KYAMBOGO UNIVERSITY

FACULTY OF ENGINEERING

DEPARTMENT OF ELECTRICAL AND ELECTRONICS

BACHELOR OF ENGINEERING IN TELECOMMUNICATIONS ENGINEERING

IMPACT OF DEVICE-TO-DEVICE COMMUNICATION ON SOFTWARE DEFINED

RADIOS (SDRs) FOR 5G

THIRD YEAR RESEARCH PROJECT

MBOGO ABDULWAHABU 19/U/ETD/18425/GV

SUPERVISOR: Mr. SSEMUJU MARK

DATE: APRIL 2022

SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE

AWARD OF BACHELOR OF ENGINEERING IN TELECOMMUNICATIONS

ENGINEERING
DECLARATION OF AUTHORSHIP
I hereby declare that the document am submitting is entirely my own original work except where
otherwise indicated and that it has not been published to any other university or institution of higher
learning for the award of degree in Bachelor of Engineering in Telecommunications Engineering. Any
direct or indirect use of the works, in any form, of any other authors has, to the best of our knowledge
been properly acknowledged as references.

NAME: MBOGO ABDULWAHABU

REG NO: 19/U/ETD/18425/GV
Signature………………………….

Date………………………………...

2
 APPROVAL
This is to prove that MBOGO ABDULWAHABU Registration number 19/U/ETD/18425/GV
carried out research on the impact of Device-to-Device communication on Software defined
radios for 5G and it’s now ready for submission to the Department of Electrical and Electronics
Kyambogo University.
Signature………………………….
Date……………………………….
Mr. SSEMUJU MARK
UNIVERSITY SUPERVISOR
KYAMBOGO UNIVERSITY

3
 ACKNOWLEDGEMENT
I would also like to thank my families for their unwavering support and love towards my academics.

I would like to extent my sincere gratitude to my supervisor, Mr. Ssemuju Mark for his time and guidance
throughout this research.

4
 LIST OF ABBREVIATIONS/ACRONYMS

5G Fifth generation

ITU International Telecommunications Union

Telecom Telecommunications

SDRs Software Defined Radios

D2D Device-to-Device

Gbps Gigabits per second

UHD Ultra High Definition

LAN Local Area Network

CUs Cellar Users

DUs Device-to-Device Users

eNB Evolved Node Base Station

SRF Slow Rayleigh Fading

FRF Fast Rayleigh Fading

IoT Internet of Things

LTE Long Term Evolution

LTE-A Long Term Evolution-Advanced

RRM Radio Resource Management

ProSe Proximity Service

QoS Quality-of-Service

QoE Quality of Experience

RRM Radio Resource Management

SNR Signal-to-Noise Ratio

SINR Signal-to-Interference-Plus-Noise Ratio

IP Internet Protocol

VoIP Voice over IP address

Wi-Fi Wireless Fidelity (Devices based on the IEEE 802.11 Standards)

5
 TABLE OF CONTENTS
ABSTRACT................................................................................................................................................2
DECLARATION OF AUTHORSHIP.........................................................................................................3
APPROVAL................................................................................................................................................4
ACKNOWLEDGEMENT...........................................................................................................................5
LIST OF ABBREVIATIONS/ACRONYMS..............................................................................................6
LIST OF FIGURES.....................................................................................................................................7
CHAPTER ONE: INTRODUCTION..........................................................................................................9
1.1 BACKGROUND OF THE STUDY..................................................................................................9
1.2 PROBLEM STATEMENT..............................................................................................................10
1.3 OBJECTIVES..................................................................................................................................10
1.3.1 General Objective.....................................................................................................................10
1.3.2 Specific Objectives...................................................................................................................10
1.4 JUSTIFICATION................................................................................................................................10
1.5 SIGNIFICANCE OF THE STUDY.....................................................................................................10
CHAPTER TWO: LITERATURE REVIEW............................................................................................12
2.1 BACKGROUND OF THE STUDY................................................................................................12
2.1.1 Understanding the series of the mobile phone generations.......................................................12
2.2 DEVICE-TO-DEVICE COMMUNICATIONS...............................................................................14
2.3 CLASSIFICATION OF D2D COMMUNICATION.......................................................................15
2.3.1 BASING ON SPECTRUM RESOURCES...............................................................................15
2.3.2 Based on the type of service offered.........................................................................................18
2.3.3 Based on Network Control........................................................................................................18
2.4 PERFORMANCE GAIN OF D2D COMMUNICATIONS.............................................................19
2.5 APPLICATIONS AND USE CASES OF D2D COMMUNICATION............................................19
2.6 CHALLENGES AND LIMITATIONS OF D2D COMMMUNICATIONS....................................20
2.7 SOFTWARE DEFINED RADIO....................................................................................................20
2.8.1 HARD REAL-TIME.................................................................................................................21
2.8.2 THE USRP SDR PLATFORM.................................................................................................21
2.9 OPEN AIR INTERFACE................................................................................................................21
2.9.1 HARDWARE REQUIREMENTS............................................................................................22
2.10 D2D COMMUNICATION IMPLEMENTATION........................................................................22
2.11 BENEFITS OF D2D COMMUNICATION...................................................................................23
2.10 REVIEW OF EXISITNG ALGORITHMS....................................................................................23

6
CHAPTER THREE: METHODOLOGY, MODEL OF SYSTEM, DESIGN AND PROBLEM
FORMULATION......................................................................................................................................26
3.1 INTRODUCTION...........................................................................................................................26
3.2 METHODS OF DATA COLLECTION..........................................................................................26
3.3 MODES OF D2D COMMUNICATION.........................................................................................26
3.3.1 CELLULAR MODE.................................................................................................................26
3.3.2 DEDICATED MODE...............................................................................................................27
3.3.3 REUSE MODE.........................................................................................................................27
3.4 SIMULATION SYSTEM MODEL.................................................................................................28
3.5 PROBLEM FORMULATION FOR D2D COMMUNICATION MODES.....................................28
3.5.1 Cellular Mode...........................................................................................................................28
3.5.2 Dedicated Mode........................................................................................................................29
3.7 PROBLEM FORMULATION FOR MODE SELECTION AND PARAMETER ANALYSIS......33
3.9 PROPOSED SCHEMES AND ALGORITHMS.......................................................................36
CHAPTER FOUR: SIMULATIONS, RESULTS AND ANALYSIS........................................................38
4.1 INTRODUCTION...........................................................................................................................38
4.2 SIMULATION TOOLS...................................................................................................................38
4.3 SIMULATION PARAMETERS.....................................................................................................38
4.4 RESULTS AND ANALYSIS..........................................................................................................40
4.4.1 Impact of Maximum Transmit Power on Overall System Throughput..................................40
4.4.2 Impact of Maximum D2D Distance on Overall System Throughput.............................41
4.4.3 Impact of the D2D Pair Numbers on Overall System Throughput.....................................42
CHAPTER FIVE: CONCLUSION, RECOMMENDATIONS AND FUTURE WORK...........................44

7
 LIST OF FIGURES

Figure 1: (a) D2D mode, (b) both cellular mode and D2D mode...............................................................16
Figure 2: classification of D2D communication........................................................................................16
Figure 3 Schematic representation of In-band D2D communication.........................................................17
Figure 4: Pictorial representation of In-band D2D communication...........................................................17
Figure 5: Schematic representation of Out-band D2D communication......................................................18
Figure 6: Out-band D2D communication..................................................................................................19
Figure 7: D2D classification based on type of service provided................................................................20
Figure 8: Sample of D2D communications concept and applications........................................................21
Figure 9: Modes of D2D communication..................................................................................................29
Figure 10: Flow chart of the algorithm......................................................................................................37
Figure 11: Snipped image from algorithm code in Matlab........................................................................38
Figure 12: showing the cell with random placed UEs and DUs in Matlab.................................................40
Figure 13:Overall system throughput for different algorithms using different number of K pairs.............45

8
LIST OF TABLES

Table 1: latency ranges for the different technologies...............................................................................16

Table 2: Simulation Parameters.................................................................................................................43

9
 ABSTRACT
Proximity user equipment (UEs) in mobile networks may be communicated directly without passing their
traffic through the  by using  (D2D) communications. This can be done using the underlaying approach in
which the D2D-UEs (DEs) are allowed to use the same resources allocated for cellular UEs (CEs), which
can enhance the spectral efficiency. The D2D communication has been largely exploited in non-cellular
technologies such as Wi-Fi or Bluetooth but it has not yet been fully incorporated into existing cellular
networks. This is formulated to maximize the network performance in terms of achieved throughput while
ensuring the quality-of-service (QoS) constraints for CEs and DEs. A two-step algorithm is proposed in
which the admission control is performed firstly to determine the set of possible D2D connections and
their CE partners that achieve the minimum QoS demands. Then, the optimal power for each permissible
DE and its possible partners in different cells are allocated to maximize the network throughput while
guaranteeing the signal-to-noise and signal-to-interference ratio both D2D and Cellular users. Simulation
results illustrate that the suggested algorithm can remarkably enhance the performance of the network in
terms of throughput gain and access rate. In this report, optimal joint mode selection, channel assignment,
power control, throughput, latency, other parameters in D2D underlying communication are addressed.
The three communication modes are considered for D2D users: cellular mode, dedicated mode and reuse
mode and they were evaluated numerically via MATLAB-based simulation.

10
CHAPTER ONE: INTRODUCTION
This chapter presents the background of the study, the problem statement, the objectives that is general
and specific objectives, justification of the study and the scope of the study.

