Wave Forms and Channel Coding For 5

FACULTY OF INFORMATION TECHNOLOGY AND ELECTRICAL ENGINEERING
Heshani Niyagama Gamage
WAVEFORMS AND CHANNEL CODING FOR 5G
Master’s Thesis
Degree Programme in Wireless Communications Engineering
March 2017
Niyagama Gamage H. (2016) Waveforms and Channel Coding for 5G. University
of Oulu, Faculty of Information Technology and Electrical Engineering, Department
of Communications Engineering. Master’s thesis, 63 p.
ABSTRACT
The fifth generation (5G) communication systems are required to perform sig-
nificantly better than the existing fourth generation (4G) systems in data rate,
capacity, coverage, latency, energy consumption and cost. Hence, 5G needs to
achieve considerable enhancements in the areas of bandwidth, spectral, energy,
and signaling efficiencies and cost per bit. The new radio access technology (RAT)
of 5G physical layer needs to utilize an efficient waveform to meet the demands
of 5G. Orthogonal frequency division multiplexing (OFDM) is considered as a
baseline for up to 30 GHz. However, a major drawback of OFDM systems is
their large peak to average power ratio (PAPR). Here in this thesis, a simple
selective-mapping (SLM) technique using scrambling is proposed to reduce the
PAPR of OFDM signals. This technique selects symbol sequences with high PAPR
and scrambles them until a PAPR sequence below a specific threshold is gener-
ated. The computational complexity of the proposed scheme is considerably lower
than that of the traditional SLM. Also, performance of the system is investigated
through simulations and more than 4.5 dB PAPR reduction is achieved.
In addition, performance of single carrier waveforms is analyzed in multiple-
input multiple-output (MIMO) systems as an alternative to OFDM. Performance
of a single carrier massive MIMO system is presented for both uplink and down-
link with single user and multiple user cases and the effect of pre-coding on the
PAPR is studied. A variety of channel configurations were investigated such as
correlated channels, practical channels and the channels with errors in channel
estimate.
Furthermore, the candidate coding schemes are investigated for the new RAT in
the 5G standard corresponding the activities in the third generation partnership
project (3GPP). The schemes are evaluated in terms of block error rate (BLER),
bit error rate (BER), computational complexity, and flexibility. These parameters
comprise a suitable set to assess the performance of different services and appli-
cations. Turbo, low density parity check (LDPC), and polar codes are considered
as the candidate schemes. These are investigated in terms of obtaining suitable
rates, block lengths by proper design for a fair comparison. The simulations have
been carried out in order to obtain BLER / BER performance for various code
rates and block lengths, in additive white Gaussian noise (AWGN) channel. Al-
though polar codes perform well at short block lengths, LDPC has a relatively
good performance at all the block lengths and code rates. In addition, complexity
of the LDPC codes is relatively low. Furthermore, BLER/BER performances of
the coding schemes in Rayleigh fading channels are investigated and found that
the fading channel performance follows a similar trend as the performance in the
AWGN channel.
Keywords: 5G, RAT, OFDM, Channel coding, Polar, LDPC, Turbo, eMBB
TABLE OF CONTENTS
ABSTRACT
TABLE OF CONTENTS
FOREWARD
ABBREVIATIONS
1. INTRODUCTION 8
2. BACKGROUND AND LITERATURE 12

2.1. Energy Efficient Waveforms . . . . . . . . . . . . . . . . . . . . . . 12
2.1.1. OFDM waveforms . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.2. Single Carrier waveforms . . . . . . . . . . . . . . . . . . . 15
2.2. Channel coding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1. Turbo codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.2. LDPC codes . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.3. Polar codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3. PAPR REDUCTION OF OFDM WAVEFORMS 19

3.1. System model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.1.1. Scrambler and descrambler . . . . . . . . . . . . . . . . . . . 20
3.2. Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4. PERFORMANCE OF SINGLE CARRIER WAVEFORMS IN MIMO

SYSTEMS 28
4.1. Single carrier - Uplink . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1.1. System model . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.1.2. Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2. Single carrier-Downlink . . . . . . . . . . . . . . . . . . . . . . . . 33
4.2.1. System model . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.2.2. Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.2.3. PAPR Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2.4. Correlated Channels . . . . . . . . . . . . . . . . . . . . . . 37
4.3. Practical Channel Models . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3.1. Simulations: Uplink . . . . . . . . . . . . . . . . . . . . . . 40
5. CHANNEL CODING 42
5.1. Candidate Coding Schemes . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.1. Turbo codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.2. LDPC codes . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1.3. Polar codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.2. Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.3. Simulation results in AWGN channel . . . . . . . . . . . . . . . . . . 44
5.4. Coding performance in fading channel . . . . . . . . . . . . . . . . . 45
5.4.1. System model . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.4.2. Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6. CONCLUSION 57
7. RECOMMENDATIONS FOR FUTURE 59
8. REFERENCES 60
FOREWARD
This thesis is focused on energy efficient waveforms and channel coding schemes for
5G New Radio air-interface as a part of the 5Gto10G project at the Center for Wireless
Communications (CWC) of University of Oulu, Finland. I express my sincere gratitude
to my supervisor and mentor Prof. Nandana Rajatheva for his support and guidance
throughout my period of masters studies. Also I express my gratitude to Prof. Matti
Latva-aho, the head of the RAT group of CWC for his guidance. Also, I thank Dr.
Pekka Pirinen, the project manager and the other colleagues of 5Gto10G project for
their support in my research.
Also, I thank my parents and my siblings for their immense support in my life.
Finally, I thank my dear husband Udaranga for for his endless support.
ABBREVIATIONS
Acronyms
1G First Generation
2G Second Generation
3G Third Generation
3GPP Third Generation Partnership Project
4G Fourth Generation
5G Fifth Generation
AR Augmented Reality
AWGN Additive White Gaussian Noise
BER Bit Error Rate
BLER Block Error Rate
BS Base Station
CCDF Complementary Cumulative Distribution Function
CDMA Code Division Multiple Access
CRC Cyclic Redundancy Check
CRCA-SCL CRC Aided Successive Cancellation List decoding
CSI Channel State Information
CSIT CSI at the transmitter
DAB Digital Audio Broadcasting
DFT Discrete Fourier Transform
DVB-T Terrestrial Digital Video Broadcasting
eMBB Enhanced Mobile Broadband
Exp Exponential Correlation
FDMA Frequency Division Multiple Access
FFT Fast Fourier Transform
FHD Full High Definition
GFDM Generalized Frequency Division Multiplexing
GOFDM Generalized Orthogonal Frequency Division Multiplexing
IFFT Inverse Fast Fourier Transform
IMT-Advanced International Mobile Telecommunications-Advanced
IP Internet Protocol
ISI Inter Symbol Interference
LDPC Low Density Parity Check
LTE Long Term Evolution
M2M Machine to Machine Communication
MAC Medium Access Control
MBB Mobile Broadband
MIMO Multiple-Input Multiple-Output
mMTC Massive Machine-Type Communications
NR New Radio
OFDM Orthogonal Frequency Division Multiplexing
OMS Offset Min-Sum
PA Power Amplifier
PAPR Peak-to-Average Power Ratio
PDP Power Delay Profile
PHY Physical Layer
PSK Phase Shift Keying
PTS Partial Transmit Sequence
QAM Quadrature Amplitude Modulation
QC Quasi Cyclic
QPSK Quadrature Phase Shift Keying
RAT Radio Access Technologies
RF Radio Frequency
SC Successive Cancellation
SC-FDMA Single Carrer - Frequency Division Multiple Access
SCL Successive Cancellation List decoding
SLM Selective Mapping
TBCC Tail Biting Convolutional Codes
TDMA Frequency Division Multiple Access
ULA Uniform Linear Array
URLLC Ultra-Reliable Low-Latency Communications
UT User Terminal
VR Virtual Reality
WLAN Wireless Local Area Network
Symbols
a Exponential correlation coefficient

Dl Vector of PDPs of all the users
D Length of ULA array
d distance from the mo antenna to the mth antenna in ULA
array
dc Average check degree
dl Power delay profile
dv Average variable degree.
Eb Bit energy
Ec Symbol power
fk k th subcarrier
H [l] Fading channel coefficients matrix of the lth path
I Number of iterations
K Number of users
L Number of channel taps -1
Ls Oversampling factor
Lp List size of the SCL decoder
M Number of receive antennas at the base station
Ma Alphabet size
Mb Memory length
No Noise power spectral density
Ns Number of sub carriers
Nb Coded block length
P Scrambler polynomial
Pb Number of parity bits
R Code rate
S Length of the scrambler polynomial
T Symbol period
t Exponential correlation matrix of a channel
Th Threshold value
U Number of scramblers
w [n] Additive white noise vector
X Transmit signal vector
X Block of data symbol
x [n] Transmitted signal
Xk k th data symbol
xrms Root mean square value of waveform x
x(t) OFDM signal at time t
Y Received signal vector
y [n] Received signal
W AWGN vector
∆f Subcarrier space
Estimation error
λ Threshold in decibels
λl PAPR of the lowest PAPR sequence
ρ Antenna spatial correlation
σ2 Variance
Operators
|·| Absolute value

E [·] Expected value
J0 (·) Zeroth order Bessel function of the first kind
log2 (·)
P Logarithm of base 2
Summation operation
8
1. INTRODUCTION
Services offered by cellular communication systems have evolved from the first gener-
ation(1G) to the fourth generation (4G) adding further enhanced services in each gen-
eration. 1G was only for voice calls followed by the second generation (2G), which
added text messaging services. The third generation (3G) added mobile internet ser-
vices to 2G. Currently 4G offers high capacity mobile multimedia service at 1 Gbps
data rate when stationary and 100 Mbps rate when mobile, making it 250 times bet-
ter than the 3G services. 5G New Radio (NR) is the forthcoming evolution of mobile
technology expected to be in use by the year 2020 with a wide range of usability be-
yond the uses of 4G [1]. Performance parameters of 5G technologies are expected
to be tens and thousands times better than in 4G. Although the performance parame-
ters for 5G are not standardized yet, typical parameters [2] are stated in Table 1 with
4G international mobile telecommunications-advanced (IMT-Advanced) performance
parameters for reference.
Table 1: Suggested performance parameters for 5G.
5G suggested Perfor- IMT-Advanced Perfor-

Parameter
mance mance
10 000 times the capacity 2.2 downlink and 1.4 up-
Network capacity
of current network link (bps/Hz/cell)
3 Gbps Downlink and 1.5
Peak data rate 10 Gbps
Gbps Uplink
Latency Less than 1 ms 10 ms
2/3/5 x greater than IMT-
Peak spectral Efficiency 6.75 bps/Hz
Advanced
Furthermore, low energy consumption as well as cost are also desired. In addition 5G
should enable machine to machine communication (M2M) at ultra low cost and ultra
high reliability while supporting 10 years battery life. To facilitate these needs spec-
tral efficiency, signaling efficiency, bandwidth and coverage should be significantly
enhanced compared to 4G.
Number of mobile devices such as mobile phones, tablets and smart watches has
been exponentially increasing in the last few years and at the same time more and
more applications such as video streaming/downloading, multi-player gaming which
requires a large bandwidth are emerging. In addition more and more household items
such as fridges, ovens, heating systems, and even clothes are being developed as "smart
devices" which can be connected with other devices via internet. Majority of these re-
quire a small amount of data to be transferred between the devices in a large network.
The networks like these, called Internet of Things (IoT) are increasing adding different
kind of devices in household as well as industrial environments. Current mobile net-
works should be enhanced by several factors in terms of capacity, data rate, latency and
9
other performance aspects in order to facilitate the increasing demand through above
mentioned technologies.
Therefore the two most significant use cases, which are the main drivers of the de-
velopment of 5G network technology can be recognized as [3]
1. Mobile broadband (MBB) services.

