Interference Management in Nb-Iot For Heterogeneous Wireless Network

TALLINN UNIVERSITY OF TECHNOLOGY
School of Information Technologies
Jeffrey Leonel Redondo Sarmiento

177325IVEM
INTERFERENCE MANAGEMENT IN NB-

IOT FOR HETEROGENEOUS WIRELESS
NETWORK
Master’s thesis
Supervisors: Muhammad Mahtab

Alam
PhD
Co-Supervisor Hassan Malik
PhD
Tallinn 2019
TALLINNA TEHNIKAÜLIKOOL
Infotehnoloogia teaduskond
Jeffrey Leonel Redondo Sarmiento

177325IVEM
NB-IOT MÜRA HALDAMINE

HETEROGEENSETES JUHTMEVABADES
VÕRKUDES
Magistritöö
Juhendaja: Muhammad Mahtab

Alam
Doktorikraad
Kaasjuhendaja: Hassan Malik
Doktorikraad
Tallinn 2019
Author’s declaration of originality
I hereby certify that I am the sole author of this thesis. All the used materials, references
to the literature and the work of others have been referred to. This thesis has not been
presented for examination anywhere else.
Author: Jeffrey Leonel Redondo Sarmiento
06/05/2019
3
Abstract
The third-generation partnership project (3GPP) introduces in release 13 the new radio
access technology called narrowband internet of things (NB-IoT), which arouses in
different operation modes with the idea to re-use the same LTE spectrum. These modes
are stand-alone, guard band, and in-band. Consequently, the interference between these
two technologies will affect the service to the end users to some extent within a
heterogeneous network (HetNet) scenario. There are some studies which show that in-
band mode causes more interference due to the use of the same resource block by NB-
IoT and LTE technology. Therefore, in this master thesis, an extensive investigation of
interference caused by the new NB-IoT devices, in different HetNet deployment
strategies is presented together with a cooperative power management approach. Those
scenarios include the presence of NB-IoT either in Macro or in Small cell and considering
if both technologies are synchronous or asynchronous. From the simulations is
demonstrated that scenario 4 (Macro and Small Cells with NB-IoT) introduces more
interference because both cells enable the power boosting of 6 dB [1]. Additionally, the
cooperative method is implemented and studied, as a result, the method increases the
performance of edges users approximately 14% to 70%.
This thesis is written in English and is 59 pages long, including 6 chapters, 36 figures,
and 10 tables.
Key Words: NB-IoT, LTE, small cells, macro cells, interference, HetNet.
4
Annotatsioon
[Thesis title in Estonian]
[Tekst]
Lõputöö on kirjutatud [mis keeles] keeles ning sisaldab teksti [lehekülgede arv]
leheküljel, [peatükkide arv] peatükki, [jooniste arv] joonist, [tabelite arv] tabelit.
5
List of abbreviations and terms
3GPP Third Generation Partnership Project

ABS Almost blank subframe
BPSK Binary Phase-shift Keying
BS Base-station
CAPEX Capital expenditures
CDF Cumulative distribution function
CRC Cyclic Redundancy Check
DCI Downlink Control Information
DL Downlink
DRX Discontinuous Reception.
ECG Electrocardiography
eDRX Extended DRX
eICIC Enhanced Intercell Interference Coordination.
eNB Evolve Node Base-station
EPC Evolve Packet Core
FDMA Frequency Division Multiple Access
GSM Global System for Mobile Communications
HARQ Hybrid Automatic Repeat Request
HetNet Heterogeneous Network
IEEE Institute of Electrical and Electronics Engineers
IoT Internet of Things
IP Internet Protocol
ITU International Telecommunication Union
KPI Key Performance Indicator
LoRa Long Range
LPWAN Low-power Wide Area Network
LTE Long Term Evolution
M2M Machine to Machine
6
MCL Maximum Coupling Loss
MME Mobility Management Entity
MTC Machine type communication
NB Narrowband
NB-IoT Narrowband internet of things
NPBCH NB Physical Broadcast Channel
NPDCCH NB Physical Downlink Control Channel
NPDSCH NB Physical Downlink Shared Channel
NPRACH NB Physical Random Access Channel
NPSS/NPSS NB Synchronization Signal
NPUSCH NB Physical Uplink Shared Channel
OFDMA Orthogonal Frequency Division Multiple Access
OPEX Operational Expenditure
ORA Optimal Resource Allocation
PDCCH Physical Downlink Control Channel
PDN Public Data Network
PGW PDN Gateway
PRB Physical Resource Block
PSM Power Saving Mode
QoS Quality of Service
QPSK Quadrature phase-shift keying
RAN Radio Access Network
RRC Radio Resource Control
RU Resource Unit
SBS Signaling Radio Bearer
SC-FDMA Single Carrier Frequency Division Multiple Access
SGW Serving Gateway
SINR Signal to Interference Plus Noise Ration
SMS Short Message Service
SNR Signal to Noise Ratio
Taltech Tallinn University of Technology
TB Transport Block
TBS Transport Block Size
UE User Equipment
7
UL Uplink
VUE Victim UE
8
Table of contents
Author’s declaration of originality ................................................................................... 3

Abstract ............................................................................................................................. 4
Annotatsioon [Thesis title in Estonian] ............................................................................ 5
List of abbreviations and terms ........................................................................................ 6
Table of contents .............................................................................................................. 9
List of figures ................................................................................................................. 11
List of tables ................................................................................................................... 13
1 Introduction ................................................................................................................. 14
1.1 Background ........................................................................................................... 15
1.2 Problem Statement ................................................................................................ 16
1.3 Motivation and Research Contribution................................................................. 16
1.4 Chapter Review .................................................................................................... 17
2 NB-IoT Overview ........................................................................................................ 19
2.1 LTE Network ........................................................................................................ 19
2.1.1 Radio Resource Organization ........................................................................ 20
2.1.2 Interfaces ....................................................................................................... 21
2.2 NB-IoT standard, specifications, and performance .............................................. 21
2.2.1 Performance ................................................................................................... 22
2.2.2 Resource grid for NB-IoT ............................................................................. 23
2.2.3 NB-IoT Physical channels and Subframe...................................................... 24
2.2.4 Signals and Channels in Downlink................................................................ 25
2.2.5 Signals and channels in the uplink ................................................................ 26
2.2.6 Narrow Band IoT modes ............................................................................... 27
3 State of The Art ........................................................................................................... 29
3.1 Evaluation of NB-IoT guard-band mode .............................................................. 29
3.1.1 Evaluation of NB-IoT in-band mode ............................................................. 30
3.2 Resource Management in Cellular Network ........................................................ 31
3.2.1 Conventional Frequency Reuse ..................................................................... 32
9
3.2.2 Almost blank subframe (ABS) ...................................................................... 32
3.2.3 Uplink Resource Scheduling for NB-IoT and LTE Hybrid Transmission .... 34
3.2.4 Interference awareness .................................................................................. 35
3.2.5 Cooperative Approach ................................................................................... 36
4 Different Deployment Strategies in an HetNet Environment ...................................... 38
4.1.1 Scenario 1 – Small cell coverage only with NB-IoT enabled ....................... 39
4.1.2 Scenario 2- Macro Cell LTE and Small Cell NB-IoT ................................... 40
4.1.3 Scenario 3- Macro Cell NB-IoT and Small Cell LTE ................................... 41
4.1.4 Scenario 4 – Macro and Small Cell NB-IoT ................................................. 41
4.1.5 Scenario 5- Macro cell LTE and Small cell randomly assign NB-IoT ......... 42
5 Performance Evaluation of NB-IoT in HetNet Scenario ............................................. 44
5.1 Simulation Setup................................................................................................... 44
5.1.1 Simulation Software and Parameters ............................................................. 44
5.1.2 Models and Formulas .................................................................................... 45
5.2 Simulation Process ............................................................................................... 49
5.2.1 Scheduling of None Optimization Algorithm Implemented ......................... 49
5.2.2 Scheduling of Cooperative Algorithm (Optimized) ...................................... 53
5.3 Simulation Results ................................................................................................ 56
5.3.1 None Cooperative Case (DL) ........................................................................ 56
5.3.2 None Cooperative Case (UL) ........................................................................ 57
5.3.3 Cooperative Case (DL) .................................................................................. 59
5.3.4 Cooperative Case (UL) .................................................................................. 62
6 Conclusion ................................................................................................................... 65
References ...................................................................................................................... 67
Appendix A – Simulation with different penetration loss .............................................. 69
10
List of figures
Figure 1. LTE Network Diagram. .................................................................................. 19

Figure 2. Physical Resource Block equivalent to twelve subcarriers. ............................ 21
Figure 3. NB-IoT Frame for 15 kHz spacing [18]. ......................................................... 23
Figure 4. NB-IoT Frame for 3.75 kHz spacing [18]. ...................................................... 24
Figure 5. Physical Resource Block NB-IoT a) Downlink [18] 15kHz and b) Uplink [19].
........................................................................................................................................ 24
Figure 6. Channels of NB-IoT plotted on the subframe [15]. ........................................ 25
Figure 7. NB-IoT Operational Modes [21]..................................................................... 27
Figure 8. SINR when LTE is the victim [22].. ............................................................... 29
Figure 9. SINR when NB-IoT is the victim [22].. .......................................................... 30
Figure 10. Co-channel interference in an asynchronous network. ................................. 31
Figure 11.Frequency Reuse factor 1 and 3 (reuse-1 and reuse-3). ................................. 32
Figure 12. ABS propose frame [5]. ................................................................................ 33
Figure 13. The scenario of the victim and non-victim UEs (VUE and NVUE) [5]. ...... 33
Figure 14. CDF of macro UE data rates for the demand of 0.5Mbps. .......................... 34
Figure 15. Establishment of a hybrid transmission strategy [6].. ................................... 35
Figure 16. Graph Results: a) Average information rate comparison. b) Average latency
vs penetration loss. ......................................................................................................... 35
Figure 17. Three procedure described on patent [7]. ...................................................... 36
Figure 18. CDF average information rate and Average Energy consumption [8]. ........ 37
Figure 19. First Scenario – Only Small Cell with NB-IoT Enabled............................... 39
Figure 20. Second Scenario – Macro Cell with NB-IoT disabled, and Small Cell with
NB-IoT Enabled. ............................................................................................................ 40
Figure 21. Third Scenario – Macro Cell with NB-IoT enabled, and Small Cell with NB-
IoT disabled. ................................................................................................................... 41
Figure 22. Fourth Scenario – Macro and Small Cell with NB-IoT Enabled. ................. 42
Figure 23. Fifth Scenario- Macro Cell and Small Cell with NB-IoT enabled randomly.
........................................................................................................................................ 43
Figure 24. Algorithm I, scheduling of None Optimization Algorithm Implemented..... 50
11
Figure 25. Algorithm II, Calculation of SINR DL. ........................................................ 50
Figure 26. Algorithm III, Calculation of SINR UL. ....................................................... 51
Figure 27. MATLAB Downlink SINR with respect to BS. ........................................... 51
Figure 28. MATLAB Uplink SINR with respect to UE. ................................................ 52
Figure 29. Flow Diagram of the simulation for the throughput calculation. .................. 53
Figure 30. Algorithm IV, Cooperative Method. ............................................................. 54
Figure 31. Flow Diagram of Cooperative Algorithm. .................................................... 55
Figure 32. Downlink throughput and SINR results for all scenarios. ............................ 57
Figure 33. Uplink throughput and SINR results for uplink scenarios. ........................... 58
Figure 34. Downlink scenario comparison between no cooperative and cooperative.. . 61
Figure 35. Scenario 1 and 2 Uplink comparison between no cooperative and cooperative
........................................................................................................................................ 63
Figure 36. Radar Graph for each of the scenarios .......................................................... 64
12
List of tables
Table 1. LTE physical layer bandwidth options and bandwidth specific parameters [12].
........................................................................................................................................ 20
Table 2. The time-frequency size of RU in NPUSCH [21]. ........................................... 24
Table 3. NB-IoT channels and Signals [14]. .................................................................. 25
Table 4. RF Parameters for Simulation. ......................................................................... 45
Table 5. MCL vs Repetitions [8]. ................................................................................... 45
Table 6. Parameters for Throughput Calculation. The total resource element for DL and
UL is [8]. ........................................................................................................................ 45
Table 7. MCS vs SINR for PDSCH [34]. ....................................................................... 47
Table 8. MCS vs SINR for PUSCH single tone [34]. .................................................... 47
Table 9. Mapping Between MCS, RUs, and TBS for PDSCH [33]. .............................. 48
Table 10. Mapping Between MCS, RUs, and TBS for PUSCH [32]. ............................ 48
13
1 Introduction
Communication systems have evolved rapidly over the past two decades allowing to
develop new technologies to guarantee connectivity to end users, which are either people
using wearables or devices deployed in the industrial or rural sector, constantly
monitoring crucial data for their proper operation (e.g. environmental measurement) and
sending the data through the internet. By 2023 Ericsson has predicted that there will be
more than 3.5 billion cellular internet of things (IoT) connections[2]. Much of those
connections will be wearable sensors located indoor doing machine type communication
(MTC) either MTC device to MTC server or machine-to-machine communication
(M2M). MTC technology requirements are low-cost devices, ubiquitous coverage, and
ultra-low battery life to achieve 10 years durability [3]; long term evolution-MTC (LTE-
MTC) was launch in LTE release 11 [3] aiming to use the cellular network to support the
massive IoT. Few examples of IoT devices and applications are electricity meter; smart
watches; biomedical electrocardiography (ECG) wearable; temperature, gas, and water
meter sensors [4], smart farm, smart cities, smart grids, smart supply chain, etc. These
devices generate a huge amount of data exhibiting a saturation for cellular operator,
subsequently, the capacity of the network requires to be increased obliging telecom
operators to deploy new technologies; and small cells which its implementation becomes
easier diminishing operational cost compared to macro cell offering the advantage of
increasing the capacity of the network and indoor coverage.
To fulfil above mention requirement, high efficiency energy, low data rate, and support
to massive IoT; a wireless connectivity technology which supports low power
transmission, low bandwidth, long coverage is narrowband internet of thing (NB-IoT)
launched by 3rd generation partnership project (3GPP) in long term evolution (LTE)
release 13. NB-IoT offers more advantages than his predecessor LTE-MTC, offering
higher energy efficiency and reduce bandwidth. However, by deploying new base stations
either macro or small cell utilizing NB-IoT technology to satisfy the network capacity
(massive IoT) and low-power devices requirements, raises a new question; how the
14
interference in different deployment strategies in a heterogeneous network, with NB-IoT,
affects the performance of the cellular network.
1.1 Background
Low-power wide area network (LPWAN) technology is characterized for offering