1.1 BACKGROUND OF THE STUDY

Substantial growth of demand for mobile data and rapidly increasing spread of personal communication
devices such as smart gadgets, phones and tablets is the major challenge faced by the mobile service
providers in recent years [1]. With the advent of Internet of Things (IoT), Big Data and realistic Ultra
High Definition (UHD) services, it is foreseen that mobile traffic will witness 1000-fold increase by the
year 2023[2]. This spectacular rise in demand for mobile data calls for drastic improvement in mobile
network capacity beyond the current 3G/4G networks to the next generation of wireless radio standards
[2]. This has urged the need for more effective ways of utilizing the available spectrum for the different
cellular networks without overloading the cellular networks. Their first actions were increasing the
number of base stations and buying more spectrum, to improve the cellular network capacity.

However, these measures did not help the operators to catch up with the traffic demand of mobile users.
Next step was to exploit the cellular resources in a more efficient manner. 5G networks aim to provide
new type services prominent among these are industrial automation, vehicle-to-vehicle communication.
Such new cases demand lower latencies, higher reliability along with greater flexibility than other
generations of cellular technologies.

5G networks intend to combine radical solutions to assure more capacity, lower latency and higher
reliability. Such solutions include several emerging technologies such as Network function Virtualization
(NFV), Software-Defined Networking (SDN), massive MIMO, Device-to-device communication and
many others. D2D communication recently has emerged as a promising technique for improving system
performance in terms of energy efficiency and spectrum utilization which in turn enables very many
numbers of users’ connections even with limited resources.

Basically, the idea is to enable two mobile devices in proximity of each other to establish a direct
connection and to bypass the base station (BS). By incorporating D2D communication into cellular
networks, mobile users can operate on D2D mode (i.e., two user equipment (UEs) communicate through
BS), which may be controlled by either network or users. D2D communication gains many benefits, like
spectrum/energy efficiency improvement, coverage extension, cellular offloading.

However, D2D communication poses a set of new technical challenges, such as interference management
and mode selection which are quite different from those of traditional cellular networks. The decision
when the D2D mode is preferred over the other communication modes(D2I) is the mode selection which
is tackled in this thesis by proposing a dynamic optimal joint mode selection low-complexity algorithm
for D2D systems. The algorithm can directly select the optimal communication mode using both uplink
and downlink channels by considering the distance between two D2D users, the network load and the
received channel state information (CSI) from the BS.

11
1.2 PROBLEM STATEMENT

The increasing mobile traffic on cellular networks which has reduced reliability of the cellular networks,
increased communication latency and caused congestion on cellular networks mainly attributed to
increasing number of devices due to applications like automation, gaming and video streaming. This
creates the need for sustainable solutions to offload the traffic from the base stations. D2D
communication comes in to offer a solution to the above. Few experimentations and studies of D2D
communication have been tried out in the real world with focus mainly on the unlicensed spectrum using
the Software Defined radio and stimulations using software like MATLAB. This project seeks to
implementation of device-to-device communication on SDR for 5G network to address the above
problems.

1.3 OBJECTIVES
1.3.1 General Objective
To implement device-to-device communication on Software Defined Radio (SDR) for 5G cellular
networks communication using Open Air Interface software and Universal Software radio Peripheral
(USRP) software defined radio.

1.3.2 Specific Objectives

 To implement a 5G cellular network to be used for D2D
 To implement D2D communication using the 5G cellular network implemented
 To analyze the best protocol to realize D2D communication on a 5G cellular network.

1.4 JUSTIFICATION
It reduces on channel interference among Cellular and Device-to-Device users as well as allowing
frequency reuse and better signal-to-noise ratio, better signal-to-interference ratio. It is a promising
technique for offloading local traffic from cellular base stations by allowing local devices, in close
proximity with allowable channel gain to communicate directly with each other. Through relaying, D2D
is also a promising approach to enhancing service coverage at cell edges or in black spots, improving
network performance and service quality.

1.5 SIGNIFICANCE OF THE STUDY

The benefits of the algorithm are;

 Improved channel capacity by considering reuse of both uplink and downlink channels.
 Improved resource and energy efficiency
 Minimized channel capacity interference through power control and controlling of signal-to-noise
ratio and signal-to-interference ratio.
 Minimized power consumption and power control of devices in D2D

12
CHAPTER TWO: LITERATURE REVIEW
This chapter presents the summarized review of literature related to device-to-device communication for
5G communication

2.1 BACKGROUND OF THE STUDY

2.1.1 Understanding the series of the mobile phone generations

There has been a vast advancement in mobile wireless communication since the last few decades. This
innovation consists of a number of generations and is still going on. Cellular technology is divided into
analog (0G – 1G) and digital i.e., 2G (2.5G – 3.75G), 3G (3.5G-3.7G-3.9G-3.95G), 4G(4G-4.5G/4.9G),
5G and then 6G (currently still in research) [1]. The journey of mobile wireless communication began
with 1G followed by 2G, 3G, 4G and under research upcoming and new generation 5G.

First generation, 1G

These phones were the first mobile phones to be used, which was introduced in 1982 and completed in
early 1990. It was used for voice services and was based on technology called Advanced Mobile Phone
System (AMPS). The AMPS system was frequency modulated and used frequency division multiple
access (FDMA) with a channel capacity of 30 MHz and frequency band of 824-894MHz.[15]

It introduces mobile technologies such as Mobile Telephone System (MTS), Advanced Mobile Telephone
System (AMTS), Improved Mobile Telephone Service (IMTS) and Push to Talk (PTT). It has low
capacity, unreliable handoff, poor voice links, and no security at all since voice calls were played back in
radio towers, making these calls susceptible to unwanted eavesdropping by third parties.[14]

Second Generation, 2G

This was launched on GSM standard in 1991, it brought along benefits such as digitally encrypted phone
conservations, more efficient use of radio frequency spectrum, data services for mobile, starting with
SMS messaging then text messaging, picture messaging. It uses the bandwidth of 30 to 200 KHz. Next to
2G, 2.5G system uses packet switched and circuit switched domain and provide data rates up to 144kbps.
E.g., GPRS, CDMA and EDGE.[15]

2.5G and 2.75G

In order to support higher data rate, General Packet Radio Service (GPRS) was introduced and
successfully deployed. GPRS was capable of data rate up to 171kbps (maximum). EDGE was capable to
support up to 473.6kbps (maximum). Another popular technology CDMA2000 was also introduced to
support higher data rate for CDMA networks. This technology has the ability to provide up to 384 kbps
data rate (maximum).

Third Generation, 3G

It’s the third generation of wireless mobile technology, it’s an upgrade or improvement of 2G, 2.5G,
GPRS and 2.75G EDGE network. This network is based on a set of standards used for mobile devices and
mobile telecommunication use services and networks that comply with the international
telecommunication union. 3G finds applications in wireless voice telephony, mobile internet access, fixed

13
wireless internet access, video calls and mobile tv. 3G supports services that provide transfer of
information at rates of at least 144 kbps. Later, 3G releases 3.5G and 3.75G also provide mobile
broadband access of several Megabits to smart gadgets, modems in pcs. This ensures it can be applied to
wireless voice calls, mobile internet access, fixed wireless internet access, video calls, mobile tv
technologies [1].

Fourth Generation, 4G

4G provides similar features as 3G and additional services like Multi-Media Newspapers, with more
emphasis on clarity and send data much quicker than previous generations, Long-term Evolution is
considered as 4G technology. 4G was developed to accommodate the QoS and rate requirements set by
forthcoming applications like wireless broadband access, multi-media messaging service (MMS), video
chat, mobile TV, HDTV content, Digital Video Broadcasting (DVB), minimal services like voice and
data, and other services that utilize bandwidth. [16]

Fifth Generation, 5G

it’s the fifth-generation technology standard for broadband cellular networks which cellular phone
companies began deploying worldwide in 2019. It also came along with benefits such as provide
connectivity to most current cellphones. 5G networks are divided into small geographical areas (cells), all
devices are connected to the internet and telephone network by radio waves through a local antenna in the
cell providing greater bandwidth, giving higher download speeds eventually up to 10 gigabits. It can
ensure connectivity for a crowded area thus improving the quality of internet services. It increases the
possibility of new applications in internet of things (I.o.T) and machine-to-machine [2].

In recent developments in cellular technology, device-to-device communication underlaying cellular

network has widely been considered as a potential resolution for both spectral and energy efficiency for
the LTE-Advanced systems. This technique has been proved to have the hop gain, the proximity gain, and
the reuse gain [5]. The hop gain comes from the fact that D2D communications only need one hop by
enabling users communicating directly instead of accessing the evolved NodeBs infrastructure (eNBs).
The proximity gain is achieved due to communications between nearby users. The reuse gain is
introduced when D2D users reuse the channels of existing cellular users with proper interference
management strategies. Therefore, the mode selection is quite essential in the D2D communications setup
as there is still no defined method to determine when to select the optimal communication mode in
various contexts.

Quick reminder that 5G is an evolution of the existing 4G therefore offers quite a number of benefits as
compared to the existing 4G as follows;

 It is designed to increase the speed and responsiveness of wireless network

 It allows device to device communications (peer-to-peer) communication
 It is faster and more reliable communication
 It reduces end-to-end delay in transmission to as low as 1ms as compared to 3G and 4G which
lies around 40 to 60ms.

Latency

Latency is the time taken by packets to travel from the Cellular user to Base station.

