Javier Sancho Vázquez

DESIGN AND TESTING OF A

TRANSMITTER-CHANNEL-RECEIVER
MODEL USING MATLAB 5G TOOLSET.

Science and Engineering

Bachelor’s Thesis
May 2019
 i

ABSTRACT

Javier Sancho Vázquez: Design and Testing of a Transmitter-Channel-Receiver Model

Using Matlab 5G Toolset
Bachelor’s Thesis
Tampere University
Science and Engineering
May 2019

Make possible new incoming services and enhance the current ones in order to improve our
lives, is the common objective of the engineers. In this way, for improving our communications,
we have passed through four generations, each of them with its corresponding enhancements.
However, the fourth generation has some limitations that must be solved to keep evolving our
communications. Hence, the fifth generation (5G) is forecast to be launched in 2020. This thesis
is based on 5G and it has approached three different objectives by using Matlab 5G Toolset. First,
the study of 5G waveforms has been carried out. These waveforms are W-OFDM, F-OFDM and
CP-OFDM and their spectrums have been compared for different bandwidths and subcarrier
spacings. Then, a basic transmitter-channel-receiver chain model has been designed and tested
in terms of bit error rate. Finally, a new configuration for improving the reliability of 5G communi-
cations, based on 3GPP, has been designed and proved. This configuration includes some vari-
ations related with the spatial diversity, the HARQ protocol and the code rate.

The analysis of the results has shown that F-OFDM and W-OFDM achieve better spectral
efficiency than CP-OFDM. However, W-OFDM presents more out of band emissions. Thus, F-
OFDM with a small value of subcarrier spacing presents better performance than CP-OFDM and
W-OFDM in terms of spectral efficiency and out of band emissions. In addition, during the testing
of the designed model, CP-OFDM has demonstrated better bit error rate than F-OFDM and W-
OFDM. Nevertheless, F-OFDM and CP-OFDM present very similar results. In addition, the new
5G configuration proposed was applied into the transmitter-channel-receiver model commented
before for CP-OFDM. The new result was compared with the previous one in terms of bit error
rate. Finally, this comparation shows that better bit error rate values are achieved with this new
configuration. Thus, the reliability of the communications is improved.

Keywords: 5G, communications, reliability

The originality of this thesis has been checked using the Turnitin OriginalityCheck service.
PREFACE
This Bachelor’s Thesis has been written during my stay in Tampere as an exchange
student at Tampere University. It represents the final work of my bachelor’s degree in
Telecommunications Systems Engineering at the Polytechnic University of Madrid.

These years have been a challenging journey, where I had to face several difficulties.
Thus, I would like to thank many people for their encouragement. I am so grateful for the
advices, guidance, patience, help and feedback provided during the development of this
thesis by my supervisors Elena-Simona Lohan and Jukka Talvitie. My gratefulness is
also to all the people I have met during my student period, to my family and to my friends,
because they have always supported me.

Tampere, 13 May 2019

Javier Sancho Vázquez

CONTENTS
1. INTRODUCTION .................................................................................................. 1
1.1 Background.......................................................................................... 1
1.2 Thesis objectives ................................................................................. 2
1.3 Thesis structure ................................................................................... 2
2. THEORETICAL BACKGROUND........................................................................... 3
2.1 Data rate .............................................................................................. 3
2.2 Massive connectivity ............................................................................ 3
2.3 Ultra-reliable and low latency communications ..................................... 3
2.4 Frequency ranges ................................................................................ 3
2.5 Frame structure.................................................................................... 4
2.6 Channel coding .................................................................................... 4
2.7 Channel model ..................................................................................... 5
2.8 5G waveforms ...................................................................................... 5
2.8.1 CP-OFDM ..................................................................................... 5
2.8.2 W-OFDM ...................................................................................... 6
2.8.3 F-OFDM........................................................................................ 6
3. RESEARCH METHODOLOGY AND IMPLEMENTATION .................................... 7
4. ANALYSIS AND RESULTS................................................................................. 10
4.1 Waveform analysis ............................................................................. 10
4.2 New 5G waveform ............................................................................. 13
4.3 5G TX-Channel-RX model with AWGN + CDL/TDL ........................... 14
4.4 5G proposal for more reliable communications .................................. 18
4.5 Limitations of 5G Library .................................................................... 21
5. CONCLUSIONS.................................................................................................. 22
REFERENCES....................................................................................................... 23
LIST OF FIGURES
Figure 3.1. TX-Channel-RX model ........................................................................... 7
Figure 4.1.1. Spectrums with different SCS .............................................................. 11
Figure 4.1.2. Spectrums with same SCS................................................................... 12
Figure 4.1.3. Spectrums of same waveforms with different NRB and SCS ................. 12
Figure 4.3.1. BER with AWGN + CDL of CP-OFDM.................................................. 16
Figure 4.3.2. BER with AWGN + CDL of F-OFDM .................................................... 17
Figure 4.3.3. BER with AWGN + CDL of W-OFDM ................................................... 17
Figure 4.3.4. BER comparation for 16QAM ............................................................... 18
Figure 4.4.1. Tradeoff between reliability and latency ............................................... 19
Figure 4.4.2. BER comparation ................................................................................. 21
LIST OF TABLES
Table 2.4. Frequency ranges .................................................................................. 4
Table 2.5. Proposed frame structure in 5G ............................................................. 4
Table 4.1.1. Transmission bandwidth configuration NRB for FR1 .............................. 10
Table 4.1.2. Transmission bandwidth configuration NRB for FR2 .............................. 10
Table 4.3. Bits per each subframe ........................................................................ 10
Table 4.4. Different configurations ........................................................................ 20
LIST OF SYMBOLS AND ABBREVIATIONS

