LTE Downlink Scheduling Algorithms

A Thesis

Presented to the

Graduate Faculty of the

University of Louisiana at Lafayette

In Partial Fulfillment of the

Requirements for the Degree

Master of Science

Sindhura Sarepalli

Spring 2016
 ProQuest Number: 10163293

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

ProQuest 10163293

Published by ProQuest LLC (2016). Copyright of the Dissertation is held by the Author.

All rights reserved.

This work is protected against unauthorized copying under Title 17, United States Code
Microform Edition © ProQuest LLC.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
© Sindhura Sarepalli

2016

All Rights Reserved

 LTE Downlink Scheduling Algorithms

Sindhura Sarepalli

APPROVED:

Michael A. Pratt, Chair Mohammad Madani

Assistant Professor of Electrical Engineering Associate Professor of Electrical Engineering

Afef Fekih Mary Farmer-Kaiser

Associate Professor of Electrical Engineering Dean of the Graduate School
 Acknowledgements

First, I would like to express my sincere gratitude to my advisor, Dr. Michael A Pratt,

who is always a great support to me in all aspects. Without him my thesis would not be

successful. His guidance helped me in doing the research activity and writing the thesis.

Sincere thanks for his motivation and patience and for being a mentor to me all the time.

I would like to thank the rest of my thesis committee for encouraging me in doing

my research. Last but not the least, I would like to thank my parents and friends for

supporting me throughout my graduate studies and my life in general.

 Table of Contents

Acknowledgements ................................................................................................................ iv

List of Tables ........................................................................................................................ viii

List of Figures ......................................................................................................................... ix

List of Abbreviations ............................................................................................................ xii

Chapter 1: Introduction ..........................................................................................................1

Chapter 2: LTE Overview.......................................................................................................4

2.1 LTE Requirements.............................................................................................................4

2.2 LTE Physical Layer Overview ..........................................................................................6

2.3 Enabling Technologies of LTE .......................................................................................10

2.3.1 OFDM (Orthogonal Frequency Division Multiplexing)............................................10

2.3.2 MIMO (Multi Input Multi Output).............................................................................15

2.3.3 Channel Coding.............................................................................................................21

2.3.4 Link Adaptation ............................................................................................................24

Chapter 3: Related Work ......................................................................................................31

3.1 Scheduling Factors ...........................................................................................................31

3.2 Sequence of Operations in a Packet Scheduler .............................................................32

3.3 Downlink Scheduling Algorithms ...................................................................................33

3.3.1 Best CQI Scheduler.......................................................................................................33

3.3.2 Round Robin Scheduler ...............................................................................................34

3.3.3 Proposed Scheduling Algorithm ..................................................................................36

Chapter 4: Simulations and Results .....................................................................................41

4.1 System Model ...................................................................................................................41

4.2 Algorithm Encoding.........................................................................................................44

4.3 Simulation Coding ...........................................................................................................47

4.3.1 Best CQI Algorithm: .....................................................................................................47

4.3.2 Round Robin Scheduling: ............................................................................................48

4.3.3 Proposed Scheduling:....................................................................................................50

4.4 Fairness index ...................................................................................................................53

4.5 Simulation Results and Analysis: ...................................................................................54

Performance versus average SNR or BS power ..................................................................54

Performance versus average SNR or BS power for large number of users .....................57

Performance versus average SNR or BS power for MIMO ..............................................58

Performance versus average SNR or BS for smaller operating system ............................61

Performance for large system bandwidth............................................................................62

Performance of the scheduling algorithms vs. number of users (low SNR) .....................64

Performance of the scheduling algorithms vs. number of users (high SNR) ...................66

Performance of the algorithms vs. total number of antennae (low SNR) .........................68

Performance of the algorithms vs. total number of antennae (high SNR) .......................69

vi
Chapter 5: Conclusion and Future Work ............................................................................71

5.1 Conclusion ........................................................................................................................71

5.2 Future Work .....................................................................................................................72

References ...............................................................................................................................73

ABSTRACT .............................................................................................................................77

Biographical Sketch ...............................................................................................................78

vii
 List of Tables

Table 1 Paired Frequency bands ...............................................................................................8

Table 2 Unpaired Frequency bands ..........................................................................................9

Table 3 Channel Bandwidths in LTE......................................................................................10

Table 4 Prefix Configuration ..................................................................................................14

Table 5 Four bit CQI ...............................................................................................................28

Table 6 Simulation Paramaters ...............................................................................................38

Table 7 List of system parameters and their default values ....................................................41

Table 8 Effeciency of each resource block as a function of CQI ...........................................52

 List of Figures

Figure 1: Flat LTE Architecture ................................................................................................5

Figure 2: OFDM Functionality Steps ......................................................................................12

Figure 3: Time Domain Structure of LTE ...............................................................................13

Figure 4: Resource Block, Resource Grid ...............................................................................15

Figure 5: Single input Single output (SISO) ............................................................................18

Figure 6: Single input Multiple Output (SIMO) .....................................................................18

Figure 7:Maximum Ratio Combining (MRC) .........................................................................19

Figure 8: Multiple Input Single output (MISO) .......................................................................20

Figure 9: Multiple input Multiple Output ................................................................................21

Figure 10: Turbo Encoder ........................................................................................................23

Figure 11: Turbo Decoder........................................................................................................24

Figure 12: Signal generation chain ..........................................................................................26

Figure 13: Reference Signals ...................................................................................................27

Figure 14: Resource Element Mapping ...................................................................................29

Figure 15: Packet Scheduler ....................................................................................................32

Figure 16: Best CQI Scheduler ................................................................................................34

Figure 17: Round Robin Scheduler..........................................................................................35

Figure 18: Proposed Scheduling Algorithm Flowchart ...........................................................37

Figure 19: Resource Block Mapping in slots ...........................................................................38

Figure 20: Throughput for Different Scheduling Algorithms..................................................39

Figure 21: Fairness for Three Scheduling Algorithms ............................................................40

Figure 22: One Snapshot of the network setup with 15 users..................................................43

Figure 23: MATLAB code to calculate CQI as a function of SNR[db] ..................................45

Figure 24: An example of a network setup with 8 users..........................................................45

Figure 25: MATLAB code of Best CQI algorithm ..................................................................47

Figure 26: MATLAB code of round robin algorithm ..............................................................49

Figure 27: MATLAB code of Proposed algorithm ..................................................................51

Figure 28: Network Throughput vs. BS transmit power..........................................................54

Figure 29: Network Throughput vs. User’s average SNR [db] ...............................................55

Figure 30: Fairness index of different algorithms....................................................................56

Figure 31: Performance of the algorithms with large number of users ...................................57

Figure 32: Fairness index for a network with large number of users K=20 ............................58

Figure 33: Network throughput for 2x2 MIMO systems, K= 6...............................................59

Figure 34: Fairness index for 2x2 MIMO systems, K=6 .........................................................60

Figure 35: Network Throughput for a 300 by 300 operating area ...........................................61

Figure 36: Fairness index for a 300 by 300 operating area......................................................62

Figure 37: Network Throughput for B=10MHz ......................................................................63

Figure 38:Fairness index for a network with B=10MHz .........................................................64

Figure 39: Network Throughput vs. number of users (Low SNR) ..........................................65

x
Figure 40: Fairness index vs. number of users (low SNR) ......................................................66

Figure 41: Network throughput vs. number of users (high SNR)............................................67

Figure 42: Fairness index vs. number of users (High SNR) ....................................................67

Figure 43:Network Throughput versus number of antennae at Tx and Rx (high SNR) ..........68

Figure 44: Fairness index versus number of antennae at Tx and Rx (low SNR).....................69

Figure 45: Network throughput versus number of antennae at Tx and Rx (high SNR) ..........70

Figure 46: Fairness Index versus number of antennae at Tx and Rx ( high SNR) ..................70

xi
 List of Abbreviations

3GPP Third Generation Partnership Project

AMC Adaptive Modulation and Coding

AWGN Adaptive White Gaussian Noise

AMC Adaptive Modulation and Coding

APP A Posteriori Probability

ARQ Automatic Repeat Request

BCQ Best Channel Quality

BER Bit Error Rate

BLER Block Error Rate

BS Base Station

CQI Channel Quality Index

CRC Cyclic Redundancy Check

CP Cyclic Prefix

DL Downlink

E-UTRAN Evolved Universal Terrestrial Radio Access Network

EPC Evolved Packet Core

EDGE Enhanced Data rates for GSM Evolution

FDD Frequency Division Duplex

GSM Global System for Mobile Communications

HARQ Hybrid Automatic Repeat Request

HSPA High Speed Packet Access

LA Link Adaptation

LTE Long Term Evolution

MAC Medium Access Control

MRC Maximum Ratio Combining

MIMO Multiple Input Multiple Output

MISO Multiple input Single output

MME Mobility Management Entity

OFDM Orthogonal Frequency Division Multiplexing

PUSCH Physical Uplink Shared Channel

PUCCH Physical Uplink Control Channel

RAN Radio Access Network

RB Resource Block

RE Resource Element

RR Round Robin

RF Radio Frequency

RM Rate Matching

SINR Signal to Interference and Noise Ratio

SC Selection Combining

SISO Single Input Single Output

SIMO Single input Multiple Output

SNR Signal to Noise Ratio

TD Transmit Diversity

TDD Time Division Duplex

xiii
QAM Quadrature Amplitude Multiplexing

W-CDMA Wideband Code Division Multiple Access (W-CDMA)

xiv
 Chapter 1: Introduction

Wireless Internet access was introduced to the world, which led to more mobile

phone subscriptions. More attention is required by mobile applications such as media

streaming, video conferencing and gaming applications, for which higher data rates are

required. Every few years, a new mobile technology built to face the challenges requires

more frequency bands. These transitions in the technologies and their infrastructure are

called generations [1]. Data services were first available in the second generation. Later after

working past a lot of challenges to achieve the needs and higher performance, the Third