MBB is the term used for wireless internet access through modems and other
mobile devices. MBB was the primary drive of the 4G and it will remain the
most critical use case of 5G. As the current networks provide faster data speeds,
many data intensive applications are emerging for both business and consumer
use. At present, a large percentage of the MBB users is for video streaming
and downloading. As higher resolution videos like 4K becomes more common,
4G networks will not be able to cope up with a large number of users at the
same time. In addition the growing use of cloud services will add more traffic
to demand for MBB. Furthermore virtual reality (VR), augmented reality (AR),
multi-player gaming are some other content-rich services which are expected to
be increasing the demand for better MBB services. All of these services are
content-rich and at the same time number of devices accessing these services are
increasing. By the nature of these applications, they have a large volume of data
the latency must be low to have a better user experience, these users need faster,
higher-capacity networks that can deliver content-rich services.
2. Internet of things(IoT)
IoT is a network of interconnected devices, such as household appliances, indus-
trial machinery, vehicles, medical devices, environmental sensors and etc em-
bedded with sensors and actuators, that are provided with unique identifiers and
the ability to compute, coordinate and communicate, data over a network with-
out requiring human intervention. There are two varieties of IoT applications
based on the nature of communication between the devices. They are
(a) Massive machine type communication (mMTC)

These are networks of large number of low cost, low power devices which
require infrequent transferring of a very small amount of data which are not
delay-sensitive. Some of the examples of these kind of systems are wireless
sensor networks, smart homes, smart cities etc. These networks needs to
be scalable, so users can add more and more devices to the network.
(b) Mission critical communication
In this scenario machine-to-machine communication needs to be ultra-
reliable with low latency. Usually these devices are relatively expensive.
Example scenarios are autonomous vehicles, remote surgeries, and indus-
try automation. These applications needs extreme reliability and low la-
tency to ensure human safety.
Based on different user requirements, new radio access technology for 5G systems
includes three scenarios; namely,
1. Enhanced Mobile Broadband (eMBB).

This is being developed to deliver a better MBB services in 5G with higher data
10
rates, higher capacity and better user experience. This is the most profitable area
in 4G and the eMBB remains the most critical scenario for 5G as well because
the ongoing growth of users demanding the eMBB services proves to be strong
and profitable. 5G is expected to deliver 10 Gbps peak throughput and up to
10,000 times the current network capacity. Based on its requirements, eMBB
needs to support a much wider range of code rates, code lengths and modulation
orders than LTE. In the current assumption, eMBB code lengths range from 100
to 8000 bits (optionally 12, 000 − 64000 bits) and code rate ranges from 1/5 to
8/9.
2. Massive machine type communications (mMTC).
This scenario is being developed to cater the massive-machine networks to en-
able IoT at home and in business. Very low power consumption, low cost, ability
to handle a large number of connections and good scalability are desired in these
type of systems.
3. Ultra reliable low latency communications (URLLC).
This is being developed to cater mission critical communication needs with ultra-
reliability and low latency.
In order to cater for the requirements of these scenarios, 5G needs to to achieve signifi-
cant enhancements in addition to existing technologies. Some of the main technologies
under research to achieve the data rate and capacity for 5G are [4]:
1. Densification of Networks
Utilization of more base stations in a heterogeneous network which is a com-
bination of macro cells in order to increase the capacity. By reducing the cell
size we can obtain a higher overall effective use of the available spectrum. This
will require efficient interference cancellation methods and other techniques to
ensure that small cells in the macro-network and deployed as femtocells can
operate satisfactorily are required.
2. Increased bandwidth
This includes additional bandwidth added to the mobile networks and increasing
spectral efficiency. This will enable the networks to cater for more traffic and de-
vices. Currently spectrum of the wireless communications are limited to 6GHz.
To increase the capacity, spectrum above 10GHz up to 100 GHz will be added
to the spectrum of the network. This will need developments in millimeter wave
technologies and other high frequency communication technologies. In addition
efficient spectrum sharing and cognitive radio will enable higher efficiency in
spectrum usage.
3. Increased spectral efficiency
Massive MIMO is a area under research for enhancing spectral efficiency.
4. New air interface
This includes waveforms, multiplexing schemes, modulation schemes and cod-
ing schemes to reduce the latency in the air-link, and to increase spectral effi-
ciency, connection density, throughput, area capacity density and energy effi-
ciency.
11
This thesis is focused on technologies for the new air-interface, waveforms and cod-
ing schemes for the 5G NR PHY Layer. In Chapter 2, some of the technologies avail-
able for waveforms and coding scheme are discussed. In the third chapter, a low com-
plex algorithm is proposed to reduce the PAPR of the OFDM waveforms followed
by single carrier waveform performance evaluation the the fourth chapter. Candidate
coding schemes for the new air-interface of 5G are evaluated in the fifth chapter. Con-
clusions and future recommendations are given in the final two chapters.
12
2. BACKGROUND AND LITERATURE
In this chapter the basic concepts and available literature of the waveforms and coding
schemes are discussed. In the first section we focus on the available waveform tech-
niques for energy efficient waveforms. We discuss about the performance aspects of
OFDM waves and single carrier waves. Next we discuss about coding schemes for 5G
eMBB scenario.
2.1. Energy Efficient Waveforms
The waveform will be a key specification of the wireless standard and will have a major
effect on the performance parameters. Wave forms used in each generation are listed
in the Table 2.
Table 2: PHY / MAC for each generation.
Generation Waveforms
1G Analog - FDMA
2G Digital FDMA/TDMA
3G CDMA
4G OFDM, FDMA/TDMA
In this thesis performance aspects of OFDM waveforms and single carrier waveforms
are investigated.
2.1.1. OFDM waveforms
OFDM is a multi-carrier modulation technique where the signal is constructed from a

number of closely spaced modulated sub-carriers. OFDM is being used in several high
data rate wireless systems due to many advantages it provides including;
1. Resilience to multipath fading, inter-symbol interference (ISI), co-channel inter-

ference and impulsive parasitic noise, due to multiple narrow band sub channels
with a lower data rate on each sub channel.
2. Lower implementation complexity compared with the single carrier solution

with usage of fast Fourier transform (FFT)/Inverse FFT (IFFT) blocks.
3. High spectral efficiency due to usage of closely spaced overlapping subcarriers.

13
Therefore, OFDM has been adopted in many wireless standards such as wireless
local area network (WLAN) radio interfaces IEEE 802.11a, g, n, ac, digital audio
broadcasting (DAB), and terrestrial digital video broadcasting (DVB-T). Also OFDM
has already been adopted as a radio access scheme for several of the latest cellular net-
work and mobile broadband standard specifications such as WiMax and 3GPP LTE 4G
mobile broadband standard. In 4G LTE, OFDM with modulated data directly mapped
to sub-carriers is used only in downlink transmission in the form of OFDM multiple
access. In the uplink, data is first transformed via a discrete Fourier transform (DFT)
before being modulated and transmitted via sub-carriers in the form of single carrier
frequency division multiple access (SC-FDMA) due to the PAPR problem.
Waveform
We consider an OFDM system with Ns subcarriers. Let X = [Xk ; k = 0, 1, . . . , Ns − 1]
be a block of Ns data symbols drawn from an alphabet of size Ma . Each of these sym-
bols will modulate one in a set of Ns orthogonal subcarriers [fk , k = 0, 1, . . . , Ns − 1]
where fk = k∆f , ∆f = 1/Ns T , and T is original symbol period. The complex
envelope of the transmitted signal x(t) can be written as
Ns −1
1 X
x(t) = Xk ej2πfk t ; 0 ≤ t < Ns T . (1)
Ns k=0
Peak-to-average power ratio (PAPR)

If |x|peak and xrms are peak value and root mean square value of waveform x respec-
tively, then PAPR of the waveform x can be defined as
|x|peak
P AP R{x} = . (2)
xrms
To better approximate the PAPR of continuous time OFDM signals, OFDM samples
are taken by Ls times oversampling. [5]. Oversampled FFT output x [n] will be
Ns −1
1 X 2πnk
x [n] = √ Xk ej Ls Ns ; 0 ≤ n < Ls Ns − 1. (3)
N s k=0
PAPR of the OFDM signal is defined as
max0≤n≤Ns Ls [|x [n] |2 ]
P AP R{x [n]} = . (4)
E [|x [n] |2 ]
Complementary Cumulative Distribution Function (CCDF)

CCDF for PAPR is the probability of OFDM symbols with PAPR exceeding some
threshold λ. Depicted in Fig. 1 is CCDF plots of 4 times oversampled OFDM PAPR
for N subcarriers. It can be seen that PAPR increases as the number of subcarriers
increases. For a small number of N , the probability that the PAPR of the OFDM
signal exceeds a certain threshold λ is given by [6]
P r{P AP R > λ} = 1 − (1 − e−λ )Ns . (5)
14
10 0
Complementary Cumulative Distribution Function