1G First Generation
2G Second Generation
3G Third Generation
4G Fourth Generation
5G Fifth Generation
AMPS Advanced Mobile Phone System
AWGN Additive White Gaussian Noise
BER Bit Error Rate
CDMA Code Division Multiple Access
CDL Clustered Delay Line
CP Cyclic Prefix
DL-SCH Downlink Shared Channel
FDMA Frequency Division Multiple Access
FR Frequency Range
GP Guard Period
IFFT Inverse Fast Fourier Transform
ICI Inter Carrier Interference
IoT Internet of Things
ISI Inter symbol Interference
LDPC Low Density Parity Check
LTE Long Term Evolution
MIMO Multiple Input Multiple Output
OFDM Orthogonal Frequency Division Multiplexing
OOBE Out Of Band Emissions
PAPR Peak to Average Power Ratio
PDSCH Physical Downlink Shared Channel
RX Receiver
SCS Subcarrier Spacing
SNR Signal-to-Noise Ratios
TDL Tapped Delay Line
TDMA Time Division Multiple Access
TX Transmitter
UFMC Universal Filtered Multi-Carrier
URLLC Ultra-Reliable Low-Latency Communications

NRB Number of resource blocks

𝐿𝐶𝑃 Length of the cyclic prefix
𝐿𝐸𝑥𝑡 Windowing extension
𝑔[𝑛] Windowing function
 1

1. INTRODUCTION

This chapter is a first approximation to the content of the thesis. This introduction is
divided into three sub-chapters and it informs about the background and the three differ-
ent objectives of the thesis. Finally, the thesis structure is presented.

1.1 Background

In order to keep improving our cellular networks and systems, we have passed
through many generations. Each of them has its corresponding enhancements for mak-
ing possible new services and for meeting new requirements. This transition started forty
years ago and since then, cellular communications have experienced several and dra-
matic upgrades during five generations.

The first generation (1G) was born in the 80’s and it was characterized by analogue
voice transmissions. It used frequency division multiple access (FDMA), frequency mod-
ulation (FM), and its frequency band was around 800 and 900 MHz. In addition, it was
based on advance mobile phone service (AMPS) and the data rate was 2 kbps. However,
the second generation (2G) provided digital signals. Due the fact that the number of
users had increased, new multiple access techniques were used, apart form FDMA: time
division multiple access (TDMA) and code division multiple access (CDMA). The princi-
pal objective of the third generation (3G) was to improve the data rate to 2 Mbps. After
that, the fourth generation (4G) arrived. 4G provides all-IP support, low latency (100 ms)
and data rate from 100 Mbps to 1 Gbps depending on the mobility of the device. Finally,
the fifth generation (5G) is coming, and it is forecast to be launched the following year,
in 2020 [15, 9, 5].

Nowadays, each cellular subscription generates 5 GB of traffic per month on average,

and there are about 8 billion of them worldwide [12]. However, there will be an increase
of data traffic and devices connected to internet in order to meet the incoming services,
such as internet of things (IoT), autonomous cars, etc. Moreover, 4G has some limita-
tions and won’t be able to support all those requirements. Even though 4G has high
spectral efficiency, it is not good enough to cover the increase in the number of con-
nected devices [7]. Thus, our current networks need several improvements.
 2

1.2 Thesis objectives

During the following sections of the thesis, three different objectives will be discussed.
These objectives are:

1) A previous study of the waveforms which 5G Library provides and the investiga-
tion of a new one.
2) Design and testing of a basic transmitter-channel-receiver chain model using
Matlab 5G toolset (“5G library functions for LTE system toolbox”)
3) Proposal of a 5G configuration which increases the reliability of the communica-
tions.

Regarding the first objective, the waveforms which have been tested are three vari-
ations of orthogonal frequency division multiplexing (OFDM). These waveforms are ex-
plained during the next chapter. The analysis includes how they can be modeled chang-
ing some of their parameters (bandwidth and subcarrier spacing (SCS)). In addition, an
investigation of the possibility of a new 5G waveform based on 3GPP specifications will
be carried out. Moreover, according to the TX-Channel-RX model, its applicability has
been proved for 5G (ultra-reliable) communications in terms of the bit error rate (BER).
Finally, during the last objective, the main parameters which can modify the reliability of
the communications have been changed in order to improve it.

1.3 Thesis structure

To develop all those purposes in a clear and optimal way, this thesis has been divided
into four chapters. Chapter two describes all the basic theoretical content needed for
approaching the objectives. Chapter three explains the research methodology and the
implementation of the objectives. Chapter four shows the achieved results and the anal-
ysis and chapter five collects all the conclusions of this thesis.
 3

2. THEORETICAL BACKGROUND

According to current forecast, 5G systems will appear on the market starting with
2020. Such systems will represent a big step forward in order to improve our communi-
cations and make possible all the incoming services. There are a huge number of
changes regarding the last generation and a lot of new enhancements.

2.1 Data rate

One of the most important enhancements is the data rate that can be achieved. 5G
is expected to achieve above 10 Gbps of peak data rate and to reach user data rates of
1 Gbps. Moreover, 5G will allow to offer a fairly good quality of service to those devices
moving at speeds above 500 km/h [13].

2.2 Massive connectivity

On the other hand, 5G will also support massive connectivity.

Hence, this new generation will be able to handle the huge increase in data traffic and
internet-connected devices expected in the incoming years. As a result, all the needed
devices for IoT, self-driving, etc., such as sensors or meters can be managed. Less en-
ergy consumption and better spectrum efficiency than 4G systems are others important
aspects that 5G will enable [7].