Generation Partnership Project (3GPP) created High Speed Packet Access (HSPA). The

mobile subscribers are increasing due to the increase in the number of smart phones being

sold, which, in turn, increases the mobile traffic data. There is a lot of demand on the mobile

applications, which then requires higher data rates and low latency. So the 3GPP further

started working on Long Term Evolution (LTE), which meets all the requirements. It is

expected that by 2018, total wireless Internet access traffic will increase to 13,000 Peta bytes

per year [2].

The evolution of LTE is governed by the standard developing body 3GPP. LTE is a

new radio access technology, where the access network’s main requirements are higher

spectral efficiency, higher data rates, and shorter round trip time. The 3GPP oversaw the

development of standards like Global System for Mobile Communications (GSM), Enhanced

Data Rates for GSM Evolution (EDGE), Wideband Code Division Multiple Access (W-

CDMA) and High Speed Packet Access (HSPA). Substantial research has been carried out
and universal terrestrial radio access network was proposed. The access network related to

LTE is the Evolved Universal Terrestrial Radio Access Network (E-UTRAN) [3]

LTE standards were established by 3GPP as a part of release 8. LTE provides downlink and

uplink peak data rates of 300 Mbps and 75 Mbps, respectively. It also provides a latency of

5ms or less in the radio access network. LTE uses both Frequency and Time Division

Duplexing whose bandwidths range from 1.4 MHz to 20MHz. Multi Input Multi Output

(MIMO) is another important technology used in LTE to improve the performance of the

system. MIMO with the help of Orthogonal Frequency Division Multiplexing (OFDM)

further improves the throughput rate, as well as the efficiency of the spectrum use [4].

LTE uses OFDM in order to avoid the multi-path fading problem in downlink

transmission. The data is transmitted from the base station to the user terminal through

multiple narrow sub carriers of 180 KHz bandwidth each, without sending all of the data over

the complete bandwidth [5]. OFDM provides spectral flexibility and can provide high peak

data rates. In OFDM, there is a time frequency grid, which is divided into a number of

resource blocks. Here, all the OFDM symbols group together to form the resource blocks.

When a user is assigned more resource blocks, then the modulation used in the resource

blocks is higher and thereby increases the bit rate [5].

All blocks are allocated to the users by a key operation called a scheduler. Each entry

in a resource block is called a resource element (RE). There are different kinds of scheduling

algorithms, from which I am considering two, namely Round Robin (RR) and Best Channel

Quality Index (CQI). In Round Robin scheduling, all the user terminals are equally scheduled

without regard for channel quality indication. In Best CQI scheduling, a channel quality

2
indication is sent from the user terminal to the base station as feedback. If the user has good

CQI then the resource block is allocated to the user terminal [6].

Before we move to the concept of scheduling algorithms, we must have a clear idea

of LTE’s requirements such as the physical layer overview of LTE, enabling technologies

and their modulation and coding schemes. All of these concepts are explained in Chapter 2.

Chapter 2 also explains the modulation schemes and how the resource elements mapping is

done.

Chapter 3 focuses on the LTE Downlink scheduling algorithms, which explains the

concept of scheduling and its purpose in the real world. Two scheduling algorithms are

introduced here to show how resource blocks are allocated dynamically to the users.

Problems are involved with these scheduling algorithms and in order to overcome them, I

propose a new scheduling algorithm. I then discuss the proposed algorithm, which is a

tradeoff between these two existing algorithms.

In Chapter 4, I show the results obtained from the proposed new scheduling

algorithm. I also show many scenarios by comparing the three scheduling algorithms with

different transmission modes and varying bandwidths and number of users. Finally, I

conclude my thesis with some implications for future work in Chapter 5.

3
 Chapter 2: LTE Overview

LTE is a fast growing technology which was developed taking into consideration all

the needs of the people across the globe. Over the past few years, various mobile standards

were introduced, but still the expectations and demands of mobile communications are far

beyond voice. Mobile applications, gaming, internet browsers, and social networking are

creating a demand for the development of next level standards to achieve the mobile

broadband requirements [3]. The new standards are being reinforced to provide peak data

rates and good capacity to support these bandwidth hungry multimedia applications. Low

latency, fewer operating costs, and efficiency of the spectrum are the other requirements that

lead to progress of LTE standards. Many technologies emerged after the 1990s to elevate the

peak data rate from 9.6 kbps. Within ten years, this data rate increased 2000 times to 1Gbps.

The mathematical formulation of the new technologies in the past ten years achieved a

phenomenal boost in the peak data rates. In this chapter, I explain the new enabling

technologies and how these are combined to attain such a performance. I also discuss the

different types of modulation schemes, the other way of delivering higher data rates within a

particular bandwidth.

2.1 LTE Requirements

LTE architecture mainly consists of two components: Evolved Universal Terrestrial

Radio Access Network (E-UTRAN) which is a combination of several eNodeBs and the

Evolved Packet Core (EPC) [5] as shown in Figure 1. To realize the requirements and goals

of attaining the mobile broad band communications, LTE should meet the following

objectives:
1. Peak Data rates: LTE should have a peak data rate of 100 Mbps in the downlink and 50

Mbps in the uplink transmission scheme.

2. Low Latency: Fewer network elements should be used for a radio access network

connection. A Mobility Management Entity (MME) is connected to the Evolved Node B

(eNB) through an S1 interface, directly, without any controller. Hence, it provides a lower

user and control plane latency of 5ms [5].

Figure 1: Flat LTE Architecture

3. Lower Operating Costs: Compared to the 3G wireless network, OFDM based LTE

architecture is simple. It has only one element known as eNodeB which is connected to the

MME. The radio network controllers are vanished from the Evolved-Universal Terrestrial

Radio Access Network (E-UTRAN). Hence, building this network is effortless and has fewer

operational costs.

4. Bandwidth: LTE has a bandwidth ranging from 1.4 MHz to 20 MHz, where the higher

bandwidths are used to attain peak data rates. Both paired and unpaired spectra are used to

support the Frequency Division Duplex (FDD) and the Time Division Duplex (TDD) [7].

5
5. Multiple Antenna Connectivity: The received signal power is directly proportional to the

data rate and hence in order to have peak data rates we need to increase the signal power. To

increase the received signal power more antennae are placed on the receiver side and to

increase the transmitter signal power more antennae are placed on the transmitter side. To

improve the overall power, multiple antennae have to be placed on both sides. This technique

is known as Spatial Multiplexing. This is the concept of Multiple Antenna Connectivity.

6. Integration with the other systems: LTE uses all the existing frequency bands of the

previous standards, and in addition to those, some more frequency bands are introduced. The

regulations of these frequency bands vary in different countries, but they can be used by any

service provider which makes a roaming mechanism easy to implement [7].

7. Mobility: For low terminal speeds, 0-15 km/h yields the better data rates than for higher

terminal speeds, such as 350 km/h to 500 km/h. The peak data is 100 Mbps/ 1Gbps for high

and low mobility rates respectively [5].

8. Coverage: The coverage area will be around 5 to 100 km. A slight breakdown of throughput

and efficiency occurs when a cell size reaches around 30 km [5].

9. Duplexing: There are three kinds of duplex, frequency division duplex (FDD), time division

duplex (TDD) and the half- frequency division duplex. Each of these three kinds of duplex is

used depending on the particular application. FDD is considered as the migration path for the

3G services and it uses the paired spectrum. TDD is used as an upgrade for the TD-SCDMA.

2.2 LTE Physical Layer Overview

In the LTE physical layer overview, I will cover the basic concepts that constitute the

LTE standard:

6
1. Frequency bands: There are a number of frequency bands, which were already defined by

the 3GPP standard. LTE’s main aim is to smoothly integrate with the earlier standards by

adding more frequency bands. These frequency bands can be used by any of the service

providers, which make the roaming operation trouble free. There are two kinds of Duplex

methodologies, namely FDD and TDD, which are used both in the uplink and the downlink

transmission. The frequency bands are listed as paired and unpaired frequency bands. FDD

uses the paired bands where the transmissions are simultaneous on two frequencies, and TDD

uses the unpaired bands sharing the same channel for both types of transmission. There are

25 frequency bands for FDD, as shown in Table 1, and 11 bands for the TDD, as shown in

Table 2 [7].