14
OR

It is the delay users experience as data makes a round trip through the network. If latency is high, the
network will feel less responsive and be slower to react to your requests. [3]
Table 1: latency ranges for the different technologies

Generation Latency (ms) Remarks

4G 36 to 48 Real world aspect
5G 17 to 26 Real world aspect
5G 1ms Theoretical aspect

Throughput

The data rates of 100+ Megabits-per-second of about 20Gbps. [4]

It comes along with further segments that look at improving the QoS and communication, these packages
that are included in 5G are;

1. Millimeter wave (mm wave)

It operates in the 30GHz to 300GHz offering more bandwidth, developed band of spectrum that can
enable high-speed LAN, broadband.

It offers higher efficiency with wavelengths ranging from 10 millimeters to 1 millimeter.

It requires line of sight for about 1 Km.

Though it faces high atmospheric attenuation therefore it can be affected by rain and humidity (rain fade)
and physical objects. This can be resolved by using small cells, works in either licensed or unlicensed
spectrum, connection range between 10m to 2km for a cell, distributed everywhere and interconnected.

Macro cell: It is by high power cellular site

Micro cell: It is large than Pico cell, with low power cell sites covering a small cell by varying power.

Pico cell: It is a small cell to extend network capacity in areas that are dense.

2. Massive Multiple Input Multiple Output (MIMO)

It increases the network capacity, the spectral efficiency where the rate of transmitting information can be
increased without changing on the given bandwidth. It supports multiple users simultaneously and
increased throughput. It faces low congestion due to more streams. This is affected by interference mainly
(signals travel in different direction).

3. Beam-forming

This improves the massive MIMO by radio frequency management, sends a focus stream of data to
individual user, less interference and more efficient, BS can handle more incoming and outgoing data
streams at once.

4. Full duplex

15
It is able to receive and transmit data at same time, on the same frequency. It is bi-directional that can
move data in both directions. Currently, BS are half-duplex.

5. Device-to-device communication

This is explained in detail below.

2.2 DEVICE-TO-DEVICE COMMUNICATIONS

Traditional cellular networks are built around entities called base stations (BS) [30],[31]. These are
responsible for a given geographical area known as a cell. All traffic within a cell has classically transited
through the BS. As telecom operators are struggling to accommodate the existing demand of mobile
users, new data intensive applications are emerging in the daily routines of mobile users (e.g., proximity-
aware services). Moreover, 4G cellular technologies (and Long-Term Evolution Advanced (LTE-A),
which have extremely efficient physical and MAC layer performance, are still lagging behind mobile
users’ booming data demand. Fig 1 illustrates the use cases of D2D communication in cellular networks.

In academia, D2D communication was first proposed in [33] to enable multi-hop relays in cellular
networks. Later the works in [34],[35] investigated the potential of D2D communications for improving
spectral efficiency of cellular networks. Soon after, other potential D2D use-cases were introduced in the
literature such as multi-casting [8][9], peer-to-peer communication [36], video dissemination [37],[38],
[39], machine to machine communication [40], cellular offloading [41], and other studies have been
carried out in regards to D2D communication.

This concept is advantageous in terms of proximity service and applications [19] comparing with current
traditional cellular mode. The devices within proximity can communicate with each other directly
without accessing the core network or any other infrastructure [20], while in the traditional cellular mode
the first device should pass on data to a core network or any other infrastructure using up-link, and then
the core network should send back to the targeted device using down-link [6]. With the dense deployment
of wireless sensor networks can be integrated with 5G networks using D2D communication, thereby
improving the spectrum utilization, system throughput, energy efficiency of the network. This can offer
higher flexibility with variety of options in choosing the proper communication scenario by setting the
involvement level of the BS in the process as well as the proper type of spectrum licensed or unlicensed
[21].

16
 Figure 1: (a) D2D mode, (b) both cellular mode and D2D mode

2.3 CLASSIFICATION OF D2D COMMUNICATION

Some research papers classify D2D communication basing on several aspects like type of spectrum
resources, the type of service, network resource control and so on. In this segment, the key aspects are
looked in detail;

2.3.1 BASING ON SPECTRUM RESOURCES

Under this, D2D communication is classified basing on the spectrum where the communication occurs in
cellular networks into the two categories; In-band (licensed part) and Out-band (unlicensed part), then the
method used to share the spectrum resources into two categories; Underlay and Overlay.
D E VIC E -TO -
D E VIC E
C OMMUNIC ATIO
NS (D 2D )

IN-B AND D 2 D OUT-B AND D 2D

Underlay Ov erlay C ontrolled Autonom ous

Figure 2: classification of D2D communication

In-band D2D communications

17
In this category, both the D2D users and Cellular users use licensed part of the spectrum for data
communication as this spectrum portion offers greater reliability and better performance. This can be
further subdivided into underlay and overlay based on the way used to share the spectrum resources.

In the underlay, the D2D user equipment use the same cellular resources at the same frequency/time of
another cellular user simultaneously.

In the Overlay, the D2D user equipment use a dedicated cellular resource at the same frequency/time of
another cellular user but not simultaneously. The motivation behind choosing this kind of communication
is high reliability since it occurs in the licensed portion of the spectrum.

Figure 3 Schematic representation of In-band D2D communication

Figure 4: Pictorial representation of In-band D2D communication

Advantages of using In-band communication

 QOS guaranteed due to the BS control to cellular spectrum.

 In underlay D2D communication, spatial diversity techniques are used to enhance spectral
efficiency.
 Any cellular device can use in-band D2D communication
 High transmission data rates up to 1 Gbps with distance up to 1km.
 No need of new interface any device can use In-band communication

Weaknesses of using in-band communication

18
  Interference problem since both DUs and CUs are using same band and infrastructure.
 Computational overhead on the BS and complex algorithm for radio resource management
(RRM).
 Resource wastage cannot be avoided in overlay D2D
 A user cannot have simultaneous cellular and D2D transmissions.


Out-band D2D communication

In this type, D2D communication uses unlicensed spectrum like 2.4 GHz ISM band, 38 GHz mmwave
band, industrial, specific, medical and a second interface; Bluetooth, wi-fi-direct to communicate.
Cellular uses its own dedicated licensed spectrum. This technique eliminates interference between D2D
users and cellular users as both cellular and D2D communication can take place simultaneously. This is
further subdivided into; controlled and autonomous based on the BS way used to control for both D2D
and cellular spectrum resources.

In controlled category, D2D radio interface is controlled by cellular network therefore the BS controls
both the CUs and DUs.

In autonomous category, cellular network controls only cellular links where as D2D links are controlled
by users themselves. The users control all the DUs and the BS controls the CUs. The intent of this to
avoid the interference problems that occur between D2D and cellular links in the in-band category.

Figure 5: Schematic representation of Out-band D2D communication

Figure 6: Out-band D2D communication

19
Advantages of using out-band communication

 No interference issue since they are using different band and infrastructure.
 No need of resource allocation
 DUs and CUs can work both simultaneous.

Weaknesses of using out-band communication

 Two interfaces are required and QoS is not guarantee.

 Not power efficient with Low data rate and transmission distance
 Unlicensed band interference is uncontrollable
 Power management between two wireless interfaces is crucial to control power consumption of
the device, hence it is not power efficient which leads to low transmitter distance and data rate.

2.3.2 Based on the type of service offered

In this classification, the type of the service is taken into consideration when the D2D communication
scenario occurred [22].

 Emergency Services: In this scenario, the D2D devices can make a connection using both cellular
resources and unlicensed resources if needed and this can be very fruitful in terms of safety.
 Commercial Services: In this category, the D2D connections take place in cellular resources only
and this makes it easy to control. Figure 7 shows the two types of D2D communications scenario
based on the provided service as we can see that in the emergency scenario when we lose the
licensed resources, we can always use a third party (unlicensed) to make a connection. On other
hand, the commercial service gives us an example of regular use for the licensed resources.

Figure 7: D2D classification based on type of service provided

2.3.3 Based on Network Control

In this category, the D2D networks configuration is built based on the control level of the BS in the
communication process [6] [20]:

 Stand-Alone D2D:

This model is similar to ad-hoc network where the devices communicate with each other in a distributed
manner without the help of cellular system BS. Moreover, in this model, fixed infrastructure such as
access points or BS is not a prerequisite the devices rely on local hardware capabilities hence the use of
the term “stand-alone D2D”. [25-26].

The BS has no control, RRM is based on both node local information.

20
It has high flexibility but no coordination between nodes.

 Network Control D2D:

In this model, the cellular system BS has the full control of the D2D connection [23]. In this model, fixed
infrastructure such as access points or BS is a requirement the devices do not rely on local hardware
capabilities and that’s why it called network control D2D [25-26].

The BS has full control, RRM is based on global information from cellular BS.

It offers high performance but insufficient scalability since everything must pass by the BS.

 Network Assisted D2D:

In here, we can see a balance between the past two categories where the cellular system BS has a partial
control of the D2D connection which is the coordination between the cellular users and the DUs operate
in self-organized way [20] [23]. To illustrate the impact of network control level, we depict the D2D
Communications schemes and compare them based on the level of the BS control and how they make
their RRM.

The BS has partial control, RRM is based on both node local information and BS global information, they
make their RRM decisions separately.

It offers high co-ordination but computation overhead of the BS.

2.4 PERFORMANCE GAIN OF D2D COMMUNICATIONS

Additional performance gain such as a higher data rates, with low latency and consume less power can be
added by D2D communication technology to what already exists in the traditional way of communication
[27-28]. First, the hop gain which denotes the ability of using one direct link instead of using two links
up-link/down-link via BS, which happens when we set a D2D communication by allowing devices to
communicate using direct link without BS. Second, the proximity gains which depends on the location of
user and denotes the ability of nearby devices to communicate [27]. The last one is the reuse gain which
denotes the DUs ability of reusing the same CUs link [27-28].

