Master Thesis 5G NR Nuria V

5G NR small cell deployment using
Beamforming
Master Thesis
submitted to the Faculty of the
Escola Tècnica d’Enginyeria de Telecomunicació de Barcelona
Universitat Politècnica de Catalunya
by
Nuria Vinyes
In partial fulfillment
of the requirements for the master in
TELECOMMUNICATIONS ENGINEERING
Advisor: Anna Umbert

Barcelona, May 2022
Contents
List of Figures 4
List of Tables 6
List of Abbreviations 11
1 Introduction 12
1.1 Statement of purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2 Methods and procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.1 ATOLL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Work plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.1 Work packages and tasks . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.2 Deviations from the initial plan and incidences . . . . . . . . . . . . 14
1.4 Outline of the report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 State of the art 16

2.1 5G NR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.1 Overall architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1.2 5G NR features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.3 Cell search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Allocation of spectrum bands to 5G . . . . . . . . . . . . . . . . . . . . . . 24
2.3 Milimiter Wave (mmW) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.4 Densification and small cells . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5 Beamforming and Multiple Input Multiple Output (MIMO) . . . . . . . . 27
3 Methodology 31
4 Project development 32
4.1 Defined scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1.1 Small venue scenario . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1.2 Big venue scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 ATOLL Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 ATOLL 3D Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.3.1 Types of beams in ATOLL . . . . . . . . . . . . . . . . . . . . . . . 39
4.3.2 Difference between Broadcast and Service Beams . . . . . . . . . . 39
4.4 Modelling Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.5 ATOLL Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.5.1 Diversity support modes . . . . . . . . . . . . . . . . . . . . . . . . 43
4.5.2 Aster and Aster mmWave propagation models . . . . . . . . . . . . 44
4.6 3D traffic map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.7 ATOLL Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.7.1 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5 Results 46
5.1 Small venue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2
5.1.1 No MIMO nor Beamforming . . . . . . . . . . . . . . . . . . . . . . 47
5.1.2 All diversity support modes active . . . . . . . . . . . . . . . . . . . 52
5.1.3 Simulations’ results using beamforming . . . . . . . . . . . . . . . . 54
5.1.4 Simulations’ results without using beamforming . . . . . . . . . . . 56
5.1.5 Simulation’s results of the deployment in the near future . . . . . . 58
5.2 Big venue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2.1 All beam indexes active . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2.2 Simulation’s results using only the first 20 beam indexes . . . . . . 64
5.2.3 No Beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2.4 Simulations’ results without using Beamforming . . . . . . . . . . . 72
5.2.5 Simulations’ results for the Video Call service as Data type . . . . . 76
6 Conclusions and future development 81
References 83
Appendices 84
A Beam Usage (DL) in the big venue scenario 84
3
List of Figures
2.1 5G will bring impactful figures . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 ITU requirements for 5G: eMBB, uRLL, Massive IoT (source: ETRI graphic,
from ITU-R IMT 2020 requirements) . . . . . . . . . . . . . . . . . . . . . 18
2.3 Non-standalone operation . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Standalone operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5 Logical, transport and physical channels . . . . . . . . . . . . . . . . . . . 21
2.6 Synchronization Signal Block . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Synchronization Signal (SS) burst set . . . . . . . . . . . . . . . . . . . . . 24
2.8 Frequency bands identified for 5G in Europe and potential use cases . . . . 25
2.9 Channel BW representation . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.10 Comparison between 4G regular antenna and 5G with beamforming . . . . 28
2.11 Analog beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.12 Digital beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.13 Beamforming architectures: i) analog (left), ii) digital (middle), iii) hybrid
(right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1 Ciutadella deployment map for the small venue scenario . . . . . . . . . . 32
4.2 Indoor User Profile Environment Traffic Map (light blue areas) . . . . . . . 33
4.3 Outdoor User Profile Environment Traffic Map (dark blue area) . . . . . . 33
4.4 Ciutadella deployment map for the big venue scenario . . . . . . . . . . . . 35
4.5 DL coverage by a 70 degrees antenna (SA Broadcast) . . . . . . . . . . . . 37
4.6 70 degrees antenna azimuth (horizontal) pattern . . . . . . . . . . . . . . . 38
4.7 70 degrees antenna elevation (vertical) pattern . . . . . . . . . . . . . . . . 38
4.8 Broadcast beams radiation pattern . . . . . . . . . . . . . . . . . . . . . . 39
4.9 Service beams radiation pattern . . . . . . . . . . . . . . . . . . . . . . . . 39
4.10 Broadcast vs Service Beam . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.11 5G NR Broadcast service properties . . . . . . . . . . . . . . . . . . . . . . 41
4.12 5G NR Video Call service properties as Data type . . . . . . . . . . . . . . 42
4.13 Diversity support in DL and UL in ATOLL’s 5G NR Cells table . . . . . . 43
5.1 DL coverage with only n257 using all diversity support modes, before using
n78 too . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 DL coverage with no SU-MIMO nor MU-MIMO . . . . . . . . . . . . . . . 48
5.3 DL coverage legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.4 DL quality with no SU-MIMO nor MU-MIMO . . . . . . . . . . . . . . . . 48
5.5 DL quality legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.6 DL coverage with no Beamforming . . . . . . . . . . . . . . . . . . . . . . 49
5.8 DL quality with no Beamforming . . . . . . . . . . . . . . . . . . . . . . . 49
5.10 DL coverage with no support modes . . . . . . . . . . . . . . . . . . . . . . 50
5.12 DL quality with no diversity support modes . . . . . . . . . . . . . . . . . 51
5.14 DL quality with no diversity support modes vs no MIMO (black) . . . . . 51
5.15 DL quality with no diversity support modes vs no Beamforming (black) . . 51
4
5.16 DL coverage with all diversity support modes . . . . . . . . . . . . . . . . 52
5.18 DL quality with all diversity support modes . . . . . . . . . . . . . . . . . 53
5.20 DL quality with diversity vs no MIMO only (black) . . . . . . . . . . . . . 53
5.21 DL quality with diversity vs no beamforming only (black) . . . . . . . . . 53
5.22 DL quality with diversity vs no diversity (black) . . . . . . . . . . . . . . . . . . . 54
5.23 Best Service Beam counter for Video Call . . . . . . . . . . . . . . . . . . . 56
5.24 Best Service Beam counter for Voice Call . . . . . . . . . . . . . . . . . . . 56
5.25 Best Service Beam counter for Broadband . . . . . . . . . . . . . . . . . . . . . . 56
5.26 Histogram of Received PDSCH Power (DL) (dBm) without Beamforming . 57
5.27 Histogram of Received PDSCH Power (DL) (dBm) using Beamforming . . 57
5.28 “Received PDSCH Power (DL) (dBm)” for ‘Activity Status’, ‘Best Server’ and ‘Diversity
Mode (DL)’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.29 Best Service Beam counter for Video Call . . . . . . . . . . . . . . . . . . . 59
5.30 Best Service Beam counter for Voice Call . . . . . . . . . . . . . . . . . . . 59
5.31 Best Service Beam counter for Broadband . . . . . . . . . . . . . . . . . . . . . . 59
5.32 Histogram of Application User Throughput (DL) (kbps) for Broadband . . 60
5.33 Beam Usage DL (%) prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.34 Best Service Beam prediction . . . . . . . . . . . . . . . . . . . . . . . . . .. . 61
5.35 Beam Usage DL with all beam indexes active . . . . . . . . . . . . . . . . . . . . 62
5.36 Best Service Beam with all beam indexes active . . . . . . . . . . . . . . . . .. . 63
5.37 Representation of figure 5.34 . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.38 Representation of figure 5.36 . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.39 Histogram of the representation above-named . . . . . . . . . . . . . . . . 64
5.40 Histogram of the representation above-named . . . . . . . . . . . . . . . . 64
5.41 Average results of indoor 3D traffic simulation . . . . . . . . . . . . . . . . . .
. . 64
5.42 Histogram of Received PDSCH Power (DL) (dBm) for Receiver Diversity . . . . . . . 66
5.43 Histogram of Received PDSCH Power (DL) (dBm) with SU-MIMO and MU-MIMO . . 66
5.44 Histogram of Application User Throughput (DL) (kbps) for Receiver Diversity . . . . 67
5.45 Histogram of Application User Throughput (DL) (kbps) with SU-MIMO and MU-MIMO 67
5.46 Received PDSCH Power (dBm) and Path Loss (dB) [*] versus Bearer . . . . . . . . . 68
5.47 User Throughput versus Bearer . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.48 Histogram of Received PUSCH & PUCCH Power (UL) (dBm) . . . . . . . . . . . . 70
5.49 Histogram of Application User Throughput (UL) (kbps) . . . . . . . . . . . . . . . 70
5.50 Coverage prediction with no beamforming . . . . . . . . . . . . . . . . . . 71
5.51 No beamforming (blue area) compared with using beamforming (black line) .. . . . . 71
5.52 DL quality prediction with no beamforming . . . . . . . . . . . . . . . . . 71
5.53 No beamforming (blue area) compared with using beamforming (black line) .. . . . . 71
5.54 Average results of no diversity support modes simulation . . . . . . . . . .
. . . . . 72
5.55 Metrics for No Service users using no Beamforming . . . . . . . . . . . .. . . . . 73
5.56 Metrics for connected users using no Beamforming . . . . . . . . . . . . .
. . . . . 73
5.57 Histogram of Received PDSCH Power (DL) (dBm) with no Beamforming . . . . . . . 74
5.58 Histogram of Received PDSCH Power (DL) (dBm) with Beamforming . . . . . . . . 74
5.59 Histogram of Application User Throughput (DL) (kbps) with no Beamforming . . . . 74
5
5.60 Histogram of Application User Throughput (DL) (kbps) with Beamforming . . . . . . 74
5.61 Histogram of Received PUSCH & PUCCH Power (UL) (dBm) without Beamforming . 75
5.62 Histogram of Application User Throughput (UL) (kbps) without beamforming . . . . . 75
5.63 Average results of video call service as data type simulation . . . . . . . . . . . . . . 76
5.64 Histogram of Application User Throughput (DL) (kbps) with video call as data . . . . 78
5.65 Histogram of Application User Throughput (DL) (kbps) with video call as voice . . . . 78
5.66 Histogram of Application User Throughput (DL) (kbps) for Video Call service . . . . . 78
5.67 Histogram of Application User Throughput (UL) (kbps) for broadband service with
video call as data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.68 Histogram of Application User Throughput (UL) (kbps) for broadband service with
video call as voice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.69 Average (blue) and counter (orange) of Application User Throughput (UL) (kbps) per
Bearer for Broadband service . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.70 Histogram of Application User Throughput (UL) (kbps) for Video Call . . . . . . . . 80
5.71 Average (blue) and counter (orange) of Application User Throughput (UL) (kbps) per
Bearer for Video Call . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
A.1 Beam Usage DL site 9 (traffic load 99,9%) . . . . . . . . . . . . . . . . . . . . . . 84
A.2 Beam Usage DL site 8 (traffic load 96,05%) . . . . . . . . . . . . . . . . . . . . . 85
A.3 Beam Usage DL site 10 (traffic load 25%) . . . . . . . . . . . . . . . . . . . . . . 85
A.4 Beam Usage DL site 17 (traffic load 41%) . . . . . . . . . . . . . . . . . . . . . . 86
Listings
List of Tables
1 5G NR numerologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 Business user properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 Standard user properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4 Small venue indoor environment . . . . . . . . . . . . . . . . . . . . . . . . 34
5 Small venue outdoor environment . . . . . . . . . . . . . . . . . . . . . . . 34
6 5G frequencies used in the small venue scenario . . . . . . . . . . . . . . . 34
7 Business user properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
8 Standard user properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
9 Big venue indoor environment . . . . . . . . . . . . . . . . . . . . . . . . . 36
10 5G frequencies used in the big venue scenario . . . . . . . . . . . . . . . . 37
11 Mobile’s rejection cause for the case of only n257 sites . . . . . . . . . . . . 54
12 Mobile’s rejection cause for the case of using n257 and n78 sites . . . . . . 55
13 Total number of connected users for the case of using n257 and n78 sites . 55
14 Total accumulated and per user throughput for the case of using n257 and
n78 sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
15 Total number of connected users for the case of not using beamfoming . . . 56
16 Total accumulated and average per user throughput for the case of not
using beamforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6
17 Total number of connected users for the case of using the first 20 beam
indexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
18 Total accumulated and average per user throughput for the case of using
the first 20 beam indexes . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
19 Mobile’s rejection cause for the case of using the first 20 beam indexes . . 65
20 Total number of connected users for the case of not using beamfoming . . . 72
21 Total accumulated and average per user throughput for the case of not
using beamfoming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
22 Mobile’s rejection cause for the case of not using beamfoming . . . . . . . 73
23 Total number of connected users for the case of video call service as data
type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
24 Total accumulated throughput and average throughput per user for the
case of video call service as data type . . . . . . . . . . . . . . . . . . . . . 77
25 Mobile’s rejection cause for the case of video call service as data type . . . 77
7
Acknowledgements
First of all, I want to thank my supervisor, Anna Umbert, for being an exemplary advisor,
for sharing with me your knowledge and dedicating your time to me whenever I need it.
On the other hand, thank you to my family and friends for all their support.
8
Revision history and approval record
Revision Date Purpose

0 14/03/2022 Document creation
1 18/04/2022 Document revision
4 17/05/2022 Document approval
DOCUMENT DISTRIBUTION LIST
Name e-mail
Nuria Vinyes nuria.vinyes@estudiantat.upc.edu
Anna Umbert anna.umbert@upc.edu
Written by: Reviewed and approved by:

Date 14/03/2022 Date 17/05/2022
Name Nuria Vinyes Name Anna Umbert
Position Project Author Position Project Supervisor
9
Abstract
5G mobile communication systems should bring significant benefits such as lower latency,
higher peak and user data rates, connection density and availability. To achieve such
benefits beamforming at the base stations and new high frequency bands will be used.
The work developed in this project is intended to contribute to demonstrating the feasibil-
ity of beamforming by means of a 5G mobile network in the Ciutadella area in Barcelona
using a professional software, for wireless network design and optimisation, ATOLL from
Forsk company.
This project aims to deploy a small cell mmWave deployment in two different scenar-
ios. The first scenario, with low user density; the second scenario, with high user density
(including 3D traffic). We achieved good performance in both scenarios when using beam-
forming in terms of throughput, received SINR and number of users connected.
10
Acronyms
5G PPP 5G Infrastructure Public Private Partnership
5GS 5G System
ADC Analog-to-Digital Converter
BH Backhaul
CIO Cell Individual Offset
DAC Digital-to-Analog Converter
DL Downlink
eNB Evolved Node B
EPC Evolved Packet Core
ETSETB Escola Tècnica Superior d’Enginyeria de Telecomunicació de Barcelona
EU European Union
gNB Next Generation Node B
IMT International Mobile Telecommunications
ITU-R International Telecommunication Union
KPI Key Performance Indicator
MNO Mobile Network Operator
mWT Millimetre Wave Transmission
NR New Radio
PBCH Physical Broadcast Channel
PDSCH Physical Downlink Shared Channel
SINR Signal to Interference and Noise Ratio
UE User Equipment
UL Uplink
11
1 Introduction
Unlike traditional antennas that could transmit and receive only on fixed radiation pat-
terns, beamforming antennas dynamically shape their beam directions according to the
location of their connected users. Beamforming antennas are unique in their ability to
effectively reduce interference, improve the signal-to-interference-and-noise ratio (SINR)
and deliver a significantly better end user experience. With the ITU-WRC 1 adoption of
new sub 6 GHz and mmWave frequency bands, and the trend towards extreme spectral
efficiencies used in 5G New Radio (NR) (the term used to refer to 5G radio interface),
this type of antennas are becoming an essential technology.
Beamforming antennas offer various configurations and capabilities; each is best suited
for a certain environment. Physically, these antennas might look very dissimilar, but all
share two main design philosophies: they use digital, analog or hybrid beamforming and
they are equipped with multiple radio transceivers—8T8R and higher [1].
Beamforming for millimeter wave communication is a crucial aspect in link budget analysis
for feasible system design. To this end, there have been notable contributions in further
decreasing the number of search steps and computational complexity in order to reduce
the delay in setting up the beamformed link. With ultra dense networks poised to be
a distinct possibility for future generation of wireless communications, millimeter wave
technology is envisaged to be the key enabler to support and leverage data rates in excess
of 10 Gbps. In this context, studies on millimeter wave mobile broadband systems and
massive MIMO systems have clearly outlined the importance of beamforming along with
spatial multiplexing techniques to enhance data rates, increase network capacity and
mitigate interference. Hybrid beamforming methods, comprising of baseband precoding
with limited number of RF chains and constant amplitude analog phase shifters, are shown
to be effective in terms of achieving the desired performance for millimeter wave MIMO
systems.
Present trends in millimeter wave beamforming research include solving complex issues
related to hybrid beamforming, polarization diversity, optimization of beam search pro-
cess, concurrent beamforming protocols, robust adaptive beamforming, exploiting channel
sparsity and 3D beamforming. These solutions must meet the objectives of reducing the
computation cost, delay and power consumption during the beamforming process while
maintaining an acceptable quality of service at multi-Gbps data rates [2].
1.1 Statement of purpose

In this context, the work to be developed in this project is intended to contribute to
demonstrating the feasibility of beamforming by means of the implementation of a 5G
mobile network deployment. In particular, we aim to prove that this technology can
be implemented using realistic case scenarios, using a professional software, for wireless
network design and optimisation, ATOLL from Forsk company.
In order to do so, we have designed two different scenarios, in the city of Barcelona, at
1
World Radio Conference
12
the UPF’s Ciutadella campus. The first scenario with a low density of users (small venue)
and the second one with a high density (big venue), using a small cell deployment with a
mmWave frequency band, n257, also referred as 28GHz.
1.2 Methods and procedures

During the duration of the thesis we have used the professional software ATOLL to test,
simulate and obtain results from. In the next lines it will be explained what it is and how
does it work.
1.2.1 ATOLL
Atoll is a multi-technology wireless network design and optimisation platform that sup-
ports wireless operators throughout the network lifecycle, from initial design to densifi-
cation and optimisation. Atoll includes advanced single-RAN multi-RAT network design
capabilities for both 3GPP and 3GPP2 radio access technologies including 5G NR, LTE,
NB-loT, UMTS, GSM, and CDMA. It supports the latest technology advances such as
massive MIMO, 3D beamforming, and mmWave propagation for the design and roll-out
of 5G networks.
Atoll uniquely combines architectural and functional features that provide operators with
a powerful, scalable, and flexible framework for streamlining their network design and
optimisation processes.
Atoll’s modular and advanced 5G NR radio technology modelling capabilities, along with
the support for mmWave propagation, massive MIMO, and 3D beamforming, provide
operators with a flexible and evolutionary framework for the design and deployment of
5G networks.
Atoll supports Aster Propagation Model, a high-performance propagation model that sup-
ports macro, micro, and small cell urban propagation scenarios. Aster and Aster mmWave,
both part of the same package, are advanced ray-tracing propagation models that support
multiple frequencies both below and above 6GHz [3].
1.3 Work plan

1.3.1 Work packages and tasks
In the paragraphs below there are the different parts into which the development of the
project has been divided, what each of these parts has consisted of and the duration of
them.
The master’s thesis was enrolled in September 2021 with the topic of multi-beam antennas
in mobility environments, but due to some setbacks explained in section 1.3.2, the new
topic and current project did not start until this year’s February and it is also the reason
why this thesis is presented in the extension period.
13
This thesis proposal began in February and the first work package was the theoretical
planning part, defining the scope and the objectives, understanding what beamforming
is and how it works, also research what has been published by university researchers and
commercial companies.
The second work package, the deployment part, began in March with the definition of
parameters for only one site, followed by the placement of the other sites using the met-
rics obtained from the predictions, i.e., to characterize the small cells, understand the
beamforming performance, and calibrate the deployment.
The third work package, the simulations part, included user traffic testing to analyze
how the network behaves and if it meets the requirements demanded by the services and
users. This part is related to the deployment part because the predictions’ results helped
improve and optimize the deployment.
As it will be explained in 1.3.2, after the update in the software in April, a new part was
created to get familiarised with the new features. Following the second and third work
packages’ tasks, the process was to start with locating one site and understanding and
defining the parameters of the site, followed by placing the other sites and deploying and
optimising the network.
The 3D traffic map part, which was introduced with the new version of ATOLL, consisted
of three tasks: learning how it works, creating a 2D traffic map to define the 3D traffic
map, and obtaining the functioning 3D traffic map.
The last work package, that started in May, was the elaboration of the final report, which
consisted of drafting the document and revising it.
1.3.2 Deviations from the initial plan and incidences

The first master’s thesis proposal was going to be implemented at a company, but later
it was changed to the university where the options where limited for the objectives of the
project given that there are no simulation tools available right now to simulate mobility
environments using multibeam antennas for mobile communications in 5G. Thus, the
second proposal, the current one, started at the beginning of the extension period.
Another change from the initial plan was that at the beginning of the project the idea
was to have a 3D indoor traffic map, but the version and licenses we had of ATOLL did
not allow it and resulted in the small venue scenario not having a 3D traffic map and
having only a 2D traffic map, ATOLL did allow to do Multi-Storey Prediction, a feature
that calculates the prediction by floor and allows to create a 3D view of the prediction.
At the middle of the project a new version of ATOLL was installed that allowed creating
a 3D traffic map, so for the big venue scenario all the traffic simulations where run using
it.
1.4 Outline of the report

The project has been structured as follows,
14
• Introduction: The project concept and objectives are stated in this chapter with a
brief summary of the contents of the thesis.
• State of the Art: It is explained the concepts involved in this project, i.e., 5G NR,
small cells, beamforming among others. Technologies such as mmWave, MIMO and
beamforming are described with its benefits and challenges, and the final decision
why beamforming is needed is justified.
• Methodology: a summary of one page explaining the contents of this project, the
phases in which the project has been divided and what has been done in each one.
• Project Development: The professional software, for wireless network design and op-
timisation, ATOLL from Forsk company is presented with a explanation of its fea-
tures. The features (ATOLL antennas, ATOLL beamforming, the services, ATOLL
predictions, 3D traffic maps, ATOLL simulations) are explained: its technology, the
applications, to finally decide how they are configured for the two scenarios pro-
posed.
• Results: The predictions and simulations carried out during this project to demon-
strate the benefits of using beamforming in an indoor mmWave deployment are
presented. Moreover, an initial change in the small venue scenario to add a n78
band, also knows as 3.5GHz band, is also demonstrated. Afterwards, the different
cases implemented are shown:
– The first scenario, the small venue, compares using diversity support modes
against using none, and analyses how the deployment will adapt to the increase
of users in the near future.
– In the second scenario, the big venue, the effect of the beamforming using 3D
traffic map is assessed, comparing the two cases on the previous test, using
beamforming and not using it, but also activating all the beam indexes of the
beamforming antenna, not only the default ones, to study the optimal case.
• Conclusions and Future Developments: Taking into account the objectives set and
looking at the experimental results of the previous section, a series of conclusions
and answers are obtained. Future challenges and updates are also exposed.
15
2 State of the art
This chapter will begin with an overview of 5G NR and its key features before moving on
to the technologies involved in this project.
2.1 5G NR
5G will enable services based in three pillars: bandwidth, latency and billions of connected
devices.
Consumer demands are shaping the development of mobile broadband services. Antici-
pated increases in traffic (estimated as 10-100 times over the period 2020 – 2030), growth
in the number of devices and services, as well as demand for enhanced affordability and
user experience will all require innovative solutions. The number of connected devices on
the Internet is projected to reach 50 billion any time from 2025 onwards.
The fifth generation of mobile technologies - 5G - connects people, things, data, appli-
cations, transport systems and cities in smart networked communication environments.
The networks transport a huge amount of data much faster, reliably connect an extremely
large number of devices and process very high volumes of data with minimal delay.
5G technologies support applications such as smart homes and buildings, smart cities,
3D video, work and play in the cloud, remote medical services, virtual and augmented
reality, and massive machine-to-machine communications for industry automation. Legacy
3G and 4G networks face challenges in supporting these services [4].
16
Figure 2.1: 5G will bring impactful figures
International Mobile Telecommunications-2020 (IMT-2020 Standard) are the require-