The transmission modes are of two types; one is the Unicast transmission mode, and

the other is the Multimedia Broadcast Multiple Services (MBMS). In the Unicast services,

the data is transferred to a particular user. In multicast, the data is transferred to different

subcarriers and combined at the user terminal [8]. This results in the improvement of Signal

to Noise Ratio (SNR). MBMS provides services like video and audio streaming.

7
Table 1: Paired Frequency Bands [9]

Uplink Operating Downlink Operating

Operating band Frequency Band Frequency Duplex
Band Index range (MHz) range(MHz) Mode
1 1920-1980 2110-2170 FDD
2 1850-1910 1930-1990 FDD
3 1710-1785 1805-1880 FDD
4 1710-1755 2110-2155 FDD
5 824-849 869-894 FDD
6 830-840 875-885 FDD
7 2500-2570 2620-2690 FDD
8 880-915 925-960 FDD
9 1749.9-1784.9 1844.9-1879.9 FDD
10 1710-1770 2110-2170 FDD
11 1427.0-1447.9 1475.9-1495.9 FDD
12 699-716 729-746 FDD
13 777-787 746-756 FDD
14 788-798 758-768 FDD
15 Reserved Reserved FDD
16 Reserved Reserved FDD
17 704-716 734-746 FDD
18 815-830 860-875 FDD
19 830-845 875-890 FDD
20 832-862 791-821 FDD
21 1447.9-1462.9 1495.9-1510.9 FDD
22 3410-3490 3510-3590 FDD
23 2000-2020 2180-2200 FDD
24 1626.5-1660.5 1525-1559 FDD
25 1850-1915 1930-1995 FDD

8
 Table 2: Unpaired Frequency Bands [9]

Uplink and Downlink

Operating operating band frequency
band index range(MHz) Duplex mode
33 1900-1920 TDD
34 2010-2025 TDD
35 1850-1910 TDD
36 1930-1990 TDD
37 1910-1930 TDD
38 2570-2620 TDD
39 1880-1920 TDD
40 2300-2400 TDD
41 2496-2690 TDD
42 3400-3600 TDD
43 3600-3800 TDD

2. Bandwidth Allocation: The channel bandwidth of the LTE technology ranges from 1.4 MHz

to 20MHz. There are a number of resource blocks which are integrated to form the frequency

spectra. These resource blocks consist of twelve subcarriers in the frequency domain and one

time slot in the time domain, which form a bandwidth of 180 MHz for each block. For a

single frequency carrier, the channel bandwidth ranges from 1.4 MHz to 20 MHz and allows

transmission from 6 to 110 resource blocks as listed in the Table 3 [7].

9
Table 3: Channel Bandwidths in LTE

Channel
Bandwidth(MHz) Number of Resource Blocks
1.4 6
3 15
5 25
10 50
15 75
20 100

2.3 Enabling Technologies of LTE

The phenomenal performance of LTE is due to the mathematical foundation of four

enabling technologies. We are going to discuss all the enabling technologies of the LTE

standard in detail. All the techniques together combine to form the LTE requirements and

attain good performance.

2.3.1 OFDM (Orthogonal Frequency Division Multiplexing)

OFDM is a novel and promising air interface technology used in LTE standards. The

factor that impacts the peak data rate is the wider transmission bandwidth. The most

important problem associated with the wider band transmission is multipath fading. There are

many channel equalization techniques to overcome multipath fading. These equalization

techniques provide better performance over a transmission bandwidth greater than 5MHz.

Since LTE transmits over wider bandwidths of 10, 15, 20 MHz or higher, where the

complexity of the channel equalization techniques increases, OFDM is chosen. In order to

master the problems related to equalization techniques, the LTE standard uses the OFDM

transmission [10].

10
 OFDM is a multi-carrier transmission scheme, which has the ability to yield

adaptable transmission bandwidths. Its main purpose is to divide the wider signal into a

number of narrow orthogonal signals. These narrow band signals are known as subcarriers.

OFDM portrays a group of narrow band channels into a frequency selective fading channel,

where the frequency spacing between the subcarriers is less. This helps the transmission

scheme to estimate the frequency response of a channel based on the reference signals or data

that is transmitted [11]. With a good estimate of the frequency response we can easily

recover the transmitted signal at the receiver using the frequency domain equalizers.

OFDM is mainly used in the downlink transmission scheme, which is based on many

multi carrier operations. There are two steps involved in the OFDM transmission. They are

Resource element mapping and OFDM signal generation, as shown in Figure 2. On the time-

frequency resource grid, modulated symbols of each layer are aligned using the resource

element mapping. The data is aligned with subcarriers on the frequency axis of the grid. In

the OFDM signal generation, by applying the inverse Fourier transform, OFDM symbols are

generated. These OFDM symbols are used to represent the data transmitted in time, and these

are sent to multiple transmit antennae for transmission [11].

11
Figure 2: OFDM Functionality Steps

The clear understanding of how data is transmitted is given by the figure 3. It is a

time- frequency representation of data which is known as a resource grid. The transmission

of LTE is in the form of frames whose length is 10ms. These frames are transmitted in

sequence. Each frame is sub divided into ten sub-frames of 1ms length each. There are two

slots in each sub-frame of 0.5ms each [12]. Six or seven OFDM symbols are in each slot

depending on whether it is an extended cyclic prefix or the normal cyclic prefix. There are 7

OFDM symbols, if the slot is designed as the normal cyclic prefix, and 6 OFDM symbols if it

is an extended cyclic prefix. Figure 3 demonstrates this.

12
 Figure 3: Time Domain Structure of LTE [9]

In OFDM signal generation, cyclic prefix generation is an important function that

copies the last part of the OFDM symbol to the beginning and prevents interference from the

previously generated symbols. For the multi carrier transmission, the length of the cyclic

prefix is an important parameter. In order to increase the spectral efficiency, there should be

a very small cyclic prefix [7]. There are three kinds of cyclic prefix values. They are

1. Normal cyclic prefix (4.7µs)

2. Extended cyclic prefix (16.6µs) for subcarrier spacing of 15 KHz.

3. Extended cyclic prefix (33µs) for subcarrier spacing of 7.5 KHz.

13
The Extended cyclic prefix of subcarrier spacing 7.5 KHz is used only in the broadcast

context. All these prefixes are explained in detail in Table 4 shown below.

Table 4: Prefix Configuration [7]

Subcarrier Number of Subcarriers Number of OFDM symbols

Configuration Spacing per resource block per resource block
Normal cyclic
prefix 15 12 7
Extended cyclic
prefix 15 12 6
Extended cyclic
prefix 7.5 24 3

In the frequency domain, a radio frequency carrier consists of a minimum of 6

resource blocks to a maximum of 110 resource blocks, which is equivalent to a transmission

bandwidth ranging from 1.4 to 20MHz. This total bandwidth is fragmented to 15 KHz band

and permits high flexibility in bandwidth. In both the uplink and the downlink transmissions,

the resource blocks are equally used by the users. The main difference between these two is

the location of carrier center frequency to the subcarriers. In downlink, users does not use the

subcarrier that intersects with the carrier center frequency. In uplink transmission, there are

no unused subcarriers [13].

A subcarrier space of 15 KHz is considered in OFDM, which turns a frequency

selective channel into a number of flat sub-channels, as shown in Figure 4. Due to the effects

of Doppler shift and phase noise, the subcarrier spacing decreases, which results in low

performance. In the LTE standard, the degradations caused by the Doppler shift and the

phase noise can be minimized by considering a fixed subcarrier spacing of 15 KHz [8].

14
Figure 4: Resource Block, Resource Grid [9]

2.3.2 MIMO (Multi Input Multi Output)

One of the important technologies used in the LTE standard is the Multi Input Multi

Output (MIMO) technology. Multiple antennae are used in the MIMO system in order to

encounter LTE requirements such as peak data rates and throughput. MIMO technology

helps to surpass the mobile communication in two ways: by stimulating the peak data rate

and by growing the fidelity of the communication link [14]. These multiple antennae are used

at the transmitter and receiver ends which render perceptible multiplexing and Diversity

15
gains. MIMO is a fast growing Radio Frequency (RF) technology that is used in many

technologies like Wi-Fi and LTE to provide good spectral efficiency.

In mobile communication, when the data is transmitted in a channel, fading occurs,

which affects the channel and influences the signal to noise ratio. This impacts the error rate

as well. In order to reduce the error rate, improvise the capability and sustain the link, several

diversity modes are used. Time, Frequency and Space Diversities are the three kinds of

diversities, which yield many advantages. In Time Diversity, a message signal is transferred

from one place to another place in different time slots using channel coding. In Frequency

Diversity, distinct frequencies are used in the form of different channels. In addition to these

two diversities, there is a special diversity which acts as a basis for MIMO known as Space

Diversity. In this diversity, antennae that are located in various positions of a mundane

environment are taken into consideration to make use of different radio paths. MIMO is an

effective antenna technology that uses multiple antennae at the transmitter and the receiver

end, to transport the data in several radio paths. In order to improve the wireless technology,

MIMO is used as a logical expansion.