2.5 APPLICATIONS AND USE CASES OF D2D COMMUNICATION

D2D is a simple, powerful approach that offers many advantages over conventional approaches in
addition to various use cases, and applications such as local data service (i.e., bit torrent, online gaming,
proximity services, streaming services, social networking, community services), local voice traffic,
Multiuser cooperative communication (MUCC) and public safety [9] [28] [19]. Figure 6 shows a sample
of D2D use cases and applications [28].

21
Figure 8: Sample of D2D communications concept and applications

2.6 CHALLENGES AND LIMITATIONS OF D2D COMMUNICATIONS

With the promising prospects, D2D communications still faces some main technical challenges that
require resolutions and need to be emphasized. Though, the connection setup between D2D users and
Cellular users is not optimal as it requires like any other communication technology key functions to
work:

 Designing new peering and service discovery methods (i.e defining network role,
synchronization, and designing reference signal) [28].
 Designing physical layer techniques (i.e., Encoding, Signaling, Data transmission and reception,
etc.) [28].
 D2D poses quite a few security threats such as eavesdropping
 Solving the RRM algorithms (i.e., mode selection, scheduling, resource allocation, power
control, and interference management problem) [8] [15] [16] [28].

2.7 SOFTWARE DEFINED RADIO

A software-defined radio (SDR) system is a radio communication system which uses software for the
modulation and demodulation of radio signals [42]. An SDR performs significant amounts of signal
processing in a general-purpose computer, or a reconfigurable piece of digital electronics. The goal of this
design is to produce a radio that can receive and transmit a new form of radio protocol just by running
new software [43]. Software-defined radios are communication systems implemented using software.
Unlike traditional radio systems, they utilize general-purpose processors (GPP) to perform the necessary
computations and operations to achieve the transfer of information. Typically, the only other piece of
hardware involved in addition to the computer is known as a radio front-end (RF–front-end) performs the
analogue-to digital (ADC) and digital-to-analogue (DAC) conversions, modulation to and from the carrier
frequency and the transmission and reception of radio frequency (RF) signals. SDR platform may further
be classified into hard real-time and soft real-time systems.

22
2.8.1 HARD REAL-TIME

Hard real-time have bounded execution time on all functions. Verification of timing requirement
compliance during the design phase enables hard real-time systems to offer greater reliability. This
however comes at the cost of rigidity as all changes must be carefully considered so as not to disturb the
timing of the execution flow. As such, any feature that may potentially exceed its allotted computation
time may not be implemented at all even when such lengthy execution times are rare.

2.8.2 THE USRP SDR PLATFORM

The USRP platform implements an extended version of the soft real-time SDR concept, an SDR
framework. SDR frameworks provide a supporting infrastructure for the development of various radio
technologies without the need to re-write every part of the necessary code [44]. Functionality typically
found in every piece of software and every radio protocol exist within the framework and are shared by
the various RF technologies implemented in a manner kin to the use of libraries in software development.
The fundamental task of the platform is to provide a radio research tool enabling rapid testing of concepts
while minimizing the amount of code needed to be written for each experiment. Additionally, the
platform solves performance, timing and other computer engineering related problems that are not
germane to radio technology research. Having been solved once, they need not be re-implemented for
each new experiment.

2.9 OPEN AIR INTERFACE

Open Air Interface (OAI) is an open-source project that implements 3GPP technology on general purpose
x86 computing hardware and off-the-shelf software defined radio cards like the Universal Software Radio
Peripheral (USRP). At its base OAI implements 4G LTE, but recently the implementation 5G New Radio
(NR) has started with the launch of cloud Native 5G core network [45],[46]. Open Air Interface is an
open-source initiative that today provides a 3GPP compliant reference implementation of enodeB (eNB),
User Equipment (UE), and evolved packet core (EPC) that runs on general purpose computing platforms
(x86) together with off-the-shelf software defined radio (SDR) cards like the ETTUS USRP, Lime SDR,
and ExpressMIMO2 [47]. It allows users to set up a compliant 4G LTE network and inter-operate with
commercial equipment. The Open-Air Interface team has been working on an implementation of 5G-NR
since 2017 with a focus on the eMBB use case. A first pre-5G demonstration was given at Mobile World
Congress 2018 where they demonstrated the feasibility of maintaining a throughput of 300Mbps over an
80MHz channel using the new 5G-NR LDPC channel coding. This was achieved by offloading the most
computationally expensive task, the LDPC decoder, to an FPGA. Using a software-only implementation
of the LDPC decoder, a throughput of150Mbps can be supported [48].

2.9.1 HARDWARE REQUIREMENTS

Open Air Interface requires the following hardware specifications for it to run successfully: Current
OpenAirInterface.org software requires Intel architecture-based PCs for the eNB or UE targets. This
requirement is due to optimized DSP functions which make heavy use of integer SIMD instructions (SSE,
SSE2, SSS3, SSE4, and AVX2). The software has currently been tested on the following processor
families [49]:

 Intel Atom Rangeley, E38xx, x5-z8300

23
  Generation 2/3/4 Intel Xeon
 Intel Atom Rangeley, E38xx, x5-z8300

In order to support full SISO throughput (75Mbps DL/ 18Mbps UL) at least 4 cores are required. Here are
some example configurations that work well for both eNBs and UE:

 Intel® CoreTM i5-6600K CPU @ 3.50GHz x 4

 Intel® CoreTM i5-6600 CPU @ 3.30GHz x 4
 Intel® CoreTM i7-8700 CPU @ 3.20GHz x 6

OAI supports the following Radio frequency components [49].

 USRP B210
 USRP X310
 BladeRF
 LimeSDR
 EURECOM EXPRESSMIMO2 RF

2.10 D2D COMMUNICATION IMPLEMENTATION

Experimental setups for cellular communications have always been a rare commodity in academic with
D2D communication being no different [50]. Most of which implementations are done via simulations for
example work by the author in [16]. [27] Discusses an SDR-based experimental study of out band D2D
communications where the authors provide an experimental analysis of out band D2D schemes and they
design DORE a complete framework for handling channel opportunities offered by out band D2D relay
nodes. DORE consists of resource allocation optimization tools and protocols suitable to integrate QoS-
aware opportunistic D2Dcommunications within the architecture of 3GPP Proximity-based Services.
They implement DORE using an SDR framework to profile cellular network dynamics in presence of
opportunistic out band D2D communication schemes. The experiments reveal that out band D2D
communications are suitable for a large variety of delay-sensitive cellular applications. And that DORE
enables notable gains even with a few active D2D relay node. With the group number of cellular devices
due to multiple cellular applications, an in-band implementation of D2D communication is required.
Other real-world implementation includes LTE Standard-Compliant D2D Communication: Software-
defined Radio Implementation and Evaluation where the authors try to implement D2D communication
compliant LTE standard release 12 [28]. The project seeks to address that gap for the implementation of
D2D communications and give a suggestion of the best protocol to achieve D2D communication.

2.11 BENEFITS OF D2D COMMUNICATION

 The end-to-end latency for the D2D devices can be substantially reduced [16],[18]. Eliminating
the transit through the BS removes one hop, which is the up-link from the sending node.
Additionally, the BS requires a certain amount of time to process the received burst and to
prepare the downlink one. This further increases the latency cost of the extra hop. Further delay
may also be introduced by a lack of transmission opportunities for the up- and down-link hops
not being available at the optimal time due to scheduling load

24
  Direct communication can offload data from the treasured spectrum to out-of-band technologies
(i.e., Wi-Fi, Bluetooth, etc.), improving spectral efficiency.
 Higher energy efficiency can be achieved. The extra hop through the BS wastes energy compared
to a direct transmission [16],[19]. This is especially so when the two D2D nodes are situated
close to each other but far from the BS hence they are forced to use full or close to full transmit
power. Even with a low extra energy requirement for transmission, energy must still be expended
on the processing, scheduling and re-sending of the information at the BS. Another opportunity
for energy saving comes from D2D nodes relaying popular content. The information is first sent
from the BS and then relayed through user devices with an opportunity to achieve better energy
efficiency [20].

2.10 REVIEW OF EXISTING ALGORITHMS

Quite a number researchers have examined this mode selection problem such as [6] [7] [8] [10] [13]
literature describes several techniques that study a set of mode selection scenarios such as interference
management [13], power efficiency [14] and resource allocation [10] [15] [16]. As listed below;

 Doppler et al. in [13] proposed a heuristic mode selection algorithm based on the quality of the
cellular and D2D links considering the different interference situations when sharing cellular up-
link /down-link resources. In case of multi-cell, it also considers the inter-cell interference impact
on the sum-rate of the D2D users. They studied two cellular network scenarios: namely, single
cell environment, and multi-cell environment. The authors proposed a strategy that assumed a BS
with full channel state information (CSI) of all users’ links, and calculate the throughput of the
three D2D modes based on path loss model and distance. Then after that, their algorithm selects
the mode with highest sum-rate that fulfills the cellular signal to interference plus noise ratios
(SINR) constraint. Their study shows that the location of the devices can be very influence and
the optimal mode can largely depend on it not only on the links quality or interference situation.
In a simulated setting, they achieved 50% more gain on system throughput compared to the
traditional cellular communications.
 Similarly, in [40] Hakola et al. examined the mode selection by proposing comprehensive system
equations to indicate the optimal mode for all devices. The proposed algorithm uses abstract
information from the network to define the system equations such as the
 quality of link, the interference and noise levels. They assumed that DUs use the uplink resources
only of the cellular network and BS with full CSI. In their study, they compared the performance
gain of proposed algorithm with three other modes: cellular mode, forced D2D mode and path
loss mode.
 Arash Asadi in [6] provided an extensive survey on the available literature on D2D
communications in cellular networks. They categorized the available literature based on the
communication spectrum of D2D transmission into two major groups, namely, in-band and out-
band. The works under in-band D2D were further divided into underlay and overlay. Out-band
D2D related literature was also sub-categorized as controlled and autonomous. The major issue
faced in underlay D2D communication is the power control and interference management
between D2D and cellular users. Overlay D2D communication does not have the interference
issue because D2D and cellular resources do not overlap. However, this approach allocates
dedicated cellular resources to D2D users and has a lower spectral efficiency than underlay. In

25
 Out-band D2D, there is no interference and power control issue between D2D and cellular users.
Nevertheless, the interference level of the unlicensed spectrum is uncontrollable, hence, QoS
guaranteeing in highly saturated wireless areas is a challenging task. They also discussed the
weaknesses and strength of the existing literature. They pointed out the shortcomings of current
works and proposed potential future research directions. Their survey showed that D2D
communication in cellular networks is immature and there are still numerous open issues such as
interference management, power control, etc. We also shed light on some possible research
directions needed to improve the understanding of D2D potentialities for real world applications.