N=64
N=128
N=256
N=512
10 -1 N=1024
10 -2
10 -3
10 -4
4 5 6 7 8 9 10 11 12 13
Peak to Average Power Ratio ( dB )
Figure 1: CCDF of PAPR in 4 times over sampled OFDM system with N subcarriers.
High PAPR in original OFDM signals will lead to distortion of the signal if the
transmitter has nonlinear components such as power amplifiers (PAs). Nonlinear de-
vices will cause spectral spreading, inter modulation and constellation distortion. To
avoid this, usage of expensive high power amplifiers in a very large linear range, ap-
proximately equal to the dynamic range of the OFDM signal is required. Also high
PAPR signals prevent PA from operating in a near saturation region reducing the PA
efficiency. There are several alternative waveforms being proposed to avoid the disad-
vantages of OFDM. They are time–frequency packing, nonorthogonal Signals, filter-
bank multicarrier, generalized frequency division multiplexing (GFDM) and tunable
OFDM.
Many different techniques have been proposed to reduce PAPR problem. The sim-
plest and the most used technique is clipping and filtering [7], [8]. However, this
method introduces signal distortion resulting in-band noise and out-of-band leakage,
which leads to BER degradation. Coding schemes such as block coding [9], cyclic cod-
ing [10] and complement block coding [11] can also be used to reduce PAPR. However,
PAPR reduction capability of these schemes is limited to OFDM systems with small
number of subcarriers and there is a data rate loss and a high computational complexity.
Partial transmit sequence (PTS) [12], [13] is a probabilistic scheme to reduce PAPR. It
carries a high computational complexity due to the calculation of PAPR value at each
step of the optimization algorithm. Also, this technique results in a BER degradation
and a data rate loss. Tone reservation and tone injection [14] are efficient techniques,
which reduce the PAPR by reserving a subset of tones for generating PAPR reduction
15
signals. These techniques, however, have a limited PAPR reduction capability, high
complexity, data rate loss and a power increase. Nonlinear companding, mainly µ-law
and exponential companding [15] are effective techniques for PAPR reduction. Espe-
cially exponential companding is an attractive method due to its simplicity and very
high PAPR reduction capability by minimizing the dynamic range of the OFDM sig-
nal. We found out through simulations that this method has a poor BER performance
due to amplification of noise at the receiver. Generalized OFDM (GOFDM) [16] is
a recent method, which lowers PAPR by reducing the number of subcarriers within
OFDM symbols. Although it produces significantly low PAPR, there are major draw-
backs such as high computational burden and the requirement of perfect knowledge of
channel transfer function etc.
Selective Mapping (SLM) technique is a probabilistic scheme, where U different
OFDM signals are generated as candidates for a specific signal and then one with the
lowest PAPR is selected for the actual transmission. At the receiver, it is necessary
to know which one of U candidates is transmitted. Hence transmission of side infor-
mation is needed in conventional methods resulting a data rate loss. There are several
techniques that has been proposed to generate different OFDM signals including, phase
rotation [17] and turbo coding [18]. Usually, complexity of SLM method is very high
due to its requirement of calculating U × N IFFTs whenever SLM is applied for an
OFDM system with N subcarriers.
Apparently none of these techniques have all the desired features of a OFDM system
including;
1. Large PAPR reduction.
2. Low computational complexity.
3. No degradation in BER Performance.
4. Low average power and no requirement of additional transmit power.
5. No bandwidth expansion or spectral spillage.
All of the above mentioned PAPR reduction techniques tend to compromise one or
more of those wanted objectives in reducing BER. Hence research of PAPR reduction
is still very intensive. A simple modified SLM technique using an array of scramblers
to reduce the PAPR of OFDM signals [19] is discussed in the following chapter of this
thesis. Unlike most of the schemes mentioned above, it has a simple architecture and
does not have a BER degradation.
2.1.2. Single Carrier waveforms
Single carrier waveforms regain the interest recently. In 5G some filtered transmissions
for the lower frequency region below 6 GHz are in the development of NR air interface
[20]. Given the way 3GPP standardization is evolving massive MIMO is going to
be a reality sooner rather than later [1, 21]. This also takes into account the fact that
there will be new frequency ranges being explored in the mmWave range [22] where
work is going on to develop new radio access technologies (RATs). One disadvantage
16
of a large number of antennas in a multi-carrier system will be the large number of

IFFT/FFT blocks that will be needed. Given both these issues it will be good to have
a system which can take advantage of the higher number of antennas being deployed
in base stations. One direct implication from Marzetta’s work [23] is that there is the
possibility of a reduced complexity receiver utilizing the large dimension and hence
realizable orthogonality due to that. It is also important to see the performance of such
systems with closer to reality channel models. In our case some results are obtained
using measured channels obtained by our CWC colleagues [24, 25].
The main reason why multi-carrier OFDM based system is universally adopted is
due to the simplicity of the receiver implementation, which circumvents the need for
sophisticated and complex equalization. With a single carrier system the issue of equal-
izer can be managed provided that one of the transmission link ends has a large number
of antennas. Another practical thing to be dealt with given the large number of chan-
nels to be measured is the accuracy of the estimate. Attention must be paid therefore
to have a realistic performance analysis.
A survey of single-carrier systems has been presented by Yang et. al. [26]. With the
incorporation of MIMO methods in practical wireless systems, e.g., LTE, 4G, it has
been possible to improve the link reliability and provide a higher capacity with better
spectral efficiency. Moving further along this path there has been a tremendous amount
of research in massive MIMO techniques. Marzetta [23] showed that it is possible to
mitigate the effects of fast fading and uncorrelated receiver noise through orthogonal-
ity arising out of large dimensions of matrices due to the large number of antennas.
Pitarokoilis et. al. [27] considered the application of single-carrier transmission in a
large scale antenna system in the downlink, where they showed that the receiver side
can essentially work without an equalizer. This effect is due to the properties as shown
by [23]. The base station transmitter uses a pre-coding scheme matched to the fre-
quency selective channel. Though single-carrier method here eliminates the use of
an equalizer, it is worthwhile to analyze the effect on the PAPR. Sheng et. al. [28]
investigated a single-carrier modulation uplink scheme for a Rician channel with max-
imum likelihood (ML) equalization expanding on [27]. A more detailed study is given
in [29]. In the case of a Rayleigh fading channel the receiver will become simpler
with a one tap equalizer similar to that used in an OFDM system. There has also been
a lot of work for single-carrier schemes, which employ a cyclic prefix and frequency
domain equalization [30]. Together with [26] therefore the literature on single-carrier
systems has been gaining significant momentum.
As mentioned previously, there are several issues specific to downlink and uplink
respectively. In the case of downlink with multiple RF chains for a large antenna
system, it is prudent to have as distortion-less as possible signal for transmission since
this has direct consequences to the spectral mask, which is tightly regulated. As the
number of carriers increases this poses a considerable challenge in OFDM systems. If,
however, this can be resolved with a single carrier scheme, which does not have the
same PAPR properties it will be quite useful. Naturally, in this light the fundamental
reason for preferring OFDM in the first place was the need for a simpler receiver with
one tap equalize. While it can be shown that with a higher number of antennas the
receiver of a single carrier system can still be less complex as opposed to a full fledged
time domain equalizer needed in a frequency selective channel with multiple taps and
a system employing high level modulation for increased bit rates, the performance of
17
such a system under less ideal conditions needs further investigation in order to bench
mark against the traditional OFDM system.
2.2. Channel coding
Channel coding plays an important role in achieving higher data rate in order to have
a fast and error free communication since data transmission occurs in an imperfect
channel environment. The selected channel code must have an excellent BLER per-
formance in a specific range of block lengths and code rates. Low computation com-
plexity, low latency, low cost and higher flexibility are desired for the coding scheme.
Furthermore, reduced energy per bit and improved area efficiency are required to sup-
port higher data rates. Tail biting convolutional codes (TBCC), turbo codes, LDPC
codes and polar codes are the competing coding schemes considered as the candidates
for 5G. However, TBCC is not considered for the eMBB scenario since it has a poor
performance in large block lengths and low code rates. In the fifth chapter of this the-
sis, we review candidate error correction coding schemes for 5G eMBB scenario and
evaluate their performances.
Although many coding schemes with capacity achieving performance at large block
lengths are available, many of those do not show consistent good performance in a
wide range of block lengths and code rates as the eMBB scenario demands. But turbo,
LDPC and polar codes show promising BLER performance in a wide range of coding
rates and code lengths and hence are being considered for 5G physical layer.
2.2.1. Turbo codes
Introduced in [31], turbo code encoder is built using a parallel concatenation of two
recursive systematic convolutional codes and the associated decoder, using a feedback
decoding rule. Due to the low error probability performance within a 1dB fraction
from the the Shannon limit and relatively low complexity, turbo codes are being used
in a variety of applications such as deep space communications, 3G/4G mobile com-
munication in Universal Mobile Telecommunications System (UMTS) and Long Term
Evolution (LTE) standards and Digital Video Broadcasting (DVB). Although it is be-
ing used in 3G and 4G, it may not satisfy the performance requirements of eMBB for
all the code rates and block lengths as the implementation complexity is too high for
higher data rates. In addition an error floor is observed in turbo code BER. In [32, Fig.
8.18] the error floor is at BER of 10−6 .
2.2.2. LDPC codes
LDPC codes were originally invented and published in [33] in 1962, they were not in
use until it was rediscovered by Mackay [34] in 1997. He showed that the empirical
performance of LDPC codes can approach the Shannon limit similar to turbo codes or
even better. LDPC codes are linear codes and as the name suggests, has a sparse parity
check matrix consisting low density 0 10 s. Due to the sparsity of the parity check matrix,
18
LDPC codes have relatively simple and practical decoding algorithms. Decoding is
done by iterative belief propagation decoding algorithms which estimates bit values
and parity check values in bit and check nodes respectively and passing values between
them in each iteration. The accuracy of the estimates will be improved in each iteration
and the number of iterations is decided based on the requirement of the application.
Trade-offs are possible between the bit error performance, latency and the complexity.
Modern LDPC decoders work with soft decision algorithms which further enhance the
decoder gain. Due to their excellent ability to achieve theoretical limits of channel
capacity, LDPC codes are currently being used in many communication systems such
as DVB-S2, 802.11n (Wi-Fi allowing MIMO) and 802.16e (Mobile WiMAX) etc.
Turbo codes and LDPC codes were competing against each other in various use
cases and application as these codes both show good performance. Turbo codes gen-
erally have a low encoding complexity and high decoding complexity whereas LDPC
codes have a high encoding complexity but low decoding complexity.
2.2.3. Polar codes
Polar codes were introduced by Arikan [35] in 2009 and they are the first provably
capacity-achieving codes with low encoding and decoding complexities. Polar codes
outperform turbo codes in large block lengths while no error flow is present. The
concept behind channel polarization in polar codes is transforming N copies (trans-
missions) of the channel (e.g., say additive white Gaussian noise channel) with a sym-
metric capacity of I(W ) into extreme channels of capacity close to one or zero. Out
of N channels, I(W ) fraction will become perfect channels and 1 − I(W ) fraction of
channels will become completely noisy channels. Then the information bits are sent
only through good channels while inputs to other channels are made frozen into one
or zero. The amount of channel polarization increases with the block length. Some
of the issues with polar codes are, that the code design is channel independent hence
not versatile for mobile fading channels and the limited application experience due to
its immaturity. Polar codes have better energy-efficiency for large block lengths than
other codes.
19
3. PAPR REDUCTION OF OFDM WAVEFORMS
In this section a low complex method to decrease the OFDM PAPR is suggested. The
proposed technique is explained in Section 3.1 followed by the simulations of the high
PAPR sequences scrambling scheme and analysis of the simulation results in Section
3.2.
3.1. System model
In SLM it is assumed that, U statistically independent alternative sequences, which

represent the same information are generated by some suitable means. The sequence
with the lowest PAPR is selected for transmission. The probability that the lowest
PAPR λl exceeds a certain threshold λ is given by [17]
P r{λl > λ} = (P r{P AP R > λ})U . (6)
As stated in (6), if the threshold λ is set to a reasonable value, there is a signifi-
cant number of symbol sequences with PAPR lower than the threshold after OFDM
modulation. For example, if a system with N = 128 subcarriers is considered, from
Fig. 1, it can be seen that about 85% of sequences will exceed a 7dB threshold and
about 42% sequences will exceed an 8dB threshold. Hence doing a transformation to
all the incoming symbol sequences is not necessary. Therefore a technique to reduce
the PAPR of only the sequences exceeding a pre-set threshold is proposed rather than
searching a minimum PAPR for all the sequences which will reduce the computational
complexity significantly. A predefined array of U scramblers is used to change high
PAPR sequences.
Scrambled bit Sequence
𝑆𝑘 𝑆𝑛
MQAM/MPSK S/P IFFT Selective P/S