When the data is transferred from one transmitter to the receiver, the signal can take

any number of paths. The obstructions that appear in the direct path determine the number of

paths. Initially, several paths are used to introduce interference. But MIMO considers this as

an advantage to enhance the signal to noise ratio to make the radio link more resilient.

MIMO uses many antennae to increase the capacity of the channel. With every increase in

the pair of antennae at the transmitter and the receiver, there is an increase in the channel

throughput [15]. As the spectral bandwidth is very crucial in the wireless system, we need

many techniques to use the bandwidth productively. MIMO is one such kind of technique.

16
 There are different kinds of MIMO configurations that attempt to produce flawless

applications. These configurations refer to a number of inputs and outputs in the radio link.

Here the input is the transmitter and the output is the receiver. The different forms of antenna

links are as follows:

• Single Input Single Output (SISO)

• Single Input Multiple Output (SIMO)

• Multiple Input Single Output (MISO)

• Multiple Input Multiple Output (MIMO)

Single Input Single Output (SISO)

The simple radio link in the MIMO system between a transmitter and a receiver with

a single antenna on both sides is termed a SISO system. A SISO system has no diversity or

additional process requirements [16]. The capability of a SISO channel is very limited.

Interference and channel fading have greater influence on this type of system. Shannon’s law

is defined as the maximal rate at which the data is transmitted in the presence of noise over a

particular bandwidth.

C = W log2 (1+ S/N)

Where W is the bandwidth, S/N is the Signal to Noise Ratio and C is the channel capacity.

Shannon’s law limits the bandwidth and thus influences the throughput of the SISO system.

The SISO structure is shown in the Figure 5.

17
Figure 5: Single Input Single Output (SISO) [17]

Single Input Multiple Output (SIMO):

This transmission system has one antenna at the transmitter end and multiple

antennae at the receiver end, as shown in Figure 6. This system is also known as the receive

diversity. The main purpose of using a SIMO transmission system is to resist the fading

conditions in a channel [21]. SIMO’s are used with the short wave broadcasts in order to

fight the effects of interference and the fading conditions in the ionosphere.

Figure 6: Single Input Multiple Output (SIMO)

At the receiver of SIMO transmission systems there are two combining methods used:

Maximum Ratio Combining (MRC) and the Selection Combining (SC). In Maximum Ratio

18
Combining (MRC) [19], all the received signals are combined, and by averaging them we

find the approximate transmitted signal. This is represented in Figure 7. In Selection

Combining (SC), we check the signal to noise ratio (SNR) of the received signals. The

received signal that has the higher SNR will be selected to calculate the transmitted signal. In

a fading channel, all the signals show similar strengths as the interference is high. In these

conditions MRC is the best MIMO technique to be used. Though it is used in many

applications, the receiver of the transmission system requires some processing, which is a big

disadvantage. The processing level is very restrained in its area. The life of a battery is

important as the receiver is situated in the mobile handset.

Figure7: Maximum Ratio Combining (MRC) [17]

Multiple Input Single Output (MISO):

In the MISO transmission system, there are multiple antennae on the transmitter side

and only one antenna on the receiver side as shown in Figure 8. It is also known as transmit

diversity [18]. Similar data is transmitted from both the transmitting antennae, and the

receiver receives the required signal from the transmitter. The main purpose of proposing this

mode of transmission is to avoid the processing requirements at the receiver end. The

19
receiver end is the user equipment. There is low battery consumption due to fewer processing

requirements.

In the MISO system, the channel link standard is upgraded, and the channel fading is

reduced. The main drawback in this system is that the data rates remain sedative. In order to

overcome the drawback in this system, various spatial multiplexing modes were developed.

Figure 8: Multiple Input Single Output (MISO) [17]

Multiple Input Multiple Output (MIMO):

MIMO transmission system has multiple antennae at the transmitter and the receiver

ends, shown in Figure 9. MIMO system was developed to avoid all the drawbacks from the

previous modes. Spatial Multiplexing is a very important concept used in the MIMO system,

providing inflation in the data capacity [20]. The data is carried through multiple paths which

are used as the additional channels. According to Shannon’s law, the large amount of data

that travels through a radio link is limited. Beyond a certain limit, it is not possible to

progress the data that is travels across a channel due to boundaries defined by the law.

From the Shannon’s expression we can say that, the Channel Capacity is always

limited by the bandwidth and the Signal to Noise Ratio. According to this expression, we can

decide how the transmission has to be made. To increase the channel capacity, different

20
modulation schemes are used. Higher order modulation schemes are used to increase the

capacity of the channel, but these approaches need to have good signal to noise ratio. In order

to maintain the signal to noise ratio and the modulation schemes, some improvements are

made, which are not quite easy [21]. One such improvement in the wireless communication

is the MIMO system, which helps in improving the throughput of the data. The data is

separated from different paths with the help of channel coding. The data capacity and the

robustness are increased with the help of this coding, but it requires some processing.

Figure 9: Multiple Input Multiple Output (MIMO) [17]

2.3.3 Channel Coding

Error correction is one of the most important issues in the mobile digital

communications. There are many techniques to rectify the errors in the data transmission.

Automatic repeat request (ARQ) is one such method, which detects the errors in the

transmission and requests for a retransmission of packets until most of the packets are

reached. There is another method where redundancy bits are added to the data to correct the

error. This is known as Forward Error Correction (FEC). Block codes and the Convolutional

codes are two types of codes used in FEC. Convolutional codes involve D Flip-flops, which

21
have a memory of 1 bit and hold it for one clock cycle [22]. If the transmitter is very weak

and relying on a poor energy source, the data cannot be sustained for a longer time. This

constraint is satisfied by Convolutional Codes. This kind of code depends on the previous

bits and the current bits. The upgrade of the Convolutional codes is the Turbo Coding

mechanism.

2.3.3.1 Turbo coding

The main support of channel coding in the LTE standard is Turbo coding which is the

integration of two Convolutional codes parallel to one another and an interleaver that

separates them. This is illustrated in Figure 10. Previously, in order to step up the

performance of a system, this turbo coding standard was used. In the LTE standard, this is

the only channel coding method used to operate the user data. In order to make the Turbo

coding more efficient, many improvements are made, such as attaching the cyclic

redundancy check (CRC) to the input of the encoder [23]. This CRC helps the turbo decoder

to do its implementations faster. CRC checks the errors in the transmission, and if there is no

error, then the Turbo decoder stops its mechanism without going through multiple iterations.

22
Figure 10: Turbo Encoder [9]

Here we are going to discuss the turbo encoder and the decoder to determine the

complete functionality of Turbo coding. In a turbo encoder, two constituent encoders are

integrated and are placed parallel to each other, separated by an interleaver. The base rate of

1/3 is used in LTE. The turbo encoder output has three output streams, as shown in the

diagram below. The bits of the first output stream are known as Systematic bits. The outputs

of the other two streams are indicated as the parity 1 and parity 2 bits [24]. The termination

of each constituent encoder is done by the tail bits. Due to trellis termination, if we consider

k input bits, then the output of the encoder of all the streams is k+4 bits. In the LTE standard

there are 188 different values for the block size. Rationalizing the memory access by using

the Quadratic Polynomial Permutation (QPP) interleaver, the performance of the turbo

coding is improved. Each encoder has a trellis structure, which is shown as a polynomial

equation.

G0(z) = 1+ z-2+z-3

23
 G1(z) = 1+ z-1+ z-3

The turbo decoder uses the same trellis function used in the turbo encoder. The turbo

decoder is an iterative function. The performance of this decoder mainly depends on the

number of iterations made. The Turbo decoder is based on two interleavers and A Posteriori

Probability (APP) decoders [23]. This is a complete inverse of the turbo encoder in the

receiver part. The output obtained from the demodulator and the descrambler is processed,

and the good measure of the transmitted bits is regained. The Turbo decoder is shown in

Figure 11.

Figure 11: Turbo Decoder [9]

2.3.4 Link Adaptation

Link adaptation is a gathering of a group of techniques in the communication system

to change and adapt the parameters of the transmission. Adaptive Modulation and Coding

(AMC) is used in the LTE standard to match the modulation and the coding rate. In the

Modulation scheme, if the Signal to interference Noise Ratio (SINR) is high, then we

consider a higher order modulation scheme, such as 64 QAM. If the SINR is less, then we

consider a lower order modulation scheme (QPSK). Higher modulation schemes are directly

proportional to the spectral efficiency. According to the channel quality and the modulation

24
scheme, we select the coding rate. If the channel quality is good, higher code rate is chosen.

Generally in the LTE standard, the code rate of 1/3 is used in the turbo coders. In order to

vary the code rate, we use Rate Matching (RM) [25]. Puncturing and Repetition are used in

the RM module to vary the code rate. In the LTE physical layer, the chain of generating a

signal is shown in Figure 12, which uses the turbo coding and the modulation schemes,

which are also a part of link adaptation.