However, many algorithms tried to solve the combined mode selection problem with power optimization,
and resource allocation [7] [8] [10].

 For instance, in [18] Jung et al. the authors proposed a two-step algorithm using uplink resource
sharing for the D2D communications in a single-cell environment where the BS has the full CSI.
In simulation setting, their algorithm achieved up to 100% gain and minimized the overall
transmission power compared to forced D2D mode, forced cellular mode, and the algorithm
proposed in [17]. One of the drawbacks of this algorithm is executing using the brute-force
algorithm for searching in all mode combinations for all devices and thus, computationally
intractable. Moreover, in [10] Yu et al. proposed another joint algorithm based on network loads
also in a single-cell environment where the BS has the full CSI. The numerical simulations show
that, their algorithm can perform very closely to the optimal solution which is branch and bound
method.

2.11 COMPARISON OF D2D COMMUNICATION AND EDGE COMPUTING

Edge-computing is delivered by computing servers closer to the ultimate user. It reduces latency, data
traffic congestion.

Instead of having data conveyed to client via a central center, due to congestion, it strains the network.
But edge computing, storage and computing resources are shifted from the central data center to point
where the data is generated. If you can’t get data closer to the data center, get the data center closer to the
data sources. This is expected to improve response times and saves bandwidth.

26
CHAPTER THREE: METHODOLOGY, MODEL OF SYSTEM,
DESIGN AND PROBLEM FORMULATION
3.1 INTRODUCTION
This chapter discusses the technical framework and procedures on how to achieve the objectives of the
study of the research objectives. It explains the methods used in sourcing the necessary data for successful
completion of the project. We present the model of the system and the various modes of D2D
communication, formulate the joint mode selection, channel assignment, power control problem for D2D
systems then analyze the different parameters in comparison to the cellular system.

3.2 METHODS OF DATA COLLECTION

This section analyzes the methods employed in obtaining the necessary information that was required for
the implementation of the project. The main methods of data collection include: discussions, consultations
and literature review.

 Discussions: This was the first method employed. After identifying the problem, it was subjected
to thorough discussion with our supervisor in accordance to the research title and methodology.
 Consultations: This method involved acquiring and seeking for advice both technical and non-
technical from different personalities that included my supervisor, lecturers plus other
experienced experts both local and online resources were employed.
 Literature review: This involved reading of published documents over the Internet, in text books,
journals, documents from organizations and institutions regarding the problem statement and
research title in order to get a thorough overview of already done work, existing algorithms and
research gap.

3.3 MODES OF D2D COMMUNICATION

To clarify on this, with reference to [10] [13], the modes of D2D communication are;

3.3.1 CELLULAR MODE

This mode is the traditional mode where the users communicate as conventional CUs through a BS using
two orthogonal channels one from both (up-link and down-link) channels that are not currently used by
any other devices without interference. Pair number (1) in Figure 7 is to illustrate this concept as it shows
the D2D transmitter sends the data to the BS and then the BS send it back to the D2D receiver. However,
due to the spectrum limitation and to achieve better performance, this mode should not be used unless the
other two modes are not available.

3.3.2 DEDICATED MODE

Under this mode, the D2D users communicate directly using only one of the cellular users unused
channels the up-link channel, or down-link channel. Pair number (2) in Figure 9 illustrates this concept
where it shows the D2D transmitter sending data to the D2D receiver directly without sending it to BS
which can reduce channel consumption to half if we compare it cellular mode. Though, spectrum
utilization is not always rich, sometimes it can be very poor and we cannot find available resource due to

27
different factors such as network load, channel conditions, and the distance between the devices so that
the availability of free resource can be a problem.

Figure 9: Modes of D2D communication

3.3.3 REUSE MODE

In this mode, the spectrum utilization could be maximized by allowing two D2D users to make a direct
connection by reusing one of the current cellular user channels (down-link/up-link). Pair number (3) in
Figure 9 illustrates this concept where it shows an example of up-link reuse channel as it shows it cause
an interference and thus since when using same channel at the same time can cause a serious interference
problem needs to be solved with an efficient mechanism.

However, the decision itself is not the only challenge, there are two other critical challenges arise the first
one, how can we select the optimal transmission mode and the second one is when can we select the
optimal transmission mode. In fact, many factors should be considered when this decision is made for
example the aims and the type of the application can play a key role in this decision to guarantee the QoS,
or quality of experience (QoE). Moreover, network load, resources availability, distance between the
devices themselves and BS since this give an insight how much the interference situation can be managed
[12].

3.4 SIMULATION SYSTEM MODEL

In this research, the simulation scenario is modeled as a single-cell system with radius R, that consists of
potential multi DUs, which denote as K pairs, K= {D1, D2 …D K}, and potential CUs, which denote as
M= {C1, C2 …C M}, where the number of K pairs is less than M users. Each one of the CUs will
communicate using orthogonal channels one uplink, and one downlink. On the other hand, the DUs will
communicate using one of the three communication modes (cellular, dedicated, or reuse) based on the
number of the available uplink links (NU), and available downlink links (ND) channels. All this happened
in the signal coverage of the BS that has a full instantaneous CSI of all users’ links.

28
3.5 PROBLEM FORMULATION FOR D2D COMMUNICATION MODES

The channel capacity formula for D2D user k and cellular user m can be expressed as:

C=Blog ( 1+SINR k ,, m )=Bits/ s/ Hz [10]

Where;

P s x hTransmitter , Receiver
SINRk ,m = 2 [10]
Pl x hTransmitter , Receiver +σ N

Subject to:

B as the bandwidth

Ps ∧ Pl as the power, and the interference power respectively

2
σ N as the power of the additive white Gaussian noise (AWGN)

hTransmitter ,Receiver as the instantaneous channel gain between the sender and receiver expressed as:
−a
hTransmitter ,Receiver =Gx β Transmitter , Receiver x d Transmitter , Receiver

Subject to:

G as the path loss constant

β Transmitter , Receiver as the channel fading component between the sender and receiver D2D pair. Here the
channel can be modeled using two channels fading the slow Rayleigh fading and fast Rayleigh fading.

d −aTransmitter , Receiver as the distance between the transmitter and receiver and α is the path loss exponent.

3.5.1 Cellular Mode

As mentioned earlier, D2D users communicate like the traditional way through BS using two free
orthogonal channels that are not currently used by any other devices one up-link channel (N U) and one
down-link channel (ND). No interference happened here since the cellular mode use free orthogonal
channels and for QoS reasons both up-link SNRk (1UL), and down-link SNRk (1DL) should be larger than a
given threshold, ξmin. In this work, the focus on the D2D power so as assumption the SNR k (1UL) will be
always the minimum. The cellular mode channel capacity denotes a K dimensional indication vector:

T ( 1)=( T 1( 1) ,.... , T k ( 1) ),

Where T k = Blog (1 + min ( SNR k (

( 1) 1 UL ) (1DL )
, SNR k )

Where the cellular SNR k (

1UL )
, SNR k (
1DL )
can be expressed as:

( 1UL ) P k ,B (1) x hk , B D (1DL ) Pk , B(1 ) x h k ,B D

SNR k = , SNR k = [10]
σ N2 σ N2

29
Subject to:
(1 ) (1 )
 Pk and Pk , B as the power between D2D pair number k and the power from D2D pair number
k to BS, respectively.
 h k ,B D as the channel gain between D2D pair number k and BS.
 σ N 2 as the power of the additive white Gaussian noise (AWGN).

3.5.2 Dedicated Mode

This mode is just like the cellular mode from the from interference perspective where the user in here
adds no interference to the system, since all channels are orthogonal to each other. In this mode, one free
orthogonal channel either up-link or down-link channel (NU, ND) is needed. The dedicated mode channel
capacity denotes a K dimensional indication vector:

T ( 2)=T 1(2 ) , .... ,T k (2 ) where T k (2) = Blog SNR k (2) (4.6)

( 2)
Where the dedicated SNR k can be expressed as:
(2 ) ( D)
( 2) Pk × hk
SNR k = (4.7)
σ N ( 2)

Subject to:

 Pk (2 ) as the power between D2D pair number k.

(D)
 hk as channel gain between D2D pair number k.
( 2)
 σ N as the power of the additive white Gaussian noise (AWGN).

