy of 89.2%. The accuracy of the machine learning algorithms depend on the problem scenario and the size of the dataset. (Less)
Popular Abstract
Are you interested in fast download speeds and little to no buffering time when streaming your favorite online tv-show? Of course you are, and guess what? 5G can make it happen! 5G has been hyped for a few years, but this year many telecommunication companies have deployed their network in the first half of the year. The first 5G network has been turned on in the United Kingdom in May 2019.
With the new technology, more and more devices will be connected to the network, utilizing data-intensive applications. In order to deal with this high demand, more antennas are needed at the base stations, which are fixed structures on rooftops, in masts or on building walls, transmitting and receiving radio waves to communicate with mobile users. In... (More)
Are you interested in fast download speeds and little to no buffering time when streaming your favorite online tv-show? Of course you are, and guess what? 5G can make it happen! 5G has been hyped for a few years, but this year many telecommunication companies have deployed their network in the first half of the year. The first 5G network has been turned on in the United Kingdom in May 2019.
With the new technology, more and more devices will be connected to the network, utilizing data-intensive applications. In order to deal with this high demand, more antennas are needed at the base stations, which are fixed structures on rooftops, in masts or on building walls, transmitting and receiving radio waves to communicate with mobile users. In order to direct these radio waves into specific directions, 5G employs what is called beamforming, which is a way to combine antenna elements in a way that limits interference between users and enables higher data rates for everyone connected to that base station. Beamforming relies on a good estimate of the transmission channel between transmitter and receiver. The user sends a reference signal, which is known by both transmitter and receiver, to the base station, which in return estimates the transmission channel according to this signal and is then able to make intelligent decisions on how to direct its signal beams.
You can imagine that with more devices being connected, and therefore more antennas needed, more of the reference signals will have to be decoded at the base stations, increasing the computational complexity immensely. This thesis aims to investigate, whether it is possible to use Machine Learning, which is a model that focuses on the development of computer programs that can access data and learn it on their own, to classify if a user is stationary or moving at different speeds. This automatic categorization would help the base station with reducing computations, allowing for more efficient beamforming and later profitable scheduling of users to avoid interference.
This thesis built and investigated three different machine learning algorithms: Neural Networks, Support Vector Machine and Logistic Regression. All three algorithms are intended for classification use, but they differ in complexity and their accuracy for different amounts of data. Data was collected from a test lab at Ericsson AB, Lund in Sweden, where different scenarios were simulated with random routes of a user moving or stationary positions. The result of the implementation and testing of the algorithms were greatly satisfactory, with high accuracy of all machine learning methods. Additionally, neural network was used to classify a user into different speed categories.
The building of these methods could result in an implementation in real world 5G base stations, therefore enabling higher data rates and larger capacity. Happy browsing ! (Less)
Please use this url to cite or link to this publication:
author
Fakhoury, Vanessa LU and Zhou, Xin LU
supervisor
organization
course
EITM02 20191
year
type
H2 - Master's Degree (Two Years)
subject
keywords
machine learning, characterization, massive MIMO, beamforming, neural networks, support vector machine, logistic regression, multiclass neural networks
report number
LU/LTH-EIT 2019-712
language
English
id
8985936
date added to LUP
2019-07-11 15:13:14
date last changed
2019-07-11 15:13:14
@misc{8985936,
  abstract     = {With the ever increasing demand for higher data rates and reliability, efficient management of cellular networks remains a challenge. Among other technologies, fifth generation systems are expected to tackle this challenge using large electronically controllable antenna arrays operating in the time division duplex mode. In such systems, spatial beamforming is implemented with accurate channel estimates at the cellular base station (BS). As a consequence of the high channel dimensions, large amounts of data are being collected at the BS which can be further utilized in order to optimize resource allocation, and to observe trends to facilitate more efficient beamforming to user equipments (UEs). To this end, machine learning methods play a vital role to identify useful patterns in data. In this thesis, four machine learning models have been built to categorize whether an UE is moving at a given velocity, or remaining stationary. In particular, binary neural network, multiclass neural network, support vector machine and logistic regression techniques are implemented and analyzed. The uplink sounding reference signal (SRS) channel estimates values are used as input data to the machine learning techniques. The SRS data are generated from a lab simulator at Ericsson AB, Lund. A binary neural network is first built to classify if the UE is moving or remaining stationary. Furthermore, the multiclass neural network is extended to classify movement of the UE at different speeds of 30 km/h or 100 km/h. Further to this, a support vector machine and logistic regression are implemented to compare performance and computational complexity of such approaches, relative to a binary neural network. The obtained results show that the binary neural network has the highest classification accuracy (98%) compared to the support vector machine (95%) and logistic regression (93.8%). For binary classification in this thesis, a larger amount of samples are input into the neural network therefore achieving the highest accuracy. Additionally, the multiclass neural network showed an accuracy of 89.2%. The accuracy of the machine learning algorithms depend on the problem scenario and the size of the dataset.},
  author       = {Fakhoury, Vanessa and Zhou, Xin},
  keyword      = {machine learning,characterization,massive MIMO,beamforming,neural networks,support vector machine,logistic regression,multiclass neural networks},
  language     = {eng},
  note         = {Student Paper},
  title        = {User Equipment Characterization using Machine Learning},
  year         = {2019},
}