connectivity to low-power devices with low data rate. It fits in applications when long-
distance connection reliability, long battery life, high-density population of devices, and
a small amount of data rate are highly required. This is achieved by reducing the
modulation scheme and using new protocol designs to reduce the packet size increase by
headers and error-correction.
NB-IoT technology is considered LPWAN. It offers the advantage of a low power

communication system in which the devices can have a battery lifetime of approximately
10 years, increases the coverage area, reduces overhead and power consumption; in
addition, implementing this technology in the existing LTE standard facilitates the time
to market, reduces the capital expenditures (CAPEX) and operating expense (OPEX). By
only upgrading their base station with new software which includes the capabilities of
NB-IoT, no investment must be done in hardware or the evolved packet core (EPC). This
is a great advantage compared with competitor technology long rage (LoRa).
Additionally, NB-IoT comes with three different modes of operations stand-alone, guard-
band, and in-band. Each of the modes affects differently the performance of NB-IoT and
LTE co-existence. For stand-alone, narrowband IoT is deployed in the GMS technology
with a bandwidth of 180 kHz. For guard-band, the LTE guard band is used for the
deployment of narrowband IoT, these two modes produce and overcome the interference
issue better compare to in-band due to the separation in frequency with LTE. However,
in-band which is deployed inside the LTE bands generates questions regarding
interference due to the use of the same physical resource blocks between neighboring
cells. The focus of this thesis is to evaluate and present the results of such deployment
scenarios in a HetNet architecture.
15
1.2 Problem Statement
By having this new technology deployed in a production network brings new challenges
to engineers, which are constantly monitoring the performance of the network to provide
the best service to end users. Massive interference is expected in full-scale deployments
scenarios.
NB-IoT over LTE functions in three different modes, which are standalone, guard-band,
and in-band, each mode is affected differently by users which are attached to a
neighboring cell and transmitting in the same frequency allocation or time slot. As a first
step, new deployment scenarios need to be proposed and performance must be evaluated.
Such a scenario can represent future 5G heterogeneous network (HetNet) environment
between macro and small cell and NB-IoT deployment in those setups. Further
interference management should be implemented and evaluated.
1.3 Motivation and Research Contribution
Therefore, this is a motivation for researchers to re-produce those challenges in

simulations to illustrate how this new technology affects the service, with the goal to
become a tool for engineers to properly deploy NB-IoT in their network without a
negative impact over the former perception user already possess.
In this thesis, two main key performance indicators (KPIs) are evaluated, signal to
interference plus noise ratio (SINR) and throughput. Different deployment scenarios
present or experiment higher interference than others. Consequently, five scenarios are
simulated, and their results are compared. In some of the eNBs, narrowband IoT will be
enabled or disabled, as in a real environment is, depending on, if the zone covered by the
eNB has NB-IoT users. Also, it could be found that the technology is in synchronous or
asynchronous mode. The descriptions of those scenarios are in Chapter 3. The simulation
is performed in MATLAB.
Many of the previous work, does not include this deployment scenarios analysis, they
only cover one scenario without the incursion of HetNet. In addition to the comparison
of the different scenarios, there are other studies which present a new algorithm or method
to avoid interference between the technologies. One of them is the almost blank subframe
(ABS) [5] in which some specific subframes are reserved for small cells, this could be
16
used also for NB-IoT; another approach is a hybrid transmission [6] by checking the SNR
the transmission is either LTE or NB-IoT. And cooperative method, which is already
patented [7] and studied in paper [8], is evaluated in this thesis to visualized how the
SINR improves, hence the throughput.
Overall, the main concern is the interference, in this thesis, the evaluation of inter-cell
interference is presented for 5 different scenarios which generate important data for real
environment deployment strategies; subsequently, a cooperative method is adopted and
evaluated to manage or reduce or cancel the inter-cell interference in a HetNet
environment with the purpose of increasing the overall throughput for downlink and
uplink.
Furthermore, part of the activities from this thesis has resulted in two publications [9],
[10], and few are under preparation:
1.4 Chapter Review
The structure of this thesis as follows:
Chapter 2 presents a review of LTE and NB technology. It starts with the explanation of
LTE network architecture, functionalities, bandwidth parameters, and the resource blocks
which is the most important part that needs to be understood to comprehend how NB-IoT
functions. Then, the specifications, performance, new features, and channels of NB-IoT
are presented. At last, an explanation of the different modes of the technology operation.
Overall, here is the overview of the technology offering an understanding of how it works
and presenting what can be achieved.
In Chapter 3, the state of the art or previous works related to the evaluation of the different
modes regarding the deployment of NB-IoT are presented; for instance, guard-band and
in-band evaluation of interference, the results are shown in a cumulative distribution
function (CDF) graph. Besides, some methods that were proposed with the aim of solving
the interference caused by frequency reuse method for spectrum efficiency, are explained.
In addition, some studies associated with scheduling to improve the performance
concerning the interference introduces by NB-IoT are reported too. This chapter gives the
idea of what has been accomplished and what could be implemented to evaluate the new
scenarios involve in a HetNet environment with NB-IoT enabled.
17
In Chapter 4, the descriptions of the five deployment strategies scenarios are described in
detail explaining which technology is enabled and from where the interference is
expected. In each sub-session, which corresponds to each scenario, there is a figure and
SINR formula that shows which assumptions are taken for the simulation. This shows the
possible deployment scenarios telecom operators can have in their network, and in
Chapter 5 the evaluation is presented in terms of throughput, energy, and SINR.
In Chapter 5, All information regarding the simulation such as RF parameters, bandwidth,

path loss models, maximum coupling loss (MCL), noise floor, throughput and code rate
selection is specified. Then, the simulation method and process are described. Finally,
results and analysis are presented with the goal of concluding which strategy is more
convenience to deploy in a real HetNet environment.
In Chapter 6, the cooperative approach is explained and studied for all the scenarios, the
result and analysis are shown giving the conclusion of how the cooperative method
implemented in a HetNet environment improve the experience of the end user, in this
case, higher throughput and lower energy consumption.
In Chapter 7, finally, the conclusion is presented and some open questions are raised and
new investigation opportunities are presented; for instance, interference prediction. If the
prediction is used the cooperative method will be more helpful due to reducing the delay
of the communication between neighbors, also artificial intelligent can be used for this
prediction process.
18
2 NB-IoT Overview
This chapter describes in detail how NB-IoT technology works plus how the bandwidth
180 kHz is distributed, channels such as narrowband physical downlink control and
shared channel (NPDCCH and NPDSCH), and different mode of operations descriptions.
The NB-IoT technology is independent of the LTE but both run in the same spectrum, in-
band, and guard-band mode; additionally, this technology inherits many of LTE features,
though, it is necessary to describe the LTE Network before moving on details with NB-
IoT.
2.1 LTE Network

LTE network consists of the radio access network (RNA), and the evolved packet network (EPC), see
Figure 1. The enhance node base-station eNB, which is part of the RNA, is the one in
charge of managing and scheduling the resource to the end users. The EPC is the joined
of three main nodes which are mobility management entity (MME), serving gateway
(SGW), and public data network (PDN) gateway (PGW).
EPC (Evolved Packet Core)
MME
S1
eNodeB
X2
S1
SGW PGW
eNodeB
Figure 1. LTE Network Diagram.

The entity MME manages the signaling known as control plane. The signaling is related
to security and mobility for the evolve universal mobile telecommunication system
19
(UTMS) terrestrial radio access (E-UTRAN) being the air interface. Besides, it is
responsible for carrying or handling the tracking and paging of user equipment (UE) in
idle-mode [11].
The SGW and PGW control the user data known as user plane, they transport the IP data
between the external network and UE. The serving gateway is the interconnection
between the eNB and EPC and it is known as the anchor [11].
LTE is based on Orthogonal Frequency Division Multiple Access OFDMA in downlink

and single carrier frequency division multiple access SC-FDMA in the uplink.
LTE can work on different bandwidth from 1.4 MHz to 100 MHz. LTE advance can use
more bandwidth and multi-carrier set up to improve the data rate. Below a table with the
specification for each of the bandwidth option [12].
Bandwidth (MHz) 1.4 3.0 5.0 10 15 20

Sub-frame duration 1 ms
Sub-carrier spacing 15 kHz
FFT length 128 256 512 1024 1536 2048
Sub-Carriers 72 180 300 600 900 1200
Symbols per slot 7 with Short CP and 6 with Long CP
Cycle prefix (CP) 5.210 us with Short CP and 16.67 us with Long CP
Table 1. LTE physical layer bandwidth options and bandwidth specific parameters [12].
2.1.1 Radio Resource Organization
For uplink and downlink, the 180 kHz in 1ms of the sub-frame corresponds to one single
LTE Physical Resource Block (PRB), this is equal to 12 subcarriers [12], [13] of 15 kHz,
and one subcarrier and one symbol is a resource element (RE), see Figure 2.
20
1 Resource Block = 180 kHz
Subcarrier spacing = 15 kHz
= 12 Subcarriers
1 slot = 0.5 ms Frequency

= 7 OFDM UE 1 UE 3
Symbols
UE 2
1 Subframe =
1 ms = 1 TTI = 1
UE 4 UE 5 UE 6 UE 3
Resource Block Pair
QPSK, 16 QAM or 64 QAM Modulation

Time
Figure 2. Physical Resource Block equivalent to twelve subcarriers. One slot equivalent to 0.5 ms and 7
OFDM symbols, and one subframe of 1 ms equal to a resource block pair [13].
The UEs are allocated in resource blocks by the eNB depending on many criteria such as
traffic demand, channel conditions, and quality of service QoS [12]. For instance, if the
user has a low signal to noise ratio SNR the eNB assigns less RB, low modulation scheme
to guarantee the transfer of the data with translate in a low throughput or the use of
diversity.
2.1.2 Interfaces
Another important characteristic of LTE is the interfaces between the nodes.