3.5.3 Reuse Mode

In this mode unlike, the earlier modes from the interference perspective, since two devices one D2D users
and one cellular user are using the same channel simultaneously the users in the reuse mode will add an
interference to the system. In this mode, one channel (up-link or down-link) only needed from the cellular
user channels. However, for QoS reasons since the CUs have the higher priority for both up-link and
C
down-link the DUs are not allowed to reuse the channel if cellular user SNR k ,m less than a given
threshold ξmin. The reuse mode channel capacity of the D2D pair number k which is reusing an active
channel that currently being used by cellular user number m denote as K× M dimensional indication
matrix:

T 1,1
¿ ( 3) (3 )
T k ,m(3 )=¿ Where T k ,m = Blog SNR k ,m (4.8)
¿
However, as mentioned earlier since D2D pair number k is reusing an active channel of a cellular user
number m. the channel capacity of the cellular user number m will be affected by this interference. The
channel capacity of reusing channel is denoted as K × M dimensional indication matrix:

30
 T 1,1
¿ C C
T k ,mC =¿ Where T k ,m = Blog SINRk ,m (4.9)
¿
In this mode, there are three scenarios as follows below:

1. Reusing Up-link Channel

In this case, the D2D pair number k is reusing an active up-link channel that is currently being used by
cellular user number m and the channel capacity of reusing up-link channel is denoted as K×M
dimensional indication matrix:

T 1,1
¿ ( 3UL)
= Blog SINRk ,m(
3UL )
T k ,m( ) =¿ Where T k ,m (4.10)
3UL

¿
T 1,1
¿ ( C UL )
= Blog SINRk ,m(
C UL )
T k ,m( ) =¿ Where T k ,m
C UL

The SINRk ,m(

3UL ) ( C UL )
, SINRk ,m can be expressed as:

P k , m(
3UL )
P k ,m(
C UL)
( 3UL ) × hk ( D ) ( C UL ) ×h m , B(C )
SINR k ,m = , SINRk ,m = [10]
P k ,m(
CUL )
P k ,m(
3UL )
×h k ,m +σ N (2) ×h k ,B D + σ N (2 )

Subject to:

 Pk , m( ) and Pk , m( ) as the power from D2D pair number k to cellular user number m, and the
3UL C UL

power from the cellular user number m to the receiver D2D pair number k, respectively.
 h k( D ) as the channel gain between D2D pair number k similar, h k ,m is the channel gain between
( C)
D2D pair number k cellular user number m. Also, h m ,B as for the channel gain between
D
cellular user number m and the BS. And finally, h k ,B is the channel gains between theD2Dpair
number k, and BS.
( 2)
 σ N as the power of the additive white Gaussian noise (AWGN).

2. Reusing Down-link Channel

In this case, the D2D pair number k is reusing an active downlink channel that is currently being used by
cellular user number m and the channel capacity of reusing downlink channel is denoted as K× M
dimensional indication matrix:

31
 T 1,1
¿ ( 3DL )
= Blog SINRk ,m(
3DL )
T k ,m =¿Where T k ,m
( 3DL ) (4.10)
¿
T 1,1
¿ ( C DL )
= Blog SINRk ,m(
C DL )
T k ,m =¿ Where T k ,m
( C DL )

¿
The SINRk ,m(
3DL ) ( C DL )
, SINRk ,m can be expressed as:

( 3 DL ) (D) (C DL ) ( C)
( 3DL ) P k ,m × hk ( C DL ) Pk , m ×hm ,B
SINRk ,m = , SINRk ,m = [10]
Pk ,m(
C DL )
Pk , m(
3DL )
×hk , m +σ N ( 2) ×hk , B D +σ N ( 2)

Subject to:

 Pk , m( ) and Pk , m( ) : Stand for the power from D2D pair number k to cellular user number m,
3DL C DL

and the power from the cellular user number m to the receiver D2D pair number k, respectively.
 h k( D ) : Stands for the channel gain between D2Dpair number k. similar, h k ,m: is the channel gain
( C)
betweenD2D pair number k cellular user number m. Also h m ,B : Stands for the channel gain
between the BS and cellular user number m.
 h k ,B D Are the channel gains between the D2Dpair number k, and BS.
( 2)
 σ N : Stands for the power of the additive white Gaussian noise (AWGN).

3. No Channel is Reused

T 1,1
¿ ( C NOT )
= Blog SINRk ,m(
C NOT )
T k ,m( )=¿), where T k ,m
C NOT

¿
The SINR k ,m(
C NOT )
can be expressed as

( C NOT ) P , m ( C ) × hm , B ( C )
SINRk ,m =
σ N (2 )

Subject to:
(C )
 P, m as the power of cellular user number m.
 h m ,B( C ) as the channel gain between the BS and cellular user number m.
( 2)
 σ N as power of the additive white Gaussian noise (AWGN).

3.6 PROBLEM FORMULATION FOR POWER CONTROL

❑
Just like mode selection, the power matrix is denoted as Pk , m , which is a K x M indication matrix:

32
Pk , m❑ ={Pk ( 1) , Pk ( 2) , Pk , m(3 ) , Pk ,mC , Pk , m3 , Pk ,m(
C DL )
UL DL
, P mC }

Where:

 p as the power of the DUs when the cellular mode and the dedicated mode is
used, respectively. In here they are always equal to pmaxD, which is maximum power of
the DUs since both are using orthogonal channels, there is no co-channel interference in
both modes, so the maximum throughput could be achieved when both DUs and CUs
with their maximum powers.

 P k , m3 , P k , m C
uL
: Represent the power of the D2D pair number k is reusing an active
UL

uplink channel that is currently being used by cellular user number m, and the power of
this cellular user number m, respectively. However, in here the problem is little bit
different, the p (k3, m) cannot have the pmaxD , which is maximum power of the DUs due to
UL

the fact of exciting of a co-channel interference. Similarly, and for the same reason the
p(kC,m )cannot have the pC max , which is maximum power of the CUs. So, we need to find
UL

the optimal power. In this thesis, to obtain optimal

power control vector (p(k3, UL
m
),
p(kC,m )), The algorithm discussed in [5] is used.
UL

(p(k3, ULm), p(kC, mUL)) = arg (max3UL) (CUL) f k, m (p(k3,ULm), p(kC, ULm)) (4.16)
(pk, m, pk, m)
Subject to

 p(3k , UL)
m : Cannot be less than zero and above pmax .
D

 p(C UL)
k , m : Cannot be less than zero and above p max.
C

 p(C UL)
k , m : The power will be equal to p max only in the case of no reuse for the CU.
c

However, for QoS reasons the D2D pair number k is allowed to reuse the channel
(C D L)
only if the SINR pk , m larger than a given threshold ξmin.

 p(3k , Dm L), p(C

k,m
D L)
: Represent the power of the D2D pair number k is reusing an active
downlink channel that is currently being used by cellular user number m, and the power of
this cellular user number m, respectively. However, in here the same
(3 D L)
problem in the uplink reusing is happened, the pk , m cannot have the pmaxD, and
p(C
k,m
UL)
cannot have the pCmax due to the fact of exciting of a co-channel interference. So, we
need to find the optimal power. In here the algorithm used before with the reusing the uplink
channel cannot being used without modification since in paper [4] they propose it for the uplink
channels only, in this thesis one of the contributions is modified the algorithm mentioned in [4] to
consider also finding the optimal power for the downlink channel:
(p (k3 ,DLm) p(kC, mDL))=arg (max 3DL)(CDL) f k,m (p(k3,DLm) , p(kC, mDL)) (4.17)
(pk, m, pk, m)
Subject to

 p(3k , Dm L) : Cannot be less than zero and above pmaxD.

33
 (C D L)
 pk , m : Cannot be less than zero and above pCmax.
 p(C k,m
D L)
: the power will be equal to pcmax only in the case of no reuse for the CU.
 P cm : Stands for the power of the cellular user number m and there are two cases to define
the right value:
The cellular user number m Not reused by any D2D pair number k then the power will equal
to pcmax
• The cellular user number m is being reused by any D2D pair number k then the
(C UL) (C D L)
power will equal one of pk , m ,or pk , m depending on the reusing case if its
uplink or downlink. Moreover, Equation (4.15) will be used to calculate the value
(C UL) (C D L)
of pk , m . Similar, Equation (4.16) will be used to calculate the value of pk , m .
3.7 PROBLEM FORMULATION FOR MODE SELECTION AND
PARAMETER ANALYSIS
To find the optimal power for D2D pair number k and the problem formulation for the joint mode
selection, and power control can be represented as follows [10] :(4.17)

(xk, m, pk, m) =argmax x, p

Subject to:

∑ x( k1) T (k1) (4.17. a)

k =1

This equation describes the D2D users in the cellular mode, where:
 x (k1): Stands for the cellular mode when it equal to 1 that means it has
been selected by the D2D pair number k.
 T (k1): Stands for the channel capacity when the D2D pair number k is in
the cellular mode.
K

∑ x( k2) T (k2) (4.17. b)

k =1

This equation describes the D2D users in the cellular mode, where:
• x (k2): Stands for the dedicated mode when it equal to 1 that means it has
been selected by the D2D pair number k.
• T (k1): Stands for the channel capacity when the D2D pair number k is in
the dedicated mode.

(4.17.c)
This equation describes the D2D users in the uplink reuse mode, where:

 x(k3, m): Stands for the uplink reuse mode when it equal to 1 that
UL

mean the D2D pair number k is reusing the uplink of cellular user
number m.

34
  T(k3, m): Stands for the channel capacity uplink reuse mode when
UL

the D2D pair number k is reusing the uplink of cellular user number m.
 T(kC,m ): Stands for the channel capacity of cellular user number m
UL

when interfered byD2D pair number k.