25
Figure 12: Signal Generation Chain [7]

The Channel Quality is the most important parameter we need to know in order to

match the modulation and the coding schemes. The quality of the channel is measured in the

user equipment and is sent to the base station in the form of Channel Quality Indicator (CQI).

The channel quality of the signal depends on the SINR, channel and also the quality

of the receiver. It delivers higher CQI if the receiver signal has higher channel quality. Some

Reference symbols are used in order to do all the signal quality measurements, as shown in

Figure 13. In a long Cyclic Prefix (CP), these Reference symbols are placed in the first and

26
the fourth places in a Resource Block (RB), and in the short CP, these symbols are placed in

the first and the fifth positions [12]. These symbols are mainly used to calculate the channel

estimation.

Figure 13: Reference Signals [12]

In order to ensure that the Block Error Rate (BLER) is less than 0.1, we choose the

modulation and coding scheme, relying on the Signal to Noise Ratio (SNR). There are 16

combinations in the modulation and coding schemes for the 4 bit CQI. The CQI value varies

from 1 to 15. If the CQI value is 1, that refers to the worst channel quality and we assign a

QPSK modulation scheme with a coding rate of 0.07. With the rise in the CQI to 15, the

higher modulation scheme such as 64 QAM is used [26]. This can be explained clearly in the

Table 5.

27
Table 5: Four bit CQI [14]

Approximate code
CQI index Modulation rate efficiency
0 No Transmission
1 QPSK 0.076 0.1523
2 QPSK 0.12 0.2344
3 QPSK 0.19 0.377
4 QPSK 0.3 0.6016
5 QPSK 0.44 0.877
6 QPSK 0.59 1.1758
7 16QAM 0.37 1.4766
8 16QAM 0.48 1.9141
9 16QAM 0.6 2.4063
10 64QAM 0.45 2.7305
11 64QAM 0.55 3.3223
12 64QAM 0.65 3.9023
13 64QAM 0.75 4.5234
14 64QAM 0.85 5.1152
15 64QAM 0.93 5.5547

CQI values are used to select an ideal resource block and a time slot. CQI reporting is

of two types. One is the periodic reporting, and the other is the aperiodic reporting. The user

equipment reports the CQI to the base station in a periodic time interval through Physical

Uplink Control Channel (PUCCH). This is the periodic reporting [14]. When the base station

needs the CQI information at a particular time, then aperiodic reporting is used. The

aperiodic reporting uses the Physical Uplink Shared Channel (PUSCH) for reporting.

The total bandwidth is divided into N sub bands and each sub band consists of K resource

blocks. Where NDL/RB is the number of resource blocks in the total bandwidth,

28
 =[ / ]

Resource management deals with all the resources, which allocate all the blocks of

data to the resource blocks. A resource block has a time slot of 0.5ms and 12 sub carriers. All

the transport blocks are mapped to the various antennae through Antenna Mapping. Mostly

four transmit antennae are used in the LTE standard [7]. There are multiple transmit antenna

schemes used in the LTE Standard such as transmit diversity, spatial multiplexing etc. After

the antenna mapping is done, there is a Resource block mapping which helps in mapping the

data to the resource blocks according to the scheduler allocations. All the data is sent through

the antennae. The resource block mapping for downlink is shown in Figure 14 below.

Figure 14: Resource Element Mapping [7]

29
 Chapter 3: Related Work

The evolution of Universal Mobile Telecommunication System (UMTS) is LTE. LTE

aims at higher data rates and better spectral efficiency. Various enabling technologies were

developed to achieve these LTE aims, like OFDM and MIMO, which I discussed in the

previous chapter. The performance of the system can be increased by adopting the Radio

Resource Management (RRM) in the LTE atmosphere. Sharing the resources, CQI reports

and automatic transmission requests are the work done in RRM. In order to satisfy the needs,

users should efficiently use the radio resources and several methods are designed to use them

[28]. In this chapter, we are going to discuss the most important mechanism used in handling

the user traffic, which is known as the Packet scheduler.

A Packet scheduler, which is situated in the base station, shares the spectrum among

different users. There are many Packet scheduling algorithms that help in distributing the

resources to many stations considering the channel conditions. Here we consider the tradeoff

between the throughput and the fairness, while allocating these radio resources. In wireless

technology, the influence on the channel quality reduces because the packet scheduler aims

to maximize the spectral efficiency through successful resource allocation. There is a great

instability in the channel quality due to Doppler Effect and the Fading Effect. Due to these

effects, OFDM systems adopt the channel aware conditions in order to obtain better channel

conditions by assigning the higher priority channel quality to the users [33].

3.1 Scheduling Factors

1. Channel quality indicator: The user gives the base station information about the channel

quality in the downlink in the form of a four digit number. By using this
 information, the base station assigns the specific modulation and coding technique and the

suitable resource block to the user [29].

2. Buffer status report: When the user has certain data in its buffer it has to inform the base

station that there is some data and request a grant from the network to share that data. This is

the user’s way of informing base station about the data in the user terminal.

3. Quality of service: Quality of service is a measure of functionality between the PDN

gateway and the user. It defines how the data should be treated in the network. For example,

a voice packet is given more priority than web browsing traffic.

The user computes the CQI value in the downlink channel and sends it to the base station

[29]. The buffer status report (BSR) is also sent to the base station. Based on these factors

(BSR, CQI, and QoS), modulation and coding scheme values and the Physical resource block

(PRB) mapping are computed and sent to the user [33].

31
 Figure 15: Packet Scheduler [27]

3.2 Sequence of Operations in a Packet Scheduler

The radio resource management modules interact with the scheduler and this

sequence is continuously repeated [30], as shown in Figure 15.

1. The user equipment takes the reference signal and, calculates the channel quality and this is

sent to the base station.

2. The base station takes the allocation decision by using the channel quality value and fills the

resource block.

3. The modulation and coding scheme is selected by the AMC module, which the scheduled

users use in the transmission of data.

4. All the information about the user, resource block and the modulation scheme are sent

through the PDCCH channel.

32
5. The user reads the PDCCH channel payload.

3.3 Downlink Scheduling Algorithms

The frequency and time resource allocations of the users are controlled by the

scheduler in each time interval. The scheduler decides to which user the resource block has to

be assigned.

The schedulers are still in research to balance the throughput and the fairness according to

the user requirements [31].

3.3.1 Best CQI Scheduler

In this algorithm, scheduling is performed by sending the channel quality indicator to

the eNodeB. The reference signals are transmitted by the base station (eNodeB) to the user

equipment. Then the user equipment calculates the channel quality indicator to assign the

resource block to the user equipment [7]. In this scheduling, the users who have the best

radio link conditions are assigned the resource block and the others are not assigned the

resource block. The users who are far away from the base station are not assigned the

resource block since they do not have the best radio link conditions. The Flow Chart of the

B-CQI is depicted in Figure 16.

Flow chart of the B-CQI scheduler:

1. The base station sends the reference signals to the users.

2. The users compute the channel quality indicator.

3. If the channel quality indicator is the highest value, then the resource block is assigned to it;

otherwise, it is not.

4. The resource block is given to the user equipment when they have the best CQI.

33
Figure 16: Best CQI Scheduler

In this scheduler, there is a maximum amount of throughput and a minimal amount of

fairness.

3.3.2 Round Robin Scheduler

In this scheduling algorithm, the CQI is not considered, and all the resource blocks

are allocated irrespective of the channel conditions [7]. Hence, there will be lower throughput

and more fairness. The scheduling is done according to the resource blocks available. The

resource blocks are allocated to the users on a first come, first served basis. Implementation

of this round robin algorithm is very easy and fairness is guarantees fairness [32]. The

34
 channel quality indicator is not taken into account and throughput of this scheduler is less

when compared to the B-CQI scheduling algorithm. The Flow chart of the Round Robin

scheduler is represented in Figure 17 below.

Figure 17: Round Robin Scheduler

Flowchart of the round robin scheduling algorithm:

1. First, schedule the first user.

2. Schedule the second user.

3. Check whether all the users have been assigned the scheduler or not.

4. Stop.

35
There are so many schedulers developed recently which are built on considering both the B-

CQI and the Round Robin scheduling strategies. This fulfills both the throughput and fairness

parameters, and thereby, performance of the network increases. By combining both of the

algorithms, there will be a successful output, and both parameters will be balanced. The

selection of the scheduler model is based on some of the characteristics.

3.3.3 Proposed Scheduling Algorithm

In the previous sections I discussed the Best CQI and the Round Robin scheduling,

whose algorithms satisfy the throughput and fairness, respectively. Now, I will discuss the

new scheduling algorithm, which is a trade-off between both scheduling algorithms. In a sub-

carrier, there are two slot periods to which the resource blocks are allocated by the base

station scheduler. According to this algorithm, the first slot allocates the resource blocks to

the user satisfying Best CQI scheduling. And the second slot allocates according to the

Round Robin Scheduling. Thus, this scheduling algorithm satisfies both throughput and

fairness.