• S1, this interface connects the eNB to the EPC, it carries control and data plane
information.
• X2, this interface provides connectivity between eNBs allowing them to perform
a new task such as handover without involving the core and releasing the core of
tasks to avoid signaling overload.
2.2 NB-IoT standard, specifications, and performance
NB-IoT uses the same frequencies used in LTE and can be deployed in all the bandwidth
options, except for bandwidth 1.4MHz [4]. It minimizes the signal overhead, especially
over the radio interface, improve battery life, support delivery of IP and non-IP data, and
SMS support. In spite of this, it does not support all the feature built-in in LTE, for
21
instance, multiple radio access technology (Multi-RAT) and handover [4]. However,
inherited many of them that are crucial such as idle mode, power saving mode, paging,
and access control [14].
In the data transmission, includes new optimization, one of them is the ability to transmit
a small amount of data in the control plane via signaling radio bearer (SBS). Another one
is the ability to suspend and resume the radio resource control (RRC), which eliminate
the need of having a new connection at each reporting instance [14].
2.2.1 Performance
As stated at the beginning of this section, NB-IoT offers advantages as follow, these
descriptions are taken from [14], [15], [16], and [17]:
a. Coverage: enhance coverage by 20 dB [16] corresponding to a Maximum

Coupling Loss (MCL) of 164 dB. NB-IoT supports three different coverage
levels: 0, 1, and 2. Repetition is an important key performance indicator (KPI) of
coverage, less or more repetitions describe the coverage level. For 0 to 10 dB few
repetitions or none are needed to reach a high data rate, where 20 dB the coverage
can be maintained by sending more repetitions but low data rate.
b. Capacity: The target is to support 52K at least [17]; however, the system-level
simulation result given in [14], it could support 250K devices in a cell sector per
carrier. With narrowband physical downlink shared channel (NPDSCH) peak data
rate of 226.7 Kbps layer1 and narrowband physical uplink shared channel
(NPUSCH) peak data rate of 250 Kbps layer 1 [17]. However, when the time
offsets between downlink control information (DCI), NPDSCH/NPUSCH, and
hybrid automatic repeat request (HARQ) acknowledgment are taken into account
[17] the data rate for downlink and uplink are lower than the ones stated above.
c. Energy efficiency: The target is to achieve a battery life of more than 10 years at
the maximum coverage level using a battery capacity of 5 Wh [15]. For this NB-
IoT keeps the same power saving mechanisms but extending the timer values to
achieve a longer battery lifetime. Those methods are Discontinuous Reception
(DRX) and power saving mode (PSM).
22
• eDRX: This is extended DRX is required for NB-IoT UE to save power
consumption. Ant the DRX cycle maximum value of 10485.76s [14].
• PSM: “In this mode, the UE remains registered to but not reachable by
the network. The UE is in the power-off or sleep mode and will wake up
only when there is data to send after timer expiration” [14].
d. Latency: Target of a maximum latency of 10 seconds. Which indicates that this

technology should be utilized for the situation where higher latency is acceptable
[14].
2.2.2 Resource grid for NB-IoT
NB-IoT uses two different carrier spacing in uplink i.e., 3.75 kHz, and 15kHz, for
downlink the spacing is only 15kHz. However, in the uplink the spacing could be either
3.75 kHz or 15 kHz and it can have single-tone or multi-tone transmission. The 3.75 kHz
spacing is for single-tone only, and 15 kHz can be utilized for both single and multi-tone,
see Table 2. Below figures Figure 3, Figure 4, and Figure 5 (PRB DL and UL) showing
the distribution of RB [14]. For NB-IoT the resource unit is introduced as the smallest
amount of time-frequency resource [14].
Figure 3. NB-IoT Frame for 15 kHz spacing [18].
23
Figure 4. NB-IoT Frame for 3.75 kHz spacing [18].
1 frame = 10ms
1 slot = 2ms
3.75 kHz
48 Subcarrier=180kHz
a) b)
Figure 5. Physical Resource Block NB-IoT a) Downlink [18] 15kHz and b) Uplink [19].
Physical Transmission Subcarrier The # of Duration

Channel Mode Spacing Subcarriers
3.75 kHz 1 32ms
Single-Tone
15 kHz 1 8ms
NPUSCH 15 kHz 3 4ms
Multi-Tone 15 kHz 6 2ms
15 kHz 12 1ms
Table 2. The time-frequency size of RU in NPUSCH [20].
2.2.3 NB-IoT Physical channels and Subframe
As mentioned before the NB-IoT occupies a bandwidth of 180 kHz that correspond to
one PRB of LTE as described in Figure 2; on the other hand, the number of channels is
different. Table 2 contains usage information of the different channels and signals,
nevertheless they are explained in more detail in [14]. See in Figure 6 [15], the NB-IoT
subframe with the allocation of the channels and signals stated in Table 3.
24
Channels/Signals Usage
Narrowband Physical Downlink Control Channel
Uplink and Downlink scheduling information
(NPDDCH)
Narrowband Physical Downlink Shared Channel
Downlink dedicated and common data
DL (NPDSCH)
Narrowband Physical Broadcast Channel (NPBCH) Master information for system access
Narrowband Synchronization Signal (NPSS/NSSS) Time and frequency synchronization

Narrowband Physical Uplink Shared Channel
Uplink dedicated data
(NPUSCH)
UL
Narrowband Physical Random-Access Channel
Random access
(NPRACH)
Table 3. NB-IoT channels and Signals [14].
Figure 6. Channels of NB-IoT plotted on the subframe [15]. In the uplink, it is described as single and
multi-tone the channel PUSCH. The multi-tone is shown in the radio frame on the top left of the figure with
a different color; on the top left, it is shown the single tone for 3.75 kHz subcarrier spacing. On the bottom,
it is described the downlink radio frame RB with the channels and signals distributions.
2.2.4 Signals and Channels in Downlink
Part of the description of these signals and channels are extracted from [8], [14]:
a) Narrowband reference signal (NRS): To provide phase reference, the signal

NRS is transmitted in subframes that are dedicated to broadcast or downlink
transmission.
25
b) Narrowband primary and secondary synchronization signal (NPSS/NSSS):
This signal is transmitted every 10 ms, and 20 ms, respectively; these signals are
used for cell searching by using time and frequency synchronization and cell
detection.
c) Narrowband physical broadcast channel (NPBCH):
The NB master information block size is 50 bits, similar to LTE MIB supplies the
UE with the system frame number (SFN), operation mode, channel raster
(depending on the operation mode) and LTE cell-specific reference signal (CRS),
system information block (SIB).
d) Narrowband physical downlink control channel (NPDCCH):
It carries the most important information paging, downlink/uplink assignment,
random access channel response, type of modulation, and power control. The size
is fixed to 23 bits in one subframe. For extended coverage, it supports 2048
repetitions. For NB-IoT, three new DCI format are defined: N0 for NPUSCH
scheduling, N1 for NPDSCH scheduling and NPDCCH order, and N2 for paging
and direct indication.
e) Narrowband physical downlink shared channel (NPDSCH):
It is scheduled by NPDCCH and transmitted after the transmission of NPDCCH,
this delay is 4 ms. The maximum transport block size (TBS) is 680 bits and can
be mapped in a maximum of 10 subframes, the set is {1, 2, 3, 4, 5, 6, 8, 10}. Error
detection CRC of 24 bits.
2.2.5 Signals and channels in the uplink
a) Demodulation reference signal (DMRS):
It is transmitted only in RUs which contains data.
b) Narrowband physical random access channel (NPRACH):
It is based in a single-tone transmission. The NPRACH resource configuration is

divided into three different levels of coverage. It used by the UE to camp on the
base station (BS). The resource configuration is performed by the estimation of
the uplink timing with the aim of maintaining orthogonality. For extended
coverage, the maximum repetition is 128.
26
c) Narrowband uplink shared channel (NPUSCH):
This channel support single and multi-tone as aforementioned in section 2.1.1

Radio Resource Organization. The largest transport block is 10 resource units. It
provides the extended coverage by the time-domain repetition, and low peak-to-
average-power ratio modulation schemes BPSJ and QPSK. For this channel, two
formats exist, the first format is for carrying uplink data and error correction, the
second format is for the hybrid automatic repeat request (HARQ)
acknowledgment for downlink data. The maximum number of repetitions is 128.
The maximum TBS is 1000 bits and can be mapped in {1, 2, 3, 4, 5, 6, 8, 10]
resource units in time.
2.2.6 Narrow Band IoT modes
NB-IoT can be implemented in three different modes, GSM-standalone (Global System