Input bits Mapping Scrambling
⋮ ⋮ ⋮
𝑤𝑛
𝑆′𝑘 𝑆′𝑛
MQAM/MPSK P/S Selective FFT S/P

Output bits Demapping Descrambling
⋮ ⋮ ⋮
Figure 2: System model.
A block diagram of the proposed method is shown in Fig. 2 and the algorithm is
given as Algorithm 1. In this method, PAPR of OFDM modulated symbol sequence
20
Algorithm 1 PAPR reduction algorithm

1: procedure PAPR REDUCTION
2: P AP Rmin = inf ,
3: j=0
4: for each incoming symbol sequence d do
5: QAM/QPSK modulation.
6: OFDM modulation
7: Oversampling and PAPR calculation
8: if P AP Rj < P AP Rmin then P AP Rmin = P AP Rj and jmin = j
9: end if
10: if P AP Rj >Threshold then
11: if Scrambler j ≤ Scrambler array length then
12: Scramble the sequence using j th scrambler
13: j++,
14: Goto step 3.
15: else Transmit Smin , jmin
16: end if
17: else Transmit Sj , j
18: end if
19: end for
20: end procedure
is computed for each incoming symbol sequence and it is compared against a pre-
set threshold value. If it exceeds the threshold value, the input symbol sequence is
scrambled by a scrambler selected from the set of scramblers and PAPR for the new
sequence is computed again. If PAPR drops below the threshold, the scrambling will
be terminated and that sequence is transmitted with the index of the scrambler used in
specific sequence for descrambling at the receiver. If the used scrambler does not make
the PAPR to drop below the threshold, next scrambler from the array is used. This will
be repeated until all the scramblers of the scrambler array are used. If it does not drops
below the threshold, the sequence with minimum PAPR will be transmitted along with
the corresponding scrambler index. If the original OFDM signal has a PAPR below the
threshold, zero is sent as scrambler index. At the receiver, scrambler index is checked
and if it is zero, no descrambling is used and signal is demodulated in the usual way.
However, if the scrambling index is greater than zero, corresponding descrambler is
used to descramble the signal before demodulation.
3.1.1. Scrambler and descrambler
A scrambler is used to convert the N length symbol sequence which is drawn from an
alphabet of size Ma to a different S length sequence from the same alphabet. Models
of scrambler and corresponding descrambler used in this technique are depicted in Fig.
3 and Fig. 4 respectively. They consist of S linear feedback shift registers. All adders
are modulo Ma adders. Different scramblers of the same length can be defined by a
21
Scrambled bits
Input bits …
1 2 S-1 S
P1 P2 Ps -1 Ps
Figure 3: Scrambler for an S length sequence.
choosing different sequence of switches to be kept ON which is denoted by Ps . These

scramblers can be defined by the scrambler polynomial P where,
P = [ 1 P1 P2 . . . PS ] ; PS ∈ {1, 0}f or s = 1 · · · S.
Proposed technique has a lower computational complexity than the SLM technique
and will not have a BER degradation given that the scrambler index signal is decoded
accurately. A simple repetition coding can be used to ensure accurate decoding of
the scrambler index signal at the receiver. The subcarrier dedicated to scrambler side
information will be the drawback of this system. However, the effect of this is insignif-
icant when number of subcarriers used are high enough. Also there is a compromise
between PAPR reduction and amount of side information needed to transmit the scram-
bler identification. For example, in quadrature amplitude modulation (QAM), if more
than 3 scramblers are used, 2 subcarriers will be needed to send scrambler indices. In
16QAM, up to 15 scramblers can be used, with overhead of 1 subcarrier for side infor-
mation. If more scramblers are used in order to have better PAPR reduction capability,
more than one subcarrier will have to be dedicated for scrambler index signal, reducing
spectral efficiency.
3.2. Simulations
In order to verify the performance of the high PAPR sequence scrambling in reduction
of PAPR, a base band OFDM system as in Fig. 2 with the number of subcarriers
N = 128 is considered throughout the computer simulations. The oversampling factor
is L is set to 4. The randomly generated input data are modulated by 16QAM . An
array of U = 15 scramblers is used for PAPR reduction throughout the simulations, so
that only one subcarrier is needed for side information.
22
Input bits
1 2 … S
S-1
1 P1 P2 Ps -1 Ps
(-)
…
Descrambled bits
Figure 4: Descrambler for an S length sequence.
Fig. 5 has the CCDF curve of PAPR of the signal after applying proposed technique
with threshold T h = 8dB. It is compared with PAPR curve of the original OFDM. It
can be observed that this method is able to contain maximum PAPR value below 8dB
with 8dB is set as threshold and there is more than 4dB PAPR reduction from original
OFDM signal in this instant.
Fig. 6 depicts the usage of scramblers by symbol sequences. Zero means no scram-
bler is used. Although 15 scramblers are used in this case, it can be seen from the
histogram that
• More than 58% of the sequences do not need any scrambler.
• About another 25% of signals go below threshold after the 1st scrambler.
• About 10% more of signals get below threshold after 2 scramblers.
• Another 4,3% of signals go below threshold after 3 scramblers.
• Only 0,4% sequences need more than 5 scramblers.
• less than 0,007% sequences have to use more than 10 scramblers to achieve
PAPR below threshold.
Therefore, it is clear that this technique reduces the computational complexity of the
original SLM technique as it does not need to compute all K candidate sequences and
choose the optimum.
Fig.7 has the CCDF curve of PAPR of the signal after applying proposed technique
with threshold reduced to T h = 7dB. Again it is compared with PAPR curve of the
original OFDM. It can be seen that the maximum PAPR value reaches about 7.8 dB at
10−5 CCDF exceeding the threshold of 7 dB. From the histogram in the Fig. 8, it can
be seen that only about 13% of the original sequences are below the given threshold
23
10 0
Original OFDM

OFDM with Proposed technique
10 -1
10 -2
10 -3
10 -4
10 -5
10 -6
5 6 7 8 9 10 11 12 13
Figure 5: PAPR comparison of OFDM and the proposed technique, 128 subcarriers
with 8 dB threshold and 15 scramblers.
and there are sequences which have a PAPR value above the threshold even after using
all the scramblers. For those sequences, the sequence which gives the minimum PAPR
output of all scrambled and non-scrambled sequences will be used for transmission.
Thus at this threshold, computational complexity is increased from the previous instant
when T h = 8dB, but, it still is lower compared to SLM. The drawbacks of this method
are reduction in spectral efficiency associated with the side information overhead and
the time delay in selecting the most suitable scrambler for the particular sequence.
These drawbacks can be minimized by setting the threshold to a suitable value and
using a small number of scramblers. This can be a trade off among the amount of
PAPR reduction, computational complexity and side information overhead.
Fig. 8 shows the evaluation of the PAPR reduction capability of the proposed
method, when the number of subcarriers is increased. The threshold value T h is
changed according to the number of subcarriers so that computational complexity does
not increase. Table 3 has a comparison of the proposed method with three other typical
PAPR reduction techniques.
24
60
50
Percentage of sequences
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Scrambler index
Figure 6: Percentage of sequences which has used each scrambler for T h = 8dB. Zero
means no scrambler is used.
Table 3: Comparison of PAPR reduction techniques.
PAPR reduc- PAPR reduc- Implementation Data rate BER degra-

tion technique tion capability complexity loss dation
Clipping& Fil-
≤ 3dB Low No Yes
tering
SLM ≤ 4dB High Yes No
PTS ≤ 4dB High Yes No
Proposed
≤ 4,5dB Low Yes No
Method
25
10 0
Original OFDM
Proposed Technique
10 -1
10 -2
10 -3
10 -4
10 -5
10 -6
4 5 6 7 8 9 10 11 12 13
Figure 7: PAPR comparison of OFDM and the proposed technique, 128 subcarriers
with 7dB threshold and 15 scramblers.
26
14
12
Percentage of sequences
10
0
-2 0 2 4 6 8 10 12 14 16
Scrambler index
Figure 8: Percentage of sequences which has used each scrambler for T h = 7dB. Zero
means no scrambler is used.
27
10 0
10 -1
10 -2
10 -3 N=128,Th=7.5dB
N=256,Th=8dB
N=512,Th=8.5dB
N=1024,Th=9dB
10 -4
5 6 7 8 9 10
Figure 9: Performance of the system with increasing number of subcarriers N, for

threshold T h.
28
4. PERFORMANCE OF SINGLE CARRIER WAVEFORMS IN

MIMO SYSTEMS
In this chapter, performance of single carrier waveforms in MIMO systems are ana-
lyzed in terms of BER. The effects of number of users, channel taps, antennas, and
channel correlation on the BER performance are investigated. Also performance is
investigated in practical channel models for the uplink scenario. In Section 4.1 the
uplink channel is considered and in Section 4.2 the downlink channel is considered. In
addition, the effect of antenna pre-coding on the PAPR of the signal is investigated for
the downlink channel.
4.1. Single carrier - Uplink
The system model of the uplink channel is explained in the first section followed by
the simulations. The performance variation with the number of antennas, number of
users and with channel state information (CSI) error are studied in the simulations.
4.1.1. System model
A system with K users is considered, each equipped with a single antenna. The base
station has M receive antennas. The frequency selective channel has L+1 equi-spaced
taps, which are Rayleigh distributed. Let hk [l, m] be the lth tap in the discrete channel
from k th U T to mth receive antenna at the BS. The notation in [28] is followed.
Let x[n] = (x1 [n] , x2 [n] , ,,,xK [n])T be the transmitted signal vector and w[n] =
(w [n, 1] , w [n, 2] , ,,,w [n, M ])T be the additive noise vector.
The received signal y[n] = (y [n, 1] , y [n, 2] , ,,,y [n, M ])T will be
L
X
y[n] = H[l]x[n − l] + w[n], (7)
l=0
where, H[l] =  
h [l, 1] h2 [l, 1] ... hK [l, 1]
 1 
 
 h1 [l, 2] h2 [l, 2] . . . hK [l, 2] 
.
.. .. ..

..
.
 
 . . . 
 
h1 [l, M ] h2 [l, M ] . . . hK [l, M ]
Considering that channels have L + 1 paths, the transmitted symbol vector at time
n is estimated by received vectors up to time slot n + L. Therefore, assembling the
transmit, receive, and noise vectors as follows;
29
Y = (y [n + L] , y [n + L − 1] , ,,,y [n])T , (8)

X = (x [n + L] , x [n + L − 1] , ,,,x [n − L])T , (9)
W = (w [n + L] , w [n + L − 1] , ,,,w [n])T . (10)
Then defining channel matrix as
 
H[0] H[1] . . . H[L] . . .
 
 
 H[0] H[1] . . . H[L] . . . 
H= 
.. .. ..
. (11)
. . .