The figure below represents the proposed scheduling algorithm flowchart. The Base

station first checks for the highest CQI value, and the user with the highest value is selected.

If there is no highest CQI value, then the scheduler makes the selection at random. All the

users with high CQI are selected and scheduled in the first slot period. In the second time

slot, all the time slots are scheduled according to Round Robin Scheduling. After the second

time slot is done, the process is repeated. The Flowchart of the Proposed Scheduling

Algorithm is shown in Figure 18.

36
Figure 18: Proposed Scheduling Algorithm Flowchart

According to the bandwidth variations, the number of resource blocks differs. In

order to check how the Proposed Scheduling Algorithm works, I am going to consider some

simulation parameters according to Table 6.

37
Table 6: Simulation Parameters

According to the parameters in the above, Table 6, I am going to consider 6 users and

the bandwidth of 1.4MHz. The number of resource blocks is 6 for that specific bandwidth.

All the users have different CQI values. The user with highest CQI is allocated first, and the

time slot is filled sequentially. Then the second slot is filled with fair allocation without

consideration of the value of CQI. The Resource block mapping of slots is shown in Figure

19. Both throughput and fairness are satisfied in this scheduling algorithm.

Figure 19: Resource Block Mapping in slots

By considering the above scenario with 6 users, SISO and bandwidth of 1.4 MHz I

get the simulated results. Figures 20 and 21 show the throughput and fairness results for all

three scheduling algorithms. If we consider the channel conditions, Round Robin

performance is lower than the proposed and the Best CQI scheduling Algorithm, as shown in

Figure 20. If we consider fairness, Best CQI performance is low compared to the other two

38
algorithms, as shown in Figure 21. The proposed algorithm is a trade-off between both the

Best CQI and the Round Robin algorithms.

Network throughput [b/s/Hz]

Figure 20: Throughput for different scheduling algorithms

39
 Network fairness index

Figure 21: Fairness for three scheduling algorithms

40
 Chapter 4: Simulations and Results

4.1 System Model

In order to evaluate the performance of the proposed scheduling algorithms, I considered an

LTE system with one base station (BS) located in the center of the operating area, with a size

of × meters. Users (UE) are distributed in the operating area uniformly at random

places. Furthermore, the BS and the users that are equipped with and transmit and

receive antennae, respectively. The default simulation parameters are listed in Table 7, and

for all simulation results I use these parameters, unless otherwise stated.

Table 7: List of system parameters and their default values

500 operating area width (meter)

500 operating area length (meter)
[250 250] BS position
2 Number of antennae at the BS
1 Number of antennae at the UE
6 Number of users
1.6MHz System bandwidth
6 Number of resource blocks (depends on )
10 dB System noise figure
-174dBm Noise spectral density per Hz
0dBm Transmit power of the BS
random Position of the -th user
 2.2 if < 320 path loss exponent
4.0 if > 320
random distance of the -th
random path-loss between -th user and base
station
ℎ &'(0,1) Rayleigh-fading channel coefficient of the
-th user which is a complex Gaussian
random variable
, Random [x,y] channel gain of user
SNR random SNR of the -th user

One snapshot of the system setup with 15 users is shown in Figure 22, which illustrates that

the BS is located at the center of the operating area, and the users are randomly distributed in

the environment.

42
 network setup

500
7
BSs
1
8 Users
400 15
9

10 2
300
3
5
14 6
200
13

100
4
12
11
0

0 100 200 300 400 500

x axis

Figure 22: One snapshot of the network setup with 15 users

To find the channel gain between each user and the BS, first we find the mutual distance of

each user based on the following equation:

= 01
5 5
,2 − ,2 4 +1 ,7 − ,7 4

Then we find the path-loss (dB) of the user as follows:

2.2 × 10 × log> + 64.02, if < 320

[dB] = :
4.0 × 10 × log> + 18.98, if > 320

Then, the channel gain (in dB) of each user can be obtained as follows:

, [dB] = − [dB] + 10 × log> |ℎ |5

The noise power in dB can be calculated as follows:

E [dB] = [dBF + 10 × log>

43
And eventually the SNR [dB] of each user can be obtained as follows:

SNR [dB] = [dB] + , [dB] − E [dB]

Note that we have the following equalities (dBm mean dB relative to mill watt):

= 10 .>×(GH [IJK]LM )

[dB] = [dBm] − 30

Furthermore, maximal ratio combining is used at the receiver side of each user to increase the

SNR in case multiple inputs are available at the RX side. The SNR is calculated for each user

at each resource block, so for one time slot of the simulation we have SNRO , where b is the

index of the resource block, and is the user index.

4.2 Algorithm Encoding

As I mentioned in the previous chapter, I compare the performance of the proposed algorithm

with two other scheduling algorithms namely, best CQI and round robin.

In the best CQI scheduling algorithm, after calculating the SNR of each user in each resource

block, we find the CQI of the user according to the following equation.

0 if SNRSO [dB] < −6

V
14
CQISO = floor X 1SNRSO [dB] + 64 + 1Y
U 26
T 15 if SNRSO [dB] > 20

The MATLAB function of this part can be seen in Figure 23.

44
Figure 23: MATLAB code to calculate CQI as a function of SNR [dB]

In this code the parameter obj. userSNRdBAtRB is a matrix of size × , which holds

SNR values of all users for all resource blocks at the current time slot.

The following network setup in Figure 24 has 8 users, and BS transmit power of 10dBm.
y axis

Figure 24: An example of a network setup with 8 users

45
The SNR [dB] of users can be calculated as follows:

And as a result the CQI can be obtained as follows:

The observation shows that user 8 and 4 are located far from the BS, and they have the

lowest SNR [dB], and, as a result, the lowest CQI. Furthermore, although users 3 and 6 have

the highest SNR [dB], their CQI cannot exceed 15.

After findingCQISO , we have to assign the resource blocks to users. This will be done through

scheduling algorithms. The output of each scheduling algorithm is a matrix b of size × ,

where its elements are either 1 or 0. If bc,d = 1, the e-th resource block has been allocated

to the -th user. Since at each time slot, a resource block can be allocated to only one user,

we have
g

f bc,d = 1
h>

46
This means that the e-th resource block has been allocated to only one user. The number of

resource blocks that are allocated to the -th can be calculated as follows:

e = f bc,d
Eh>

4.3 Simulation Coding

4.3.1 Best CQI Algorithm:

In the best CQI algorithm, the BS sorts the users’ indices based on their CQI, and allocates

the e-th resource block to the user with the highest SNR [dB] at it. The MATLAB code of

this algorithm is shown in Figure 25.

Figure 25: MATLAB code of best CQI algorithm

In case, there is more than one user with highest CQI at a specific resource block, we select

one of them at random, and allocate that resource block to the selected user. This can be seen

47
in the MATLAB code. After the scheduling is done, we calculate users capacities and their

channel efficiencies.

For the example given in figure 24, the allocation matrix for the first time slot can be

obtained as follows:

As it can be seen in the first resource block, users 3 and user 6 have the highest CQI (here,

13). So one of them is selected at random (here the selected user is 6). We have a similar

scenario for the other resource blocks. Another representation of the allocation matrix can be

made in an array form i of size × 1. In this way each element of the array shows the

index of the user which that resource blocks is allocated to. For the above example we have

With such representation we can show the scheduled users for different time slots in a better

way. We can observe that the resource blocks are allocated only to users 3 and 6 because

they have the highest CQI among the users. This indicates that by using the best CQI

algorithm there is a chance that some users cannot use network resources.

4.3.2 Round Robin Scheduling:

In the round-robin scheduling algorithm we start with the first user, and we allocate the first

resource block to it. Then we move to the second user, and we allocate the second resource

48
bock to this user. We continue this process until we have no resource blocks left, or until we

have no users left. In case the number of resource blocks is greater than the number of users,

we start over with user one and allocate the + 1-th resource block to user 1, and so on.

Conversely, if the number of users is greater than the number of resource blocks, in the next

time slot we start with a user that has not been scheduled yet (namely user = + 1).

The MATLAB code of this algorithm is shown below in Figure 26. In this algorithm the BS

does not need to know about the SNR of the users.

Figure 26: MATLAB code of Round Robin Algorithm

For the above example in Figure 24, the allocation array can be obtained as follows:

If we continue for more time slots we will have

49
There are 8 users here, and all the users are allocated in the Round Robin scheduler by

allocating first 6 users in the first time slot. When the time slot is totally occupied then it

starts allocating the second time slot from user 7, and the allocation continues.

4.3.3 Proposed Scheduling:

The proposed scheduling algorithm combines the ideas of the best CQI and Round Robin

scheduling. In this algorithm we use best CQI for odd time slots and round-robin for the even

time slots. In this way we can keep the balance between capacity and fairness, which is

shown in Figure 27.

50
Figure 27: MATLAB code for the proposed algorithm

The scheduling array of the proposed algorithm for the above example can be represented as

51
The best CQI scheduling is used in the odd time slots and the Round Robin scheduling is

used in the even time slots.