for Mobile communications), in-band, and guard band. See Figure 7.
Figure 7. NB-IoT Operational Modes [21].
For standalone mode, the GSM technology is used by the replacement of one or more
GSM carrier. This allows efficient re-framing of GSM carriers for IoT [21].
Over in-band operation, some of the LTE PRB are reserved to NB-IoT which indicates
none LTE PRB will be transmitted where the NB-IoT PRB is allocated and vice-versa.
The use of this operation mode increases the spectrum efficiency because of the reuse of
the LTE spectrum. The total power is shared between both technologies with the
possibility to use it in advantage for power boosting on NB-IoT [21]. Additionally, by
deploying NB-IoT in this operational mode increases the interference between neighbors
for both technology either being the victim LTE or NB-IoT PRB. The interference
problematic is seen in more scale when single-tone in uplink with 3.75 kHz spacing is
used. Nevertheless, “this interference can be reduced by scheduling users with similar
27
SNR requirements in NB-IoT nearby LTE PRBs. On the other hand, if 15kHz subcarrier
spacing is used, LTE and NB-IoT orthogonality are maintained” [21] which guarantee
more stability in the system.
The guard-band mode offers a good co-existence between LTE and NB-IoT technology
because the PRB assigned to NB-IoT are the ones in the LTE guard-band. “Each carrier
is within the guard-band and the center frequency is at a most 7.5kHz offset from the
100kHz channel raster. In addition, the orthogonality with LTE is maintained” [21].
28
3 State of The Art
This chapter includes the overview of some of the research published in the evaluation of
NB-IoT modes (guard-band and in-band modes), interference management and
cancellation either in NB-IoT technology or HetNet environment.
3.1 Evaluation of NB-IoT guard-band mode
Guard-band mode offers good protection against interference in either technology, LTE
or NB-IoT. In [22], it is presented the study of the interference experienced by LTE or
NB-IoT as a victim. The biggest impact of interference is LTE over NB-IoT in the uplink;
however, the interference caused by the LTE user is not significant to deteriorate the
performance of narrowband IoT. In Figure 8, and Figure 9, it is shown the SINR when
LTE and NB-IoT are the victim respectively. The evaluation was performed under the
condition that the separation of 0 Hz between technologies.
Figure 8. SINR when LTE is the victim [22]. C/I means carrier over interference, w/o without, and w with.
This graph shows the uplink evaluation of interference over LTE when NB-IoT is considered the aggressor
technology.
In the paper describes the first scenario when NB-IoT interferes LTE name it also as NB-
IoT the aggressor and LTE the victim. As seen in Figure 8, the CDF graph shows a small
change of carrier over interference (C/I) between 5 to 15 dB, the degradation was of 0.7
dB at 95% SNR, this is interpreted in throughput loss of 4.7% [22].
29
Figure 9. SINR when NB-IoT is the victim [22]. C/I means carrier over interference, w/o without, and w
with. This graph shows the uplink evaluation of interference over NB-IoT when LTE is considered the
aggressor technology.
When NB-IoT is considered the victim, the interference increases as shown in Figure 9.
The SNR loss was of 2 dB approximately. Overall, mutual interference exists, when LTE
is the victim the interference is relatively low due to the higher bandwidth (10 MHz)
compare to NB-IoT 200 kHz bandwidth. The simulation assumptions or parameter can
be found in [22].
3.1.1 Evaluation of NB-IoT in-band mode
In [21] the evaluation involves a number of cells in which NB-IoT technology is activated
with the aim of measuring the interference caused by the neighboring cells. It is
mentioned that the impact is shown in two-folds:
• The relative sparse deployment of NB-IoT results in a larger area that needs to be
covered by each NB-IoT cell.
• The NB-IoT devices that are on the edge coverage or remote from the serving cell
could potentially be covered by an LTE cell, resulting in strong co-channel
interference.
Hence, if an NB-IoT device with the second condition stated above would have a low
SINR, because of the near LTE cell which is transmitting in the same PRB. This can be
improved by power boosting, NB-IoT standard limits it to 6 dB [21].
The simulation assumption can be found in [21]. The simulation is run for three different
NB-IoT deployment densities of 50%, 75%, and 100%, and in synchronous mode and
30
asynchronous mode between cells. Figure 10 illustrates the co-channel interference when
the cells are not synchronized. Nonetheless, When the “cells are synchronized the LTE
subcarriers from adjacent PRBs of the second cell are orthogonal to the NB-IoT
subcarriers in the first cell” [21], and when they are not synchronized this orthogonality
is lost and adjacent PRBs of the neighboring cell could potentially introduce interference.
Figure 10. Co-channel interference in an asynchronous network. On top of the figure, it is the resource
usage of a cell with NB-IoT, on the bottom, it is the resource usage of a cell with only LTE [21].
To solve the problem of co-channel interference it is proposed blanking the PRB used for
NB-IoT on the LTE cell. This can be seen in Figure 10, where the unused (blank) PRBs
on the resource usage on the bottom of the figure (LTE only) are not utilized for
transmission. Moreover, the cells should be synchronous which guarantee orthogonality.
3.2 Resource Management in Cellular Network
In HetNet, the macro and small-cell operate in different channels or co-channel. The
advantage of using different channels or dedicated channel is the interference is not an
issue and reduce the complexity of deployment, but this implies the operators should own
a license for each frequency band utilized which translates in new investment or partition
of their spectrum in few parts to assign one channel for each type of cell indicating a
decrement in their network capacity.
Therefore, it is necessary to introduce a system that shares the spectrum efficiently. By

sharing the spectrum, the interference becomes the problem to solve by having an
efficient allocation or scheduling process. The interference affects negatively the
performance of the small cell due to the higher transmission power of the macro cells.
One of the approaches to overcome this problem is the use of OFDMA systems. Even
31
though, there should be a good scheduling algorithm along the OFDMA system to
mitigate the interference in some extends to improve the performance of the network.
3.2.1 Conventional Frequency Reuse
The frequency reuse factor 1 or reuse-1 scheme is known as the simplest frequency reuse
method where the complete bandwidth is reused in each cell [23]. In this scheme, all the
cells use the same frequency band and without any power limitation, resulting in the
maximum throughput [24]. Nevertheless, reuse-1 introduces high interference. Though,
the reuse-3 or frequency reuse factor 3 divides the total bandwidth into three equals and
orthogonal sub-bands [23]. The sub-bands are assigned to adjacent cells which the
condition of not repetition. The resue-3 reduces or divides the bandwidth to avoid
interference. However, it decreases the throughput due to the use of a third of the
bandwidth. Reuse-3 becomes the first or simplest form of static interference coordination
[24]. The illustration of both frequency reuse factors are in the figure below,
Reuse 1
S1
S2
Power
S1 S2
2
S3 S3
Frequency
Reuse 3
S2 S1
S1
Power
S2
S3 2
S3
Frequency
Figure 11.Frequency Reuse factor 1 and 3 (reuse-1 and reuse-3). In reuse-1, the three sectors (S) use the
same band, therefore, the whole bandwidth. For reuse-3, each sector uses a different sub-band dividing the
total bandwidth into three parts [24].
3.2.2 Almost blank subframe (ABS)
This method allows the mobile user to ensure resources free of interference by muting
one of the transmitters. In [5] ABS has been studied in a HetNet environment, showing
the essence of using ABS in LTE cellular network to mitigate the high interference seeing
32
by the small cell due to the macro cell. It is considered two types of users, victim user
(VUE), and non-victim user (NVUE). The authors specify the victim users are those
which camps on the edge of the pico-cell, also macro cell users that are affected by
picocell users should be considered as a victim user (VUE). Consequently, both macro
and pico-cell should cooperate to reduce the inter-cell interference by introducing ABS
method into picocell besides macro-cell. In the next figure, it can be seen in the proposed
frame.
Figure 12. ABS propose frame [5]. This frame displays the ABS implemented in the picocell frame.
∝p is the number of pico − VUE, and ∝M is the number of macro − VUE.
It is proposed two solutions, based on dynamic enhanced intercell interference

coordination (eICIC). First, product-rate utility function based on [25] which maximizes
the product of bitrates of all UEs, and second, physical resource block allocation ratio-
based method. In Figure 13Figure 12, it is shown the scenario of VUE and non-VUE.
Figure 13. The scenario of the victim and non-victim UEs (VUE and NVUE) [5]. As illustrated the VUE
are located on the edge of the coverage of picocells.
In paper [26] the same idea of muting the PRB on the femtocell is used, for those macro
users considered as the victim, the authors proposed an optimal resource allocation (ORA)
33
approach to guarantee the data rate demanded by the UE. This approach produces good
results; in the below CDF graph can be observed that the ORA keeps the user data rate
demand nearly 50% and 90% of the achieved it. However, the method loses some
resources which limit the number of simultaneous users connected to the Network.
Figure 14. CDF of macro UE data rates for the demand of 0.5Mbps. Optimal Resource Allocation (ORA),
Efficient Suboptimal RB Allocation (ESRA), reuse-1 (the simplest frequency reuse factor), Orthogonal
Reuse [26].
3.2.3 Uplink Resource Scheduling for NB-IoT and LTE Hybrid Transmission
In [6], a novel uplink resource scheduling is proposed. The idea of using a hybrid
transmission to enhance the uplink throughput by categorizing the users. See the
establishment hybrid strategy in Figure 15. For those users with low SINR, LTE
technology is utilized to provide service and achieve the highest throughput feasible. On
the contrary, when the SINR is high the service is delivered by NB-IoT. Having this
scheme becomes useful when the base station supports both technologies; nonetheless, in
a massive deployment in HetNet, the interference affects the user in a higher amount,
though this approach needs to be evaluated in a dense environment. Moreover, NB-IoT
devices usually do not support LTE.
34
Spectrum Sensing
Establish the occupying

matrix
Weak-interference Strong-Interference
Non-interference Spectrum Spectrum Spectrum
Traditional
NB-IoT LTE Not to use
Technology Technology
Figure 15. Establishment of a hybrid transmission strategy [6]. The occupying matrix corresponds to the
SINR matrix for each of the RB allocations.
3.2.4 Interference awareness
In paper [27], interference aware radio resource has been proposed with the aim of
reducing the retransmission and latency. Each user is assigned RB based on the data rate
and SINR requirements. This information regarding which RBs are occupied by a certain
user is shared to neighbors via the interface X2. This helps to restrict each user to a certain
transmission power to achieve the required data rate. Having shared afore-mentioned
information, in case the data rate conditions are not accomplished, the RBs for those users
would be reassigned accordingly. The authors proposed an interference-aware radio
resource (IARR) approach. Figure 16 shows that by using this method of awareness, the
average rate and latency improve by 7% and 10% compare to round-robin scheduling
(RRS).
a) b)
Figure 16. Graph Results: a) Average information rate comparison. b) Average latency vs penetration loss
[27].
35
3.2.5 Cooperative Approach
The cooperative approach consists of cooperation between neighbor cells. This method
aims to improve the interference by reducing the transmission power of the neighboring
cell or user. There is a patent [7] which provide this solution for scheduling cell-edge
user. It consists of retrieving interference information and reporting it to the neighboring
cell for the usage of resource allocation. Then the edge users are scheduled depending on
the interference reported data. There are three procedure described as follow,
Start Start Start

Receive Interferer information Receive interferer information Receive interferer information
from UEs from UEs from UEs
Schedule cell edge UEs based on

Schedule cell edge UEs Compute Avoidance Pattern
interferer information
Send Interference information to Report cell edge resource usage Report aviodance pattern to
neighboring eNBs to neighboring eNBs neighboring eNBs
Receive Interference information Receive cell edge resource usage Receive avoidance pattern from
to neigboring eNBs from neighboring eNBs neighboring eNBs
Schedule Cell center UEs base on Schedule cell center UEs based on
Schedule cell edge UEs
received interference information received resource usage reports
Transmit to scheduled UEs per Transmit to scheduled UEs per Schedule cell center UEs based on
schedule schedule received avoidande patterns
Transmit to scheduled UEs per

end end
schedule
end
Figure 17. Three procedure described on patent [7]. In each of the three procedures include the interfere
share information action; however, the procedure two (middle) shares the resource usage, and procedure
three reports the avoidance pattern.
In addition, in paper [8], this method is implemented and showed an improvement in