 ... 
 
... H[0] H[1] . . . H[L]
Thus,
Y = HX + W. (12)
The maximum likelihood (M L) receiver is
HH Y = HH HX. (13)
Given that the number of base station antennas, M , is large enough, a simplified re-
ceiver is therefore, [23, 28]
1 H
x̃[n] = H Y. (14)
M
Up to now, ideal CSI is assumed to be available at the receiver side. Next, the
scenario where there is uncertainty in the CSI is considered. This can be modeled as
h̃[n, m] = h[n, m] + [n, m], (15)

where, the estimation error or the uncertainty, [n, m] is modeled as a circularly
symmetric complex Gaussian random variable, and is independent for each tap.
4.1.2. Simulations
Here, the performance of the system is investigated with different parameters in terms
of the number of users, antennas, and channel taps. 16QAM is used as the modulation
scheme. The Fig. 10 shows the performance with different number of users with ideal
CSI, whereas the Fig. 11 shows the performance with different number of antennas.
The results indicate that the reduced complexity receiver performs well in the given
antenna regime. This is an important observation since it is not optimal. In Fig. 12,
single and multiple user performance is considered for higher number of taps with
ideal CSI and with errors in CSI (0 dB error in variance). A higher number of users
cause additional interference. The degradation in performance compared to ideal CSI
is seen to increase with the number of users. Having a larger number of antennas can
address both these issues to a certain extent as seen in the results. Even in the case of
OFDM this is the scenario as well, that the performance is compromised with multiple
users operating within the same set of subcarriers.
30
10 0
K=1
K=2
K=5
K = 10
10 -1
BER
10 -2
10 -3
10 -4
-20 -15 -10 -5 0
Eb / N o (dB)
Figure 10: Uplink performance of a system with 300 antennas and 20 channel taps for
variable users.
31
10 0
M = 100
M = 500
M = 1000
10 -1
BER
10 -2
10 -3
10 -4
-20 -15 -10 -5 0
Eb / N o (dB)
Figure 11: Uplink performance of a system with 4 users, 20 channel taps for variable
number of receive antennas.
32
100
1 user, CSI error
1 user
9 users, CSI error
10-1 9 users
SER/BER
10-2
10-3
10-4
-20 -15 -10 -5 0
E b / N o (dB) per Antenna
Figure 12: Uplink performance of a system with 300 receive antennas, 9 users and 0
dB error in CSI.
33
4.2. Single carrier-Downlink
Performance of single carrier-downlink System model of the downlink channel is in-

vestigated in this section.The system model is explained in the first subsection followed
by the simulations of BER for different number of users, antennas and with CSI error.
In the third subsection, the effect of antenna pre-coding on the PAPR of the signal is
studied. BER for the different correlated channels are studied in the last sub section.
4.2.1. System model
In the case of downlink a similar scenario with the base station and users is assumed.
The main difference in this case is that the transmit signal will be precoded based
on the channel response or channel state information at the transmitter (CSIT). Thus
following [27] if the information symbol vector is s[i] = (s1 [i] , s2 [i] , ,,,sK [i])T , the
transmit vector is given by
r L−1
ρf X 1/2
x[i] = Hl Dl s[i + l], (16)
M K l=0
where, ρf is the long term average total power at the BS antennas. The received
signal vector at time i is therefore [27],
L−1
X
y[i] = Dl 1/2 Hl H x[i − l] + n[i]. (17)
l=0
Here, Dl = (dl [1] , d1 [2] , ,,,dl [K]), where, dl [K] ≥ 0, l = 0, 1, ,,L − 1 gives the
power delay profile (PDP) of each user K. Further PDP for each user is normalized
such that,
L−1
X
dl [k] = 1, k = 1, ,,,,, K. (18)
l=0
Due to the orthogonality in the channel matrices [23], as in the case of uplink due to
the large number of antennas at the base station side, the information symbol vector can
be estimated with asymptotic optimality. The issue of CSIT error which is normally
unavoidable in this case in any practical set up is also investigated.
4.2.2. Simulations
Here, the performance of the system is investigated with different parameters in terms
of the number of users, antennas, and channel taps. The degradation in performance
compared to single tap is not very significant implying that the reduced complexity re-
ceiver is performing quite well. Fig. 13 shows the performance with a different number
34
of antennas under ideal CSI. The performance in the presence of channel uncertainty
is investigated in Fig. 14. The degradation in the presence of CSIT error (0dB) seems
quite significant.
These also show a similar performance to the uplink curves in the sense that there
needs to be a larger number of antennas to compensate for interference and non-
idealities. As a general conclusion, therefore, one can say that the single carrier system
performs satisfactorily in the massive MIMO paradigm expected to be strengthened by
the application in the mmWave frequencies.
100
10-1
BER
10-2
K=1
K=2
K=5
10-3 K = 10
K = 20
10-4
-20 -15 -10 -5 0 5 10
E b / No (dB)
Figure 13: Downlink performance of a system with 500 antennas and 5 channel taps
for variable users.
4.2.3. PAPR Analysis
As mentioned previously, to fairly compare an OFDM system with a downlink single-

carrier system, the issue of PAPR needs to be considered. This is due to the fact
that the channel is frequency selective channel and pre-coding is carried out to match
the transmit symbols to the channel, which unlike in the case of flat fading, is a bit
more involved due to multiple taps. To better approximate the PAPR of continuous
time OFDM signals, OFDM samples are taken by several times oversampling [5, 13],
where, a survey of methods dealing with this issue is presented. Following the same
arguments and assuming a certain carrier frequency fc PAPR for single-carrier systems
can be calculated.
35
100
1 user
1 user, CSI error
4 user
4 user, CSI error
BER 10-1
10-2
10-3
10-4
-20 -18 -16 -14 -12 -10 -8
Eb/No (dB)
Figure 14: Downlink performance of a system with 300 transmit antennas and 0 dB
error in CSI.
Using (16) as the transmit signal, the PAPR is given by
max0≤n≤N [|x [n] |2 ]

P AP R{x [n]} = . (19)
E [|x [n] |2 ]
CCDF for PAPR is the probability of OFDM symbols with PAPR exceeding some
threshold λ. In the case of OFDM a typical plot is shown in Fig. 15. The Fig. 16
shows the CCDF plot of single-carrier PAPR. It can be seen that PAPR increases as the
number of channel taps increases. As the Fig. 16 shows the PAPR value can change
significantly when a pre-coded downlink signal is considered. The values are seen to
vary from 6 − 10 dB without any other modifications such as windowing etc.
In the case of OFDM systems this can be as high as 12 dB as seen in Fig. 15 [5]
and normally some methods are employed for the reduction of these values. Similar
approaches can also be considered for the single carrier system which can lower the
values to 4 − 8 dB range [19]. One can therefore say there is still a significant PAPR
advantage in using single-carrier systems in the downlink coupled with the reduced
complexity receiver.
36
10 0
Ls=16
Ls=4
Ls=2
-1
10 Ls=1
CCDF
10 -2
10 -3
2 3 4 5 6 7 8 9 10 11 12
PAPR ( dB )
Figure 15: PAPR performance of OFDM for variable oversample factor Ls .
10 0
L=2
L=9
L=20
10 -1
CCDF
10 -2
10 -3
0 2 4 6 8 10 12
PAPR(dB)
Figure 16: PAPR performance with variable channel taps L.
37
4.2.4. Correlated Channels
In the previous cases, another effect which is normally present in multiple antenna
systems were not considered. This is the correlation among antennas which also is a
significant factor as the number of antennas increases. Therefore, several cases with
two of the possibilities; exponential and uniform linear array (ULA) correlation were
considered in the following sections.
Exponential correlation
Let a be the exponential correlation (Exp) coefficient. Then exponential correlation
matrix of a channel with M BS antennas will be
 
M −1
1 a ... a
 

 a M −2 
1 ... a 
t= ... ... .
. (20)

 ..  
 
M −1
a ... 1
Uniform linear array

In a typical ULA, where, antennas are placed long a line with length D, the distance
between mth and mth0 antennas will be [29]
(m − m0 )D
d= . (21)
M −1
Then, the spatial correlation between any two antennas with distance d will be
d J0 (2πd)
ρ( ) = . (22)
λ λ
where, J0 (.) is the zeroth order Bessel function of the first kind.
Simulations
Figs. 17 and 18 show the BER curves for different correlated channels with quadra-
ture phase shift keying (QPSK) and 8 phase shift keying (8PSK) respectively with
clear degradation for higher correlation. The correlation is unavoidable given the cases
where the close spacing of antennas is likely or needed even in the mmwave region.
The application of suitable channel codes may be one way to address this [36, 37].
4.3. Practical Channel Models
It is also important to evaluate the performance of a single carrier system with a mea-
sured channel model, in this case at 10 GHz as well to get a more realistic idea about
the performance. Here a Quadriga based model modified as given in [24, 25] is em-
ployed. The coefficients of the channel matrix are in the order of 10−5 − 10−9 which
38
10 0
10 -1
BER
10 -2 Ind,0
Exp,0.9
Exp,0.5
ULA,40
10 -3 ULA,30
ULA,20
Exp,0.3
10 -4
-20 -15 -10 -5 0
E b / No (dB) per Antenna
Figure 17: Downlink QPSK performance of a system with 200 transmit antennas, 5
channel taps, and 4 users for correlated channels.
39
100
10-1
Ind,0
BER
10-2 Exp,0.9
Exp,0.5
ULA,40
ULA,30
10-3
ULA,20
Exp,0.3
10-4
-20 -15 -10 -5 0
Eb / No (dB) per Antenna
Figure 18: Downlink 8PSK performance of a system with 200 transmit antennas, 5
channel taps, and 4 users for correlated channels.
40
100
M = 20
M = 50
M = 100
M = 200
10-1
BER
10-2
10-3
10-4
-40 -30 -20 -10 0 10
Eb / No (dB) per Antenna
Figure 19: Uplink QPSK performance for a system with 9 channel taps, 4 users and
variable number of receive antennas in a practical channel model.
is very low. Thus in our case the channel matrix is normalized across M BS antennas
to produce coefficients in the order of 10−1 .
4.3.1. Simulations: Uplink
Figs. 19 and 20 show the BER curves for QPSK and 8PSK modulated signals in
practical channel models. QPSK modulated signals have acceptable performance with
50 BS antennas for four user case. In the case of 8PSK with 100 BS antennas, a higher
number of users will degrade the performance.
41
100
K=1
K=2
K=5
10-1 K = 10
BER
10-2
10-3
10-4
-25 -20 -15 -10 -5 0 5
E b / N o (dB) per Antenna
Figure 20: Uplink performance of an 8PSK system with 200 receive antennas and
variable users in a practical channel model.
42
5. CHANNEL CODING
Candidate coding schemes for the eMBB scenario of 5G NR are evaluated in this
chapter. In the first section, some performance aspects of the candidate coding schemes
are discussed along with the decoding algorithms. In the second section, simulation
parameters are stated, followed by the simulation results in the third section.
5.1. Candidate Coding Schemes
5.1.1. Turbo codes
The 3GPP LTE standard specification parameters [38] are used for the simulations [41]
and the discussions of turbo codes in this thesis. Each convolutional turbo code encoder
outputs two streams, one systematic stream and one parity stream. Input information
bits to the second encoder is fed after interleaving the input bit stream.
For iterative decoding of the parallel concatenated encoding scheme, the turbo de-
coder uses a MAX-Log-MAP algorithm as the constituent decoder component. The
internal interleaver of the decoder is identical to the one the encoder uses.
Since there are three output streams; a systematic bit stream and two parity streams
in the encoder, coding rate for this LTE system is 1/3. Other higher rates are achieved
through puncturing of some parity bits. Assume R and Nb are code rate and coded
block length respectively, for all the coding schemes. Assuming the decoder has a
memory length of Mb bits for the component code and I number of iterations, turbo
codes has 16 · I · R · Nb · 2Mb additions and 8 · I · R · Nb · 2Mb logical operations. [36]
For comparison, computational complexity of each coding scheme is expressed in
terms of equivalent additions. Number of equivalent additions for different operations
are stated in Table 4 [36].
5.1.2. LDPC codes
Quasi-Cyclic LDPC (QC-LDPC) codes are used for simulations and discussions
throughout this thesis. A QC-LDPC matrix is characterized by the parity check matrix
H. It consists of small square blocks which are the zero matrix or circulant permuta-
tion matrices. An advantage of QC-LDPC code is that it supports variable code lengths
which can be easily obtained by adjusting the circulant permutation matrices in H. Sum
product algorithm (SPA) is the optimal decoder for LDPC, and min-sum algorithm is
a sub optimal algorithm with reduced computational complexity.
While QC-LDPC codes allows reasonable flexibility in code length, puncturing is
used to achieve rate-adaptive codes. Assuming dv and dc as the average variable and
check degrees of the LDPC parity check matrix respectively and Pb as the number of
parity bits, min-sum LDPC decoder has I ·(2Nb ·dv +2Pb ) additions and I ·(2dc −1)·Pb
max processes [36]. Offset min-sum (OMS) decoder is a reduced complexity version
of min-sum decoder where algorithm converges in a smaller number of iterations and
it is used as the decoder for the simulations in this thesis.
43
Table 4: Number of equivalent additions per operation.
Operation Number of Equivalent