After the users are scheduled we calculate the capacity, modulation of efficiency of each

subcarrier and equivalently of each user for each time slot. These parameters are extracted

from the LTE standard, as shown in Table 8.

Table 8: Efficiency of each resource block as a function of CQI

CQI Modulation Efficiency Code rate

0 None 0 0
1 QPSK 78 0.1523
2 QPSK 120 0.2344
3 QPSK 193 0.3770
4 QPSK 308 0.6016
5 QPSK 449 0.8770
6 QPSK 602 1.1758
7 16QAM 378 1.4766

52
 8 16QAM 490 1.9141
9 16QAM 616 2.4063
10 64QAM 466 2.7305
11 64QAM 567 3.3223
12 64QAM 666 3.9023
13 64QAM 772 4.5234
14 64QAM 873 5.1152
15 64QAM 948 5.5547

If a user is allocated more than one RB, then their capacity would be the sum of the capacity

of the RBs, and their efficiency would be the average of the efficiency of all the allocated

RBs.

4.4 Fairness index

In order to evaluate the algorithms, we consider two factors, network throughput and

fairness. The network throughput is the average normalized (over bandwidth) capacity of all

users over all time slots and can be formulated as

j = 1/ × 1/k × 1/ ff l
O
lh> Oh>

where k is the number of timeslots that we have run the simulation, and l
O is the capacity of

the -th resource block at the m-th timeslot.

The fairness index , is an indicator that shows how the network resources are distributed

among all users. In situations where is close to unity, the network resources are distributed

among all users almost evenly. Low values of indicate that only part of the users are

allocated. The fairness index can be formulated as:

(∑gh> n )5
(n> , n5 , … , ng ) =
× ∑gh> n 5

53
Here, n is the user capacity.

If we have

n> = n5 = ⋯ = ng

then we have

=1

which means the algorithm is completely fair, and all the users have used network resources

equally.

4.5 Simulation Results and Analysis:

Performance versus average SNR or BS power

In this section I evaluate the performance of different scheduling algorithms for different BS

transmit powers.
Network throughput [b/s/Hz]

Figure 28: Network throughput vs. BS transmit power

54
The x axis of Figure 28 can be replaced by average SNR, defined as the mean of the SNR

[dB] of all users:

Network throughput [b/s/Hz]

Figure 29: Network throughput vs. User’s average SNR [dB]

As Figures 28 and 29 illustrate, higher values of transmit power lead to a better signal to

noise ratio at the receiver of the UEs, and enable the user to use higher order modulation

schemes such as 64QAM. As a result, the throughput of the network increases as the transmit

power increases. We can also see that the network efficiency of the best CQI is the best in

comparison with other algorithms, and the efficiency of the round-robin algorithm is at the

bottom. The proposed algorithm performs in the middle.

55
 Network fairness index

Figure 30: Fairness index of different algorithms

In Figure 30, we have compared the performance of the algorithms in terms of fairness. In

low SNR all the scheduling algorithms have low fairness index, because most of the time the

capacity of the RBs is zero and the users are located out of the communication range of the

BS. However, as the average SNR increases, the fairness index also increases. It is clear that

although best CQI has the highest efficiency in terms of network throughput, it introduces the

worst fairness amongst users. The proposed algorithm introduces a kind of moderate

performance. The fairness of the proposed algorithm is located in the middle, and it has

provided a balance between capacity and fairness. The fairness index of the round robin

algorithm is very large since all users have the chance of being allocated all the time.

56
Performance versus average SNR or BS power for large number of users

For this simulation I increased the number of users to = 20, and we compared the

performance of the algorithms in such a highly congested network.

Network throughput [b/s/Hz]

Figure 31: Performance of the algorithms with large number of users

57
 Network fairness index

Figure 32: Fairness index for a network with large number of users (K = 20)

According to Figure 31, when the number of users is large, network efficiency of the best

CQI gets better, because there is a higher chance that a user with higher CQI can be found

(users near BS). On the other hand, in Figure 32, the fairness index of the best CQI and the

proposed algorithm drops. Still the round robin algorithm performs the best in terms of

fairness.

Performance versus average SNR or BS power for MIMO

In the last results we have only one transmit antenna at the transmitter and one receive

antenna at the UE. In this simulation I have increased the number of antennae, and now I

have a 2 by 2 MIMO system. The number of users is set to 6, and I compare the performance

of the algorithms.

58
 Network throughput [b/s/Hz]

Figure 33: Network throughput for 2x2 MIMO systems, (K = 6)

The MIMO system and maximal ratio combining help users experience a better SNR at their

receiver. The effect of spatial diversity plays a major role in a MIMO Rayleigh fading

channel, and that can be observed from Figures 33 and 34. We can see that the network

efficiency has been increased a lot for all the algorithms, but the fairness index has not

changed that much.

Another fact evident in Figures 33 and 34 is that at BS transmit power of 35 dB, the SNR of

all users is above 20 dB, and therefore all the users are the same in terms of CQI. Under this

condition all the algorithms perform the same in terms of throughput and fairness index.

59
 Network fairness index

Figure 34: Fairness index for 2x2 MIMO systems, (K = 6)

60
Performance versus average SNR or BS for smaller operating system

In this simulation we consider a SISO system with 6 users.

Network throughput [b/s/Hz]

Figure 35: Network throughput for a 300m by 300m operating area

The length and width of the operating area are reduced to 300 meters, and as a result the

users are located closer to the BS. Under this condition, they experience a better SNR and the

efficiency will increase. This can be seen in Figure 35.

If we compare the results of this section and previous sections we observe that reducing the

operating area more or less has the same effect as using the MIMO system in terms of

network throughput.

61
 Network fairness index

Figure 36: Fairness index for a 300m by 300m operating area

In terms of fairness index, we do not see any considerable changes, as shown in Figure 36.

Performance for large system bandwidth

In this section we have increased the system bandwidth to 10MHz, and this means that in

each time slot we have 50 resource blocks. Under this condition we have evaluated the

performance of the system. Still we consider the SISO system and the number of users as 6.

As the number of resource blocks increases, we expect more diversity for each user, due to

the independent Rayleigh fading channel. However, since the number of users is kept very

low (K = 6), the path-loss plays a major role in SNR of each user. The capacity of the best

CQI algorithms slightly increases due to the fading diversity. Yet we do not expect any

throughput improvement for the round-robin scheduling, as shown in Figure 37. The

simulation results in Figure 38 show that the fairness index has not changed considerably.

62
 Network throughput [b/s/Hz]

Figure 37: Network throughput for B = 10MHz

63
 Network fairness index

Figure 38: Fairness index for a network with B = 10MHz

Performance of the scheduling algorithms vs. number of users (low SNR)

In this experiment, we evaluate the performance of the scheduling algorithms as the number

of users in the system increases. First we consider the low SNR at which the BS power is set,

15 dBm.

64
Figure 39: Network throughput vs. number of users (Low SNR)

The Figures 39 and 40 depict that, for best CQI and the proposed algorithm, the network

throughput increases as the number of users increases, while their fairness indices decrease.

For the Round Robin case the throughput and fairness index remain more or less the same for

any number of users.

65
Figure 40: Fairness index vs. number of users (low SNR)

Performance of the scheduling algorithms vs. number of users (high SNR)

In this experiment, we set the BS TX power to 25dBm, and we have high average SNR.

Figures 41 and 42, illustrates that in comparison to previous sections, the throughput and

fairness index for all algorithms has increased, but still the same behavior is observed.

66
 Network throughput [b/s/Hz]

Figure 41: Network throughput vs. number of users (high SNR)

Network fairness index

Figure 42: Fairness index vs. number of users (high SNR)

67
Performance of the algorithms vs. total number of antennae (low SNR)

In this experiment, we keep the BS TX power at 15dBm, and we change the number of

antennae at the transmitter and receiver. As explained before as the number of antennae

increases, the SNR will also increase due to the diversity and maximal ratio combining.

Therefore, a better throughput for all the algorithms is observed in Figure 43. The fairness

also gets better, but this does not happen for the low SNR case, which is observed in Figure

44.

Figure 43: Network throughput versus number of antennae at TX and Rx (high SNR)

68
Figure 44: Fairness index versus number of antennae at TX and Rx (low SNR)

Performance of the algorithms vs. total number of antennae (high SNR)

In high SNR we can see that the performance of all algorithms in terms of throughput and

fairness has increased to a point that they all work nearly the same, as shown in Figures 45

and 46.

69
Figure 45: Network throughput versus number of antennae at TX and Rx (high SNR)

Figure 46: Fairness index versus number of antennae at TX and Rx (high SNR)

70
 Chapter 5: Conclusion and Future Work

5.1 Conclusion

The main objective of this thesis was to study the LTE Downlink Scheduling

Algorithms. In order to know about these Scheduling Algorithms, The study the overview of

LTE is needed. There are many enabling technologies in the LTE standard that help in

improving the capacity of the system and their coverage. OFDM is one of the enabling

technologies used in the downlink transmission, where the total bandwidth is divided into a

number of time-frequency resource elements. The scheduler is a key element present in the

base station, which helps in allocating the time-frequency resources to the users.