ments issued by the ITU Radiocommunication Sector (ITU-R) of the International Telecom-
munication Union (ITU) in 2015 for 5G networks, devices and services [5]. ITU-R has
defined the following main usage scenarios for IMT for 2020 and beyond in their Recom-
mendation ITU-R M.2083:
• Enhanced Mobile Broadband (eMBB) to deal with hugely increased data rates, high
user density and very high traffic capacity for hotspot scenarios as well as seamless
coverage and high mobility scenarios with still improved used data rates.
• Massive Machine-type Communications (mMTC) for the IoT, requiring low power
consumption and low data rates for very large numbers of connected devices.
• Ultra-reliable and Low Latency Communications (URLLC) to cater for safety-
critical and mission critical applications which requires different key capabilities
according to ITU-R M.2083.
17
Figure 2.2: ITU requirements for 5G: eMBB, uRLL, Massive IoT (source: ETRI graphic,
from ITU-R IMT 2020 requirements)
Deployed 5G networks deliver more speed and capacity to support massive machine-to-
machine communications and to provide low-latency (delay), high-reliability service for
time-critical applications. Based on the numerous existing, commercial networks, IMT-
2020 demonstrates high performance in different scenarios such as dense urban areas and
indoor hotspots.
With its ambitious goals, 5G networks faced considerable challenges. The increased capac-
ity and data rates enabled by 5G may require more spectrum and vastly more spectrally
efficient technologies, beyond what is used in 3G and 4G systems.
Some of this additional spectrum will come from frequency bands above 24 GHz, which
pose considerable challenges. The first challenge refers to the intrinsic propagation char-
acteristics of this part of the spectrum (a.k.a. millimeter waves). These radio waves prop-
agate over much shorter distances than those of medium- (between 1-6 GHz) and low-
(below 1 GHz) frequency bands.
Hence, coverage of a given area will require a significantly increased number of base
stations that will increase the complexity of the infrastructure, including the need to
deploy radio equipment on street facilities, such as traffic lights, lampposts, utility poles
and power supplies.
Furthermore, spectrum is a scarce and very valuable resource, and there is intense –
and intensifying – competition for spectrum at the national, regional and international
levels. As the radio spectrum is divided into frequency bands allocated to different ra-
diocommunication services, each band should be used only by the allocated services with
established technical condition so that they can coexist with each other without creating
harmful interference to adjacent services.
18
2.1.1 Overall architecture
Like in previous systems (e.g. LTE), the 5G system architecture encompasses three big
constitutive blocks:
• User Equipment
• Next Generation - Radio Access Network (NG-RAN). Responsible for all the radio-
related functionality of the system
• 5G Core Network (5GC). Responsible of the non-radio access related functions such
as authentication, charging, set-up of end-to-end connections, mobility management,
etc.
The NG-RAN, the network function that connects the User Equipment with the 5G Core,
can be composed by two types of nodes [6]:
• gNode B (gNB): it operates with the 5G New Radio (NR) technology
• next generation eNodeB (ng-eNB): it operates with the LTE technology
The 5G core network is responsible for functions not related to the radio access but needed
for providing a complete network. This includes, for example, authentication, charging
functionality, and setup of end-to-end connections. Handling these functions separately,
instead of integrating them into the RAN, is beneficial as it allows for several radio-access
technologies to be served by the same core network. However, it is possible to connect
the NR radio-access network also to the legacy LTE (Long-Term Evolution) core network
known as the Evolved Packet Core (EPC). In fact, this is the case when operating NR in
non-standalone mode, where LTE and EPC handle functionality like connection set-up
and paging [7].
Thus, in the non-standalone approach, 5G NR only provides user plane connectivity.
This approach relies on the use of dual connectivity between LTE and NR (i.e. a UE is
connected to an eNB acting as master cell and to a gNB acting as secondary cell).
19
Figure 2.3: Non-standalone operation
In the standalone operation the gNB is connected directly to the 5GC and the gNB handles
both user-plane and control-plane functions. Also, the dual connectivity with other gNBs,
eNBs and ng-eNBs is possible.
Figure 2.4: Standalone operation
20
2.1.2 5G NR features
Prior to discussing the detailed NR cell search and frequencies, a description of the logical,
transport and physical channels will be provided in this section, including OFDM and
numerologies, carrier aggregation, reference signals and control signalling.
In 5G NR, the set of logical, transport and physical channels is basically the same as in
LTE.
Figure 2.5: Logical, transport and physical channels
At the PHY layer, there exists Primary and Secondary Synchronization Signals (PSS, SSS)
in the DL and different types of Reference Signals (RS) in UL and DL. Also, PDSCH is
the only shared channel.
OFDMA technique is used by 5G NR as the baseline for both DL and UL. In UL, it is
possible to use DFT-precoded OFDMA (i.e. SC-FDMA).
To support a wide range of deployment scenarios, from large cells with sub1 GHz carrier
frequency up to mm-wave deployments with very wide spectrum allocations, NR supports
a flexible OFDM numerology with subcarrier spacings ranging from 15 kHz up to 240 kHz
with a proportional change in symbol and cyclic prefix duration [7].
Table 1: 5G NR numerologies
Subcarrier spacing Symbol duration

Cyclic prefix Frequency range
(∆f) (Tu = 1/∆f )
15 kHz 66.67 µs 4.69 µs FR1 (3GHz)
30 kHz 33.33 µs 2.34 µs FR1
60 kHz 16.67 µs 1.17 µs FR1 and FR2
120 kHz 8.33 µs 0.59 µs FR2
240 kHz 4.17 µs 0.29 µs FR2
5G NR can operate in multiple bands specified in two Frequency Ranges (FR):
21
• FR1: [450– 6000] MHz
• FR2: [24.25 – 52.6] GHz
Different channel bandwidths are allowed for each Frequency Range (FR). In FR1, the
channel bandwidth can be 5 MHz up to 100 MHz. In FR2 the channel bandwidth goes
from 50 MHz up to 400 MHz.
Low values of ∆f are better because they allow having longer cyclic prefix durations,
needed at low frequencies to counteract the longer delay spread associated to large cells.
At high frequencies, high values of ∆f are useful to counteract frequency errors and phase
noise, the cyclic prefix can be shorter because cells are smaller and beamforming is ex-
tensively used, so delay spread is also smaller [6].
The possibility of carrier aggregation is part of NR from the first release. Similar to
LTE, multiple NR carriers can be aggregated and transmitted in parallel to/ from the
same device, thereby allowing for an overall wider bandwidth and correspondingly higher
per-link data rates. The same options like in LTE are: intraband carrier aggregation
with contiguous component carriers, intraband carrier aggregation with non-contiguous
component carrier, interband carrier aggregation and up to 16 carriers can be aggre-
gated (with the same or different bandwidths). The maximum possible bandwidth is
16 ∗ 400M Hz = 6.4GHz. Regarding the operational aspects: one of the carriers is the
Primary Cell, and the others are Secondary Cells and the activation/deactivation of Sec-
ondary Cells is very dynamic to adapt to traffic variations.
The next concept are the reference signals: the four main reference signals are the demodu-
lation reference signal (DMRS), phase-tracking reference signal (PTRS) and channel-state
information reference signal (CSI-RS). DMRS is used to estimate the radio channel for
demodulation. PTRS support the compensation of the phase noise generated by the oscil-
lators, which is particularly relevant at high frequencies (mmW). CSI-RS is a DL reference
signal used by the UEs for channel sounding (i.e. determine the channel state) to be used.
They are configured on a per-device basis and each UE is instructed to measure a specific
set of CSI-RS.
The last topic in 5G NR features is the physical layer control signalling. Like in LTE,
the PDCCH (Physical Downlink Control CHannel) in 5G NR is used to transmit the
information needed to control both the DL and UL transmissions, for example:
• Assigned resources by the scheduling in UL and DL
• HARQ information (timings, process number, etc.)
• Multi-antenna related information (antenna port, etc.)
The PDCCH in NR is designed to support device-specific beamforming, in line with the
general beam-centric design of NR and a necessity when operating at very high carrier
frequencies with a corresponding challenging link budget [7].
Demodulation reference-signals specific for a given PDCCH candidate are mapped onto
22
every fourth subcarrier in a resource-element group, that is, the reference signal overhead is
1/4. The use of a dedicated reference signal per PDCCH candidate is beneficial, despite the
slightly higher overhead than in LTE, as it allows for different types of device-transparent
beamforming. By using a beamformed control channel, the coverage and performance can
be enhanced compared to the non-beamformed control channels in LTE.
2.1.3 Cell search

Cell search covers the functions and procedures used by a UE to find new cells. It is
executed when the device initially enters the system (e.g. at switch-on), but also con-
tinuously when the UE moves within the system. Cell search is based on detecting the
Synchronization Signal Block (SSB) sent by each cell, and composed of:
• PSS (Primary Synchronization Sequence): The first signal that a UE will search for.
It provides synchronization up to the periodicity of the PSS (default 20ms).
• SSS (Secondary Synchronization Sequence): Detected after the PSS to determine
the Physical Cell Identity (PCI).
• PBCH (Physical Broadcast CHannel): Carries the Master Information Block (MIB)
with the minimum information needed by the UE to acquire the rest of broadcast
information (sent through the PDSCH).
Figure 2.6: Synchronization Signal Block
When beamforming is used, SSBs are transmitted in different beams at different times
by means of beam sweeping. A Synchronization Signal (SS) burst set includes L SSBs
transmitted in different times, each one through a different beam. Up to L=64 SSBs can
be transmitted for the higher frequencies (FR2). This allows that the UE identifies the
appropriate beam pair for transmitting/receiving [7].
23
Figure 2.7: Synchronization Signal (SS) burst set
2.2 Allocation of spectrum bands to 5G

The allocation of spectrum bands to 5G services is agreed at the World Radiocommu-
nication Conferences of the International Telecommunications Union (ITU). Within the
European Union, in November 2016 the Radio Spectrum Policy Group (RSPG) approved
the Opinion that identifies the frequency bands to be used in the launching of 5G across
the European Union, namely:
• The 3.4-3.8 GHz band (3400-3800 MHz) is considered to be the main band for
introducing 5G based services in Europe, even before 2020.
• 5G will have to be deployed over bands that are already harmonised below 1 GHz,
including the 700 MHz band specifically.
• The mobile industry supports the 26 GHz band (24.25-27.5 GHz) as a pioneer band
for the early implementation of 5G in Europe over frequencies above 24 GHz.
Operator Sunrise in Switzerland was the first company to launch 5G in Europe in April
2019. Vodafone launched Spain’s first commercial 5G network in 15 cities in June 2019.
Then, Orange launched commercial 5G services in Spain in September of 2020, using
equipment provided by Swedish vendor Ericsson. Telefonica confirmed in a release last
year that the first deployments of 5G SA technology will occur in 2022 [8].
24
Figure 2.8: Frequency bands identified for 5G in Europe and potential use cases [9]
This being said, the Spanish government aims for 75% of the population to have 5G
coverage by 2025 through the bands that have been identified as priorities: 694-790 MHz,
3.4-3.8 GHz and 24.25-27.5 GHz [9].
The Ministry of Economic Affairs and Digital Transformation published in December 2017
a public consultation on the demand and the management and exploitation model of the
26 GHz frequency band, one of the bands identified as priority for the development of 5G
technology, together with the 700 MHz and 3.5 GHz bands.
The 26 GHz band is the last spectrum band that remains to be assigned to Spanish
carriers for 5G.
“The 26 GHz band is appropriate for the provision of electronic communications services,
and also for multiple use cases of low coverage, high capacity and low latency, such as,
for example, the specific coverage of areas with high traffic density.”, the ministry said in
a statement.
According to the ministry, this consultation process is the previous step before launching
the auction process to award the 26GHz band, which will take place in the second half of
2022 [10].
2.3 Milimiter Wave (mmW)

In the course of several decades increased transport capacity requirements and greater and
greater site density have promoted the use of ever higher frequency bands. The physics
of radio waves propagation determine the relation among capacity, availability and link
length.
25
Since the available spectrum is proportional to the centre frequency, the highest frequen-
cies are also those that carry the most capacity, but also cover the comparatively shortest
link lengths [11].
The combination of 5G and millimeter wave bands creates a new level of mobile perfor-
mance with ultra-high speeds and low latencies. Spectrum and channel bandwidth are
critical to enable performance,
Figure 2.9: Channel BW representation
Despite its potential, the utilisation of mmWave for mobile has had to overcome major
technical challenges: mmWave signals travel relatively short distances; can be susceptible
to attenuation from trees and other obstacles; and have difficulties in penetrating through
material, in particular concrete but also glass and wood. This means that they may not
be a good solution for reaching indoors or dealing with severe obstacles.
However, as mobile data traffic continues to grow rapidly, with demand for higher data
rates to serve new applications along with a potential need for more uplink capacity,
the need for mmWave bands is only becoming more apparent. mmWave bands can ac-
commodate more capacity and bandwidth than any other band. And since spectrum in
these bands is abundant, mmWave spectrum is ideally placed to deliver high speeds, low
latency and high capacity, all at the same time. The short wavelength of mmWave al-
lows for very small antennas, which helps with beam forming for enhanced coverage and
spectral efficiency. mmWave can also be a good solution indoors where the propagation
characteristics become an advantage to avoid inter-cell interference [12].
2.4 Densification and small cells

By 2026, the Small Cell Forum expects that 38.3 million small cells will be deployed,
with the bulk of those deployments coming from enterprises, followed by urban and rural
service provider environments.
The compact size and lower power consumption of a small cell means it has many appli-
cations. For example, a 5G small cell can provide an extension from the core network into
26
densely urban areas or may be used in indoor settings to extend a private or corporate
network within an enterprise. Or, in a metropolitan area to offload users from an over-
loaded macrocell. Similarly, the small cell can improve coverage and throughput in areas
where buildings obstruct the network.
The ability to deliver multi-gigabit per second is where 5G small cells can really help
alleviate issues of densification so that multi-gigabit throughput can be maintained in
user-dense applications. In the 5G era, 5G small cells will enable the network to extend
coverage and to deliver lower latency, and also serve more users while maintaining multi-
gigabit performance.
5G small cells will play a critical role in a number of use cases. Take for example large
venues such as sports complexes or congress halls where thousands of users need to connect
within an enclosed area. In order to ensure all users maintain network coverage with good
throughput, 5G small cells can be deployed in these facilities and surrounding areas to
provide extra capacity so that the influx of users do not exhaust the network [13].
Enterprise small cells typically transmit at power levels between around 0.02 watts and
0.25 watts, with a range of up to around 200 metres, depending on the desired quality of
service and number of simultaneous multiple users. Urban small cells typically transmit
at power levels between around 0.25 and 6.0 watts, giving extra coverage and capacity to
serve multiple users. Small cells typically allow mobile phones to work at very low powers,
increasing their battery life and reducing interference.
2.5 Beamforming and Multiple Input Multiple Output (MIMO)