2.3 Ultra-reliable and low latency communications

One of the main purposes of future 5G transmissions is to achieve ultra-reliable and

high available communications. This means to offer connectivity with incredibly low error
rate even in dramatic conditions or scenarios. 5G systems are expected to guarantee
availability and reliability rates of 99.99999%. Another important new improvement re-
lated with the speed of the future communications is the low latency, which will be ≤ 1
ms. [7, 13].

2.4 Frequency ranges

Regarding to the 5G future spectrum, it will cover not only the range below 6 GHz,
which includes the LTE frequency range as well, but also the frequencies above 24 GHz.
The purpose of using this new range is to increment the data bandwidth available over
 4

reduced and crowded areas [11]. 3GPP establishes in [1] the following frequency ranges
(FR):

Table 2.4. Frequency ranges

Frequency range Frequency range (MHz)

designation
FR1 450 – 6000
FR2 24250 – 52600

2.5 Frame structure

The frame structure has also experimented changes compared with 4G. In 5G, down-
link and uplink transmissions use frames of 10 ms. Each of them consists of 10 sub-
frames of 1 ms and these subframes can contain a variable number of slots depending
on the different existing numerologies. Table 2.5. shows these new possible configura-
tions [2].

Table 2.5. Proposed frame structure in 5G

Numerology Subcarrier Symbols Slots per Slots per

spacing (kHz) per slot frame subframe
0 15 14 10 1
1 30 14 20 2
2 60 14 40 4
3 120 14 80 8
4 240 14 160 16

Thus, in contrast to the single-numerology utilization in 4G, 5G waveforms are very

flexible, allowing lower latencies and higher reliability [9]. 15, 30, and 60 kHz SCS are
used for lower frequencies, and 60, 120, and 240 kHz SCS are used for higher frequen-
cies [7].

2.6 Channel coding

According to the channel coding, there are some different candidates for 5G commu-
nications. One of them is turbo coding, which is also used in 4G. In addition, there are
two more candidates which could be used during the new generation. Those channel-
coding schemes are Polar codes and low-density parity check (LDPC). For larger blocks
of bits to be decoded, there is not difference in performance between these codes. In
contrast, for smaller blocks, Polar codes are better than LDPC and turbo code. They can
 5

get lower error rate and high reliability applications. Nevertheless, the implementation
experience of LDPC and Turbo code are less limited than polar code. Turbo code have
a great flexibility respecting the code rate and block length. On the contrary, LDPC code
needs to be designed for a specific code rate and block length. In addition, LDPC and
Polar code are less complex to implement than turbo code regarding the decoder [7].

2.7 Channel model

A necessary aspect when one wants to design and test a wireless communication
system is the channel model. There are currently two channel models proposed in 5G
for link level evaluations: Tapped Delay Line (TDL) and Clustered Delay Line (CDL) [3].
Moreover, each of them has different sub-models in order to represent different channel
profiles: TDL-A, TDL-B, TDL-C, CDL-A, CDL-B and CDL-C for non-line of sight (NLOS)
model and TDL-D, TDL-E, CDL-D and CDL-E for line of sight (LOS) model. The param-
eters of each of them can be found in [3].

2.8 5G waveforms

Another important aspect to take into account for the development of this thesis is the
type of the waveforms which will be used in 5G. There are several candidates, but in this
project just 3 of them will be employed: cyclic prefix OFDM (CP-OFDM), windowed
OFDM (W-OFDM) and filtered OFDM (F-OFDM), described in what follows.

2.8.1 CP-OFDM

Baseband OFDM can be expressed as:

𝑛
𝑗2п𝑘
𝑂𝐹𝐷𝑀 [𝑘] = ∑𝑁+1
𝑛=0 𝑑𝑛 𝑒 𝑁 , (2.8.1)

where N is the number of subcarriers and 𝑑𝑛 is the symbol for each subcarrier 𝑛. The
next step is to add the CP, which is a copy of the last part of the inverse fast Fourier
transform (IFFT) sequence. It is placed at the beginning as a guard interval and its length
is calculated depending on the maximum delay of the channel to reduce the conse-
quence of inter symbol interference (ISI). Some of the best problems of CP-OFDM is
high peak to average power ratio (PAPR) and out of band emissions (OOBE). However,
using filtering OOBE can be reduced [4, 6]. This waveform is also used in 4G.
 6

2.8.2 W-OFDM

W-OFDM signal on baseband can be described as follows:

𝑛
𝑁+1 𝑗2п𝑘
𝑊 − 𝑂𝐹𝐷𝑀 [𝑘] = ∑+∞
𝑚=−∞ ∑𝑛=0 𝑑𝑚,𝑛 𝑔[𝑘 − 𝑚(𝑁 + 𝐿𝐶𝑃 + 𝐿𝐸𝑥𝑡 )]𝑒 𝑁, (2.8.2)

where 𝑛 represents the subcarrier, 𝑚 the symbol and 𝑑𝑚,𝑛 the data symbol. In addition,
𝐿𝐸𝑥𝑡 is the windowing extension, 𝐿𝐶𝑃 is the CP size and 𝑔[𝑛] represents the windowing
function. This operation is used on both sides of the symbol. Since outer sub-carriers
have more impact on OOBE than inner sub-carriers, edge windowing is used in order to
avoid extra windowing and poor spectral efficiency. This method uses the CP duration
of the channel to create windows which maintain spectral efficiency [4].