additions per operation
Addition, Subtraction 1
Multiplication, Division 1
Comparison 2
Maximum, Minimum 1
Parallel list 1
5.1.3. Polar codes
Polar code construction can be done recursively via Kronecker products and have an
encoding complexity of O(n log n). Complexity of decoding is also the same via suc-
cessive cancellation (SC) decoding. However, the SC decoder itself is not sufficient
to achieve competitive performance as other coding schemes. Hence list decoding
is incorporated into SC decoder which results in the SC list (SCL) decoder [39]. In
addition a cyclic redundancy check bits (CRC) are used to further enhance the code
performance at an expense of increased complexity.
Construction of the polar codes requires knowledge about specific channel condi-
tions, hence not versatile. While different code rates can be achieved by changing the
number of frozen bits in polar codes, a puncturing scheme is needed to obtain some
code lengths. SC decoder has a complexity of only Nb ·Log2 Nb additions where as SCL
decoder with list size of Lp has a complexity of Lp ·Nb ·log2 Nb +(Nb −Pb )·Lp ·Log2 Lp
additions [36].
5.2. Simulations
The candidate coding schemes are simulated with QAM and AWGN channel in the
first set of simulations. Rayleigh fading channel with AWGN is considered with CSI
available in the second set of simulations.
Coding parameters are given in Table 5 and performance is evaluated in terms of
BLER and BER. For turbo codes LTE-Advanced encoder with two 8-state constituent
encoders and a block interleaver is used. At the receiver, the scaled MAX-Log-MAP
decoder with scaling factor of 7, and iteration count of 8 is used. The code rate of
original turbo code is 1/3 and higher rate codes are obtained through puncturing.
In LDPC code simulations, parity check matrices specified in IEEE 802.11n are used
for encoding. The OMS decoder with 0.3 as the offset value and 50 iterations is used
at the receiver. Different code lengths are obtained by changing the size of base matrix
of the parity check matrix.
Polar codes construction is based on construction for AWGN channel as suggested
in [40]. The performance is simulated using both SC decoder and CRC aided-SCL
44
Table 5: Parameters for simulations.
Parameters Turbo LDPC Polar

Channel AWGN, Rayleigh Fading
Modulation QPSK
Code lengths 128, 512, 1024, 2048, 4096
Code rates 1/3 , 1/2 , 2/3
CRC NA NA 8
SC,
max-
Offset- CRCA-
Decoder log-
min-sum SCL,
MAP
L=8
decoder (CRCA-SCL) with list size of 8 and 8 CRC check bits for all the code rates
and lengths.
Algorithmic complexity of the decoders used for each coding scheme is obtained for
block lengths 128 and 4096 in Table 6. This also shows the percentage of complexity
with respect to turbo code in the percentage column.
5.3. Simulation results in AWGN channel
In Fig. 21 to 24, BLER and BER of the coding schemes are plotted against the energy
per bit to noise power spectral density ratio (Eb /N0 ) for Turbo, LDPC and Polar codes
for different block lengths and coding schemes.
In summary, at the block-length of 128 bits (Fig. 21), polar codes with CRCA-SCL
decoder outperforms turbo and LDPC codes for all the 3 rates. As the block length
increases LDPC and Turbo performance comes close to polar codes with CRCA-SCL
decoder performance. For example, at the block length of 512 bits (Fig. 22) , po-
lar codes with CRCA-SCL decoder has similar performance as Turbo for rate 1/3 in
BLER. But turbo outperforms polar codes in BER at this block length and rate.
When the block length increases as in Fig. 23, turbo codes outperform polar codes
with CRCASCL decoder and LDPC for rate 1/3. For rates 1/2 and 2/3 LDPC has
better performance than polar or turbo. It should be noted that LDPC codes performs
relatively well even without the use of CRC bits, hence this performance can be further
enhanced by adding CRC at cost of increased complexity.
In terms of algorithmic complexity as computed in the Table 6, turbo and LDPC
codes shows similar complexity in rate 1/3. But for all the other rates turbo codes has
relatively higher complexity. SC decoder of Polar codes has the lowest complexity for
all rates and SCL decoder with list size of 8 has about 10 times the complexity of SC
45
Table 6: Decoding complexity of coding schemes.
Complexity (×103 ) Percentage

Block Coding
length Scheme 1/3 1/2 2/3 1/3 1/2 2/3
Turbo 65.5 98.3 131.1 100% 100% 100%
LDPC 66.0 57.2 48.5 100.7% 58% 37%
128
Polar SC 1.0 1.0 1.0 1.5% 1.0% 0.8%
Polar
11.0 11.0 11.0 16.8% 11.2% 8.4%
SCL
Turbo 2097.1 3145.7 4194.3 100% 100% 100%
LDPC 2112.0 1831.9 1551.9 100.7% 58.2% 37.0%
4096
Polar SC 53.2 53.2 53.2 2.5% 1.7% 1.3%
Polar
524.0 524.0 524.0 25.0 % 16.7% 12.5%
SCL
decoder. It should be noted that these complexities are based on the assumptions in the
Table 6 and actual implementation complexity may differ in practical scenarios.
5.4. Coding performance in fading channel
5.4.1. System model
An uncorrelated fading channel with AWGN is used as the channel. Fading gain h
is distributed as Rayleigh fading and CSI is assumed to be available at the receiver.
Quadrature amplitude modulation (QAM) is the modulation scheme used at the trans-
mitter. Since polar codes are channel dependent, code design for the fading channel is
in [42] Soft decision decoding is used in decoders of all three coding schemes using
LLR.
LLR calculation for fading channel is as follows. If M is the modulation alphabet
size, then the symbol length S is N/log2 (M ). ys and xs are assumed to be sth received
and transmitted symbols respectively, and hs is assumed to be the complex scaling
factor for sth symbol corresponding to Rayleigh channel, and ns as the AWGN channel
with ns ∼ N (0, σ 2 ). Then received signal ys can be written as
ys = h s xs + ns . (23)
√
If energy of QAM modulated symbol is Ec , then the LLR of the received symbols
in AWGN channel can be computed as [43]
√
awgn 2 Ec
LLR2s−1 = ,real(ys ), (24)
σ2
46
√
awgn 2 Ec
LLR2s = ,imag(ys ). (25)
σ2
For the fading channels, fading coefficient is incorporated into the LLR as [44]
√
f ading 2 Ec
LLR2s−1 = ,real(hs ys ), (26)
σ2
√
f ading 2 Ec
LLR2s = ,imag(hs ys ). (27)
σ2
5.4.2. Simulations
The results are presented from Fig. 25 to Fig. 30. In each plot BLER or BER is shown
against the energy per bit to noise power spectral density ratio (Eb /N0 ). Fig. 25 to
Fig. 27 present BER and BLER performances of the three coding schemes in Rayleigh
fading channel for block lengths of 128, 512 and 1024 bits. Fig. 28 to Fig. 30 illustrate
AWGN and fading channel performance of turbo, LDPC and polar codes, respectively.
The results indicate that polar codes perform better for lower block lengths compared
to the other two coding schemes, especially at lower code rates as in Fig. 25. As the
block lengths increase from Fig. 26 to Fig. 27, LDPC and turbo coding show better
performances compared to polar coding. At short block lengths, Turbo and LDPC
codes have almost similar performances. LDPC shows better performance than turbo
codes at longer block lengths for both BLER and BER, especially in higher code rates
such as 2/3, LDPC codes exhibit better performance compared to other two schemes.
It is important to note that a similar trend was observed in AWGN channels with
these three coding schemes [37]. This fact can be useful for obtaining an estimate
for the performance of a coding scheme in a fading channel given the performance of
it in AWGN channel or vise versa. Furthermore it can be seen from Fig. 28 to Fig.
30 that all three coding schemes have similar performance difference from AWGN to
Rayleigh fading channels.
47
100
Turbo 1/3
LDPC 1/3
Polar 1/3 SCL with L=8
Turbo 1/2
10-1 LDPC 1/2
Turbo 2/3
LDPC 2/3
BLER
10-2 Polar 2/3 SCL with L=8
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
Turbo 1/2
10-1 LDPC 1/2
Turbo 2/3
LDPC 2/3
BER
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
Figure 21: BLER and BER performance of coding schemes at a coded block length of
128 bits and QAM modulation for different code rates.
48
100
Turbo 1/3
LDPC 1/3
10-1 Turbo 1/2
LDPC 1/2
BLER
Turbo 2/3
10-2 LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
10-1 Turbo 1/2
LDPC 1/2
BER
10-2 Turbo 2/3

LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
49
100
Turbo 1/3
LDPC 1/3
10-1 Turbo 1/2
LDPC 1/2
BLER
Turbo 2/3
10-2 LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
10-1 Turbo 1/2
LDPC 1/2
BER
Turbo 2/3
10-2
LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
50
100
Turbo 1/3
LDPC 1/3
Turbo 1/2
LDPC 1/2
BLER