In release 13, and continued in release 14, improved support for massive antenna arrays was
introduced, primarily in terms of more extensive feedback of channel-state information.
The larger degrees of freedom can be used for, for example, beamforming in both elevation
and azimuth and massive multiuser MIMO where several spatially separated devices are
simultaneously served using the same time-frequency resource .
Massive MIMO uses a large excess of service-antennas over active terminals and time di-
vision duplex operation. Extra antennas help by focusing energy into ever-smaller regions
of space to bring huge improvements in throughput and radiated energy efficiency.
With massive MIMO, we think of systems that use antenna arrays with a few hundred
antennas, simultaneously serving many tens of terminals in the same time-frequency re-
source. Overall, massive MIMO is an enabler for the development of broadband (fixed and
mobile) networks which are energy-efficient, secure, and robust, and use the spectrum ef-
ficiently [14].
At high frequencies (FR2: [24.25 – 52.6] GHz) the bandwidth is higher (up to 400 MHz),
so capacity is not an issue and propagation conditions are substantially worse, given this
situation MIMO can exploit beamforming, to increase the directivity of the antennas and
compensate the higher propagation losses.
27
Figure 2.10: Comparison between 4G regular antenna and 5G with beamforming
Beamforming is a particular processing technique for signals that allows for directional
transmission or reception letting 5G connections to be more focused toward a receiving
device. For example, a typical 5G small cell that does not employ beamforming during
its multiple-input multiple-output (MIMO) transmission will not be able to narrowly
concentrate or focus its transmit beams to a particular area. With beamforming, the
small cell can focus the transmission in a particular direction towards a mobile device
such as a cell phone, laptop, autonomous car or IoT node. This improves the efficiency
overall of the network and saves energy. Because millimeter waves can be more prone
to disruption from interfering objects, and mmWave does not penetrate walls and other
barriers as easily as low and mid band spectrum used for 4G, beamforming can help to
create more reliable connectivity dynamics.
The small wavelength of mmWave frequencies allows us to pack many antennas in a small
area. Transmitting a signal through many antennas with specific phase offsets provides
beamforming gain by coherently combining the electromagnetic signal radiated from each
transmitter antenna. Transmit and receive beamforming power gain, coupled with re-
duced interference, boosts signal-to-interference-plus-noise ratio (SINR), bringing higher
data rates, more cellular capacity, and extended cell coverage. The realized gain in SINR
is especially useful for mmWave communication systems, given the high penetration and
propagation losses at mmWave frequencies. Furthermore, multiple beams at the trans-
mitter can be used to multiplex parallel data streams to communicate with spatially
separated multiple users (MU-MIMO). Multiple beams at the receiver can bring diversity
gain, increasing reliability via redundant data streams [15].
To realize efficient beamforming, each antenna requires an amplitude controller, phase
shifter or time delay element. A specific configuration of these elements, coupled with the
antenna array geometry, determines the beamforming radiation pattern. Either open-loop
or closed-loop beamforming estimates can be used to perform beamforming. Closed-loop
techniques consider that the channel estimated between each transmitter and receiver
antenna is known, allowing the best beamforming configuration to be obtained from the
estimated channel matrix. Open loop techniques utilize a pre-established codebook for
beamforming at both the transmitter and receiver. The best beamforming configuration
28
from the codebook is selected according to the received signal strength without explicitly
estimating the channel.
Depending on the choice of phase shifter types in the analog or digital domains, several
beamforming architectures have been considered. Two common beamforming architec-
tures for mmWave communications are: analog phase array (APA) and hybrid phase ar-
ray (HPA). HPA and APA are the preferred architectures for mmWave frequencies when
performance and cost tradeoffs are considered.
Analog beamforming implies that: it is not possible to have different beam directions for
different frequencies (RBs) in the same carrier and beams to UEs located in different
directions should be transmitted at different times. Thus, in analog beamforming at most
one beam at a time can be used for transmission. Different devices therefore need to be
time-multiplexed and with the very large bandwidths available in the mmWave range, a
few OFDM symbols can be sufficient even for relatively large payloads [7].
Figure 2.11: Analog beamforming [6]
Although fully digital architecture (FDA) is popular choice for sub-6 GHz massive MIMO
communications, it is less suitable for mmWave due to the channel sparsity, high propaga-
tion loss etc. HPA and APA are the preferred architectures for mmWave frequencies when
performance and cost tradeoffs are considered. FDA is considered as a next generation
architecture for mmWave, with pending advancements in circuit technologies.
29
Figure 2.12: Digital beamforming [6]
Hybrid phased array (HPA) architectures which use multiple pairs of ADCs/DACs (RF
chain) are also considered to support multi-user/multi-stream transmission. The number
of RF chains are usually much smaller than the number of antennas and can be as low as
the required number of data streams.
HPA utilizes both digital beamforming with multiple RF chains and analog beamforming
with analog phase shifters (thus “hybrid”). The digital beamforming can be used to further
reduce the inter-beam interference. The main benefit of HPA is it enables multi-beam
transmission to support multi-user communication using the same time and frequency
resources. In addition, the digital beamforming part of HPA can be useful for improving
beamforming gain in non-line-of-sight channels for better coverage.
Compared to analog beamforming, HPA can provide faster initial access, since a transceiver
with HPA can create multiple beams in different directions at the same time. This func-
tionality is critical especially when control channels are also beamformed to improve
mmWave cell coverage [15].
Figure 2.13: Beamforming architectures: i) analog (left), ii) digital (middle), iii) hybrid
(right) [16]
30
3 Methodology
The methodology applied in this master’s final thesis can be explained in five phases:
The first phase, the theoretical planning phase, we defined two scenarios in the city of
Barcelona, at the UPF’s Ciutadella campus with a total area of 0.052km2 . The only
difference between the two scenarios, which we will refer to as small venue and big venue, is
the density of users, having around 400 pedestrians in the small venue scenario and around
2000 pedestrians in the big venue scenario; in both, users connect to the following services:
broadband, video call and voice call by using a 5G terminal, following the performance
requirements for high data rate and traffic density scenarios by 3GPP. In both scenarios,
users can request downloads from services like video streaming and video call, usually
5Mbps per user in DL and 2Mbps per user in DL, respectively. We use for the small cells
deployment a mmWave frequency band, n257. Band n257 refers to 26.5-29.5 GHz, also
referred as 28 GHz, and is the most tested band for Millimeter Wave communication in
5G NR networks.
For the deployment phase, the second phase, we use the professional software, for wireless
network design and optimisation, ATOLL from Forsk company. We implemented first
the small venue scenario, placing the small cells on the rooftops and the open areas
surrounding the buildings of the chosen computation zone: the Ciutadella UPF campus;
and we defined the parameters of the cells, the traffic maps, the services (broadband,
video call and voice call), the type and density of users, etc. The details can be found in
the next chapter. Secondly, we deployed the big venue scenario, placing the small cells
indoors (on account of ATOLL’s update) and defining, again, the parameters mentioned
above.
The third phase, the performance analysis phase. This phase consisted on running predic-
tions regarding the coverage, the signal quality, the capacity, the beam usage, etc. to study
the best course of action for the location of the 5G sites, followed by defining a traffic
map and running simulations with users’ traffic to analyze how the network behaves and
if it meets the requirements demanded by the services and users.
The fourth phase, the optimization phase, we solved the detected problems and increased
the benefits by adding sites, re-locating the actual sites, changing the bandwidth, the
mechanical downtilt of the antennas, the parameters of the 5G cells, as we will explain in
further detail in the next chapter.
Between this phase and the previous phase, the performance analysis phase, iterations
were produced by, with the metrics and data obtained from this phase, making new
iterations until the network deployment was optimized.
Lastly, this project has the conclusions and future work phase, in this phase the conclusions
obtained through the project deployment and results are summarized and the future lines
of work and improvements that can be continued in the future related with this project
are explained.
31
4 Project development
In this chapter, the description of the scenarios is presented. Also ATOLL’s configura-
tions and parameters are explained in more detail: the smart antennas and beamforming
antennas used in both scenarios, the configuration of the services and the predictions and
simulations.
4.1 Defined scenarios

As we mentioned previously we have two scenarios: the small venue and the big venue
with particular cases in each of them.
4.1.1 Small venue scenario

Starting with the small venue scenario, the deployment map is the following,
Figure 4.1: Ciutadella deployment map for the small venue scenario
As we explained at the beginning of the report the objective is to deploy a network for
indoor users that are attending a small fair or congress in the Ciutadella campus where
most of the pedestrians are inside the buildings and a small amount are moving around
buildings. For this reason two traffic maps are defined: one, the indoor where the density
32
of users is higher, second, the outdoor traffic map with a low density of users. Both user
environment traffic maps areas are,
Figure 4.2: Indoor User Profile Environment Figure 4.3: Outdoor User Profile
Traffic Map (light blue areas) Environment Traffic Map (dark blue area)
We define two types of user profiles: business user and standard user, one with more
requirements than the other. For the business user and standard user defined in both
environments, the properties regarding the services they use and the traffic they generate
are,
Table 2: Business user properties
Uplink Downlink
Duration
User Profile Terminal Calls/hour Volume Volume
(sec.)
(KBytes) (KBytes)
Broadband 5G Smartphone 7 10 000 50 000
Video Call 5G Smartphone 2 180
Table 3: Standard user properties
Uplink Downlink
Duration
(sec.)
(KBytes) (KBytes)
Voice Call 5G Smartphone 2 240
33
The user density for each environment, indoor environment and outdoor environment
traffic maps, are (considering a surface of 0.052km2 ),
Table 4: Small venue indoor environment
Density
User Profile Mobility
(Subscribers/km2 )
Business User Pedestrian 45 000
Standard User Pedestrian 60 000
Table 5: Small venue outdoor environment
Density
(Subscribers/km2 )
For the small venue scenario, the sites use two different 5G frequencies: n257 in the
majority of the sites and n78 in one site to give coverage to the small areas where there
was none due to the range of small cells in mmWave frequencies. Thus, the frequencies
chosen for this scenario are,
Table 6: 5G frequencies used in the small venue scenario
Reference Frequency Carrier centre

Name Bandwidth (MHz)
(MHz) (ARFCN)
n257 26 700 2057499 400
n78 3 350 623334 100
For most of the sites we chose the mmWave frequency band n257 because n257 is intended
for short range transmission at high data rates. It allows 400 MHz of the spectrum to be
dedicated to 5G services as we can see from the previous table. The 3GPP band n257
refers to 26.5-29.5 GHz. It is commonly called 28 GHz.
In this project we use an Aster mmW propagation for both scenarios: the small venue
and the big venue, and also Aster propagation model in the small venue scenario because
we have one small cell using band n78.
4.1.2 Big venue scenario

After designing and implementing a 5G mmWave with beamforming deployment for the
small venue scenario, taking advantage of the update from ATOLL software in this section
34
we are deploying an indoor scenario with both the base stations and the users inside the
buildings in the UPF Ciutadella’s campus in Barcelona.
The map of the Computation zone with the base stations is the following:
Figure 4.4: Ciutadella deployment map for the big venue scenario
In the map it is displayed by colour the coverage by transmitter, that is, each colour
represents the coverage by one mmWave small cell. We can asses that, unlike the small
venue scenario, in this scenario the traffic will always be indoor.
The actual deployment has 12 base stations in total: 4 in the building Depòsit d’Aigües
or building 10, 5 BSs in Edifici Jaume I or building 20 and lastly 3 BSs in Edifici Roger
de Llúria or building 40.
Because we are using band n257 for all the base stations in the deployment, we use the
32T32R 360 degrees beamforming antenna which is specific for high frequencies. In section
4.3 more information about this beamforming antenna is given.
In this scenario instead of using a 2D traffic map following the example of the small venue
scenario, we want to create and use a 3D traffic map to have pedestrians in each floor of
the buildings, in order to do so, first we need to create a 2D traffic map to use the results
from the simulations to define the 3D traffic map. The 2D traffic map in the big venue
scenario has the following user profiles,
35
Table 7: Business user properties
Uplink Downlink
Duration
(sec.)
(KBytes) (KBytes)
Video Call 5G Smartphone 1 600
Table 8: Standard user properties
Uplink Downlink
Duration
(sec.)
(KBytes) (KBytes)
The user density for the 2D traffic map environment is (considering a surface of 0.0051km2 ),
Table 9: Big venue indoor environment
Density
(Subscribers/km2 )
To run-through, for the 3D map creation, we create a cell traffic map for which we enter,
for each cell, the number of users by importing tabulated information. Then, Atoll gets the
exact number of users per cell with the proper map as a starting point in simulations. In
the next step, we define the ratios of terminals and mobilities. This process is explained
in more detail in section 4.6. Now, we can select the traffic map which is compatible
with 3D simulations (environment traffic maps are not compatible). Therefore, for the
simulations run in ATOLL for the big venue scenario there are pedestrians in each floor
of the buildings.
Regarding the frequency band in this scenario, the sites use the n257 band using different
carriers in order to help with the intra-frequency interference for small cells located in the
same buildings. Thus, the carriers chosen for this scenario are,
36
Table 10: 5G frequencies used in the big venue scenario
Reference Frequency Carrier centre

Name Bandwidth (MHz)
(MHz) (ARFCN)
n257 26 700 2057499 400
n257 26 600 2055833 200
n257 26 800 2059166 200
4.2 ATOLL Antennas

In the indoor deployment carried out in this project there are three antennas used: a 70
degrees smart antenna, a 32T32R 360 degrees beamforming antenna for high bands and
the Default Beamformer.
Atoll models beamforming using the smart antenna equipment. The 70 degrees dedicated
Broadcast smart antenna has the following DL coverage when located in a rooftop,
Figure 4.5: DL coverage by a 70 degrees antenna (SA Broadcast)
When choosing the antenna, the process followed has been to carry out a study for a
single transmitter and then apply it to the rest of them. The chosen antenna “70deg (SA
Broadcast)” has the following radiation pattern:
37
Figure 4.6: 70 degrees antenna azimuth Figure 4.7: 70 degrees antenna elevation
(horizontal) pattern (vertical) pattern
4.3 ATOLL 3D Beamforming