K M K M M K

(4.17.d)
This equation describes the D2D users in the uplink reuse mode, where:
 x(k3, m): Stand for the downlink reuse mode when it equal to 1 that
DL

mean the

D2D pair number k is reusing the downlink of cellular user number m.

(3 D L)
 T k ,m : Stands the channel capacity downlink reuse mode when
the D2D pair number k is reusing the downlink of cellular user
number m.
 T (C D L)
k ,m : Stands the channel capacity of cellular user number m
when interfered by D2D pair number k.
3.8 THE MODE SELECTION, RESOURCE ALLOCATION, AND POWER CONTROL
BASED ON THE NETWORK LOAD

As mentioned in the earlier chapter each one of the three-communication mode needs
resources/channels to communicate. Thus, selecting the optimal mode largely depends on the
number of free resources (NU, ND). However, the number of free resources cannot be an accurate
measurement to network status and we cannot depend on it to set the optimal mode since the
network status can be affected by the number of active devices to device users but first, we have to
consider whether the two devices are in close proximity as summarized in the flow chart below in
figure 8, thus the communication scenario can be categorized as follow [10]:

35
Figure 10: Flow chart of the algorithm

Case 1: Light Load

Where The number of the available links (NU, ND) is more than or equal the number of active
device-to-device users, which means that there is at least one empty resource for each D2D pair
and thus, makes the cellular and the dedicated mode are the optimal modes in this case.

Case 2: Medium Load

In here the number of the available links (NU, ND) is less than the number of active device-to-device
users, which means the empty resource is not enough for all the D2D pair and thus makes the
dedicated mode and the reuse are the optimal modes in this case.

Case 3:Heavy Load

This case is a little bit difficult since no empty resource is available for any D2D pairs which
means the only way to communicate is to use the reuse mode.

3.9 PROPOSED SCHEMES AND ALGORITHMS

To maximize the overall system throughput while guarantee the SINR for both cellular users, and
D2D pairs links, we need to solve the optimization problem in (4.17) and find the optimal solution,

36
we will consider the base problem considered in [10]. I proposed a low-complexity, dynamic
algorithm that reuse the uplink and downlink channels of a cellular user unlike the approach
discussed in [10], which reused only the uplink.

The proposed algorithm will directly select the optimal communication mode bearing in mind the
distance between two D2D users, the availability from both uplink and downlink channels
comparing with number of D2D k pairs. It will then find the optimal power to reach optimal
solution and maximize achievable throughput. In Figure 9, a pseudo code for the main selection
process is declared. Our Optimal Algorithm will dynamically select one of these three cases:

Figure 11: Snipped image from algorithm code in MATLAB

Case 1: There are no orthogonal empty resources in the network. In this case the reuse mode will
be directly selected. However, in our algorithm the reuse mode will have two reusing cases:
reusing the uplink channel and reusing the downlink channels. First our optimal algorithm will
calculate the channel capacity for all K pairs in both cases.
For all K pairs
T3_UL (k, m) = Channel capacity for reusing uplink channel (k, m) T3_DL (k, m) = Channel
capacity for reusing downlink channel (k, m)

Then after that, the proposed algorithm for the reuse mode will take the maximum value for each
D2D pair number k and cellular user number m and set the proper mode based on the available
information.

If highest value form T3_UL then, the uplink reuse mode will be selected.
If highest value form T3_DL then, the downlink reuse mode will be selected.

Case 2: There are orthogonal empty resources in the network. In this case we will go to the
dedicated mode or reuse mode and thus due to the fact that there are resources but not for all DUs.
Simply our algorithm will distribute all the resources on the K pairs and for the ones who couldn’t
have the chance to have one of the empty orthogonal resources the reuse mode in both cases will be
selected. The question is how? Our algorithm will assume that there are no empty resources and do

37
the same procedure in (Case 1), then after that the algorithm will take the result of (Case 1) for the
all (K - (NU + ND)) and for the rest of K pairs the dedicated mode will be selected.
Case 3: There are orthogonal empty resources in the network for all K pairs. In this case we will go
to the dedicated mode or cellular mode. First our optimal algorithm will calculate the channel
capacity for all K pairs in both cases, then after that based on the result the algorithm will set the
proper mode by taking into consideration the number of K pairs and the number of the available
links without forgetting that the cellular mode needs two empty resources.

For comparison reasons and to provide an insight on the algorithm behavior and its effect on the
system performance also we examine the behavior of the proposed heuristic algorithms in [10]:

 Algorithm 1: This algorithm is the conventional cellular mode and it has been used for
declaring the difference between using the traditional way of communication all time and
using D2D technology when applicable.
 Algorithm 2: In here, this algorithm proposed to solve the medium load scenario where we
can find orthogonal empty resources but not enough for all D2D pairs. That is make the
dedicated mode and reuse mode are applicable options. However, in [10] the DUs can only
reuse the uplink channel only.

38
 CHAPTER FOUR: SIMULATIONS, RESULTS AND ANALYSIS

4.1 INTRODUCTION

A single cellular network with a radius of 500 m is considered in these simulations. The eNB is
located in the center of the cell and conventional Cellular CUs are randomly scattered with uniform
distribution in the cell. The Tk and Rk denote to the transmitter and the receiver of D2D pair k,
respectively. The Tk are randomly spread with uniform distribution in the cell. The cell network is
represented with radius R and r represent maximum distance between the Tk and Rk. In Figure 10,
we consider a single cellular user and Tx has full information to send to a single Rx.

Figure 12: showing the cell with random placed UEs and DUs in MATLAB

4.2 SIMULATION TOOLS

Due to many random components in the system model, Monte Carlo simulation is performed to
obtain statistically reliable results. In each Monte Carlo iteration, cell is created, and then CUs and
DUs are randomly placed within the cell. Path gains for all links are then calculated, and power
control and resource allocation are executed. After the last Monte Carlo iteration is executed,
parameters of interest (e.g., SINR, consumed power, throughput) can be evaluated.

4.3 SIMULATION PARAMETERS

In this research, all the links assumed to experience independent block fading. Some simulation
parameters, e.g., the number of DUs/CUs and the number of available links, are set divergently
according to the intended simulation scenario. However, some parameters are always kept at the

39
same values. Our Simulation parameters are inspired from [4] [10][13]. Table 3 summarized all
parameters.

Table 2: Simulation Parameters

Parameter Value

Cell Radius, R 500 m

D2D Distance, r 20….100 m
Uplink Bandwidth 3 MHz
Noise Spectral Density -174 dBm /Hz
Shadowing Standard Deviation 10 dB
SINR Threshold, (ε min) 10 dB

Maximum Number of Uplink 20

Channels
Maximum number of Downlink 20
Channels
Cellular
Number of Cellular Users, M 10-20
Maximum Tx power of CU, 18-27 dBm
(pmax ) c

Path Loss Model 128.1+37.6 log (d[km])

SINR Minimum Threshold, (ε ci,min Uniformly Distributed in [0, 25]

dB
D2D
Number of D2D Pairs, K 1-15
Maximum Tx Power of DU, (pdmax) 18-27 dBm

Path Loss Model 148+40 log (d[km])

SINR Minimum Threshold, (ε di, min Uniformly Distributed in [0, 25]
dB
4.4 RESULTS AND ANALYSIS

4.4.1 Impact of Maximum Transmit Power on Overall System Throughput

This section studies the impact of changing the transmit power on the overall system throughput.
The goal of this scenario is to illustrate the impact of increasing the maximum powers (pdmax, pcmax )
on the overall performance for the uplink and downlink D2D underlay with different channel

40
fading. The x-axes represent Pmax, where the maximum powers of both DUs and CUs is increased
from 18 dBm to 27 dBm. Similarly, y-axis represents the total system throughput [bits/sec/Hz].

Figure 11 shows the performance of (Algorithm 1) which is the worst of all, since it indicates only
the cellular mode in which the users use the maximum power and consumes a lot of power to
communicate with BS. Algorithm 2 shows a better performance than the (Algorithm 1) which is
obvious due to the fact of using the reusing mode compared to (Algorithm 1). However, my
proposed algorithm (Optimal Algorithm) superior all the algorithm since it selects the proper
mode directly and reuse both the uplink and downlink channels.

Moreover, my (Optimal Algorithm) outperform the other algorithm even with low transition
power. As a summary, from the simulation results, it shows that the proposed scheme (Optimal
Algorithm) in all scenarios has the best performance.

Figure 11: Overall system throughput for different algorithms using different maximum power
(where Nu = Nd = 2, r=0...30m, K = 10, M = 4, P_max^d=P_max^c=18-24 dBm)

4.4.2 Impact of Maximum D2D Distance on Overall System Throughput

This section studies the impact of changing the distance between the DUs. The aim of this scenario
is to show the effect of the maximum distance between the D2D pairs on the overall performance
of the uplink and downlink. The x-axes represent r (m), which is the mean distance between the
D2D pair where the distance increases from 20 -100 meters. Similarly, y-axis represents the total
system throughput in [bits/sec/Hz].

Figure 12 shows our (Optimal Algorithm) and (Algorithm 2) from [10] have the best
performance which is intuitive, since the D2D K pairs are more than the available orthogonal links,
the cellular mode by itself like in (Algorithm 3) will not solve the problem. However, only the

41
reuse mode will help us in this case and since our (Optimal Algorithm) and (Algorithm 2) from
[10] they are only one who have this mode therefore they outperform the other.