10-2 Turbo 2/3
LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
Turbo 1/2
LDPC 1/2
BER

10-2 Turbo 2/3
LDPC 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
51
100
10-1
Turbo 1/3
BLER
LDPC 1/3
10-2
Polar 1/3
Turbo 1/2
LDPC 1/2
Polar 1/2
10-3
Turbo 2/3
LDPC 2/3
Polar 2/3
10-4
0 1 2 3 4 5
E b /N 0 (dB)
100
10-1
10-2 Turbo 1/3

BER
LDPC 1/3
Polar 1/3
10-3 Turbo 1/2
LDPC 1/2
Polar 1/2
10-4 Turbo 2/3
LDPC 2/3
Polar 2/3
0 1 2 3 4 5
E b /N 0 (dB)
128 bits and QAM modulation for code rates 1/3, 1/2 and 2/3.
52
100
Turbo 1/3
LDPC 1/3
Polar 1/3
10-1 Turbo 1/2
LDPC 1/2
Polar 1/2
BLER
Turbo 2/3
10-2
LDPC 2/3
Polar 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
Polar 1/3
10-1
Turbo 1/2
LDPC 1/2
10-2 Polar 1/2
BER
Turbo 2/3
LDPC 2/3
10-3 Polar 2/3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
53
100
Turbo 1/3
LDPC 1/3
Polar 1/3
10-1 Turbo 1/2
LDPC 1/2
Polar 1/2
BLER
10-2 Turbo 2/3

LDPC 2/3
Polar 2/3
10-3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
100
Turbo 1/3
LDPC 1/3
10-1 Polar 1/3
Turbo 1/2
LDPC 1/2
10-2 Polar 1/2
BER
Turbo 2/3
LDPC 2/3
10-3
Polar 2/3
10-4
0 1 2 3 4 5 6
E b /N 0 (dB)
54
100
10-1
BLER
10-2
Turbo 1/3-AWGN
Turbo 1/3-fading
Turbo 1/2-AWGN
10-3
Turbo 1/2-fading
Turbo 2/3-AWGN
Turbo 2/3-fading
10-4
0 1 2 3 4 5
E b /N 0 (dB)
100
Turbo 1/3-AWGN
Turbo 1/3-fading
10-1 Turbo 1/2-AWGN
Turbo 1/2-fading
Turbo 2/3-AWGN
10-2 Turbo 2/3-fading
BER
10-3
10-4
0 1 2 3 4 5
E b /N 0 (dB)
Figure 28: BLER and BER performance of Turbo coding in AWGN and fading chan-
nels at a coded block length of 512 bits and QAM modulation for code rates 1/3, 1/2
and 2/3
55
100
LDPC 1/3-AWGN
LDPC 1/3-fading
LDPC 1/2-AWGN
LDPC 1/2-fading
10-1 LDPC 2/3-AWGN
LDPC 2/3-fading
BLER
10-2
10-3
0 1 2 3 4 5
E b /N 0 (dB)
100
Turbo 1/3-AWGN
Turbo 1/3-fading
Turbo 1/2-AWGN
10-1 Turbo 1/2-fading
Turbo 2/3-AWGN
Turbo 2/3-fading
BER
10-2
10-3
10-4
0 1 2 3 4 5
E b /N 0 (dB)
Figure 29: BLER and BER performance of LDPC coding in AWGN and fading chan-
and 2/3
56
100
10-1
BLER
10-2
Polar 1/3-AWGN
Polar 1/3-fading
Polar 1/2-AWGN
10-3 Polar 1/2-fading
Polar 2/3-AWGN
Polar 2/3-fading
10-4
0 1 2 3 4 5
E b /N 0 (dB)
100
10-1
10-2
BER
Polar 1/3-AWGN
10-3 Polar 1/3-fading
Polar 1/2-AWGN
Polar 1/2-fading
10-4
Polar 2/3-AWGN
Polar 2/3-fading
0 1 2 3 4 5
E b /N 0 (dB)
Figure 30: BLER and BER performance of Polar coding in AWGN and fading chan-
and 2/3
57
6. CONCLUSION
New air-interface technologies, waveforms and coding schemes for the 5G NR are dis-
cussed in this thesis. First, a novel algorithm is proposed to reduce PAPR problem of
OFDM waveforms to be considered in 5G RAT. The proposed selective mapping tech-
nique uses an array of scramblers and it is evaluated through simulations. It selects
the input symbol sequences which result in a PAPR greater than a threshold value and
changes those by sending them through scramblers until PAPR gets under the thresh-
old. Search for a low PAPR sequence will be terminated when a sequence is generated
with PAPR lower than the preset threshold. Hence computational complexity of this
method is significantly decreased from the conventional SLM method. It can be seen
from the simulation results that the proposed technique is effective in reducing PAPR
from more than 4 dB of the original signal for an arbitrary number of subcarriers. Also
we can find a compromise between PAPR reduction and computational complexity by
selecting an appropriate threshold value and the overhead of scrambler index side in-
formation can be used for trade off in PAPR reduction capability by choosing the size
of the scrambler array to a suitable value.
Next we analyzed performance of Single carrier waveforms in MIMO systems as
an alternative to OFDM. Performance of a single carrier massive MIMO system is
presented for both uplink and downlink with single user and multiple user cases. A
variety of channel configurations were investigated, from errors in channel estimate,
correlated to measured channels. As opposed to traditional single carrier systems we
did not consider any cyclic prefix or frequency domain equalization. A simplified re-
ceiver structure is used. It is shown to be effective as the number of antennas increases.
The degradation in performance is evident with the number of users. However, this will
be the same for any other multi-user system as well. The PAPR performance shows
that higher values are obtained with a higher number of taps in the case of the downlink
due to beamforming needed. On the other hand in the case of uplink the main process-
ing will be done in the base station. Therefore the proposed single carrier system is a
good match there. In general as an access method for mmWave region for new radio
in 5G, this is seen to be a viable alternative to multicarrier techniques currently being
standardized for the lower frequency range.
Finally we investigated several performance aspects of candidate coding schemes
for 5G NR. Performance is analyzed in terms of BER and BLER in both AWGN and
fading channels. For block lengths around 100 bits, turbo codes have better perfor-
mance than polar and LDPC codes. However, LDPC codes exhibit relatively good
performance in all the coding rates and block lengths. Furthermore we should note
that LDPC codes show this performance without the aid of CRC. Hence LDPC per-
formance can be further enhanced by using a CRC. On the other hand, polar codes
have the benefit of not having an error floor compared to LDPC and turbo, both of
which have error floors. However, to achieve the optimal performance of polar codes,
code construction should be done based on the channel. Hence, polar codes are not yet
versatile. Further research should be conducted to achieve channel independent code
design.
Although SC decoder of polar codes shows the lowest complexity, CRC aided SCL
decoder exceeds the complexity of LDPC codes and turbo codes. Actual costs of SLC
decoder is uncertain due to lack of implementations. There are many other factors
58
to be considered when choosing a coding scheme, such as latency for encoding and
decoding, energy efficiency, and area efficiency. In current implementations, LDPC
codes show relatively good performance in both area and energy efficiency. Turbo
codes consume highest energy per bit and has a very low area efficiency. Turbo codes
have good energy efficiency but low area efficiency. Exact area and energy efficiency
of SCL decoder of polar codes are not known yet due to lack of implementations. By
comparing code performance in AWGN and fading channels, it can be seen that fading
channel performance follows the same trend as the AWGN channel performance.
59
7. RECOMMENDATIONS FOR FUTURE
Performance of single carrier waveforms was analyzed in AWGN, correlated, and prac-
tical channel models in Chapter 2. However,the effect of hardware imperfections also
need to be taken into account in order to get a more practical performance. Further-
more, other performance aspects such as sum rate, efficient power allocation tech-
niques should also be investigated.
Turbo, LDPC and Polar codes were evaluated in terms of BLER/BER performance
and computational complexity in Chapter 3. We considered max-log-MAP, Offset-
min-sum and CRCA-SCL decoding algorithms respectively for each coding scheme.
In addition, other efficient decoding algorithms need to be taken into account. For
example, sequential decoding algorithm for polar decoding should be compared with
SCL decoder to choose an efficient decoding scheme for polar codes. Also, polar codes
need to be researched further in the areas of performance of higher order modulation
schemes, design of interleavers to be used with polar codes, and cost of hardware
implementation.
60
8. REFERENCES
[1] N. Rajatheva, S. Suyama, W. Zirwas, L. Thiele, G. Fodor, A. Tolli, E. De Car-

valho, J. H. Sorensen, "Massive multiple input multiple output (MIMO) systems,"
in 5G Mobile and Wireless Communications Technology, Cambridge University
Press, Chapter 8, pp. 208-247, June 2016.
[2] "5G Use Cases and Requirements White Paper," in Nokia White Paper, July
2014, [online]. Available: http://resources.alcatel-lucent.com/
asset/200010.
[3] "The Road to 5G: Drivers, Applications, Requirements and Technical Develop-
ment," in Global Mobile Suppliers Association(GSA) Executive Report, Novem-
ber 2015, [online]. Avalable: http://www.huawei.com/minisite/5g/
img/GSA_the_Road_to_5G.pdf.
[4] J.G. Andrews, S. Buzzi, W. Choi, S.V. Hanly, A. Lozano, A.C.K. Soong, and
J.C. Zhang, "What Will 5G Be?," in IEEE Journal on Selected Areas in Commu-
nications, vol. 32, no. 6, pp. 1065-1082, June 2014.
[5] T. Jiang and Y. Wu, "An Overview: Peak-to-Average Power Ratio Reduc-
tion Techniques for OFDM Signals," in IEEE Transactions on Broadcasting ,
vol. 54,no. 2, pp. 257-268, 2008.
[6] S. Shepherd, J. Orriss, and S. Barton, “Asymptotic limits in peak envelope power
reduction by redundant coding in orthogonal frequency division multiplex mod-
ulation,” in IEEE Transactions in Communications, vol. 46, no. 1, pp. 5–10, Jan.
1998.
[7] X. Li and L. J. Cimini, "Effects of Clipping and Filtering on the performance of