The 3D beamforming antenna used in the small venue scenario has 60 beam indexes and
it is called “Default Beamformer” and its main characteristics are listed below:
It has 8 columns (N) and 8 rows (M), a 64-array with 64 transmission ports and 64
reception ports, as we said, it has 60 beam indexes. The azimuth ranges from -45º to 45º
and the tilt ranges from -5º to 5º. Lastly, the antenna gain is 26.08dBi.
After explaining the broadcast smart antenna and the Default Beamformer used in the
small venue scenario, we will define the 3D beamforming antenna used in the big venue
scenario: 32T32R 360 degrees beamforming antenna for high bands.
The physical parameters of this antenna are the following: the frequency ranges from
24.25GHz to 52.6GHz, it has 4 rows (M) and 8 columns (N) obtaining a 32-array with
32 transmission ports and 32 reception ports, lastly it uses cross-polar polarisation. The
beamforming antennas has two types of beams: broadcast beams and service beams. More
information about them is given afterwards. For the two types of beam the radiation
pattern are:
38
Figure 4.8: Broadcast beams radiation
pattern Figure 4.9: Service beams radiation pattern
Both have the same radiation pattern but the dedication is different as we can see in the
following section 4.3.1.
4.3.1 Types of beams in ATOLL

• Beamforming for broadcast channels: Fixed-weight beamforming is used to
process the broadcast channel, control channel, and reference signals of all user
devices in a cell. Different fixed antenna weights (amplitude and phase) can be
configured to change the coverage scope of the broadcast beams, improving cell
coverage performance.
• Beamforming for service channels: PDSCH dynamic adaptive beamforming
is used for user devices. Narrow beams are formed in order to concentrate energy
on target user devices and reduce interference to surrounding user devices. The
beamforming is based on TDD uplink and downlink channel reciprocity and uplink
channel measurements. This increases the demodulation SINR for target user devices
and improves cell-edge performance [17].
4.3.2 Difference between Broadcast and Service Beams

Broadcast beams determine the coverage area, whereas service beams are used for data
transmission.
39
Figure 4.10: Broadcast vs Service Beam [17]
[*] DMRS is short for dedicated demodulation reference signal, which is a reference signal for the PUSCH
4.4 Modelling Services

There are various services, such as voice, mobile internet access, etc., available to sub-
scribers. These services can be either circuit-switched (voice) or packet-switched (data)
depending on the radio access technology and the type of application.
The service properties dialog box consists of multiple tabs, depending on the radio access
technologies that you are using. For Single-RAT documents, the dialog box displays a
General tab and a Parameters tab. For Multi-RAT documents, there is a General tab and
a tab for each technology.
You can select either Voice, Data, IoT, or Broadband as the service type. Broadband type
services support carrier aggregation. Voice and Data type services do not support carrier
aggregation [3].
For the Broadband service supporting 5G NR used in the indoor deployment, the param-
eters are:
40
Figure 4.11: 5G NR Broadcast service properties
For the Video Call service supporting 5G NR used in the particular case in the big venue
scenario where we change the technology type from Voice to Data, the parameters are:
41
Figure 4.12: 5G NR Video Call service properties as Data type
For the Voice Call service which uses Voice type, we select the default configuration
given by ATOLL. Due to the type of technology used, this service does not affect the
throughputs of the other services described.
4.5 ATOLL Predictions

ATOLL allows the generation of prediction plots, based on calculated cell load figures
[18]:
• Cell and network coverage analysis
• Cell and beam acquisition plots
• Intra- and inter-technology interference analysis
• Downlink and uplink service areas
• Downlink and uplink throughputs
42
4.5.1 Diversity support modes
There are various diversity modes when using 5G NR in ATOLL: Transmit and Receive
Diversity, Single-User MIMO or Spatial Multiplexing and Multi-User MIMO or Collabo-
rative MIMO.
Figure 4.13: Diversity support in DL and UL in ATOLL’s 5G NR Cells table
Transmit or receive diversity uses more than one transmission or reception antenna to send
or receive more than one copy of the same signal. The signals are constructively combined
(using optimum selection or maximum ratio combining) at the receiver to extract the
useful signal. As the receiver gets more than one copy of the useful signal, the signal level
at the receiver after combination of all the copies is more resistant to interference than
a single signal would be. Therefore, diversity improves the C/(I+N) at the receiver. It is
often used for the regions of a cell that have insufficient C/(I+N) conditions.
During calculations in Atoll, a user (pixel, mobile, or subscriber) using a MIMO-capable
terminal, and connected to a cell that supports transmit or receive diversity, will benefit
from the downlink or uplink diversity C/(I+N) gains if the received SU-MIMO criterion
(RS C/N, RS C/(I+N), or PDSCH or PUSCH C/(I+N)) is less than the SU-MIMO
threshold defined in the reception equipment of the terminal or cell, respectively.
SU-MIMO uses more than one transmission antenna to send different signals (data streams)
on each antenna. The receiver can also have more than one antenna to receive different
signals. Using spatial multiplexing with M transmission and N reception antenna ports,
the throughput over the transmitter-receiver link can be theoretically increased M or N
times, whichever is smaller, M or N. SU-MIMO improves the throughput (channel capac-
ity) for a given C/(I+N), and is used for the regions of a cell that have sufficient C/(I+N)
conditions. SU-MIMO (single-user MIMO) is also referred to as SM (spatial multiplexing)
or MIMO.
During calculations in Atoll, a user (pixel, mobile, or subscriber) using a MIMO-capable
terminal, and connected to a cell that supports SU-MIMO, will benefit from the SU-MIMO
gain in its throughput depending on its PDSCH or PUSCH C/(I+N) if the received SU-
MIMO criterion (RS C/N, RS C/(I+N), or PDSCH or PUSCH C/(I+N)) is higher than
or equal to the SU-MIMO threshold defined in the reception equipment of the terminal
or cell, respectively.
MU-MIMO (Multi-User MIMO) or collaborative MIMO is a technique for spatially multi-
plexing users in good radio conditions. A cell with more than one antenna port can serve
different users over the same frequency-time allocation. This technique provides consider-
able capacity gains and can be used with single-antenna user equipment, i.e., it does not
43
require more than one antenna at the user equipment as opposed to SU-MIMO, which
only provides considerable gains with more than one antenna at the user equipment.
MU-MIMO can only work under good radio conditions and if the cell has more than
one reception antenna port. Therefore, the RS C/N, RS C/(I+N), or PDSCH or PUSCH
C/(I+N) must be higher than the MU-MIMO threshold defined in the reception equip-
ment in order for the scheduler to be able to multiplex users.
4.5.2 Aster and Aster mmWave propagation models

Aster is a high-performance propagation model for Atoll that supports macro, micro,
and small cell urban propagation scenarios. Aster also includes a dedicated mmWave
propagation model for 5G frequencies above 6GHz. Aster is based on advanced ray-tracing
propagation techniques and combines high accuracy with superior calculation speed.
Aster can take all the main radio propagation effects into account, leading to highly accu-
rate coverage prediction results. Its accuracy applies to antennas above rooftops (where
signal levels are mainly due to vertical diffractions) as well as antennas under rooftops
(where signal levels are mainly due to horizontal diffractions).
For high-frequency propagation modelling, it takes into account for vegetation the diffrac-
tion and through loss, also atmospheric absorption, rain attenuation, LoS propagation at
different heights, diffraction and specular reflections [19].
4.6 3D traffic map

ATOLL allows to create sector maps that then they can be used to create simulations
using a 3D distribution of the traffic.
The parameter used to create the sector map for this project is the following: using a
prediction “Coverage by Transmitter” with a sensitivity of -120 dBm.
In a current LTE network, the transmitter at the device end may be up to 23 dBm output
power, and requires below -113 dBm receiver sensitivity. In 5G now, on cellular technology,
the receiver sensitivity is up to -120 dBm.
Using the coverage obtained from the prediction and using the percentage of traffic load
in DL, we can extract the percentage of users in each sector and create a 3D traffic map
where we define that the 70% of the users will be using Broadcast, 20% Voice Call and
the last 10% will be in a Video Call following the parameters defined in each of these
services where all the users are using a 5G Terminal.
4.7 ATOLL Simulations

4.7.1 Simulation results
After you have created a simulation you can either display the results as a distribution
map or you can access the actual values of the simulation. Actual values can be displayed
either for a single simulation or as average values for a group of simulations.
44
The Simulation Results window displays results for single simulations and for groups of
simulations. When displaying results for a group of simulations, the Sites and Cells tabs
display average results.
In the Statistics tab, one parameter is the Total number of users not connected (rejected),
in this project we have two types: No Coverage and No Service.
If users are rejected during server determination, the cause of rejection is ”No Coverage”.
If users are rejected because quality is too low to obtain any codec mode or coding scheme,
the cause of rejection is ”No Service” [3].
In regards to the best serving cell selection in 5G, in order to avoid a 5G transmitter’s
cell to be overloaded from DL or UL traffic load, we can configure the parameter ”Cell
Individual Offset (CIO) (dB)”. The CIO adds the specified number of dBs to the signal
level detected by a mobile terminal, so that even if the actual received signal is lower
than that emitted by the overloaded base station, the terminal will connect to the small
cell with less traffic load. This makes the terminals prioritize connecting to one particular
small cell instead of the neighbouring cell which could have been the better serving cell,
i.e. we can tune or bias the ranking of potential servers for best serving cell selection in
connected mode [3].
Another parameter we use in this project is the SSS EPRE (dBm): The SSS energy per
resource element. This value is used to calculate the transmission power corresponding
to the secondary synchronisation signal (SSS) [3]. For the small venue and big venue to
avoid the intra-frequency interference, one option was to choose different carriers with
different bandwidth or change this parameter to decrease the maximum power allowed by
the sites.
45
5 Results
In chapter 5, the results obtained from the predictions and simulations done in the project
are collected here. The results can be divided in the small venue scenario and the big venue
scenario.
5.1 Small venue

In this section we will analyze the predictions and simulations made during the realization
of this scenario with the intention of demonstrating the improvements that beamforming
brings with respect to the use of MIMO technology or without using any type of diversity
support. More information about the diversity support modes is found in 4.5.1.
As we said, the small cells are using millimeter wave bands and providing broadband,
video call and voice call services to users within the Ciutadella campus buildings.
The map of the computation zone (as referred in ATOLL) with the base stations can be
found in section 4.1.1, as well as, the parameters of the base stations and the services.
When deploying the small cells in the rooftops and areas around the campus buildings, we
decided to add one small cell using n78 band due to the fact that on the one hand using
only mmWave for the deployment was not enough to cover the whole computation zone
and on the other hand adding mmWave small cells for only the outdoor areas without
coverage was not practical (given the great amount of small cells needed in areas with
low density of users). The choice made was to add only one site using band n78 that
has more coverage range compared with mmWave which has only up to 100m. Band n78
(3500 MHz), or commonly referred to as the 3.5 GHz 5G band, is the most commonly
tested and deployed 5G frequency.
In the next figure, we can observe the areas without coverage, mostly open spaces at the
edge of the computation zone where the antennas placed on the map do not reach.
46
Figure 5.1: DL coverage with only n257 using all diversity support modes, before using
n78 too
5.1.1 No MIMO nor Beamforming

Before analyzing the results of the predictions and simulations run in ATOLL for the
scenario of a 5G NR mmWave with small cells indoor deployment using Beamforming,
we will study the most basic case where the diversity support modes are deactivated, this
means, no MIMO and no Beamforming.
The way ATOLL is configured, on one hand, you can deactivate the beamforming antenna
and, on the other hand, you can deactivate the diversity support modes, explained each one
in further detail in 4.5.1, where there is the transmit and receive diversity that helps with
interference and SINR, usually at the edges of the cells where the quality is worse, there
is the SU-MIMO and the MU-MIMO that help with the throughput or channel capacity.
When doing a comparison between them in this section, we study three cases, deactivating
them one by one until we reach the most basic case with everything deactivated:
• The first case, only deactivating SU-MIMO and MU-MIMO
• The second case, only deactivating the beamforming antenna
• The third and last case, deactivating all the diversity support modes including trans-
mit and diversity support, as well as, beamforming
To begin with, we will focus on the first case of having no SU-MIMO and no MU-MIMO
but still using Beamforming, these technologies have been explained in section 2.5,
47
Figure 5.2: DL coverage with no SU-MIMO
nor MU-MIMO Figure 5.3: DL coverage legend
Figure 5.4: DL quality with no SU-MIMO

Figure 5.5: DL quality legend
nor MU-MIMO
As we can observe from the figures, in both cases the results are not too bad, except in
some areas in the buildings where we can see a decrease in the quality, overall the results
are quite good.
48
Now, onto the second case: no beamforming but still using SU-MIMO AND MU-MIMO,
we will observe that for our scenario not using beamforming greatly decreases both the
coverage and the quality in DL.
Figure 5.6: DL coverage with no