Figure 12: Overall system throughput for different algorithms using different maximum
D2D distance (where NU = ND = 2, r=20-100 m, K = 10, M =
4, P_maxd = P_max^c=24 dBm)

By comparing my algorithm with those already mentioned in terms of distance between D2D pair
we can observe that the optimal algorithm again beats algorithm 1(Cellular) and Algorithm 2 For
example, at r = 50 which is the maximum allowable we can see that my algorithm is at 1.3e8
compared to 1.25E8 and 0.86e8Bits/S/Hz

As a summary, the distance between the D2D users can play a significant role in selecting the
proper mode not only the number of available links as it can be seen from the results, when the
distance between the D2D transmitter and the D2D receiver is very big the possibility for losing
the connection is increased due to the interference factor for example. Also, the location of the CUs
can play an important role and give us the chance to take advantage of reuse downlink channel not
only the uplink one so that reusing the downlink channel can be fruitful in some cases.

4.4.3 Impact of the D2D Pair Numbers on Overall System Throughput

This section studies the impact of changing the number of D2D pairs comparing to exist resources
on the overall system performance. The aim of this scenario is to show the effect of the network
load on the overall system throughput. The x-axes represent K, which is the number of D2D pairs

42
while y-axis represents the total system throughput. In this scenario by comparing the number of
the available channels (network load) and K pairs, the result is shown in figure 13 below.

Figure 13:Overall system throughput for different algorithms using different number of K pairs

When we consider the number of D2D pairs, it is observed that the cellular mode (Algorithm 1)
throughput increases until it maintains a constant value at 5 since the number of free channels are
finished.

Also, Algorithm 2 [22] which reuses only uplink maintains a constant value at 10 due to no more
empty resources but my Optimal Algorithm’s throughput continues to increase with increase
number of D2D pairs due to reusing both uplink and downlink channels therefore it gives optimal
system throughput.

CHAPTER FIVE: CONCLUSION, RECOMMENDATIONS AND

FUTURE WORK
In this research, D2D communication for software defined radios to set up optimal mode for
network assisted D2D communications underlying cellular networks using uplink and downlink
resources is introduced. My scheme maximized the spectrum utilization efficiency and the system
throughput while at the same time minimizing the impact of interference and guaranteeing the
target SINR of both cellular and D2D links. By reusing the downlink link, we have to control the
power from the BS which is quite hard to compute therefore uplink reuse is preferred for DUs that
are far away from the BS (i.e., near the edge of the cell) since interference caused by the power
from the BS will be much reduced compared to those devices near the BS. The algorithm’s

43
practicality and performance enhancement were exhibited through numerical evaluations via
MATLAB-based simulations. As a result, a significant improvement to overall system throughput
was achieved by using this dynamic smart mode selection and considering the distance between the
D2D pairs, the total number of available downlink and uplink links and the allowable SINR caused
by reusing the uplink and downlink channels.

However, studying a single-cell scenario can be unrealistic scenario from real-world perspective.
Thus, studying multi-cell scenario with more than one D2D pair can reuse the CUs users can be
more realistic and promising future research direction. Though, these additions can make the
optimization problem be more difficult and complex.

REFERENCES
[1] K. Doppler, M. Rinne, C.Wijting, C. B. Ribeiro, and K. Hugl, “Device-to-device communication as an
underlay to LTE-Advanced networks,” IEEE Communication. Mag., vol. 47, no. 12, pp. 42–49, Dec.
2009.

[2] K. Doppler, M. P. Rinne, P. Jänis, C. Ribeiro, and K. Hugl, “Device-to-Device communications:

Functional prospects for LTE-Advanced networks,” in Proc. IEEE Int. Conf. Communication, Dresden,
Germany, Jun. 2009, pp. 1–6.

[3] P. Jänis et al., “Device-to-Device communication underlaying cellular communications systems,” Int.
J. Communication, Network, Syst. Sci., vol. 2, no. 3, pp. 169–178, Mar. 2009.

[4] Feng, D., Lu, L., Yuan-Wu, Y., Li, G.Y., Feng, G. and Li, S., 2013. Device-to-device communications
underlying cellular networks. IEEE Transactions on Communications, 61(8), pp.3541-3551.Meibergen,

44
M., 2011. Device-to-device communications underlying a cellular network. Delft University of Technol,
pp.1-117.

[5] C.-H. Yu, O. Tirkkonen, K. Doppler, and C. Ribeiro, “On the performance of device-to-device
underlay communication with simple power control,” in Proc. IEEE VTC— Spring, Barcelona, Spain,
Apr. 2009, pp. 1–5.

[6] Asadi, A., Wang, Q. and Mancuso, V., 2014. A survey on device-to-device communication in cellular
networks. IEEE Communications Surveys & Tutorials, 16(4), pp.1801-1819.

[7] NIU, L. and SALMAN, M., 2013. Resource allocation and power control for device-to-device (D2D)
communication. Master of Science Thesis, CHALMERS UNIVERSITY OF TECHNOLOGY, Goteborg,
Sweden.

[8] Pradini, A., 2013. Power control and resource allocation for device-to-device communications in
cellular networks.

[9] Xu, C., Song, L. and Han, Z., 2014. Resource management for device-to-device underlay
communication (pp. 1-79). Berlin: Springer.

[10] Yu, G., Xu, L., Feng, D., Yin, R., Li, G.Y. and Jiang, Y., 2014. Joint mode selection and resource
allocation for device-to-device communications. IEEE Transactions on Communications, 62(11),
pp.3814-3824.

[11] Shalmashi, S., 2015. Device-to-Device Communications for Future Cellular Networks: Challenges,
Trade-Offs, and Coexistence (Doctoral dissertation, KTH Royal Institute of Technology).

[12] Chen, F. and Wang, Z., 2015. Research on D2D Communication.

[13] Doppler, K., Yu, C.H., Ribeiro, C.B. and Janis, P., 2010, April. Mode selection for device-to-device
communication underlying an LTE-advanced network. In Wireless Communications and Networking
Conference (WCNC), 2010 IEEE (pp. 1-6). IEEE

[14] Doppler, K., Rinne, M., Wijting, C., Ribeiro, C.B. and Hugl, K., 2009. Device-to-device
communication as an underlay to LTE-advanced networks. IEEE Communications Magazine, 47(12).

[15] Janis, P., Koivunen, V., Ribeiro, C., Korhonen, J., Doppler, K. and Hugl, K., 2009, April.
Interference-aware resource allocation for device-to-device radio underlying cellular networks. In
Vehicular Technology Conference, 2009. VTC Spring 2009. IEEE 69th (pp. 1-5). IEEE.

[16] Zulhasnine, M., Huang, C. and Srinivasan, A., 2010, October. Efficient resource allocation for
device-to-device communication underlying LTE network. In Wireless and Mobile Computing,
Networking and Communications (WiMob), 2010 IEEE 6th International Conference on (pp. 368-375).
IEEE.

[17] Hakola, S., Chen, T., Lehtomaki, J. and Koskela, T., 2010, April. Device-to-device (D2D)
communication in cellular network-performance analysis of optimum and practical communication mode
selection. In Wireless Communications and Networking Conference (WCNC), 2010 IEEE (pp. 1-6).
IEEE.

45
[18] Jung, M., Hwang, K. and Choi, S., 2012, May. Joint mode selection and power allocation scheme for
power-efficient device-to-device (D2D) communication. In Vehicular Technology Conference (VTC
Spring), 2012 IEEE 75th (pp. 1-5). IEEE.

[19] Huawei Smartphone ecosystem R&D support team. 2012 “Future smartphone solution white paper”,
Issue 2.0, September 17, 2012.

[20] Mumtaz, S., Huq, K.M.S. and Rodriguez, J., 2014. Direct mobile-to-mobile communication:
Paradigm for 5G. IEEE Wireless Communications, 21(5), pp.14-23.

[21] Rodriguez, J. ed., 2014. Smart device to smart device communication. Springer International
Publishing. 40

[22] Networks Lab, TCS Research, (2015). [online] Available at: http://www. tsds
i.org/media/Meeting/2015-07-22/D2D_TSDSI_F2F_Bangalore_July_2015_v1 .pdf [Accessed 22 Apr.
2016].

[23] Wang, W. and Lau, V., 2014. Delay-aware cross-layer design for device-to-device communications
in future cellular systems. IEEE Communications Magazine, 52(6), pp.133-139.

[24] Haus, M., Waqas, M., Ding, A.Y., Li, Y., Tarkoma, S. and Ott, J., 2017. Security and Privacy in
Device-to-Device (D2D) Communication: A Review. IEEE Communications Surveys & Tutorials.

[25] Sun, H., Sheng, M., Wang, X., Zhang, Y., Liu, J. and Wang, K., 2013, August. Resource allocation
for maximizing the device-to-device communications underlying LTE-Advanced networks. In
Communications in China-Workshops (CIC/ICCC), 2013 IEEE/CIC International Conference on (pp. 60-
64). IEEE.

[26] Phunchongharn, P., Hossain, E. and Kim, D.I., 2013. Resource allocation for device-todevice
communications underlying LTE-advanced networks. IEEE Wireless Communications, 20(4), pp.91-100.

[27] Fodor, G., Dahlman, E., Mildh, G., Parkvall, S., Reider, N., Miklós, G. and Turányi, Z., 2012.
Design aspects of network assisted device-to-device communications. IEEE Communications Magazine,
50(3).

[28] Feng, D., Lu, L., Yuan-Wu, Y., Li, G., Li, S. and Feng, G., 2014. Device-to-device communications
in cellular networks. IEEE Communications Magazine, 52(4), pp.49-55

46