2.8.3 F-OFDM

Baseband F-OFDM is represented as:

𝑛−𝑙−𝑚𝐿𝐶𝑃
𝑀−1 𝐿𝑏 −1 𝑁+1 𝑏 𝑗2п𝑘
𝐹 − 𝑂𝐹𝐷𝑀 [𝑘] = ∑𝐵−1
𝑏=0 ∑𝑚=0 ∑𝑙=0 ∑𝑛=0 𝑑𝑚.𝑛 𝑔𝑏 [𝑙]𝑒 𝑁 , (2.8.3)

𝑏
Where 𝑑𝑚,𝑛 denotes the data corresponding to block 𝑏, subcarrier 𝑛 and symbol 𝑚. 𝐿𝐶𝑃
means the CP length and 𝑔𝑏 [𝑙] represents the frequency equivalent windowing function
of block 𝑏. This waveform allows to reduce OOBE since, after the modulator, it uses a
transmit filter cascaded. F-OFDM also requires CP, but if CP is longer than the channel
impulse response, it reduces inter symbol interference (ISI) and inter carrier interference
(ICI). However, F-OFDM systems are more complex than CP-OFDM. [4, 6]

In addition, these waveforms are supported by the Matlab ”5G Waveform Spectral
Analysis” library, which has been used during the first objective. Also, they are used
during the second and third objectives as the transmitted waveforms through the chain
model.
 7

3. RESEARCH METHODOLOGY AND IMPLEMEN-

TATION

As a methodology to develop the thesis, all the objectives have been covered started
from the first one, to the last one, incrementally and in ascending order. The first study
has been based on the example called “5G Waveform Spectral Analysis” from “5G library
for LTE system toolbox”. However, some modifications have been added to the code in
order to perform the analysis in more depth. Then, after checking the 3GPP specifica-
tions related with 5G waveforms, an explanation about the possibility of creating a new
one has been given.

For the design of the chain model, the transmitter (TX) has been implemented in first
instance. When the TX part was totally designed, the channel model was added, and
then, the receiver part (RX). It has been done incrementally to facilitate the debugging of
the code. This model is based also on “5G library for LTE system toolbox”, as was com-
mented during the introduction of the thesis.

At the end, regarding the second objective of the project, this is the block diagram:

TX Channel

X-OFDM Channel
DL-SCH PDSCH AWGN
modulation model

RX

DL-SCH PDSCH Channel X-OFDM

Synchronization
decoding decoding estimation demodulation

Figure 3.1. TX-Channel-RX model

As you can see in figure 3.1, this model is divided into three parts. TX part is com-
posed by three blocks:
 8

• Downlink shared channel (DL-SCH) generation

• Physical downlink shared channel (PDSCH) generation and mapping
• Waveform modulation.

For implementing this modulation, h5gOFDMModulate function has been used. After
that, the generated signal is sent through a noisy propagation channel, which can be
CDL or TDL. AWGN is added to the generated signal, as we can see in the next block
of the chain. Regarding the propagation model, nr5gCDLChannel and nr5gTDLChannel
functions have been used respectively for implementing it. They are available in Matlab
5G Toolset.

On the other hand, the RX is formed by 5 blocks:

• Synchronization
• waveform demodulation
• Channel estimation
• PDSCH decoding
• DL-SCH decoding.

According to the synchronization, perfect synchronization is assumed. Furthermore,

for implementing the demodulator’s block, a function of Matlab is used. It is called,
h5gOFDMDemodulate. In addition, the channel estimation block must be added. Perfect
channel estimation will be assumed. The perfect channel estimation knowledge will be
applied to equalize the received signal through h5gPerfectChannelEstimate function,
eliminating gains and the ISI introduced by the propagation channel. Later, during the
simulation, the BER is calculated for various Signal-to-Noise ratios (SNR) in order to
prove the model. Then, when the simulations were finished, a figure with the results is
showed for different modulations. All these results have been discussed and explained
in the next chapter.

Finally, as commented in the first chapter, the last objective is to propose a new 5G
configuration (TX-Channel-RX) which will increase the reliability of 5G communications.
Once again, it will be based on the “5G library for LTE Systems”. Also, it is necessary to
check 3GPP specifications. First, I have checked all the properties and parameters that
influence in the reliability and then, all of them have been applied to the previous design
in order to see how the model is affected. At the end of this objective, a result of this
more reliable configuration will be shown. Moreover, it will be compared with a similar
 9

result of the previous configuration in order to see the differences between them. They
will be compared in terms of BER during the next chapter.
 10

4. ANALYSIS AND RESULTS

This chapter discusses about the analysis and results of the three different objectives
described in the previous chapters. However, the objective one has been divided into
two parts. First, the analysis of the waveforms will be approached. Then, the investigation
of a new one, based on 3GPP, will be discussed. Hence, this chapter is divided into four
sub-chapters.

4.1 Waveform analysis

During this analysis, various scenarios have been tested in order to know the
behaviour of the 3 different waveforms commented above: CP-OFDM, W-OFDM and F-
OFDM. The way that their behaviour has been tested is by variating the following
parameters:

• Bandwidth, measured in number of resource blocks (NRB)

• SCS

The transmission bandwidth configuration NRB for each base station (BS) channel
bandwidth and SCS is specified in tables 4.1.1 and 4.1.2 [1], where the gray areas are
the ones that have been analyzed.

Table 4.1.1. Transmission bandwidth configuration NRB for FR1

SCS 5 10 15 20 25 30 40 50 60 70 80 90 100
(kHz) MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz MHz
NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB NRB
15 25 52 79 106 133 160 216 270
30 11 24 38 51 65 78 106 133 162 189 217 245 273
60 11 18 24 31 38 51 65 79 93 107 121 135

Table 4.1.2. Transmission bandwidth configuration NRB for FR2

SCS 50 MHz 100 MHz 200 MHz 400 MHz

(kHz) NRB NRB NRB NRB
60 66 132 264
120 32 66 132 264

For evaluating these different scenarios, different cases were considered. The results
are shown in figures 4.1.1, 4.1.2 and 4.1.3, which represents each case. On the left
 11

column, picture a) of the figures 4.1.1 and 4.1.2, represent the double side spectrum for
20 MHz, while on the left column picture b) of figure 4.1.1 shows the spectrum for 50
MHz and picture b) of figure 4.1.2 represents the spectrum for 40 MHz. However, it is
important to highlight that “5G Waveform Spectral Analysis” library uses CP-OFDM as
an LTE waveform and this means that there is only one value for its SCS, 15 kHz. In
addition, LTE bandwidth has an occupancy limit of 90% and that is the reason why CP-
OFDM uses 100 resource blocks (RB) instead of 106 RB as it is specified in table 4.4.1.
for 20 MHz.