Beamforming
Figure 5.7: DL coverage legend
Figure 5.8: DL quality with no

Figure 5.9: DL quality legend
Beamforming
Regarding the coverage we appreciate almost no difference because for short range situa-
tions there is very little to no additional performance boost effect. In comparison, when
we compare figure 5.4, only deactivating SU-MIMO and MU-MIMO, and figure 5.8, only
49
deactivating the beamfoming antenna, we can observe that it is very different, in figure
5.8 only the areas really close to the antennas have good quality (red), the other areas
have bad quality (yellow/green), the reason why is mmWave are more prone to disruption
from objects than interference, such as walls so beamforming helps create a more reliable
connectivity by allowing a transmitter to focus the transmission in a specific direction,
thus without beamforming all the reflections created by the building walls in the map
greatly worsens the signal quality at receiver.
The third case, the most basic case, is the worst possible case because there is no diversity
support mode at all, not even transmit or receive diversity. As we can see in the following
figure 5.12, the DL quality is very unsatisfactory, most of the area has a lower value than
10dB PDSCH SINR (Signal to Interference and Noise Ratio) and some areas now have
no signal at all.
Figure 5.10: DL coverage with no support Figure 5.11: DL coverage legend

modes
50
Figure 5.12: DL quality with no diversity Figure 5.13: DL quality legend
support modes
In terms of gains, we will first compare the first case, the single case where we did not use
MIMO but continued to use Beamforming, against the worst case, the third case, where
we did not use any type of diversity support mode. Secondly, we will do a comparison
between the second case, not using Beamforming but using MIMO, against the third case,
the worst case,
Figure 5.14: DL quality with no diversity Figure 5.15: DL quality with no diversity
support modes vs no MIMO (black) support modes vs no Beamforming (black)
In figure 5.14 when not using SU-MIMO nor MU-MIMO around 70% of the area has 30dB
of PDSCH SINR compared with not using any diversity support which is less than 10%,
and whereas without using MIMO the percentage of computation with good quality signal
(≥ 20dB) stays higher than 70%, for the case of not using diversity support around 0dB
of SINR the % of area decreases quite quickly down to 20% for a good value of PSDCH
SINR of 20dB.
Regarding 5.15, we can see that, as we mentioned previously in the comparison between
figure 5.4 and figure 5.8, not having beamforming affects more than not using MIMO. In
51
the comparison between not using beamforming versus no diversity support, in both cases
around 0dB they start the negative slope reaching 15dB of SINR now with 70% and the
other with 30%, more than double the percentage (in figure 5.14 we had 70% at 30dB not
using MIMO). At 30dB of SINR again the percentage is doubled with a 20% and 10%.
In essence, the worst possible case is having no diversity support at all as we can see from
the previous figures and graphs where it creates areas with no signal at all and the slope
of the graphs is very steep.
5.1.2 All diversity support modes active

After comparing the effects of not using any technological advantages of the different
mobile communications generations, we will analyze the predictions using diversity versus
the previous figures,
Figure 5.16: DL coverage with all diversity Figure 5.17: DL quality legend
support modes
52
Figure 5.18: DL quality with all diversity Figure 5.19: DL quality legend
support modes
At the edges of the map we can see a lower value of coverage (blue) which is related in
figure 5.18 with the lower quality level (green) of PDSCH SINR. For all the buildings of
the Ciutadella campus we have around 30dB of SINR, a very good signal. Taking into
account that the users will be indoors and only a small amount of users will be around the
buildings as we defined in table 4 and table 5 in chapter 4.1.1, we can consider that this
deployment matches the requirements of the scenario we want to study and implement.
If we compare this case with the three restricting cases we have seen previously, the results
are,
Figure 5.20: DL quality with diversity vs no Figure 5.21: DL quality with diversity vs no
MIMO only (black) beamforming only (black)
53
Figure 5.22: DL quality with diversity vs no diversity (black)
As we have seen previously, when we worsen the conditions there is a penalty in levels
of quality, in figure 5.20 the difference is very small compared with the situation where
all diversity modes are activated. Next, in figure 5.21 there is only a slight decrease in
the area for 30dB, instead with no beamforming starting at 0dB there is a continuous
decrease of the SINR. Lastly, when there is no diversity support mode, there is a decrease
of the area covered because as we have seen in figure 5.12 it creates areas without signal,
again the SINR starts decreasing at 0dB with a steeper slope than in figure 5.21.
In summary, when comparing these graphs we can see the gains of using beamforming
both in the overall coverage and signal level of the deployment, after all beamforming is
an improvement regarding the MIMO technology.
5.1.3 Simulations’ results using beamforming

As we mentioned at the start of this section 5.1 we needed to add one site using n78 to
give coverage to the areas at the edge of the computation zone. The results of only using
small cells with n257 and not yet the small cell with band n78; and using the traffic maps
defined in section 4.1.1 is,
The total number of users not connected (rejected) is 13,6 (3,3%) from 406,33 users.
Table 11: Mobile’s rejection cause for the case of only n257 sites
No Coverage 13,2 users

No Service 0,4 users
As defined at the start of the deployment of the small venue the number of users outside
the buildings is quite small, the number of users with no coverage is not very high, being
less than 5% of the total number of users, nonetheless our objective is to have all users in
the computation zone with coverage.
54
When deploying the n257 sites with the added n78 one, the results of the number of users
rejected is,
The total number of users not connected (rejected) is 0,2 (0%) from 475,53 users.
Table 12: Mobile’s rejection cause for the case of using n257 and n78 sites
No Coverage 0,13 users

No Service 0,07 users
The results obtained from the rejected users are very good, being 0% for 475 users. But,
also the results obtained from the broadband service, which is the only one being used that
has throughput demand in DL and UL, are also great. The maximum throughput demand
in DL is 1,893.33Mbps and the peak RLC cumulated throughput in DL is 1,822.74Mbps.
For UL, the maximum throughput demand is 336.07Mbps and the peak RLC cumulated
throughput is 335.74Mbps.
Regarding the total users obtained classified by activity status is:
Table 13: Total number of connected users for the case of using n257 and n78 sites
Users 475.53
Downlink 140
Uplink 110,87
Downlink + Uplink 150,33
Inactive 74,33
With these users, the total accumulated throughput and the average throughput per user,
both in the downlink and in the uplink is:
Table 14: Total accumulated and per user throughput for the case of using n257 and n78
sites
Total Per User

Downlink 1731.2Mbps 5.96Mbps
Uplink 318.39Mbps 1.21Mbps
After showing the average statistics of the simulation using beamforming and the n78 and
n258 frequency bands, we analyze for each service provided in this deployment (broad-
band, video call and voice call) the metrics for the best service beam. As we mentioned
in section 4.3, the Default Beamformer used in this scenario uses 60 beam indexes, taking
into account the traffic distribution depicted in figure 4.2 and figure 4.3, the graphs for
the best service beam used in each site or cell for each service is:
55
Figure 5.23: Best Service Beam counter for Figure 5.24: Best Service Beam counter for
Video Call Voice Call
Figure 5.25: Best Service Beam counter for Broadband
From the figures we can infer that the site with more service beam usage is site5 and the
one with less usage is site2, given that site2 is an outdoor cell between buildings which
gives coverage to the pedestrians in that area, the density of users compared with the users
in the building where site5 is located is much lower. We also have to take into account
that more that one beam can be used by only one user but ATOLL metrics only allows
to know the best service beam used.
5.1.4 Simulations’ results without using beamforming

After analyzing the beamforming in the small venue scenario, in the next simulations’
results we will study the effect of not using beamforming when users are involved and
how it translates in limitations with the throughputs obtained and other metrics.
The total users obtained from the simulation average are the following:
Table 15: Total number of connected users for the case of not using beamfoming
Users 423
Downlink 148
Uplink 106
Downlink + Uplink 95
Inactive 74
56
With these users, the total accumulated throughput and average per user throughput,
both in the uplink and in the downlink, is:
Table 16: Total accumulated and average per user throughput for the case of not using
beamforming
Total Per User

Downlink 1194.16 Mbps 4.91Mbps
After comparing this table with table 14, we observe a decrease in the total accumulated
throughput both in DL and UL due to the decrease in the signal quality, the received
SINR, and for the users in DL they have in average 1Mbps less.
Regarding, the received PDSCH power without using beamforming versus using it ,
Figure 5.26: Histogram of Received PDSCH Figure 5.27: Histogram of Received PDSCH
Power (DL) (dBm) without Beamforming Power (DL) (dBm) using Beamforming
Even though both graphs have a similar distribution, the lowest value in figure 5.26 is
-104.4dBm and in figure 5.27 is -74.57dBm, a 30 dBm difference. Also, the average in the
first one is -70dBm and in the second one -48.9dBm, which means that because of the
beamforming there is an average increase of around 21dBm.
After comparing the limitations caused by not using beamforming in regards to the re-
ceived PDSCH power, where the noise and interference cause a decrease in the values
obtained. Now, we analyze the application user throughput for the broadband service
defined in this scenario which an average of requested throughput of 4000kbps in DL
and 1000kbps in UL. The average DL throughput for broadband without beamforming is
8.40Mbps (obtaining the maximum throughput using only the n257 sites not the n78 site)
and the average DL throughput for broadband with beamforming is 10.70Mbps (obtaining
the maximum throughput when using only the 38 (64QAM 5/6) bearer).
57
5.1.5 Simulation’s results of the deployment in the near future
After performing the deployment for a specific number of users, in this case 400, we
multiply the global factor of the simulations by 2 to see how the network would behave
in a couple of years when more users attend the congress, having around 800 users in this
case, now we will study how the sites behave, the throughput demanded by the services
and the throughput obtained, and the behavior of the network for each connected indoor
user.
At first sight, after multiplying the global factor by two to double the number of users and
traffic, we can confirm from the cells tab from the average simulation results that all the
cells can manage the increase in traffic load, not having any overloaded cell. Regarding
the demanded throughput for DL, the total maximum throughput is 6992.4Mbps and
the obtained peak RLC cumulated throughput (DL) is 4406.54Mbps. Even though the
obtained throughput is lower than the maximum demanded throughput, it is enough to
serve all users in the computation zone, that is, it fulfills the average throughput requested
by the users. The total number of users not connected (rejected) for 877 users in total is
still 0 (0%).
If we analyze the results from the 5G terminals simulation:
Figure 5.28: “Received PDSCH Power (DL) (dBm)” for ‘Activity Status’, ‘Best Server’ and ‘Diversity
Mode (DL)’
In this first figure, we can observe the received PDSCH Power in DL by the activity status:
‘Active DL’ and ‘Active DL+UL’ and the best server and lastly, by diversity mode: ‘SU-
MIMO and MU-MIMO’ and ‘Transmit Diversity’. With these parameters we can check
the best serving cell which is site7 in terms of best received power value with -29.54dBm
and the worst one is site3 with -70.53dBm, as well as, observe, between using MIMO
58
and not using it, there is an average of 20dBm of difference. Nonetheless, all sites have a
measured PDSCH power above -71 dBm and an average of -46.43dBm that translates in a
very good signal quality at receiver. Even though all have good values of received power, if
we observe the good values of site8 (≥ −65dBm) and taking into account site8 is located
between buildings which create a lot of reflections, this can also be used as an example of
the benefits of using beamforming, helping decrease the limitations the buildings would
create in the communication link.
After analyzing the received power, now we will study the best service beam for each
service: video call, voice call and broadband,
Figure 5.29: Best Service Beam counter for Figure 5.30: Best Service Beam counter for
Video Call Voice Call
Figure 5.31: Best Service Beam counter for Broadband
Compared with figures 5.23, 5.24, 5.25 the number of users using the same beam index
has increased, the most clear example is in figure 5.25 where site1 has 31 users using
beam index 8. Taking into account the traffic load in DL in each site the values ob-
tained are correct as site1 has the highest traffic use around 95% and the least used site,
site8, has a 15% of traffic load, thus the usage of beam indexes depends on the traffic load.
As the last metric to analyze from the simulation with a global factor of 2 to demonstrate
future traffic situations, we will study the DL throughput for the Broadband case,
59
Figure 5.32: Histogram of Application User Throughput (DL) (kbps) for Broadband
The average throughput measured for broadband is 11.42Mbps, more than enough for the
requirements that the service asks of 4Mbps as average requested throughput.
5.2 Big venue

After designing and implementing a 5G mmWave with beamforming deployment with
outdoor cells, taking advantage of the update from ATOLL software in this section we
are deploying an indoor scenario with both the base stations and the users inside the
buildings in the UPF Ciutadella’s campus in Barcelona providing broadband, video call
and voice call services to the users.
The map of the Computation zone with the base stations can be found in section 4.1.2,
as well as, the parameters of the base stations and the services.
Therefore, in this section we will analyze the predictions and simulations made during
the realization of this scenario with the intention of demonstrating if for this particular
deployment we need to use all the beam indexes from the beamforming antenna or not,
as well as, the improvements that beamforming brings with respect to the use of MIMO
technology or without using any type of diversity support. More information about the
beamforming antenna and the diversity support modes is found in 4.3 and 4.5.1.
Hereafter, we study the prediction map of beam usage in DL. In the following figure we
can observe a map legend where each colour represents the percentage of utilization of
each beam index. In most of the map, the usage is around 0% and 10%, some beams are
used around 10% and 20% and only site 7 and site 8 in building 10 exceeds the 20%.
60
Figure 5.33: Beam Usage DL (%) prediction
In the next prediction we obtain the best service beam for the indoor deployment,
Figure 5.34: Best Service Beam prediction
The service beam is the PDSCH (or traffic) beam which is formed towards the UE dy-
namically based on its position and traffic. More information about the types of beams is
found in 4.3.1.
Here we can see how each beam has been distributed in the map for each mmWave small
cell in each building. We can also distinguish each coverage area by each transmitter and
how each antenna uses the beams depending if the site only uses the omnidirectional
61
beamforming antenna or the broadcasting antenna, as in the case of building 10. More
information about the types of antennas used is found in 4.2.
More information and figures related with the beam usage in DL can be found in appendix
A, where it can be demonstrated how the beam indexes are used in order to help with
the interference created by neighbouring cells and the buildings’ structure.
5.2.1 All beam indexes active