throughput of 9%, see Figure 18. The simulation is run by allocating the slots by using
the maximum data rate achievable this is for non-cooperative. On the other hand, for
36
cooperative, this is improved by optimizing the transmission power of the BS (DL) or UE
(UL) using the water-filling and considering the interference threshold. The simulation
parameters and algorithm proposed can be found in [8]. Nevertheless, the author
considered the synchronous network, thri-sector sites with an inter-site distance of 500m
(adjacent to each other).
Figure 18. CDF average information rate and Average Energy consumption [8].
However, this paper does not evaluate HetNet scenarios. Another point is that cells are
not deployed randomly within a radius coverage, instead, they are deployed adjacent to
each other. which in this thesis I will cover it giving a more realistic experiment in HetNet
environment.
37
4 Different Deployment Strategies in an HetNet Environment
As known, the growth of data transfer increases in huge steps every year, and small cells
are now the best approach to fulfill that requirement of end user because of their easier
implementation in infrastructure and operational cost compare to macro cells. Moreover,
they increase the capacity of the network and improves indoor coverage. Mentioned the
advantage of having small cells, still, it is needed to evaluate the performance of the NB-
IoT in a HetNet environment. Consequently, in this chapter, five scenarios involving the
deployment of NB-IoT, which could be seen in a real environment, are described. For
instance, scenario one only involves small cell with NB-IoT, hence the inter-cell
interference is only caused by the neighboring small cells. Below, it is the general formula
for SINR calculation used for all the scenarios. Afterward in the next sessions, it will be
modified depending on the scenario; besides, the simulations procedures and results will
be explained in detail i.e. the parameter and conditions stated for the calculation and
evaluation of the deployment strategies.
𝑗 𝑗
𝑃𝑡𝑥 𝐺𝑖
𝑆𝐼𝑁𝑅 𝑈𝐸𝑁𝐵𝑖𝑗 = (1)
𝑃𝐿+𝐼𝑀𝑎𝑐𝑟𝑜 +𝐼𝑆𝑚𝑎𝑙𝑙 +𝜎
𝑘 𝑘 𝑘 𝑘
Where, 𝐼𝑀𝑎𝑐𝑟𝑜 = ∑𝑘∈{𝛺𝑀𝑁𝐵 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖 + ∑𝑘∈{𝛺𝑀𝐿𝑇𝐸 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖 , and
𝑘 𝑘 𝑘 𝑘
𝐼𝑆𝑚𝑎𝑙𝑙 = ∑𝑘∈{𝛺𝑆𝑁𝐵 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖 + ∑𝑘∈{𝛺𝑆𝐿𝑇𝐸 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖
PL = Pathloss
σ = Noise floor
Ptx = Trnasmission power.
G = Gain.
Ismall = Interference within small cells.
Domains:
ΩMLTE − Macro Cell LTE
ΩSLTE − Small Cell LTE
ΩMNB − Macro Cell NB
ΩSNB − Small Cell NB
i → UE, j → BS, k → neighbour cells
38
There are two terminologies that need to be explained before continuing with the
description of the scenarios, synchronous and asynchronous. When the cells are
synchronized the dedicated NB-IoT PRBs are the same in all cells, and when they are not
synchronized the neighboring PRBs could be potentially an LTE PRB or NB-IoT PRB
due to the no synchronization between the cells [8].
4.1.1 Scenario 1 – Small cell coverage only with NB-IoT enabled
This is the scenario (Figure 19Figure 19), in which macro cells are not involved, gives us
the first result for comparison with the rest of the scenarios that are HetNet. It can
illustrate the normal interference behavior when none HetNet scenario is evaluated. In
this particular scenario, the small cells have NB-IoT enabled and are synchronous This
states that the same RB use for transmission in NB-IoT over the small cells are the same
for the whole network and the interference is coming only from narrowband IoT
technology. Therefore, interference is evaluated with the following formulas,
𝑗 𝑗
𝑆𝐼𝑁𝑅 𝑈𝐸𝑁𝐵𝑖𝑗 = ; (2)
𝑃𝐿+𝐼𝑆𝑚𝑎𝑙𝑙 +𝜎
𝑘 𝑘
Where , 𝐼𝑆𝑚𝑎𝑙𝑙 = ∑𝑘∈{𝛺𝑆𝑁𝐵 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖
Above formula is used to calculate the SINR for downlink and uplink, the only
consideration that needs to be taken into account is that the transmission power should be
changed accordingly to get the proper calculation.
Figure 19. First Scenario – Only Small Cell with NB-IoT Enabled.
39
4.1.2 Scenario 2- Macro Cell LTE and Small Cell NB-IoT
In this second scenario (Figure 20), now macro cells are taken into account. The macro
cell has NB-IoT disabled, and the small cells are deployed with NB-IoT. Those Macro
cell users that are assigned with the same physical resource block (PRB) interferes the
small cell NB-IoT users and vice versa. From this, the interference expected on the NB-
IoT small cell should be higher, due to the neighboring macro cell involved, compare
with the former scenario described in 4.1.1
Figure 20. Second Scenario – Macro Cell with NB-IoT disabled, and Small Cell with NB-IoT Enabled.
For this scenario, because the macro cell does not have NB-IoT enabled the SINR formula
only considers two domains macro cell with LTE (MLTE) and small cell with narrowband
(SNB),
𝑗 𝑗
𝑘 𝑘 𝑘 𝑘
Where, 𝐼𝑀𝑎𝑐𝑟𝑜 = ∑𝑘∈{𝛺𝑀𝐿𝑇𝐸 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖 ; 𝐼𝑆𝑚𝑎𝑙𝑙 = ∑𝑘∈{𝛺𝑆𝑁𝐵 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖
As mentioned in the previous section, the power needs to be considered at the moment to
calculate the SINR either for downlink or uplink direction. For instance, the transmission
power in the small cells is boosted by 6 dB.
40
4.1.3 Scenario 3- Macro Cell NB-IoT and Small Cell LTE
In this scenario, the macro cell supports NB-IoT technology, and the small cell only
supports LTE (Figure 21). The NB-IoT users allocated in the macro cell are affected by
those LTE users attached to the small cell which share the same PRB. The main difference
between scenario two is that in this scenario the Tx power of the small cell does not have
the 6dB boosting power, for the contrary Tx power of the macro cell is boosted.
Figure 21. Third Scenario – Macro Cell with NB-IoT enabled, and Small Cell with NB-IoT disabled.
Thus, it is expected that the interference of those NB-IoT users over the macro cell will
be lower than the interference experienced by the NB-IoT user on the previous scenario.
Below the formulas,
𝑗 𝑗
𝑘 𝑘 𝑘 𝑘
𝐼𝑀𝑎𝑐𝑟𝑜 = ∑𝑘∈{𝛺𝑀𝑁𝐵 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖 ; 𝐼𝑆𝑚𝑎𝑙𝑙 = ∑𝑘∈{𝛺𝑆𝐿𝑇𝐸 }𝑘≠𝑗 𝑃𝑡𝑥 𝐺𝑖
4.1.4 Scenario 4 – Macro and Small Cell NB-IoT
In this scenario (Figure 22), both macro and small cell support NB-IoT. It also emerges
two new sub-scenarios, one when the technologies are synchronous, and the other one
when they are asynchronous. When synchronous is used, all cells reserve the same PRB
for NB-IoT making easier the evaluation of this case and avoiding interference from LTE
41
users. Nevertheless, in asynchronous, a combination of interference takes place, LTE or
NB-IoT users from neighboring cells could simultaneously affect NB-IoT users.
Figure 22. Fourth Scenario – Macro and Small Cell with NB-IoT Enabled.
Therefore, it is expected to see a result very close or similar to the second scenario; though
they are slightly different due to the fact, that scenario two differs by having the same
transmission power compared to an asynchronous mode where the PRB is either LTE or
NB-IoT meaning two different transmission power. Below the formula, for this scenario,
all the domains are included due to the synchronous and asynchronous mode.
𝑗 𝑗
4.1.5 Scenario 5- Macro cell LTE and Small cell randomly assign NB-IoT
In this scenario (Figure 23), macro and small cells are deployed with both technologies;
however, not all of them will radiate or support NB-IoT, the selection of which cell
supports NB-IoT is random. As the fourth scenario, the fifth scenario raises two sub-
scnarios which are synchronous and asynchronous.
42
Figure 23. Fifth Scenario- Macro Cell and Small Cell with NB-IoT enabled randomly.
It is expected that the result of both sub-scenarios will be similar, because even if the cells
are synchronous some of them will have NB-IoT disabled translating in an LTE RB.
Below the formula, for this scenario, all the domains are included due to the synchronous
and asynchronous mode
𝑗 𝑗
𝑆𝐼𝑁𝑅 𝑈𝐸𝑁𝐵𝑖𝑗 = (6)
43
5 Performance Evaluation of NB-IoT in HetNet Scenario
Each of the scenarios and simulations varies in the transmission power used either for the
macro, small cell or in the uplink by the UE. In addition, when NB-IoT technology is
enabled, a power boosting of 6 dB is injected to help in the performance against LTE
users, that also incurs in interference due to closed neighboring cells. The simulation
includes the calculation of the path loss with the Hata model, SINR, MCL, and downlink
(DL) and uplink (UL) throughput. In the following subsections parameters, the software
selected for the simulations, formulas, and models are documented.
5.1 Simulation Setup
The parameters were taken from the papers [8], [26], [28] and International
Telecommunication Union (ITU) recommendation [29]. The radius coverage was
selected regarding a small city and following the range stated in [29].
5.1.1 Simulation Software and Parameters
For the simulation of presented scenarios, MATLAB software is used, and the parameters
utilized are in Table 4, Table 5, and Table 6.
Parameter Value
Tx Power (Macro Cell) 46dBm (LTE) Max Power
29dBm +6dB Boosting (NB-IoT RB)
Tx Power (Small Cell) 40dBm (LTE) Max
23dBm +6dB Boosting (NB-IoT RB)
UE Tx Power 23 dBm
Radius Coverage 1 Km (Macro Cell)
200 meters (Small Cell)
Frequency 900 MHz
LTE Bandwidth 10 MHz
NB-IoT Bandwidth 180 kHz
Pathloss Model – Small Cells 𝐿 = 120.9 + 37.6 log10 (𝑅) [8]
, R in kilometers
Pathloss Model – Macro Cells Hata Model [30]
Described below.
Macro cell height (BS) 20 meters
Mobile antenna height (UE) 2 meters
44
Shawoding 8 dB
Correlated Shadowing 0.5 dB
BS antenna gain 18 dBi
UE antenna gain -4 dBi
BS cable loss 3 dB
Building penetration loss 40 dB
Noise figure BS 5 dB
Noise figure UE 3 dB
Noise power spectral density -174 dBm/Hz
Table 4. RF Parameters for Simulation [8].
MCL (dB) Repetition

Below 145 1
145 – 148 2
149 – 151 4
152 – 154 8
155 – 157 16
158 – 160 32
161 – 163 64
Above 164 128
Table 5. MCL vs Repetitions [8].
Parameter Value
Transport Block size (TB) 680 bits (Downlink)
1000 bits (Uplink)
Resource Element (RE) 100 (Downlink)
148 (Uplink)
CRC 24 bits
Header 65 Bytes= 520bits
Time Subframe 1ms (Downlink)
1ms (1 RU, 12 sub-tones,
Uplink)
Code Rate See Sub-section 3.2.5
# bits per Modulation 2 bits QPSK
Table 6. Parameters for Throughput Calculation. The total resource element for DL and UL is [8].
5.1.2 Models and Formulas
For the calculation of pathloss for macro cell, the Hata model [30] is selected because it
includes the height of the BS and UE antenna giving a more realistic model, see Table 4,
the formula used is,
𝐿 = 69.55 + 26.16 log10 (𝑓) − 13.82 log10 (ℎ𝐵 ) − 𝐶𝐻 + [44.9 −

6.55 log10 (ℎ𝐵 )] log10 (𝑑) (7)
𝐶𝐻 = 0.8 + (1.1 log10 (𝑓) − 0.7)ℎ𝑀 − 1.56 log10 (𝑓) , 𝑤ℎ𝑒𝑟𝑒
L: Pathloss in urban areas. Unit in decibel.
45
hB : Height of base station antenna. In meter.
hM : Height of mobile Station antenna. In meter.
f: Frequency of transmission. In Megahertz.
CH : Antenna height correction factor.
d: Distance between the base and mobile stations. In kilometer.
For the noise floor, the following formula is used,
𝑃 = 𝑘𝑇𝐵, 𝑡ℎ𝑎𝑡 𝑓𝑜𝑟 𝑟𝑜𝑜𝑚 𝑡𝑒𝑚𝑝𝑒𝑡𝑢𝑟𝑒 𝑜𝑓 290 𝑑𝑒𝑔𝑟𝑒𝑒 𝑘𝑒𝑙𝑣𝑖𝑛 𝑖𝑠 𝑒𝑞𝑢𝑎𝑙 𝑡𝑜

− 174 𝑑𝐵𝑚/𝐻𝑧
Noise floor = −174 + NF + 10 log10 (Bandwidth) (8)
NF = Noise Figure; P = Power in watts

K = Boltzmann′ s constant = 1.380649 ∗ 10−23 J/K
B = Bandwidth in Hz
The maximum coupling loss formula is,
𝑀𝐶𝐿 = 𝑃𝑡𝑥 − 𝑁𝑜𝑖𝑠𝑒 𝐹𝑖𝑔𝑢𝑟𝑒 − 𝑁𝑜𝑖𝑠𝑒 𝐹𝑙𝑜𝑜𝑟 − 10 ∗ log10 (𝐵) − 𝑆𝐼𝑁𝑅 [8] (9)
For the throughput calculation, the formula is gathered from [8]