The spectrum that can be obtained selecting the properly NRB, as it is indicated in
table 4.1.1. for different SCS for achieving 20 and 50 MHz is presented in figure 4.1.1.
The occupacy limit of 90% of CP-OFDM has been taken into account.

a) 20 MHz b) 50 MHz

Figure 4.1.1. Spectrums with different SCS.

Figure 4.1.1 shows how F-OFDM and W-OFDM achieves bigger bandwidth than CP-
OFDM, proving that these two 5G waveforms increase the spectral efficiency beyond the
90%. In addition, CP-OFDM and F-OFDM exhibit a more rapid drop-off in power outside
of the useful band. Thus, W-OFDM has more OOBE.

In figure 4.1.2 it is possible to appreciate the spectrums of the three waveforms com-
mented before, but for the same SCS. In this case, the bandwidths are 20 MHz and 40
MHz. The occupacy limit of 90% of CP-OFDM has been taken into account again.
 12

a) 20 MHz b) 40 MHz

Figure 4.1.2. Spectrums with same SCS.

In figure 4.1.2 it is possible to appreciate how CP-OFDM keeps showing less band-
width than F-OFDM and W-OFDM for the same SCS. In addition, F-OFDM reduces bet-
ter the amount of OOBE. So that, for the same SCS, W-OFDM and F-OFDM perfor-
mance is better than CP-OFDM in relation to the bandwidth for the same SCS. However,
W-OFDM keeps showing more OOBE.

On the other hand, figure 4.1.3 compares the spectrum of the same waveform for 200
MHz, different SCS and diffrent NRB. F-OFDM and W-OFDM are presented. However,
CP-OFDM cannot be included in this case, since achiving 200 MHz of bandwidth with 15
kHz of SCS is not specified in table 4.1.2.

a) 200 MHz and W-OFDM b) 200 MHz and F-OFDM

Figure 4.1.3. Spectrums of the same waveforms with different NRB and SCS

The result of figure 4.1.3 shows that less OOBE for the same bandwidth can be
 13

achieved for a specific waveform if the option with less SCS and more NRB is chosen.
Moreover, F-OFDM presents less OOBE than W-OFDM.

To sum up and based on the results shown in figures 4.1.1, 4.1.2 and 4.1.3, it is
possible to conclude that regarding the waveforms, F-OFDM shows better performance
than W-OFDM and CP-OFDM. This waveform also presents a more rapid drop-off in
power outside of the useful band than CP-OFDM and W-OFDM. This means that less
OOBE is produced with F-OFDM. On the other hand, W-OFDM can achieve more band-
width for the same SCS than CP-OFDM, but CP-OFDM produce less OOBE. Hence, F-
OFDM presents the best performance overall. Moreover, according to the SCS, after
analysing the figure 4.1.3, it is possible to conclude that if less SCS is used, less OOBE
is achieved. So that, for the analysed cases, F-OFDM with a small SCS value is the best
option.

4.2 New 5G waveform

In order to design waveforms that fulfil all the 5G requirements it is necessary to follow
some waveform design principles. Among them, it is important to highlight flexible nu-
merology, time and frequency localization, MIMO and high order modulation support, low
OOBE and low PAPR. All the waveforms supported by 5G library and that have been
commented above, meet 5G requirements, but there are more candidates which could
be implemented. Universal Filtered Multi-Carrier (UFMC) is one of them, as it is included
by 3GPP TSG-RAN WG1 #84b as we can observe in [18].

UFMC is composed by a group of sub-bands that contains different subcarriers. An

IFFT is computed for each sub-band and then, all of them are filtered. The length of this
transmission filter is usually set to the same size of the guard interval (GI). Finally, these
sub-bands are added to get the final signal to transmit. Also, different filters can be ap-
plied to different sub-bands [20].

This waveform is a variant of the traditional OFDM, but it presents some differences.
It uses a GI instead of CP, as mentioned above. This GI is filled with zeros and added
between the IFFT symbols to avoid ISI [18]. Moreover, it has some advantages, such as,
lower OOBE than CP-OFDM and the possibility to use different numerology or signal
format to each sub-band. However, it also has some important drawbacks as, for exam-
ple, the complexity that it requires. This complexity increases because of several facts.
 14

The need to use zero padding and double-sized fast Fourier transform (FFT) at the re-
ceiver are two of them [20]. In addition, CP-OFDM and W-OFDM have lower PAPR than
UFMC. Nonetheless, F-OFDM has higher PAPR than UFMC [6].

4.3 5G TX-Channel-RX model with AWGN + CDL/TDL

During this sub-chapter the second objective of the thesis is carried out. This objective
consists on the design and testing of the TX-Channel-RX chain model with AWGN and
propagation channel model. In addition, this channel will be estimated. First, the most
important parameters will be presented for giving an approximation about the behaviour
of the model. Then, the achieved results will be shown. These results consist on three
BER simulations of the different waveforms supported by 5G library:

• CP-OFDM
• W-OFDM
• F-OFDM.