In this section we have activated all the beam indexes from 0 to 99 where the electrical
tilt ranges from -30º up to 30º to compare it with the results obtained with only using
the beam indexes with electrical tilt of 0º which corresponds to indexes 0 to 19.
The prediction of the beam usage using all beam indexes is,
Figure 5.35: Beam Usage DL with all beam indexes active
In the figure most of the map has more than a 0% usage and only in some areas there is
more than a 10% usage.
In the next prediction we obtain the best service beam for the indoor deployment using
all the beam indexes,
62
Figure 5.36: Best Service Beam with all beam indexes active
Compared with figure 5.34 where only the first 20 indexes where used, there is not much
change between the best beams chosen, the only exception is near site 15 and 16 in building
10 and site 7 in building 20 where indexes higher than 20 are used as it can be appreciated
in figure 5.38 and in figure 5.40. With both these predictions (beam indexes 0-19 and all
beam indexes) we can infer that there is no much difference in this deployment from using
only the first 20 indexes with no electrical tilt or using all beam indexes given that all
the buildings in this campus are lower than 30m. Thus, for this particular 5G mmWave
deployment the smart choice is to only use the first 20 beam indexes.
Figure 5.37: Representation of figure 5.34 Figure 5.38: Representation of figure 5.36
63
Figure 5.39: Histogram of the Figure 5.40: Histogram of the
representation above-named representation above-named
5.2.2 Simulation’s results using only the first 20 beam indexes

We run a simulation using a 3D traffic map using a Coverage by Transmitter prediction
with a receiver sensitivity of -120dBm. The average results are:
Figure 5.41: Average results of indoor 3D traffic simulation
The total users obtained are the following:
Table 17: Total number of connected users for the case of using the first 20 beam indexes
Users 2137
Downlink 29.6
Uplink 30.8
Downlink + Uplink 2027.8
Inactive 48.8
With these users, the total accumulated throughput and the average per user throughput,
both in the uplink and in the downlink is:
64
Table 18: Total accumulated and average per user throughput for the case of using the
first 20 beam indexes
Total Per User

The total number of users not connected (rejected) is 8,4 (0,4%).
Table 19: Mobile’s rejection cause for the case of using the first 20 beam indexes
No Coverage 1.2 users

No Service 7.2 users
With these metrics we can observe the power of densification that beamforming mmWave
deployments offer. The rejection percentage is below 1% which is really good for an area
of 1000m2 and buildings with up to 8 floors filled with pedestrians. The reason for having
rejected users is because in order to not overload site 8 in building 20, we activated the
“Cell Individual Offset (dB)” for the neighboring sites to help reroute the traffic load, as
a consequence on the cell edge of site 8 users connect to the next serving cell even though
site 8 had the best serving cell and the second best cell does not have enough power to
offer the requested service. It is a tradeoff between not overloading the small cells and
having a few users without service. In regards to the no coverage status occurs with the
users that are in higher floors from where the small cell is located and there is almost no
coverage, but still the percentage of the number of no coverage users is very low compared
with the total connected users.
More information about the two rejected status and the Cell Individual Offset (dB) pa-
rameter is in 4.7.1.
If we analyze the results for the 5G Terminals focusing on the results we obtain when we
use the different diversity support modes, i.e. we use only receiver diversity or SU-MIMO
and MU-MIMO. More information about the diversity support modes can be found in
4.5.1,
65
Figure 5.42: Histogram of Received PDSCH Power (DL) (dBm) for Receiver Diversity
Figure 5.43: Histogram of Received PDSCH Power (DL) (dBm) with SU-MIMO and MU-MIMO
Comparing in both graphs the received PDSCH power we can observe that the lower
the received power, the simpler the diversity support mode is used, as explained in 4.5
if the PDSCH C/(I+N)) is less than the SU-MIMO threshold defined in the reception
equipment of the terminal or cell, the user will benefit from the DL diversity and when
the received SU-MIMO criterion or PDSCH C/(I+N)) is higher or equal, the user will
benefit from the SU-MIMO gain in its throughput. That is the reason why for figure
5.42 the highest measured power is -63,36dBm whereas in figure 5.43 the highest value is
-30.11dBm.
66
Figure 5.44: Histogram of Application User Throughput (DL) (kbps) for Receiver Diversity
Figure 5.45: Histogram of Application User Throughput (DL) (kbps) with SU-MIMO and MU-MIMO
After analyzing the received PDSCH power in DL, we analyze the application user
throughput in DL using only receiver diversity or also SU-MIMO and MU-MIMO. The
values are consistent with the fact that the simpler the diversity support mode the lower
the values, in figure 5.44 most of the users have around 3.81 and 6.13Mbps wheres in
67
figure 5.45 most of the users have a throughput of 11.70Mbps. The average being for both
cases: 8.50Mbps and 13.60Mbps, almost doubling the throughput when using SU-MIMO
and MU-MIMO.
We have analyzed the received PDSCH power when using receiver diversity or SU-MIMO
and MU-MIMO. Now we will obtain some insights looking into the relation between the
path loss (dB) and the received PDSCH power (dBm) depending on the bearer for the
broadband service,
Figure 5.46: Received PDSCH Power (dBm) and Path Loss (dB) [*] versus Bearer
[*] Bearer 23 was filtered because it matches only with voice call and video call services.
We can check that the minimum modulation and highest modulation matches with the
Broadband parameters selected in 4.4. The path loss, the blue bars, decrease as the mod-
ulation improves meaning a better channel quality, the same happens with the received
PDSCH power where it is around -100dBm with bearer 20 and improves to less than
-50dBm using bearer 46. This graphs is an average of the measurements but if we filter
by the maximum value, the maximum received power matches with the maximum value
in figure 5.43: -11.75dBm.
68
Figure 5.47: User Throughput versus Bearer
When measuring the User Throughput it is equivalent to measuring the throughput of

the broadband service which is the only active service that uses data. In this chapter, we
will also analyze the deployment when both broadband and video call use data.
We can see a relation between a lower throughput in the bearers which in figure 5.46 have
a worse path loss, for example bearer 29 or 38 or 44. Likewise as the channel improves, so
does the modulation used and therefore the average user throughput. The penalization we
see from the application user throughput versus the other two throughputs is because it
is calculated from the effective RLC throughput and the throughput scaling factor (95%).
Next, we will briefly talk about the UL regarding the received power in the base station
and the user throughput,
69
Figure 5.48: Histogram of Received PUSCH & PUCCH Power (UL) (dBm)
The average UL received power is -76.9dBm, but a high amount of users, 102 users, are
measuring [-65.1, -58.9]dBm.
Figure 5.49: Histogram of Application User Throughput (UL) (kbps)
The average UL application user throughput is 4.47Mbps, most of the users have the
maximum throughput and a few of them are scattered around low values of throughput.
70
5.2.3 No Beamforming
We have analyzed the results of the predictions and simulations run in ATOLL for the
scenario of a 5G NR mmWave with small cells indoor deployment using Beamforming.
Now, we will study the case where there is no Beamforming.
Figure 5.50: Coverage prediction with no

beamforming Figure 5.51: No beamforming (blue area)
compared with using beamforming (black line)
For the case of having no Beamforming there are a lot of areas in figure 5.50 with coverage
lower than -100 dBm (blue), that means a very bad quality, as staring from -90 dBm the
signal can be considered acceptable. In figure 5.51 we observe that there is a larger area
covered with good coverage, still in both cases the best received power is around -60 dBm.
Also, where in some areas there was no coverage, due to using beamforming even with
low values of received power there is coverage. Looking into these figures we can notice
the benefits that beamforming has in the quality and range of the coverage.
Figure 5.52: DL quality prediction with no

Figure 5.53: No beamforming (blue area)
beamforming compared with using beamforming (black line)
Looking into figure 5.52, the DL quality is quite bad, there are very few areas with a good
quality of signal (orange/red) and also there are quite a few areas without signal. We have
seen in the previous figures (5.50, 5.51) that there is a bigger coverage area when using
71
beamforming, also as seen in 5.53 the area with higher quality of signal is bigger when
using beamforming, for the case of the blue area the area decreases quite constantly until
it reaches the 30 dB of PDSCH C/(I+N) whereas with beamforming it remains with a
better quality of signal until reaching the 30 dB. With these figures the benefits of using
beamforming when deploying a mmWave 5G NR network are proven.
5.2.4 Simulations’ results without using Beamforming

Analyzing now the simulation’s results for the no Beamforming case and comparing the
results with the indoor deployment case, we obtain:
Figure 5.54: Average results of no diversity support modes simulation
Table 20: Total number of connected users for the case of not using beamfoming
Users 2667
Downlink 36.1
Uplink 38.6
Inactive 55.1
With these users, the total accumulated throughput and the average throughput per user,
Table 21: Total accumulated and average per user throughput for the case of not using
beamfoming
Total Per User

72
The total number of users not connected (rejected) is 188.1 (6.6%).
Table 22: Mobile’s rejection cause for the case of not using beamfoming

In table 21, when comparing the throughput per user in DL with the value in table 18,
when using Beamforming, we obtain a benefit x4 in the throughput per user. In the UL,
if we compare the throughput per user in table 21 with table 18, it is up to 3 times more.
In regards to the rejected users it changes from 8,4 (0,4%) in table 19 to 188.1 (6.6%)
users in table 22. If we study the metrics for the ”No Service” users versus the connected
users:
Figure 5.55: Metrics for No Service users using no Figure 5.56: Metrics for connected users using no
Beamforming
Beamforming
In figure 5.55 the received power is under -100dBm and compared with figure 5.56 the
path loss is around 30-40dB worse, thus users are rejected because the quality is too low
to obtain any codec mode or coding scheme.
Regarding the no coverage rejection cause, as we explained in 5.2.2, it occurs with the
users that are in higher floors from where the small cell is located and there is almost no
coverage.
If we analyze the results from the 5G Terminals simulation now for the connected users,
we analyze the PDSCH received power and the application user throughput for the Broad-
band user to compare it with the indoor deployment case using beamforming,
73
Figure 5.57: Histogram of Received PDSCH Figure 5.58: Histogram of Received PDSCH
Power (DL) (dBm) with no Beamforming Power (DL) (dBm) with Beamforming
In figure 5.57 the histogram ranges from -100dBm to -19dBm and in figure 5.58 it ranges
from -72dBm to -12dBm, we can asses the benefits of beamforming help in bad channel
conditions given that the deployment is the same the only difference is the power perceived
in the user equipment and in the good channel conditions there is an increase of 7dBm
between the two use cases.
After analyzing the received PDSCH power, now we will obtain some insights looking into
the comparison of the application user throughput,
Figure 5.59: Histogram of Application User

Throughput (DL) (kbps) with no Beamforming Figure 5.60: Histogram of Application User
Throughput (DL) (kbps) with Beamforming
The average throughput is for both cases: 7.70Mbps and 13.60Mbps, almost doubling the
throughput when using Beamforming. Even though the maximum throughput achieved
does not change. For the minimum throughput, in figure 5.60 it is 5.30Mbps with 14 users
and in figure 5.59 it decreases to 1.18Mbps and increases the number of users to 208. The
outlook is that with beamforming the majority of the users have high throughput and
in figure 5.59 the majority of the users have been moved towards the minimum values of
throughput.
Regarding the UL channel a few considerations when comparing the performance with
having Beamforming,
74
Figure 5.61: Histogram of Received PUSCH & PUCCH Power (UL) (dBm) without Beamforming
The average UL received power is -86,49dBm, but a high amount of users, 146 users, are
measuring [-98.66, -93.76]dBm. The average UL received power in figure 5.48 is -76.9dBm,
there is a 10dB difference between both figures.
Figure 5.62: Histogram of Application User Throughput (UL) (kbps) without beamforming
The average UL application user throughput is 3.26Mbps, compared with figure 5.49 with
4.47Mbps, there is no big penalisation in UL when we are not beamforming because it is
designed to help in the DL given that the base station has the antenna’s array with the
75
beam sweeping, also the services that we are using now have almost no requirements in
the UL which translates to very little penalty on the part of the system when removing
the the beamforming.
5.2.5 Simulations’ results for the Video Call service as Data type
In the following section, we will analyze the changes in the use cases when instead of
having the Video Call service as Voice, meaning there is no packet transmission, i.e., no
throughput, we have it as Data type, having traffic requirements both in UL and DL.
We will focus on the simulations because the coverage and quality predictions have not
changed, only the results related with the traffic have now changed given that the service
properties have changed. The new properties defined appear in section 4.4.
Figure 5.63: Average results of video call service as data type simulation
Table 23: Total number of connected users for the case of video call service as data type
Users 2119
Downlink 33.5
Uplink 28.8
Inactive 42.7
76
With these users, the total accumulated throughput and the average throughput per user
(obtained by dividing the total accumulated throughput by the total number of users)
Table 24: Total accumulated throughput and average throughput per user for the case of
video call service as data type
Broadband Video Call

Total Per user Total Per user
Downlink 9673.73 Mbps 11.11Mbps 306.8Mbps 0.272Mbps
Uplink 3909.65Mbps 4.48Mbps 318.16 Mbps 0.282Mbps
The total number of users not connected (rejected) is 12.1 (0.6%).
Table 25: Mobile’s rejection cause for the case of video call service as data type