𝑇𝐵
𝑇ℎ𝑟 = (10)
𝑇𝑇
Where,
1
(𝑇𝐵 + 𝐶𝑅𝐶 + ℎ𝑒𝑎𝑑𝑒𝑟) ( )
𝑇𝑇 = ( 𝐶𝑅 ) ∗ 𝑅𝑒𝑝 ∗ 𝑇𝑆𝐹
#𝑏𝑖𝑡𝑠 ∗ 𝑅𝐸
TT = Transmission time.
CR = Code Rate.
Thr = Throughput.
Rep= number of Repetitions.
Tsf= Subframe time
Code Rate:
For the code rate, first, the MCS needs to be stipulated. To gather the correct MCS, the
SINR is essential. By having the SINR, the MCS is extracted from Table 7 and Table 8
depending on which channel is utilized. These tables are extracted from the standard [31],
[32] and master’s thesis [33]. After the MCS is declared, the code rate should be
calculated by using Table 9 and Table 10. The table contains the number of TBS that can
be transmitted depending on the MCS and how many RU needs to be used to transmit
46
certain TBS or the number of bits. For PDSCH the maximum TB size is 680 bits and
1000 bits for PUSCH. For instance, if 600 bits need to be transmitted with an MCS of 5,
then 8 RUs allocated in scheduling.
From the book [18], the code rate is calculated as follow, “First, a 24-bit CRC is
calculated and attached to the TB. The CRC-attached TB is encoded using the TBCC
encoder and rate-matched according to the code-word length determined jointly by the
number of NPDSCH subframes allocated to the TB and the number of REs per subframe.
Thus, the combination of TB size and the number of NPDSCH subframes allocated to the
TB determines the coding rate.” [18]
SINR MCS
-3 0
-2 1
-1 2
0 3
1 4
2 5
3 6
4 8
5 9
6 10
Table 7. MCS vs SINR for PDSCH [33].
SINR MCS
-4 0
-3 1
-2 2
-1 3
0 4
1 5
2 6
3 7
4 8
5 9
6 10
7 11
8 12
Table 8. MCS vs SINR for PUSCH multi-tone [33].
Therefore, by having the correspondent TBS and number of RUs, the code rate is,
47
𝑇𝐵𝑆+24
𝐶𝑅 = 𝑀𝑜𝑑∗𝑅𝑈∗𝑅𝐸 (11)
Mod: number of bits for the modulation in use (2 bits for QPSK).
RU: number of resource units.
RE: number of resource elements.
MCS Number of Resource Units

1 2 3 4 5 6 8 10
0 16 32 56 88 120 152 208 256
1 24 56 88 144 176 208 256 344
2 32 72 144 176 208 256 328 424
3 40 104 176 208 256 328 440 568
4 56 120 208 256 328 408 552 680
5 72 144 224 328 408 504 680 -
6 88 176 256 392 504 600 - -
7 104 224 328 472 600 680 - -
8 120 256 392 536 680 - - -
9 136 296 456 616 - - - -
10 144 328 504 680 - - - -
Table 9. Mapping Between MCS, RUs, and TBS for PDSCH [32].
MCS Number of Resource Units

1 2 3 4 5 6 8 10
0 16 32 56 88 120 152 208 256
1 24 56 88 144 176 208 256 344
2 32 72 144 176 208 256 328 424
3 40 104 176 208 256 328 440 568
4 56 120 208 256 328 408 552 680
5 72 144 224 328 424 504 680 872
6 88 176 256 392 504 600 808 1000
7 104 224 328 472 584 712 1000 -
8 120 256 392 536 680 808 - -
9 136 296 456 616 776 936 - -
10 144 328 504 680 872 1000 - -
11 176 376 584 776 1000
12 208 440 680 1000
Table 10. Mapping Between MCS, RUs, and TBS for PUSCH [31].
48
5.2 Simulation Process
The superlative simulation method used is the Monte Carlo method with 1000 samples to
get enough information to obtain the correct behavior of the interference for different
radio conditions. The simulation is run first for all the scenarios without any scheduling
optimization algorithm, after having this result the next step is to run the simulation with
the cooperative approach to evaluate and see how the scheduling by cooperative method
improves the performance of the cells in an HetNet environment.
5.2.1 Scheduling of None Optimization Algorithm Implemented
In Figure 29, a flow diagram illustrates all the simulation steps. The simulation starts by
creating a macro cell within 1 Km radius coverage which is common in an urban
environment. Second, the small cells are created randomly inside the macro cell coverage
with the conditions of being inside the macro cell without touching the edges, and a radius
of 200 meters coverage. Another restriction is that the small cells cannot overlap with
each other.
Third, the UEs are randomly allocated in the macro cell, and then into the small cells. In
the MATLAB script, the number of users can be easily changed. The macro cell is
populated with 100 users, seven small cells, and 25 users per small cell (The macro cells
are created as a structure variable).
Fourth, the SNR is calculated for each of the users for macro cell and small cell. This is
used for the scheduling process. For the SNR calculation, it is taking into consideration
the white Gaussian noise, and the model path loss described in above section 5.1.2. For
macro cell coverage, the Hata model is used to consider the height of the antenna. And
for small cell the model stated in Table 4 [8] is used.
The fifth step is the execution of the scheduling by using the maximum rate by measuring
the SNR.
The SNR is sort and user’s allocation or scheduling is performed, see algorithm I (Figure
24). The interference is calculated as follow; for SINR downlink calculation relate to
algorithm II (See Figure 25), the simulation is performed by calculating the contribution
of all the base station within the coverage area with respect to the evaluated UE, see
Figure 27. The uplink SINR is calculated (Algorithm III, Figure 26) by calculating the
49
contribution of the other UEs allocated in the same resource unit (RU) with respect to the
evaluated cell, see Figure 28. The procedure is executed 100 times to get the average of
path loss which takes into account the shadowing and co-shadowing of 8 dB.
Algorithm I: Scheduling of None Optimization Algorithm Implemented

Initialization:
1: Set RB Ptx;
2: Set cell coordinates x, y
3: Set Lt; total lost (cables and penetration)
4: Set GainBS and GainUE
5: Set TS; # of time slot available
Start:
6: for all TS do
7: dist=calculate distance (cell and UE coordinates)
8: PL=calculate path loss (dist, Lt);path loss model
9: SNR(i)=calculate SNR(PL)
10: end for
Scheduling:
11: index = sort(SNR)
12: New TS allocation (index)
Figure 24. Algorithm I, scheduling of None Optimization Algorithm Implemented.
Algorithm II: SINR DL

Initialization:
1: Set RB Ptx;
2: Set UE coordinates x, y
3: Set Lt; total lost (cables and penetration)
6: Set Noise=Noise_down; (calculate with Noise floor formula)
Start:
7: for all UE do
8: for all Scells do ; # of small cells
9: dist=calculate distance (neigboring cell and UE coordinates)
10: for j=1 to 100 do
11: pl(j)=calculate path loss (dist, Lt); path loss model
12: end for
13: PL=average(pl);
14: Int += Calculate interference (Ptx, PL, GainBS, GainUE) ;
Interference
15: end for
16: SINR(UE)=calculate SINR(Ptx, PLi, GainBS, GainUE, Noise, Int)
17: end for
Figure 25. Algorithm II, Calculation of SINR DL.
50
Algorithm III: SINR UL
Initialization:
1: Set UE Ptx;
2: Set Cell coordinates x, y
3: Set Lt; total lost (cables and penetration loss)
6: Set Noise=Noise_up; (calculate with Noise floor formula)
Start:
7: for all UE do
8: for all Scells do; # of small cells
9: dist=calculate distance neighboring UE and Cell coordinates)
10: for j=1 to 100 do
11: pl(j)=calculate path loss (dist, Lt) ;path loss model
12: end for
13: PL=average(pl);
14: Int += Calculate interference (Ptx, PL, GainBS, GainUE) ;
Interference
15: end for
17: end for
Figure 26. Algorithm III, Calculation of SINR UL.
Figure 27. MATLAB Downlink SINR with respect to BS.
51
Figure 28. MATLAB Uplink SINR with respect to UE.
Subsequently, the throughput is calculated which is the main KPI of evaluation for the
scenarios in this thesis. To calculate the throughput, the next procedures state the steps to
get a reliable calculation:
I. With the calculated SINR, the MCL is calculated to get the number of repetitions
that need to be added to the transmission by following the formula presented in
section 5.1.2 and Table 5.
II. The code rate is calculated as stated in section 5.1.2, the MCS is extracted from
Table 7 or Table 8 by using the SINR, after obtaining the value of MCS, in the
table the number of RUs and TBS are selected and the throughput formula is
applied.
III. By having followed I, II, and III steps, the throughput is calculated by using the
formula in section 5.1.2 where the TB size is chosen as the maximum, 680 bits
and 1000 bits for PDSCH and PUSCH, respectively.
Above procedure or methodology is described for one UE. However, the evaluation is
performed on the cell; therefore; the SINR and throughput are averaged per cell as shown
in the above flow diagram.
52
Create Macro cell Calculate average Throughput
for each cell and scenario
Iteration (1000)
Calculate MCL
Create Small Cell and UEs

Obtain MCS
Loop 100 for Pathloss for
each UE within the cell
Calculate Code rate

Calculate average SINR, for each
cell and scenario
Calculate Throughput
Calculate average
Throughput for each
cell and scenario End
Final data
Figure 29. Flow Diagram of the simulation for the throughput calculation. The throughput is calculated
overall cell; therefore, the throughput is averaged of all the users within the small cell.
5.2.2 Scheduling of Cooperative Algorithm (Optimized)
Figure 31 shows the flow diagram of the cooperative algorithm. The cooperative method
is implemented with the aim of reducing the interference caused by neighboring cells or
user by communicating between them. This means that the macro and small cells are
communicating with each other the current interference status in every time slot; this
communication is performed via the interface X2. Consequently, for the simulation, all
the procedures stated on sub-section 5.2.1 are followed until the fifth step allocation by
SNR.
In this part, the process differs. The interference is calculated for each user or timeslot
occupied. An SINR or interference threshold is set. Before calculating the throughput or
assigning the resource, the interference is measured as explained in the sub-section 5.2.1
depending on DL or UL.
53
Algorithm IV: Cooperative Algorithm
Initialization:
1: Set UE Ptx;
2: Set Cell coordinates x, y
3: Set Lt; total lost (cables and penetration loss)
6: Set Noise=Noise_up; (calculate with Noise floor formula)
7: Set SINR threshold;
Start:
8: for all UE do
9: for all Scells do; # of small cells
10: dist=calculate distance (neighboring UE and Cell coordinates)
11: for j=1 to 100 do
12: pl(j)=calculate path loss (dist, Lt) ; using the corresponding
path loss model
13: end for
14: PL=average(i);
15: Int += Calculate highest interference (Ptx, PL, GainBS, GainUE) ;
Interference
16: end for
18: if SINR < threshold
19: I=calculate interference (threshold); min interference to aim the
threshold
20: Pngh = calculate power (I); power that neighbor should transmit.
21: SINRngh= calculate SINR(Ptx, PLi, GainBS, GainUE, Noise, Int);
neighbor SINR
22: if SINRngh<threshold
23: Pngh= calculate power(threshold); reduce the power is
possible
24: else
25: Next UE for scheduling
26: sch=0; whether the UE is schedule (1) or not (0).
27: break
28: end if
29: Int = Calculate interference (Pngh, PL, GainBS, GainUE)
30: SINR(UE)= calculate SINR(Ptx, PLi, GainBS, GainUE, Noise, Int);
again considering new Pngh
32: else
33: Next UE for scheduling
35: end if ; end for
Figure 30. Algorithm IV, Cooperative Method.
54
When the interference is calculated is compared with the threshold, see algorithm IV
(Figure 30), if this interference is higher than the set value. The cell communicates with
the neighboring cell requesting to reduce the Tx power either of itself or UE. If it is
possible to reduce the power and keep the interference of the neighbor above the
threshold, the neighbor will agree to reduce the power. But if unfortunately, it is not
possible to reduce the power requested, the neighboring cell or UE will reduce its power
to the minimum taking into account the interference threshold limit.
This procedure is followed in every UE allocation either UL or DL. After the allocation
is concluded the throughput is calculated as former sub-section 5.2.1.
Cooperative Method
SINR measurement for YES

the highest Interference
Next
SINR<
UE?
Threshold
NO
YES NO
Communicate with the End