The following part of Matlab’s code shows the main parameters of the model:

sp = []; % Simulation parameters initialization (sp)

2 sp.NDLRB = 106; % NRB
sp.CyclicPrefix = 'Normal';
4 sp.SubcarrierSpacing = 15; % SCS in kHz
sp.WaveformType = 'F-OFDM';
6
sp.Alpha = 0.0115; % Roll off for W-OFDM
8 sp.FilterLength = 424; % Length of the filters for F-OFDM
sp.ToneOffset = 1.9; % Tone offset for F-OFDM
10 sp.CellRefP = 4; % Number of transmitting antennas
%Creation of PDSCH substructure
12 sp.PDSCH.TxScheme = 'TxDiversity';
sp.PDSCH.CodingType = 'LDPC';
14 sp.PDSCH.PRBSet = (0:sp.NDLRB-1)'; % Allocation of PDSCH
sp.PDSCH.TargetCodeRate = 1/2; % Code rate
16 sp.PDSCH.CSI = 'On'; % LLR scaling after demodulation
sp.PDSCH.NLayers = 2;
18
ncw = 1; % Active codewords
20 nRxAntennas = 2; % Number of receiving antennas
nTxAntennas = sp.CellRefP; % Change of variable

Program 1. Main parameters of the simulation

It must be remarked that the modulation type is not specified in program 1. This is
because the modulation scheme is defined later, during a loop, in order to not have to
 15

set all other parameters again for each different modulation, since three modulations
have been used. Nonetheless, the waveform type is specified in program 1 as F-OFDM,
but the reason is that this parameter has been changed manually to CP-OFDM and F-
OFDM later for obtaining the three different results that are shown in figures 4.3.1, 4.3.2
and 4.3.3. Also, we can see that the NRB, which in the code is represented by the variable
NDLRB, is 106 and the SCS is 15 kHz. This configuration, as we can see in the table
4.1.1, corresponds to a signal’s bandwidth of 20 MHz. In this way, is only necessary to
define the new modulation and set the rest of the necessary parameters such as, the
gNodeB and the transport block size.

GNB = lteRMCDL(sp,ncw); % Generate a default gNodeB prameters

2 % not specified in sp
pdsch = GNB.PDSCH; % Separate the PDSCH parameters from gNodeb
4 TBS = pdsch.TrBlkSizes; % Transport block size
waveformInfo = h5gOFDMInfo(GNB); % Provides dimensional
6 % information related to the OFDM modulation schemes
% implemented by h5gOFDMModulate

Program 2. Setting parameters

The next step is to add the AWGN to the transmitted signal. Once the chain model
with AWGN is designed, the next step is to add the propagation channel models and the
channel estimation blocks. In order to implement these new blocks, I have used the func-
tions commented in the previous chapter. For the CDL channel, the nr5gCDLChannel
function was used. It returns a CDL channel object. On the other hand, the nr5gTDL-
Channel function is used for implementing the TDL channel. Finally, the channel estima-
tion is designed by the nr5gPerfectChannelEstimate function.

In addition, during the transmission, 35 frames have been sent for a SNR range from
-2 to 15 dB. These SNR values have been selected because they represent a critical
scenario for wireless communications, since SNR is the difference between the noise
and the received signal measured in decibels. Also, the modulations that have been used
are:
• 16QAM
• 64QAM
• 256QAM

Furthermore, it is important to highlight that for this evaluation the CDL-A channel model
has been used. In addition, the SCS used, as it is possible to see in program 1, is 15
 16

kHz. So that, each subframe is composed by one slot (14 symbols), as it is specified in
table 2.5. In addition, the number of subframes sent for 35 frames is 350, since each
frame has 10 subframes. Table 4.3 presents the size in number of bits of each subframe
and transport block used for the three different modulations and for each waveform.
Transport block sizes have been calculated as given by TS 36.101 Annex A.3.1.

Table 4.3. Bits per each subframe

Waveform 16QAM 64QAM 256QAM
CP-OFDM 22920 bits/subframe 32856 bits/subframe 63776 bits/subframe
F-OFDM 24496 bits/subframe 37888 bits/subframe 66592 bits/subframe
W-OFDM 24496 bits/subframe 37888 bits/subframe 66592 bits/subframe

However, when 64QAM was used for F-OFDM and W-OFDM, the first subframe of
every each ten subframes contained 34008 bits, while the other nine subframes con-
tained 37888 bits, as it is specified in table 4.3. On the other hand, figure 4.3.1 shows
the BER results that have been achieved for CP-OFDM by using the previous parame-
ters:

Figure 4.3.1. BER with AWGN + CDL of CP-OFDM

As one can see in figure 4.3.1, the BER results show a good behaviour of the model
when CP-OFDM is used. In figure 4.3.2, the BER of F-OFDM is represented.
 17

Figure 4.3.2. BER with AWGN + CDL of F-OFDM

In figure 4.3.2 it is possible to see the behaviour of the transmission when F-OFDM is
sent through the chain model designed before. Similar results are achieved for CP-
OFDM and F-OFDM. The result of the figure 4.3.3 is based on W-OFDM. The same
modulations are used.

Figure 4.3.3. BER with AWGN + CDL of W-OFDM

As one can see in figure 4.3.1, 4.3.2 and 4.3.3, for testing which waveform achieves
better results, different BER graphics are represented for a SNR from -2 to 15 dB, as it
was commented before. Regarding the three waveforms tested, it is possible to conclude
 18

that CP-OFDM achieves a similar result than F-OFDM. This conclusion is based on the
values achieved for each modulation for the SNR. However, when CP-OFDM is used,
the BER presents better results for all the SNR range. Thus, in general it can be con-
cluded that CP-OFDM has better results than F-OFDM in terms of BER. In addition, W-
OFDM presents even a worse BER than F-OFDM. In figure 4.3.4 the BER values of each
waveforms are presented for 16QAM in order to see the differences among these three
waveforms.