In this thesis, I discussed two scheduling algorithms, namely the Best CQI and the

Round Robin Schedulers. The two algorithms are studied in detail, where the Best CQI

schedules the users that have good channel quality and thus increases the throughput of the

system. Round Robin schedules the users sequentially without considering the channel

quality and thus maximizes the fairness of the system. After scrutinizing the effects of

throughput and fairness, I have proposed a new scheduling algorithm. There are two slots in

a sub-frame. According to the new scheduling algorithm, the first slot of each sub-frame is

filled with the highest CQI value, and the second slot is scheduled sequentially. Hence, the

new scheduling algorithm is an adjustment between the fairness and the throughput.

These scheduling algorithms have been implemented in MATLAB™. By considering

different transmission schemes and different scheduling algorithms, the throughput analysis

is performed. A better throughput is observed in the new scheduler compared to the Round

Robin Scheduler and better fairness compared to the Best CQI. In order to improve the

throughput, multiple antenna transmission schemes are used.

5.2 Future Work

Packet scheduling algorithms play a very important role in scheduling the resource

blocks to the users considering throughput and fairness. According to our requirement we can

optimize throughput or fairness. In my thesis, I made a compromise between the throughput

and the fairness. If we want to optimize the throughput, we can develop the Best CQI

algorithm, and thus the new algorithm is proposed. If we want to optimize the fairness, we

can develop either Round Robin or the Proposed Algorithm. By considering different

transmission schemes like MIMO we can optimize the throughput of the system.

72
 References

[1] European Telecommunications Standards Institute, ETSI TS 136 300 V13.2.0

(2016-01), 2016.

[2] International Mobile Telecommunications-2000 (IMT-2000), Recommendation ITU-

R M. 687-2, Feb 1997.

[3] International Mobile Telecommunications-2000 (IMT-2000), RECOMMENDATION

ITU-R M.1457-10, ITU-R (2010), Detailed Specifications of the Radio Interfaces of

International Mobile Telecommunications-2000, May 2010.

[4] ITU-R, Resolution ITU-R 57, Principles for the Process of Development of IMT

Advanced, October 2007.

[5] H. Myung and D. Goodman, Introduction to single carrier FDMA. Hoboken, N.J.:

Wiley, 2008.

[6] S. Schwarz, C. Mehlfuhrer and M. Rupp, "Low complexity approximate maximum

throughput scheduling for LTE", 2010 Conference Record of the Forty Fourth

Asilomar Conference on Signals, Systems and Computers, 2010.

[7] E. Dahlman, S. Parkvall and J. Skold, 4G LTE/LTE-Advanced for mobile broadband.

Oxford: Academic Press, 2011.

[8] A. Ghosh and R. Ratasuk, Essentials of LTE and LTE-A. Cambridge UK: Cambridge

University Press, 2011.

[9] H. Zarrinkoub, Understanding LTE with MATLAB.

[10] Y. Cho, MIMO-OFDM wireless communications with MATLAB. Singapore: IEEE

Press, 2010.
[11] A. Goldsmith, Wireless communications. Cambridge: Cambridge University Press,

2005.

[12] J. Zyren, Overview of the 3GPP Long Term Evolution Physical Layer, 1st ed. Free

scale semiconductor, 2007.

[13] S. Parkvall, E. Englund, A. Furuskar, E. Dahlman, T. Jonsson and A. Paravati, "LTE

evolution towards IMT-advanced and commercial network performance", 2010 IEEE

International Conference on Communication Systems, 2010.

[14] F. Khan, LTE for 4G mobile broadband. Cambridge, U.K.: Cambridge University

Press, 2009.

[15] "A Mathematical Theory of Communication", Claude E. Shannon, 2009.

[16] H. Jafarkhani, Space-time coding. Cambridge: Cambridge University Press, 2005.

[17] S. Megajolla, "Interference Cancellation Technique with Orderings in Rayleigh

Fading Channel for MIMO Systems", M.Tech, Lovely Professional University, 2014.

[18] S. Alamouti, "A simple transmit diversity technique for wireless communications",

IEEE J. Select. Areas Commun., vol. 16, no. 8, pp. 1451-1458, 1998.

[19] H. Lee, S. Park and I. Lee, "Orthogonalized Spatial Multiplexing for Closed-Loop

MIMO Systems", IEEE Transactions on Communications, vol. 55, no. 5, pp. 1044-

1052, 2007.

[20] J. Akhtar and D. Gesbert, "Spatial multiplexing over correlated MIMO channels with

a closed-form precoder", IEEE Transactions on Wireless Communications, vol. 4,

no.5, pp. 2400-2409, 2005.

[21] G. Tsoulos, "Adaptive antennas and MIMO systems for wireless communications:

part II", IEEE Commun. Mag., vol. 42, no. 12, pp. 61-61, 2004.

74
[22] European Telecommunications Standards Institute, ETSI TS 136 212 V11.1.0 (2013-

02), 2013

[23] J. Proakis and M. Salehi, Digital communications. Boston: McGraw-Hill, 2008.

[24] European Telecommunications Standards Institute, ETSI TS 136 211 V10.0.0 (2011-

01), 2011.

[25] P. Tan, Y. Wu and S. Sun, "Link adaptation based on adaptive modulation and coding

for multiple-antenna OFDM system", IEEE J. Select. Areas Commun., vol. 26, no. 8,

pp. 1599-1606, 2008.

[26] S. Falahati, A. Svensson, M. Sternad and T. Ekman, "Adaptive modulation systems

for predicted wireless channels", GLOBECOM '03. IEEE Global

Telecommunications Conference (IEEE Cat. No.03CH37489).

[27] F. Capozzi, G. Piro, L. Grieco, G. Boggia and P. Camarda, "Downlink Packet

Scheduling in LTE Cellular Networks: Key Design Issues and a Survey", IEEE

Communications Surveys & Tutorials, vol. 15, no. 2, pp. 678-700, 2013.

[28] International Telecommunication Union (ITU), Overall network operation, telephone

service, service operation and human factors, ITU-T Recommendation E.800 Annex

B, Aug. 2008.

[29] N. Kolehmainen, J. Puttonen, P. Kela, T. Ristaniemi, T. Henttonen and M. Moisio,

"Channel Quality Indication Reporting Schemes for UTRAN Long Term Evolution

Downlink", VTC Spring 2008 - IEEE Vehicular Technology Conference, 2008.

[30] R. Love, R. Kuchibhotla, A. Ghosh, R. Ratasuk, W. Xiao, B. Classon and Y.

Blankenship, "Downlink Control Channel Design for 3GPP LTE", 2008 IEEE

Wireless Communications and Networking Conference, 2008.

75
[31] R. Kwan, C. Leung and J. Zhang, "Multiuser Scheduling on the Downlink of an LTE

Cellular System", Research Letters in Communications, vol. 2008, pp. 1-4, 2008.

[32] D. Liu and Y. Lee, "An efficient scheduling discipline for packet switching networks

using earliest deadline first round robin", Proceedings 12th International Conference

on Computer Communications and Networks (IEEE Cat. No.03EX712).

[33] H. Ramli, R. Basukala, K. Sandrasegaran and R. Patachaianand, "Performance of

well known packet scheduling algorithms in the downlink 3GPP LTE system", 2009

IEEE 9th Malaysia International Conference on Communications (MICC), 2009.

76
Sarepalli, Sindhura. Bachelor of Engineering and Technology, Acharya Nagarjuna
University, Spring 2014; Master of Science, University of Louisiana at Lafayette,
Spring 2016
Major: Telecommunications
Title of Thesis: LTE Downlink Scheduling Algorithms
Thesis Director: Dr. Michael A. Pratt
Pages in Thesis: 92; Words in Abstract: 164

ABSTRACT

In this thesis, an Overview of the LTE Downlink Scheduling Algorithms is presented.

LTE provides high data rates and is expected to be used by many users in future since it has

good coverage and has access to internet through their devices. The main mechanism of this

LTE network in handling the user traffic is the scheduler. A scheduler is a key functionality

in which the base-station decides which user should be given resources to receive or send the

information. We focused on LTE Downlink schedulers and their functionality by studying

their algorithms. The users are assigned with the shared resources by the network scheduler.

We considered two scheduling algorithms, Round Robin and the Best CQI. The impact of

each scheduler in terms of throughput and fairness and their abilities are known. Throughput

and fairness are the two performance measures in the scheduler. By studying these two

schedulers, a new scheduler is proposed which is a compromise between the throughput and

the fairness.
 Biographical Sketch

Sindhura Sarepalli was born in Vijayawada, Andhra Pradesh, India on March 25,

1991. She graduated from Acharya Nagarjuna University in 2013 with a Bachelor of

Engineering and Technology in Electronics and Communications Engineering. In spring

2014, she joined the Electrical Engineering program at the University of Louisiana at

Lafayette. She earned a Master of Science degree in Electrical Engineering in spring 2016.