neighbor cell to reduce the
Tx power
NO
NO
Check if it is possible
SINR(neig to reduce the power
hbor)< and keep the SINR
Ok?
Threshold above threshold.
YES YES
Tx power of the Neighbor

is reduced
Figure 31. Flow Diagram of Cooperative Algorithm.
55
5.3 Simulation Results
It is important to mention that all the results are shown in a cumulative distribution
function (CDF) graph. Furthermore, all the measurements are considered for NB-IoT
technology. As explained in section 5.2.1, and illustrated in Figure 27 and Figure 28; for
downlink, the interference measurement is performed over the UE which is listening to
the signals coming from the neighboring cells; and for uplink, the interference
measurement is performed over the cell that is now listening to the UEs. In section 4, the
description of each of the scenarios explains that some of them produce a similar result,
consequently, the analysis is focused on those which present a more significant
difference.
5.3.1 None Cooperative Case (DL)
Figure 32 shows the result of the simulation for each of the scenarios and illustrates that
a few of the scenarios present similar behaviors or SINR conditions.
Therefore, only the following four scenarios are considered for analysis:
• Scenario 1 (SC1 Only Scell(NB))
• Scenario 2 (SC2 MCell(LTE)-SCell(NB))
• Scenario 4 synchronous mode (SC4 Synch MCell(NB)-SCell(NB)),
• Scenario 5 synchronous mode (SC5 Synch MCell(LTE or NB)-SCell(LTE or
NB))
Figure 32 shows the result of the achievable downlink throughput and SINR. Where the
third scenario excels having exceptionally SINR condition than the rest of the scenarios.
Consequently, the achievable throughput results are 60% better than the other scenarios.
This is because the interference coming from the small cells is lower due to the no
boosting in the transmission power considering that none of the small cells have NB-IoT
enabled.
It is noted that the maximum throughput does not match the one on [17] 226.7 kbps
because of the CRC and header inclusion in our calculation, plus the repetitions. From
the results, the throughput range is between 2.5 and 57 kbps approximately. The minimum
throughput of 2.5 kbps correspondent to SC5, 3.28 kbps to SC4, and 5.98 kbps to SC2.
56
As expected the first scenario which does not involve macro cells, offers a higher
throughput compared to the other scenarios, and this is because the interference is lower.
And the most affected scenario is SC2 where macro cells and small cells support NB-IoT.
In this particular scenario, the SINR is higher than the other scenarios due to the power
boosting of 6 dB used, besides the power used over the macro cell is 6 dB higher than the
small which incurs in a higher interference from small cells.
Figure 32. Downlink throughput and SINR results for all scenarios. SC stands for the scenario, SCell for
Small cell, MCell for macro cell. When LTE or NB is stated, it means that the cells could or not have NB-
IoT enabled during each iteration in the simulation, and this was selected randomly. Synch and Asynch
states for synchronous and Asynchronous respectively. For synch, it means that the two technologies are
synchronous and works in the same PRB. For Asynch, the technologies utilize different PRB.
From Figure 32, The gap between the scenarios is small. For instance, between SC1 and
SC4 the gap is 3.49 kbps which indicates that SC1 offers 11% higher information rate;
SC1 is 5.5% and 7.5% better than SC2, and SC5 respectively.
5.3.2 None Cooperative Case (UL)
With respect to uplink there are only two scenarios possible without and with the presence
of macro cell:
• Scenario 1 only small cells (SC1 Only SCell (NB))
• Scenario 2 HetNet (SC2 MCell(LTE or NB)-SCell(LTE or NB))
57
The reason for having above-mentioned uplink scenarios is because the uplink power (23
dBm) is the same when evaluating the rest of the scenarios, the UE transmission power
does not vary.
Figure 33. Uplink throughput and SINR results for uplink scenarios. SC stands for the scenario, SCell for
Small cell, MCell for macro cell. When LTE or NB is stated, it means that the cells could or not have NB-
IoT enabled during each iteration in the simulation, and this was selected randomly.
Figure 33 shows the result of achievable uplink throughput and SINR. From the results,
the maximum achievable throughput was 100.4 kbps, and 105 kbps for SC1 and SC2
correspondingly. The minimum uplink throughput for SC1 was 19.89 kbps and SC2 21.57
kbps. This implies or corroborates that the presence of macro cells users collaborates
negatively in the SINR hence obtaining in low throughput. Between SC1 and SC2, the
gap is just 2.3%.
The simulation was also run with different penetration loss to evaluate diverse indoor
coverage, those penetration loss values utilized were 20 dB and 30 dB. It is clear from the
results that throughput improves due to lower interference. For instance, the SINR for
SC1 in uplink and downlink never dropped below zero. See Appendix A – Simulation
with different penetration loss.
58
5.3.3 Cooperative Case (DL)
As stated in section 5.3, the results and analysis selected are four, refer to that section.
Firstly, it seen that the maximum through reached by cooperative approach is higher
compared to former results with none optimization methods; however, this is caused due
to the fact that in this result only the users, which are using cooperative method or in
which time slot the method is applied, are considered; hence an increase in the average
throughput. Additionally, the throughput seen in the graphs is only for those users and
not for the overall cell. Secondly, Throughput and energy consumption are used as KPIs
in this analysis. Figure 34 presents the result of achievable throughput, SINR, and energy
consumption.
Figure 34 (a) illustrates the cooperative result for the first scenario. As expected the gap
between no cooperative and cooperative results is not significant. Nevertheless, it
improves the energy consumption as seen in the third graph in Figure 34 (a), even though,
both cooperative and none cooperative achieve the same maximum energy consumption.
The gap percentage between cooperative and no cooperative are 14% increment and 23%
decrement for throughput and energy consumption, respectively. However, the maximum
throughput in cooperative is lower than no cooperative. Which indicates that the
cooperative approach improves the interference for some user by decreasing the SINR to
others.
The straight line, seen in the graph for all scenarios, corresponds to the SINR threshold,
as many of the cells reduce its power to achieve the threshold, the CDF shows a rapidly
change to 1.
Figure 34 (b), (c), and (d) shows that the cooperative approach presents an enhancement
when it is utilized in an HetNet environment.
In Figure 34 (b), the second scenario shows an improvement in throughput of 78% and a
reduction of 64% of energy consumption, and the maximum energy consumption for
cooperative is 71% reduced.
59
a)
b)
60
c)
d)
Figure 34. Downlink scenario comparison between no cooperative and cooperative. (a) Scenario 1 (Only
small cell environment). (b) Scenario 2 (Macro cell NB-IoT disabled and Small cell NB-IoT enabled). (c)
Scenario 4 comparison between no cooperative and cooperative. (d) Scenario 5 comparison between no
cooperative and cooperative. The evaluation is shown for throughput, SINR, and energy consumption. SC:
scenarios.
61
Figure 34 (c) shows that the fourth scenario the improvement in throughput is 80% and a
reduction of 73% in energy consumption. In Figure 34 (d), the fifth scenario presents a
gap of 80% and 68% for throughput and energy consumption, respectively. By comparing
the cooperative method applied to the different scenarios, it points out that the best result
in cooperative approach is when the HetNet strategy deployment both macro and small
cells (scenario 4 and 5) have NB-IoT enabled.
For Scenario 4 and 5 with a cooperative approach, the maximum energy consumption for
cooperative is 71% and 78% reduced respectively.
5.3.4 Cooperative Case (UL)
Figure 35 presents the result of achievable throughout, SINR, and energy consumption.
Figure 35 (a) exposes the same behavior seen in the downlink. For the first uplink
scenario, the cooperative method does not improve the throughput significantly.
However, in the uplink, the average power consumption was reduced; still for both
cooperative and no cooperative the maximum achievable power consumption are equal.
(a)
62
(b)
Figure 35. Scenario 1 and 2 Uplink comparison between no cooperative and cooperative (a) Scenario 1 (b)
Scenario 2. The evaluation is shown for throughput and energy consumption.
Figure 35 (b) illustrates how the cooperative approach performs better in a HetNet
environment. As expected, it reduces the interference which leads to an improvement in
the throughput and energy consumption, the gaps between no cooperative and cooperative
are 66% and 67% correspondingly. The maximum energy consumption for cooperative
is 73% reduced.
Figure 36 shows in a more qualitative and descriptive view of how the cooperative
method enhances the performance of the cell in each of the scenarios. The measured
indicators are minimum power, average throughput, minimum SINR, the maximum
number of repetitions, and maximum energy consumptions.
The values go from 0 to 1. This scale is taken as a qualitative measurement which 0 means
the worst case and 1 the best. For instance, in scenario 2 (HetNet Macro with NB-IoT
disabled and Small with NB-IoT enabled), in absent of cooperative approach there is more
energy consumption, the SINR and throughput are lower compared to the cooperative.
Because the average throughput is very close it cannot be seen the improvement on a big
63
scale. As dictated before, the SC1 DL and SC1 UL there is not a big difference between
cooperative and none cooperative approach.
Figure 36. Radar Graph for each of the scenarios. As all the parameters have a different unit, each of those
parameters is shown in the rank between 0 to 1 to have a better comparison. Thus, those values represented
are not the real values obtained in the simulation. For each parameter, it was selected minimum, maximum
or average to show the improvement. For instance, the minimum power is chosen because the cooperative
approach reduces the power of the interference cell or user.
64
6 Conclusion
This study presents the interference concern of the different deployment strategies of NB-
IoT technology in an HetNet environment and an interference management method to
improve the performance of the cells or the overall network. Different scenarios which
involve enabling and disabling NB-IoT were evaluated; the results prove that the best-
case scenarios, which present a lower interference hence higher throughput, are scenarios
2 and 5, macro cell without NB-IoT and small cell with NB-IoT, and Macro cell and small
cell with either LTE and NB-IoT in synchronous mode, respectively. These two scenarios
are characterized in controlling a mix LTE and NB-IoT activation, which gives the
advantage of having some cells transmitting with a lower power which corresponds to
LTE that has not boosted power. All these results are shown in a CDF graph of 1000
samples using the Monte Carlo method.
By having performed this simulation gives us a perspective of how a real HetNet

environment behaves when NB-IoT technology is deployed. Moreover, it helps to
understand the root of the interference and opens the door to implement an interference
management method with the aim of enhancing the performance of the cells. For the
cooperative, the idea was to have communication between neighbors before scheduling
the user. Results have proved that the interference management works as expected
showing better throughput and lower energy consumption which contributes to having a
longer battery lifetime.
This thesis produces an extensive study of how the performance of the new technology
will perform in 5 possible deployment strategies. This offers crucial information to the
researcher to see how the technology could perform in a real environment, more than that,
the study is helpful to telecom engineer which are working in maintaining high standards
performance to end user. For instance, before implementing or deploying this technology
in a certain region or zone, the engineers can verify what kind of user and how many are
covered by macro or small cells. After, they can utilize the above results of the scenarios
and choose one accordingly. In addition, they can see the cooperative results over that
specific scenario and decide if this method should be activated. However, the cooperative
65
approach showed a good result in all the scenarios. It is important to mention that this
thesis covers the gaps of previous work which did not include any HetNet scenario
involving NB-IoT.
There are more open research fields after this work, for example, how the cooperative
method could be improved by implementing a sort of prediction either in the cell or
devices. It could be enough to get a model or a simple prediction algorithm or it is
necessary to involve artificial intelligent or learning machine to predict the interference
and act accordingly without asking to measure the inference continuously and employ the
cooperative method.
Another future research is the cancellation of interference in hardware, diversity