4.3.4. BER comparation for 16QAM

Hence, to sum-up, CP-OFDM presents better results in terms of BER than F-OFDM
and W-OFDM. In addition, W-OFDM presents the worst result compared with F-OFDM
and CP-OFDM.

4.4 5G proposal for more reliable communications

There are several ways to improve the reliability, since this measure is affected by
various parameters, such as diversity, modulation, retransmission mechanisms, etc.
[10]. Nevertheless, an improvement in the reliability usually causes an increase of the
latency because of retransmissions, as it is shown in figure 4.4.1:
 19

Reliability
*URLLC: Ultra-reliable low
latency communications

URLLC Latency

Figure 4.4.1. Tradeoff between reliability and latency

However, there are cases where reliability and latency are optimized. Regarding the
data rates, it is demostrated that lower rates incur higher reliability and vice-versa [16].
This section discusses about those parameters based on 5G library which can improve
the reliability of the TX-channel-RX chain model proposed. These parameters are:

• Spatial diversity
• HARQ
• Coding rate

Regarding spatial diversity, it is important to highlight the possibility of using multiple

input and multiple output systems (MIMO) as it is used in the chain model of section 4.3.
However, by incrementing the number of transmitting and receiving antennas, it is pos-
sible to achieve more reliability [19]. In addition, this spatial diversity technique reduces
the power requirements. These enhancements are possible since the transmission in-
formation is divided into multiple data streams and sent by each transmitting antenna.
Then, the multiple signals reach the receiving antennas. Depending on the different
paths the signals have travelled, each one will have been affected by different factors.
However, the receiving antennas can reconstruct the data by taking into account the
noise, the interferences and the slight time difference between the received signals.
Moreover, 5G library can manage this type of transmission, and allow to set different
antenna configurations, such as 4x2, that is 4 transmitting and 2 receiving antennas. This
can be done by changing the transmission scheme of gNodeB configuration to ‘TxDiver-
sity’ and modifying the number of antennas.

On the other hand, the previous chain model also can be improved in terms of relia-
bility by adding a hybrid automatic repeat request (HARQ) protocol. The purpose of this
protocol is to provide high reliability and efficiency by retransmitting those packets which
have been received with errors. Those improvements are possible because HARQ de-
creases the error rate and avoid the retransmission of the whole packet, only the wrong
 20

part of the packet is retransmitted [17]. In addition, this protocol can be employed in 5G
library.

Vary the coding rate is another option for improving the reliability of 5G communica-
tions. It is important to increment the reliability of every packet for the purpose of
minimizing the error rate and the number of retransmissions. This aspect can be man-
aged by variating the coding rate, since it adds extra error checking bits for making that
the packets are received correctly. Also, it is possible to select different coding types
such as LDPC, Turbo code and Polar code. As it is showed in [14], when the code rate
decreases, better reliability is achieved. Thus, for improving the reliability of an TX-chan-
nel-RX chain model based on 5G library, the coding rate needs to be decreased. This
parameter can be changed in 5G library by variating the variable ‘TargetCodeRate’ of
the PDSCH configuration. Then, the transport block sizes will be calculated as given by
TS 36.101 Annex A.3.1. Moreover, as it is demonstrated in [14], it is possible to achieve
better reliability for different coding rates for LDPC and turbo code than polar code when
the block length increases. On the contrary, for smaller block lengths polar code gets
better reliability than LDPC and turbo code.

As a result, and because of the aforementioned, in order to improve the reliability of

the 5G communications that has been achieved through the previous TX-Channel-RX
chain model, some changes are proposed. As the first improvement, more antennas will
be added to the system. During the section 4.3, it has been used a 4x2 configuration.
Nevertheless, since 5G library can manage 4x4 antennas, this configuration will be used
in this case. Moreover, the HARQ protocol will be added and the coding rate will be
decreased from 1/2 to 1/3. In figure 4.4.2, one can see the new result in comparison with
the previous one for 16QAM. The number of frames that have been sent are 35. These
frames have been sent for a SNR range from -2 to 15 dB and the waveform that has
been employed is CP-OFDM. In addition, the propagation channel is the same as has
been used for all the simulations of the thesis, CDL-A. Table 4.4 compares the differ-
ences between the new configuration and the previous one:

Table 4.4. Different configurations

Parameters Previous configuration New configuration
Antennas 4x2 4x4
HARQ protocol disenabled enabled
Code rare 1/2 1/3
 21

The following figure shows the result achieved by this new configuration in compari-
son with the result achieved in chapter 4.3 for 16QAM when CP-OFDM is used.

Figure 4.4.2. BER comparation. The red BER is the modified and improved version
of the blue one.

As one can see in figure 4.4.2, the red BER is the one that has been achieved by the
configuration proposed during this section of the thesis, while the blue BER was achieved
by the previous configuration. It is possible to appreciate that for the same SNR, the BER
corresponding to the new configuration achieves better values than the previous one.
Thus, it is possible to conclude that the reliability of the communications has been im-
proved.

4.5 Limitations of 5G Library

During the analysis of the results, some limitations have been found in “5G library for
LTE Systems” of Matlab, library which has been used for the development of this thesis.

• The antenna’s configuration can be 4x4 as much. This limitation does not al-
low to obtain better reliability in the results.
• The model cannot be designed without perfect channel estimation when
AWGN is added, because it produces a strange behaviour when the BER is
calculated.