There is an increase in the percentage of rejected users compared with table 19 where
it was 0.4% and even though the no coverage users stays in 1 user, for the no service
the number of users almost doubled, because as previously mentioned in order to avoid
overloading site 8 and 9 from building 20, the neighbouring cells have the “Cell Individ-
ual Offset” (dB) activated and for a small amount of users who are not able to connect
in exchange we avoid overloading the site. It must be stressed here that the increase in
percentage is still under 1%, so the change from voice type to data type for the video call
does not affect too much.
If we analyze the results from the 5G Terminals simulation now for the connected users,
we compare the application user throughput for Broadband with the indoor deployment
case where the video call used voice. We do not analyze the PDSCH received power in
this section because the change in service properties does not affect it, but we add the
application user throughput for the Video Call user:
77
Figure 5.64: Histogram of Application User Figure 5.65: Histogram of Application User
Throughput (DL) (kbps) with video call as data Throughput (DL) (kbps) with video call as voice
The average for both cases is 11.10Mbps in figure 5.64 and 13.60Mbps in figure 5.65. The
difference in throughput is due to the fact that now the video call service consumes data,
although a much smaller amount, around 300kbps, as we will see next. The outlook is
that in both cases the majority of the users have high throughput.
The novelty now, as we mentioned before, is the application user throughput for the Video
Call service:
Figure 5.66: Histogram of Application User Throughput (DL) (kbps) for Video Call service
Due to the fact that now video call has data instead of broadband, the video call uses
some particular bearers that translates to specific throughput values, so the users are us-
ing bearers the majority of which use a 16QAM modulation. Lastly, most of the users have
78
the maximum allocated throughput, resulting in an average throughput of 274.46kbps.
As previously mentioned, the measured power is not affected by the change of technology
type of the services, when analysing the UL, we will focus again in the application user
throughput of both broadband and video call service. First, we compare the broadband
service with the results from figure 5.68, and then we do a comparison for the video call
between the DL results and the UL results,
Figure 5.67: Histogram of Application User Figure 5.68: Histogram of Application User
Throughput (UL) (kbps) for broadband service Throughput (UL) (kbps) for broadband service
with video call as data with video call as voice
The behaviour in both cases is quite similar, i.e., most of the users are using the maximum
allocated throughput. In figure 5.67 the average is 4.49Mbps and in figure 5.68 the average
is 4.47Mbps. The reason why the values do not change too much is because, as we will
see when analysing the video call service below, most of the users in the video call case
are connected to bearer 29 because the maximum modulation in UL for this service is
16QAM, then the bearers using 64QAM are only used by the broadband service. Also, the
requirements in UL for the broadband service are quite low compared with DL, then when
the traffic in UL for the video call service is added, the unused resources are reallocated
to this service, so at user level they do not perceive any difference. In the next figure we
can see how the broadband users are distributed per bearer (UL),
79
Figure 5.69: Average (blue) and counter (orange) of Application User Throughput (UL) (kbps) per
Bearer for Broadband service
After comparing in UL the broadband service in the case of having the video call service
as data type versus voice type, now we compare the video call service in UL versus DL,
figure 5.66, (given that we did not have any throughput for this service when it had voice
as the technology type),
Figure 5.71: Average (blue) and counter (orange)

Figure 5.70: Histogram of Application User of Application User Throughput (UL) (kbps) per
Throughput (UL) (kbps) for Video Call Bearer for Video Call
Looking at figure 5.71 we observe that all the users, except 1%, are in the highest bearer
allowed, indicating the channel has enough resources. Also, there is a symmetry between
the UL and DL because unlike broadband where the heavy requirements are in the DL, the
video call service has the same requirements for UL and DL, this is the reason why if we
compare both DL and UL the average for this service does not change, in DL is 274.46kbps
per user and in UL is 282.63kbps per user in average. As mentioned previously the small
difference between the two values is due to the demanding requirements for broadband in
DL.
80
6 Conclusions and future development
In this work we use the professional software ATOLL to deploy a 5G mobile communica-
tion network in order to better understand beamforming, how it works, and the advan-
tages of using it over not using it. To do so, we create two scenarios (a small venue and
a big venue) in which we deploy mmWave band small cells and analyze the benefits of
beamforming in both.
As a result, ATOLL was used to design and implement a deployment, which included
service and user modeling, transmitter and cell configuration, predictions that can be used
as a support to setup or an instrument to asses the deployment, and simulations that add
the user traffic to create realistic situations to fully examine the network’s behavior.
In the first scenario, the small venue, the deployment we implemented was in the city
of Barcelona, at the UPF’s Ciutadella campus, and had low density of users inside the
buildings. The results were satisfactory since the deployment’s behavior was above the
demanded average requirements in all the services when using beamforming, as we will
see in the following paragraphs.
Broadband was the most demanding service in DL, and in regards to received power
and average throughput per user, when using beamforming there was a 30dBm increase
in received power versus not using it; and an increase of 1.3 in the average throughput
per user in DL and in UL when using beamforming, obtaining 10.70Mbps in DL and
2Mbps in UL. Beamforming is a technology for gNBs to establish the best propagation
direction to individual users, using narrow antenna beams. Its most important aspect is its
interference reduction to other users, thus, increasing the achievable data rates. Therefore,
with beamforming we can correct the losses caused by the path loss, for example the
reflections with the walls.
On the particular case of increasing the traffic global factor by 2 to do the provisioning
of the deployment in the near future, the results showed, in particular for broadband
(the service with the most requirements in DL), that the average throughput per user
was 7.5Mbps in DL and 2Mbps in UL. Taking into account that the average demanded
throughput per user for broadband was 4Mbps in DL and 1Mbps in UL, this means, that
the requirements in terms of throughput were fulfilled.
The second scenario, the big venue, gave coverage to the same zone, via, this time, a 5G
indoor deployment with a high density of users inside the buildings.
In the high density scenario not using beamforming implies a 6.6% of rejected users,
mostly because the quality is too low to obtain any codec mode or coding scheme. In
average, the throughput per user (DL) is 3.24Mbps, lower than the average requested
throughput of 4Mbps, thus, when beamforming is not used the deployment does not have
enough resources to provide a satisfactory service to all the users in the buildings.
When using beamforming, the majority of the users of the various services have in average
a higher throughout that the demanded average. For the broadband service, the network
provides in DL: 11.11Mbps and in UL: 4.48Mbps. Hence, when using beamforming, in DL
there is four times more throughput and three times more throughput in UL. For the video
81
call service, given that the service consists of a symmetric link, the average throughput for
DL and UL is quite similar: 274kbps and 282kbps; taking into account that the average
demanded throughput is 128kbps it perfectly satisfies the requirements. Additionally the
percentage of rejected users is only 0.6%. To keep it under 1%, as explained in 4.7.1, we
can tune or bias the ranking of potential servers for best serving cell selection in connected
mode.
In view of the situation, in order to improve the deployment developed during the imple-
mentation of this project, it is necessary to highlight possible improvements to be taken
into account. During this project, we have detected some promising future directions
which can be summarized as follows:
• In ATOLL’s new version it will be possible to create beam sets and beam bursts,
instead of only using the given beamforming models and patterns.
• The In-Building feature, right now only floors and outer walls are taken into account.
This feature enables better in-building wireless network design and we could do an
import of floor plans and see the effect of the propagation losses having corridors,
doors and walls.
• Do a survey of different beamforming antennas by different manufacturers or creat-
ing our own (focused on helping with interfering signals or instead the useful signal
from the serving cells). Right now, ATOLL only offers one 3D beamforming antenna
for high bands. Another option would be to create a densification environment in
areas with skyscrappers to have the opportunity to use all the beam indexes of the
3D beamforming antenna ATOLL already offers.
82
References
[1] Mohamed Nadder Hamdy. Beamformers explained. Commscope, WP-114491-
EN,(06/00), 2020.
[2] Shajahan Kutty and Debarati Sen. Beamforming for millimeter wave communica-
tions: An inclusive survey. IEEE communications surveys & tutorials, 18(2):949–973,
2015.
[3] Forsk. Atoll 3.4.1 User Manual for Radio Networks. Forsk, 2 edition, 2021.
[4] 5g - fifth generation of mobile technologies, 2022.
[5] SDxCentral Studios. What is imt-2020?, Mar 2018.
[6] Jordi Perez. Chapter 5 5g new radio (5g nr), 2021.
[7] Erik Dahlman, Stefan Parkvall, and Johan Skold. 5G NR: The next generation
wireless access technology. Academic Press, 2020.
[8] Juan Pedro Tomás. The current state of 5g in spain, Jan 2022.
[9] MINECO. Spain’s 5g national plan 2018 - 2020, Dec 2017.
[10] Juan Pedro Tomás. Spanish government starts process to award additional 5g spec-
trum in 2022, Dec 2021.
[11] Renato Lombardi. Microwave and millimetre- wave for 5g transport, 2018.
[12] GSMA Intelligence. The economics of mmwave 5g, 2021.
[13] Rex Chen. An introduction to the 5g small cell, Aug 2021.
[14] Erik G Larsson, Ove Edfors, Fredrik Tufvesson, and Thomas L Marzetta. Mas-
sive mimo for next generation wireless systems. IEEE communications magazine,
52(2):186–195, 2014.
[15] 5G Americas. Understanding mmwave for 5g networks, 2020.
[16] Anokiwave David Corman. A comprehensive guide to active antennas (or, ”beam-
forming 101”), 2018.
[17] Huawei. 8t8r antenna beamforming technology introduction, 2020.
[18] Forsk. Atoll 5g nr planning software, 2022.
[19] Forsk.com. Aster propagation model.
83
Appendices
A Beam Usage (DL) in the big venue scenario
We will begin analyzing first the beam usage of the first two sites with the highest traffic
load in DL and the two with the lowest traffic load in DL:
Figure A.1: Beam Usage DL site 9 (traffic load 99,9%)
Because all the buildings in the Ciutadella campus are under 30 meters, there is no need
to use a value different than 0º for the electrical tilt. If we were in an area with skyscrapers
then we would have to use the beam indexes from 20 to 99 which covers an electrical tilt
of -30º, -15º, º15 and 30º.
In the case of Site 9 (with a traffic load (DL) of 99,9%), located in building 10, index 0
has a beam usage of 20% and beam 7, 8, 10, 12 and 16 are not used, it is interesting that
beam 16 is not used when both beam 15 and 17 has a 10% of beam usage.
In the case of Site 8 (with a traffic load (DL) of 96,05%), located in building 10, index 0
has only a beam usage of 7% and beam 4, 8, 9, 10, 11 and 15 are not used, compared with
the beam usage of site 9, which is a neighbour of site 8, beam 16 now has a beam usage
of 11% and beam 15 is not used and beam 17 has only a 3% usage. This is an example of
how beamforming works trying to avoid interference between neighboring base stations
and also taking into account the building infrastructure and the surroundings. For this
site the most used beam is beam 3 with a 21.24%.
84
Figure A.2: Beam Usage DL site 8 (traffic load 96,05%)
In the case of Site 10 (with a traffic load (DL) of 25%), located in building 10, index 6 has
a beam usage of 30.27% and beam 10, 12-14 and 16-19 are not used, it is worth noticing
that beam 6 is the most used in site 10 whereas in site 9 it is only used a 0.99% (both
being neighbouring sites).
Figure A.3: Beam Usage DL site 10 (traffic load 25%)
85
In the case of Site 17 (with a traffic load (DL) of 41%), located in building 40, index 0
has only a beam usage of 1.12% and beam 10 and 11 are not used, the beams with the
highest use are beam 19 with a 12.2% usage and beam 13 with a 12.03% beam usage. All
beams have less than a 20% usage.
Figure A.4: Beam Usage DL site 17 (traffic load 41%)
86

Master Thesis 5G NR Nuria V

Uploaded by

Copyright:

Available Formats

Master Thesis 5G NR Nuria V

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Master Thesis 5G NR Nuria V

Uploaded by

Copyright:

Available Formats

5G NR small cell deployment using

Advisor: Anna Umbert

2 State of the art 16

6 Conclusions and future development 81

A Beam Usage (DL) in the big venue scenario 84

Revision Date Purpose

DOCUMENT DISTRIBUTION LIST

Anna Umbert anna.umbert@upc.edu

Written by: Reviewed and approved by:

1.1 Statement of purpose

1.2 Methods and procedures

1.3 Work plan

1.3.2 Deviations from the initial plan and incidences

1.4 Outline of the report

International Mobile Telecommunications-2020 (IMT-2020 Standard) are the require-

Figure 2.4: Standalone operation

Figure 2.5: Logical, transport and physical channels

Subcarrier spacing Symbol duration

5G NR can operate in multiple bands specified in two Frequency Ranges (FR):

2.1.3 Cell search

Figure 2.6: Synchronization Signal Block

2.2 Allocation of spectrum bands to 5G

2.3 Milimiter Wave (mmW)

Figure 2.9: Channel BW representation

2.4 Densification and small cells

2.5 Beamforming and Multiple Input Multiple Output (MIMO)

Figure 2.11: Analog beamforming [6]

4.1 Defined scenarios

4.1.1 Small venue scenario

Table 2: Business user properties

Table 3: Standard user properties

Table 4: Small venue indoor environment

Table 5: Small venue outdoor environment

Table 6: 5G frequencies used in the small venue scenario

Reference Frequency Carrier centre

4.1.2 Big venue scenario

Table 8: Standard user properties

Table 9: Big venue indoor environment

Reference Frequency Carrier centre

4.2 ATOLL Antennas

Figure 4.5: DL coverage by a 70 degrees antenna (SA Broadcast)

4.3 ATOLL 3D Beamforming

4.3.1 Types of beams in ATOLL

4.3.2 Difference between Broadcast and Service Beams

4.4 Modelling Services

4.5 ATOLL Predictions

Figure 4.13: Diversity support in DL and UL in ATOLL’s 5G NR Cells table

4.5.2 Aster and Aster mmWave propagation models

4.6 3D traffic map

4.7 ATOLL Simulations

5.1 Small venue

5.1.1 No MIMO nor Beamforming

Figure 5.4: DL quality with no SU-MIMO

Figure 5.6: DL coverage with no

Figure 5.8: DL quality with no

Figure 5.10: DL coverage with no support Figure 5.11: DL coverage legend

5.1.2 All diversity support modes active

5.1.3 Simulations’ results using beamforming

No Coverage 13,2 users

No Coverage 0,13 users