implementation in NB-IoT. Besides, the investigation of device 2 device communication
could be implemented to guarantee reliability and zero outage to those users which are
presenting bad RF conditions.
Nevertheless, a limitation in this research was the absence of a simulation tool for NB-
IoT, for instance, a MATLAB tools that allows to the researcher to set certain parameters
such as a number of users, packet size, etc... and this could accelerate the research. Thus,
I recommend for further research on this field to build a complete NB-IoT communication
system which can guarantee better progress in this field and giving more time for
proposing solution or method to guarantee a better service.
As this thesis was funded by one of the telecom operators in Tallinn, which has already
deployed NB-IoT technology in their network. This result will give them the advantage
of sharing this data to their engineer to run some optimization in their network.
Overall, the thesis offers a conclusive result that cooperative approach guarantees a lower
energy consumption, and high throughput by reducing the interference generated by
neighboring cells or UEs. New challenges are raised related to prediction either by
modeling or implementation of artificial intelligence. And finally, extends the knowledge
of NB-IoT deployment in a HetNet environment which could help telecom engineer in
their daily work.
66
References
[1] 3rd Generation Partnership Project, “3gpp tr 36.802,” vol. 0, no. Release 13, 2016.
[2] Fredrik Jejdling, “Ericsson Mobility Report, June 2018,” no. June, 2018.
[3] S. Persia and L. Rea, “Next generation M2M Cellular Networks: LTE-MTC and NB-IoT
capacity analysis for Smart Grids applications,” AEIT 2016 - Int. Annu. Conf. Sustain.
Dev. Mediterr. Area, Energy ICT Networks Futur., pp. 1–6, 2016.
[4] J. Schlienz and D. Raddino, “Narrowband Internet of Things Whitepaper
NarrowBand_IoT ─ 1MA266_0e.”
[5] R. Singh and C. Siva Ram Murthy, “Techniques for Interference Mitigation Using
Cooperative Resource Partitioning in Multitier LTE HetNets,” IEEE Syst. J., vol. 12, no.
1, pp. 843–853, 2018.
[6] Z. Zhuang, L. Feng, P. Yu, and W. Li, “Uplink Resource Scheduling for Power Wireless
Private Network Based on NB-IoT and LTE Hybrid Transmission,” Proc. - 2017 IEEE
Int. Conf. Comput. Sci. Eng. IEEE/IFIP Int. Conf. Embed. Ubiquitous Comput. CSE
EUC 2017, vol. 2, pp. 77–84, 2017.
[7] T. (US) Weimin Xiao, Hoffman Estates, IL (US); Oghenekome Oteri, San Diego CA
(US); Cornelius Dawid Janse van Rensburg, Wylie, “System and Method For Inter-Cell
Interference Coordination,” US9787445B2, 2017.
[8] H. Malik, H. Pervaiz, M. Mahtab Alam, Y. Le Moullec, A. Kuusik, and M. Ali Imran,
“Radio Resource Management Scheme in NB-IoT Systems,” IEEE Access, 2018.
[9] Y. L. M. Sikandar Zulqarnain Khan, Hassan Malik, Jeffrey Leonel Redondo Sarmiento,
Muhammad Mahtab Alam, “ScienceDirect DORM : Narrowband IoT Development
Platform and Indoor Deployment Coverage Analysis,” vol. 00, pp. 0–7, 2019.
[10] S. P. Hassan Malik, Muhammad Sikandar Zulqarnain Khan, Jeffrey Leonel Redondo
Sarmiento, Muhammad Mahtab Alam, Yannick Le Moullec and Arand, “NB-IoT
Network Trial : Indoor , Outdoor and Underground Coverage Campaign,” 2019.
[11] 3GPP MCC Frédéric Firmin, “The Evolve Packet Core.” [Online]. Available:
https://www.3gpp.org/technologies/keywords-acronyms/100-the-evolved-packet-core.
[Accessed: 27-Apr-2019].
[12] A. Toskala, “LTE Release 8 and 9 Overview,” LTE Adv. 3GPP Solut. IMT-Advanced,
pp. 14–29, 2012.
[13] Rohde & Schwarz, “UMTS Long Term Evolution ( LTE ) Technology Introduction,”
White Pap., vol. 5, pp. 1–55, 2008.
[14] R. Ratasuk, N. Mangalvedhe, Y. Zhang, M. Robert, and J. P. Koskinen, “Overview of
narrowband IoT in LTE Rel-13,” 2016 IEEE Conf. Stand. Commun. Networking, CSCN
2016, 2016.
[15] P. Andres-Maldonado, P. Ameigeiras, J. Prados-Garzon, J. Navarro-Ortiz, and J. M.
Lopez-Soler, “Narrowband IoT Data Transmission Procedures for Massive Machine-
Type Communications,” IEEE Netw., vol. 31, no. 6, pp. 8–15, 2017.
[16] A. El Mahjoubi, T. Mazri, and N. Hmina, “First Africa and Morocco NB-IoT
experimental results and deployment scenario,” Proc. Mediterr. Symp. Smart City Appl.
- SCAMS ’17, pp. 1–6, 2017.
[17] H. S. R. Y.-P. Eric Wang, Xingqin Lin, Ansuman Adhikary, Asbjörn Grövlen, Yutao
Sui, Yufei Blankenship, Johan Bergman, “A Primer on 3GPP Narrowband Internet of
Things,” IEEE Commun. Mag., vol. 55, no. 3, pp. 117–123, 2017.
[18] J. Sachs, Olof Liberg, Mårten Sundberg, Y -P Eric Wang, Johan Bergman, Cellular
Internet of Things. Academic Press, an imprint of Elsevier, 2017.
[19] S. Ali, Y. Li, K. H. Jewel, O. J. Famoriji, and F. Lin, “Channel Estimation and Peak-to-
67
Average Power Ratio Analysis of Narrowband Internet of Things Uplink Systems,” vol.
2018, 2020.
[20] Y. Gao, N. Zhang, and G. Kang, “Multiple Resource Units Allocation and Scheduling
with QoS Guarantees for NB-IoT systems,” 2018 IEEE/CIC Int. Conf. Commun. China
(ICCC Work., pp. 243–248, 2018.
[21] N. Mangalvedhe, R. Ratasuk, and A. Ghosh, “NB-IoT deployment study for low power
wide area cellular IoT,” IEEE Int. Symp. Pers. Indoor Mob. Radio Commun. PIMRC, no.
1, pp. 1–6, 2016.
[22] R. Ratasuk, J. Tan, N. Mangalvedhe, M. H. Ng, and A. Ghosh, “Analysis of NB-IoT
Deployment in LTE Guard-Band,” IEEE Veh. Technol. Conf., vol. 2017–June, 2017.
[23] N. Rani and S. Kumar, “Comparative Analysis of Traditional Frequency Reuse
Techniques in LTE Network,” vol. 5, no. 8, pp. 139–142, 2016.
[24] A. Saeed, “Efficient Resource Allocation for Joint Operation of Large and Small-cell
Heterogeneous Networks,” no. December, 2015.
[25] S. Vasudevan, R. N. Pupala, and K. Sivanesan, “Dynamic eICIC -A proactive strategy
for improving spectral efficiencies of heterogeneous LTE cellular networks by
leveraging user mobility and traffic dynamics,” IEEE Trans. Wirel. Commun., vol. 12,
no. 10, pp. 4956–4969, 2013.
[26] M. A. Imran, E. Katranaras, M. Dianati, and A. Saeed, “Dynamic femtocell resource
allocation for managing inter-tier interference in downlink of heterogeneous networks,”
IET Commun., vol. 10, no. 6, pp. 641–650, 2016.
[27] H. Malik, M. M. Alam, and Y. Le Moullec, “Interference-Aware Radio Resource
Allocation for 5G Ultra-Reliable Low-Latency Communication.”
[28] A. Adhikary, X. Lin, Y. E. Wang, H. Way, and S. Jose, “Performance Evaluation of NB-
IoT Coverage,” 2016 IEEE 84th Veh. Technol. Conf., pp. 1–5, 2016.
[29] R. I.-R. M1035, “Rec. ITU-R M.1035,” Framework, vol. 2000, 2001.
[30] M. HATA, “Empirical formula for propagation loss in land mobile radio services,” IEEE
Trans. Veh. Technol., vol. 29, pp. 317–325, 1980.
[31] 15th - 19th February 2016 3GPP TSG RAN WG1 NB-IoT Meeting #84, St Julian’s,
Malta, “R1-160480 - Consideration on uplink data transmission for NB-IoT,” no.
February, pp. 15–19, 2016.
[32] 22nd - 24th March 2016 Meeting, 3GPP TSG RAN WG1 NB-IoT Ad-Hoc, Sophia
Antipolis, France, “R1-161860 - Further considerations on NB-PDSCH design for NB-
IoT,” no. March, 2016.
[33] V. Nair, “Evaluating the suitability of Narrowband-Internet of Things ( NB-IoT ) for
smart grids,” 2017.
68
Appendix A – Simulation with different penetration loss
Here the result for different penetration loss.
Penetration loss of 20 dB
69
Penetration loss of 30 dB
70

Interference Management in Nb-Iot For Heterogeneous Wireless Network

Uploaded by

Copyright:

Available Formats

Interference Management in Nb-Iot For Heterogeneous Wireless Network

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

Interference Management in Nb-Iot For Heterogeneous Wireless Network

Uploaded by

Copyright:

Available Formats

TALLINN UNIVERSITY OF TECHNOLOGY

School of Information Technologies

Jeffrey Leonel Redondo Sarmiento

INTERFERENCE MANAGEMENT IN NB-

Supervisors: Muhammad Mahtab

Jeffrey Leonel Redondo Sarmiento

NB-IOT MÜRA HALDAMINE

Juhendaja: Muhammad Mahtab

Author: Jeffrey Leonel Redondo Sarmiento

3GPP Third Generation Partnership Project

Author’s declaration of originality ................................................................................... 3

Figure 1. LTE Network Diagram. .................................................................................. 19

Low-power wide area network (LPWAN) technology is characterized for offering

NB-IoT technology is considered LPWAN. It offers the advantage of a low power

1.3 Motivation and Research Contribution

Therefore, this is a motivation for researchers to re-produce those challenges in

1.4 Chapter Review

The structure of this thesis as follows:

In Chapter 5, All information regarding the simulation such as RF parameters, bandwidth,

2.1 LTE Network

EPC (Evolved Packet Core)

Figure 1. LTE Network Diagram.

LTE is based on Orthogonal Frequency Division Multiple Access OFDMA in downlink

Bandwidth (MHz) 1.4 3.0 5.0 10 15 20

2.1.1 Radio Resource Organization

1 slot = 0.5 ms Frequency

QPSK, 16 QAM or 64 QAM Modulation

Another important characteristic of LTE is the interfaces between the nodes.

2.2 NB-IoT standard, specifications, and performance

a. Coverage: enhance coverage by 20 dB [16] corresponding to a Maximum

d. Latency: Target of a maximum latency of 10 seconds. Which indicates that this

2.2.2 Resource grid for NB-IoT

Figure 3. NB-IoT Frame for 15 kHz spacing [18].

Physical Transmission Subcarrier The # of Duration

2.2.3 NB-IoT Physical channels and Subframe

Narrowband Synchronization Signal (NPSS/NSSS) Time and frequency synchronization

Table 3. NB-IoT channels and Signals [14].

2.2.4 Signals and Channels in Downlink

a) Narrowband reference signal (NRS): To provide phase reference, the signal

2.2.5 Signals and channels in the uplink

a) Demodulation reference signal (DMRS):

It is transmitted only in RUs which contains data.

b) Narrowband physical random access channel (NPRACH):

It is based in a single-tone transmission. The NPRACH resource configuration is

This channel support single and multi-tone as aforementioned in section 2.1.1

2.2.6 Narrow Band IoT modes

NB-IoT can be implemented in three different modes, GSM-standalone (Global System

Figure 7. NB-IoT Operational Modes [21].

3.1 Evaluation of NB-IoT guard-band mode

3.1.1 Evaluation of NB-IoT in-band mode

3.2 Resource Management in Cellular Network

Therefore, it is necessary to introduce a system that shares the spectrum efficiently. By

3.2.1 Conventional Frequency Reuse

3.2.2 Almost blank subframe (ABS)

It is proposed two solutions, based on dynamic enhanced intercell interference

Establish the occupying

3.2.4 Interference awareness

Start Start Start