OFDM," in IEEE Electronic Letters, vol. 2, no. 5, pp. 131-133, 1998.
[8] J. Armstrong, "Peak-to-Average power reduction for OFDM by repeated clipping

and frequency domain filtering," in IEEE Electronic Letters, vol. 38, no. 5, pp.
246-247, February 2002.
[9] A. E. Jones, T. A. Wilkinson, S. K. Barton, "Block coding scheme for reduction

of peak to mean envelope power ratio of multicarrier transmission schemes," in
IEEE Electronics Letters, vol. 30, no. 25, pp. 2098-2099, 1994.
[10] D. Wulich, “Reduction of peak to mean ratio of multicarrier modulation using

cyclic coding,” in IEE Electronics Letters, vol. 32, no. 29, pp. 432-433, Feb.
1996.
[11] T. Jiang and G. Zhu, "Complement block coding for reduction in peak-to-average
power ratio of OFDM signals" in IEEE Communications Magazine, vol. 43,no. 9,
pp. s17-s22, September 2005.
[12] S. H. Muller and J. B. Huber, "OFDM with reduced peak-to-average power ra-
tio by optimum combination of partial transmit sequences,” in IEE Electronics
Letters, vol. 33, no. 5, pp. 36-69, Feb. 1997.
61
[13] L. Yang, R.S. Chen,Y. M. Siu,K. Soo, “PAPR reduction of an OFDM signal by
use of PTS with low computational complexity,” in IEEE Transactions on Broad-
casting, vol. 52, no. 1, pp. 83–86, Mar. 2006.
[14] J. Tellado,“Peak to Average Power Ratio Reduction for Multicarrier Modulation,”
in PhD thesis,University of Stanford, 1999.
[15] T. Jiang,Y. Yang,Y. Song, "Exponential companding technique for PAPR reduc-
tion in OFDM systems," in IEEE Transactions on Broadcasting, vol. 51, no. 2,
pp. 244-248, June 2005.
[16] H. Gacanin, S. Takaoka, F. Adachi, "Generalized OFDM for bridging between
OFDM and single-carrier transmission," in The Ninth International Conference
on Communication Systems, ICCS 2004, pp. 145-149, Sept. 2004.
[17] R. W. Bauml, R. F. H. Fisher,J. B. Huber, “Reducing the Peak-to-Average Power
Ratio of Multicarrier Modulation by Selected Mapping” in IEE Electronics Let-
ters, vol. 32, no. 22, pp. 2056-2057, Oct.1996.
[18] A. A. Abouda, "PAPR reduction of OFDM signal using turbo coding and selec-
tive mapping," in Proceedings of the 6th Nordic Signal Processing Symposium,
NORSIG 2004, pp. 248-251, June 2004.
[19] H. Gamage, N. Rajatheva and M. Latva-aho, "High PAPR Sequence Scrambling
for Reducing OFDM Peak-to-Average Power Ratio" in European Wireless 2016;
22th European Wireless Conference, Oulu, April 2016.
[20] "3GPP Contributions[Online]". Available: http://www.3gpp.org/
DynaReport/TDocExMtg--R1-86--31663.htm.
[21] Verizon, "Verizon 5G Technical Forum," 2016, [Online]. Available: http://
www.5gtf.org.
[22] M. Tercero, P. Wrycza, A. Amah, J. Widmer, M. Fresia, V. Frascolla, J.Lorca,
T. Svensson, M. Hamon, S. D. Roblot, A. Vijay, M. Peter, V. Sgardoni,
M. Hunukumbure, J. Luo and N. Vucic, "5G systems: The mmMAGIC project
perspective on use cases and challenges between 6–100 GHz," in IEEE Confer-
ence on Wireless Communications and Networking WCNC, April 2016.
[23] T. L. Marzetta, "Noncooperative Cellular Wireless with Unlimited Numbers of
Base Station Antennas," in IEEE Transactions on Communications, vol. 9, no. 11,
pp. 3590-3600, November 2010.
[24] A. Roivainen, P. Kyösti, C. F. Dias, V. Hovinen, N. Tervo, M. Sonkki, and
M. Latva-aho, "Geometry-Based Stochastic Channel Model Parametrization for
Urban Small Cells at 10 GHz," in IEEE Transactions on Antennas and Propaga-
tion, under revision, May 2016.
[25] A. Roivainen, C. F. Dias, N. Tervo, and M. Latva-aho, "Geometry-Based Stochas-
tic Channel Model for Two-Story Lobby Environment at 10 GHz," in IEEE
Transactions on Antennas and Propagation, vol. 64, no. 9, pp. 3990-4003, Sept.
2016.
62
[26] P. Yang et al., "Single-Carrier SM-MIMO: A Promising Design for Broadband

Large-Scale Antenna Systems," in IEEE Communications Surveys and Tutorials,
vol. 18, no. 3, pp. 1687-1716, Third Quarter 2016.
[27] A. Pitarokoilis, S. K. Mohammed and E. J. Larsson, "On the Optimality of Single-

Carrier Transmission in Large-Scale Antenna Systems," in IEEE Wireless Com-
munications Letters, vol. 1, no. 4, pp. 276-279, Aug. 2012.
[28] Y. Sheng, Z. Tan and G. Ye Li, "Single-Carrier Modulation with ML Equaliza-

tion for Large-Scale Antenna Systems over Rician Fading Channels," in IEEE
Conference on Acoustic, Speech and Signal Processing, ICASSP, pp. 5759-5763,
September 2014.
[29] Y. Liu, G. Ye Li, Z. Tan and D. Qiao, "Performance Analysis of Single-

Carrier Modulation with Correlated Large-Scale Antennas" [Online]. Available:
https://arxiv.org/pdf/1508.00109.
[30] Yu Zhu, Peng Zhe, Hong Zhou and D. Huang, "Robust Single Carrier Frequency
Domain Equalization with Imperfect Channel Knowledge," in IEEE Transactions
on Wireless Communications, vol. 15, no. 9, pp. 6092-6103, September 2016.
Turbo
[31] C. Berrou, A. Glavieux and P. Thitimajshima, "Near Shannon limit error-

correcting coding and decoding: Turbo-codes," in IEEE International Confer-
ence on Communications, 1993.
[32] A. Goldsmith, "Chapter 8: Coding for wireless channels," in Wireless Communi-

cations, Cambridge University Press, 2005.
[33] R.G. Gallarger, "Low Density Parity Check codes," in IRE Transactions on In-
formation Theory, 1962.
[34] D.J.C. MacKay and R.M. Neal, "Near Shannon limit performance of low density
parity check codes," in Electronic Letters, Vol. 33, No. 6, March 1997.
[35] E. Arikan, "Channel Polarization: A Method for Constructing Capacity-

Achieving Codes for Symmetric Binary-Input Memoryless Channels," in IEEE
Transactions on Information Theory, vol. 55, no. 7, pp. 3051-3073, July 2009.
[36] M. Sybis, K. Wesolowski, K. Jayasinghe, V. Venkatasubramanian, and V. Vuka-

dionovic, "Channel coding for ultra-reliable low-latency communication for 5G
systems," in IEEE Vehicular Technology Conference(VTC Fall 2016), 2016.
[37] H. Gamage, N. Rajatheva and M. Latva-aho, "Channel Coding for Enhanced Mo-
bile Broadband Communication in 5G Systems," in submitted to European Con-
ference on Networking and Communications, 2017.
[38] ETSI, "Evolved Universal Terrestrial Radio Access (E-UTRA) Multiplexing and
Channel Coding," in 3GPP Technical specification, version 12.6.0, Release 12.
63
[39] I. Tal and A. Vardy, "List Decoding of Polar Codes," in IEEE Transactions on
Information Theory, vol. 61, no. 5, pp. 2213-2226, May 2015.
[40] H. Vangala, E. Viterbo and Y. Hong , "A Comparative Study of Polar Code Con-
structions for the AWGN Channel," in IEEE Transactions on Information Theory,
vol. 61, no. 5, pp. 2213-2226, Jan. 2015.
[41] C. Studer, C. Benkeser, S. Belfanti, and Q. Huang, "Design and Implementation

of a Parallel Turbo-Decoder ASIC for 3GPP-LTE," in IEEE Journal of Solid-
State Circuits, 2011.
[42] A.Bravo-Santos, "Polar Codes for the Rayleigh Fading Channel," in IEEE Com-
munication Letters, vol. 17, issue. 12, December 2013.
[43] S. Le Goff, A. Glavieux and C. Berrou, "Turbo-codes and high spectral efficiency
modulation," in IEEE International Conference in Communications, vol. 2, pp.
645-649, May 1994.
[44] R. Yazdani and M. Ardakani, "Optimum Linear LLR Calculation for Iterative De-
coding on Fading Channels," in IEEE International Symposium on Information
Theory (ISIT), June 2007.

Wave Forms and Channel Coding For 5

Uploaded by

Copyright:

Available Formats

Wave Forms and Channel Coding For 5

Uploaded by

Document Information

Original Description:

Original Title

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

Wave Forms and Channel Coding For 5

Uploaded by

Copyright:

Available Formats

FACULTY OF INFORMATION TECHNOLOGY AND ELECTRICAL ENGINEERING

Heshani Niyagama Gamage

WAVEFORMS AND CHANNEL CODING FOR 5G

2. BACKGROUND AND LITERATURE 12

3. PAPR REDUCTION OF OFDM WAVEFORMS 19

4. PERFORMANCE OF SINGLE CARRIER WAVEFORMS IN MIMO

7. RECOMMENDATIONS FOR FUTURE 59

a Exponential correlation coefficient

|·| Absolute value

Table 1: Suggested performance parameters for 5G.

5G suggested Perfor- IMT-Advanced Perfor-

1. Mobile broadband (MBB) services.

(a) Massive machine type communication (mMTC)

1. Enhanced Mobile Broadband (eMBB).

2. BACKGROUND AND LITERATURE

2.1. Energy Efficient Waveforms

Table 2: PHY / MAC for each generation.

2.1.1. OFDM waveforms

OFDM is a multi-carrier modulation technique where the signal is constructed from a

1. Resilience to multipath fading, inter-symbol interference (ISI), co-channel inter-

2. Lower implementation complexity compared with the single carrier solution

3. High spectral efficiency due to usage of closely spaced overlapping subcarriers.

Peak-to-average power ratio (PAPR)

Complementary Cumulative Distribution Function (CCDF)

Complementary Cumulative Distribution Function

1. Large PAPR reduction.

2. Low computational complexity.

3. No degradation in BER Performance.

4. Low average power and no requirement of additional transmit power.

5. No bandwidth expansion or spectral spillage.

2.1.2. Single Carrier waveforms

of a large number of antennas in a multi-carrier system will be the large number of

2.2. Channel coding

2.2.1. Turbo codes

2.2.2. LDPC codes

2.2.3. Polar codes

3. PAPR REDUCTION OF OFDM WAVEFORMS

3.1. System model

In SLM it is assumed that, U statistically independent alternative sequences, which

Scrambled bit Sequence

MQAM/MPSK S/P IFFT Selective P/S

MQAM/MPSK P/S Selective FFT S/P

Figure 2: System model.

Algorithm 1 PAPR reduction algorithm

3.1.1. Scrambler and descrambler

Figure 3: Scrambler for an S length sequence.

choosing different sequence of switches to be kept ON which is denoted by Ps . These

Figure 4: Descrambler for an S length sequence.

• More than 58% of the sequences do not need any scrambler.

• About 10% more of signals get below threshold after 2 scramblers.

• Another 4,3% of signals go below threshold after 3 scramblers.

• Only 0,4% sequences need more than 5 scramblers.

Complementary Cumulative Distribution Function

Table 3: Comparison of PAPR reduction techniques.

PAPR reduc- PAPR reduc- Implementation Data rate BER degra-

Figure 9: Performance of the system with increasing number of subcarriers N, for

4. PERFORMANCE OF SINGLE CARRIER WAVEFORMS IN

4.1. Single carrier - Uplink

h̃[n, m] = h[n, m] + [n, m], (15)