Due to these limitations, the results are limited. In addition, there are some variations
of the model that cannot be implemented and tested.
 22

5. CONCLUSIONS

During the development of this thesis several conclusions have been reached. First,
during the study of the different waveforms supported by 5G Toolset of Matlab, it has
been proved how 5G waveforms achieve better spectral efficiency than LTE waveforms.
In addition, the study has revealed that F-OFDM presents the best performance for the
different scenarios that have been tested. Also, it has been shown that selecting less
SCS it is possible to obtain less OOBE. Then, it has been explained that UFMC can be
applied for 5G communications.

On the other hand, after the study and investigation commented in the previous par-
agraph, the design and testing of the TX-Channel-RX chain model based on 5G Toolset
was carried out. It has demonstrated that in general, CP-OFDM achieves better perfor-
mance than F-OFDM and W-OFDM in terms of BER. However, the behaviour of F-OFDM
and CP-OFDM is very similar. On the contrary, W-OFDM has shown the worst BER
result of the three waveforms tested.

The final objective of this thesis has been to propose a new configuration in order to
improve the reliability of the communications. For achieving this challenging objective,
the proposal was to develop a new configuration based on the spatial diversity, the
HARQ protocol and the code rate. The result achieved by this new configuration was
compared with the result obtained by the previous model in order to see how the relia-
bility varied. Regarding the spatial diversity, the new configuration proposed to add more
antennas. In this way, two more receiving antennas were added. Also, the code rate was
changed to 1/3 and the HARQ protocol was enabled. The achieved result shows how
this new configuration improves the reliability of the previous model.

On the other hand, it is important to highlight that as has been commented previously,
this thesis is based on 5G Toolset. This means that the characteristics of the designed
model are subject to the parameters that this library support. So that, there is a limitation
in the results that can be obtained. Hence, it is necessary a more complete research in
order to achieve more realistic results.
 23

REFERENCES

[1] 3rd Generation Partnership Project (3GPP) (Dec. 2018). NR; Base station (BS)
radio transmission and reception. Technical Specification 38.104, ver. 15.4.0.

[2] 3rd Generation Partnership Project (3GPP) (Jan. 2019). NR; Physical channels
and modulation. Technical Specification 38.211, ver. 15.4.0.

[3] 3rd Generation Partnership Project (3GPP) (Jun. 2018). Study on channel model
for frequencies from 0.5 to 100 GHz. Technical Specification 38.901, ver. 15.0.0.

[4] A. F. Demir, A. Fatih, M. Elkourdi, M. Ibrahim and H. Arslan (2018). Waveform

Design for 5G and Beyond. DOI: 10.1002/9781119333142.ch2.

[5] A. Garg (2014). Digital Society from 1G to 5G: A Comparative Study. International
Journal of Application or Innovation in Engineering & Management. Volume 3. 186
to 193.

[6] A. Ijaz, L. Zhang, P. Xiao and R. Tafazolli (Dec. 2016). Analysis of Candidate
Waveforms for 5G Cellular Systems. Towards 5G Wireless Networks - A Physical
Layer Perspective. Hossein Khaleghi Bizaki, IntechOpen, DOI: 10.5772/66051.
Available at: https://www.intechopen.com/books/towards-5g-wireless-networks-a-
physical-layer-perspective/analysis-of-candidate-waveforms-for-5g-cellular-sys-
tems

[7] A. Othman (2019). 5G System Architecture & Implementation. DOI:

10.13140/RG.2.2.11793.68966.

[8] A. Yazar, B. Pekoz and H. Arslan (2018). Flexible Multi-Numerology Systems for
5G New Radio.

[9] B. Kalra and D. K. Chauhan (2014). A Comparative Study of Mobile Wireless Com-
munication Network: 1G to 5G. International Journal of Computer Science and In-
formation Technology Research. 2. 430-433.

[10] B. Soret, P. Mogensen, K. I. Pedersen and M. C. Aguayo-Torres (2014). Funda-

mental tradeoffs among reliability, latency and throughput in cellular net-
works. 2014 IEEE Globecom Workshops (GC Wkshps), Austin, TX pp. 1391-1396.
DOI: 10.1109/GLOCOMW.2014.7063628
 24

[11] F. Amato (2019). A Primer on 5G.

[12] F. Sharevski (2018). Towards 5G cellular network forensics. EURASIP Journal on

Information Security. 2018. DOI: 10.1186/s13635-018-0078-7.

[13] G. Nguyen, A. Brunstrom, K. Grinnemo and J. Taheri (2017). 5G Mobile Networks

Requirements, Enabling Technologies, and Research Activities.

[14] H. Gamage, N. Rajatheva and M. Latva-aho (2017). Channel Coding for Enhanced
Mobile Broadband Communication in 5G Systems.

[15] I. Balapuwaduge and F. Li (2018). Cellular Networks: An Evolution from 1G to 4G.

[16] J. Park and P. Popovski (Dec. 2017). Coverage and Rate of Downlink Sequence
Transmissions with Reliability Guarantees. in IEEE Wireless Communications Let-
ters, vol. 6, no. 6, pp. 722-725. DOI: 10.1109/LWC.2017.2735974

[17] J. Yeo, S. Park, J. Oh, Y. Kim and J. Lee (2017). Partial Retransmission Scheme
for HARQ Enhancement in 5G Wireless Communications. 2017 IEEE Globecom
Workshops (GC Wkshps), Singapore, pp. 1-5.
DOI: 10.1109/GLOCOMW.2017.8269132

[18] P. Gaal (2016). Waveform candidates. Qualcomm Incorporated, R1-162199.

[19] P. Patil, N. Muchhal and R. Shankar (2014). An Extensive Review on Spatial Di-
versity Techniques in Wireless Communication. International Journal of Computer
Technology and Electronics Engineering. Volume 4

[20] X. Zhang, L. Chen, J. Qiu, J. Abdoli (2016). On the Waveform for 5G. IEEE
Communications Magazine, vol.54, issue 11, pp. 74-80.