5G Off Road: Coverage for Rural Applications

5G Off Road

Coverage for Rural Applications

YASIN EL GUENNOUNI

Master of Science Thesis

Stockholm, Sweden 2013

5G Off Road

Coverage for Rural Applications

YASIN EL GUENNOUNI

Master of Science Thesis performed at the Radio Communication Systems Group, KTH.

November 2013

Examiner: Ben Slimane

KTH School of Information and Communications Technology (ICT) Radio Communication Systems (RCS)

TRITA-ICT-EX-2013:245

⃝ Yasin El Guennouni, November 2013 c

Tryck: Universitetsservice AB

i

Abstract

The attention for rural coverage was increasing in media. The need for faster mobile internet speeds was growing, not least for the forest industry. This industry was becom- ing more updated and sufficient wireless communication could make it more efficient.

Additionally, rural coverage could be used by people living in remote areas, such as farmers, and tourist organizations as well.

In order to enhance the rural coverage, firstly the existing coverage had to be determined. There were coverage maps done by the operators; however people were not satisfied with them since the stated coverage not always was available. Hence, one goal of this study was to verify how much coverage that was available for the forest industry.

The other goals were to verify this coverage and suggest how to improve it.

Coverage maps from operators were compared with harvesting locations for the forest industry. This would give an illustration of how the situation looked like, accord- ing to the available coverage maps. In addition a couple of methods that verified the coverage maps were done. These methods were then backed up by theoretical calcula- tions.

The results showed that the best coverage according to the operator coverage maps

was given by Net1. Furthermore the best verification of coverage is given by Ascom Tems

products, if several details are wanted. However, if a budget solution would be sought,

less advanced applications could be enough.

Acknowledgements

Verily, all praise is for Allah; as none of this would have been possible without him enabling it for us. I would like to thank my family for supporting me and having patients with me during this thesis.

I would like to thank Ben Slimane for pointing me into the direction of this thesis and for his advice during it.

I would like to express my gratefulness to my advisor Mats Nilson for his dedication and willingness to help as well as the time he offered to help me. I have learned a lot from his knowledge and guidance during this thesis.

I would also like to thank my external thesis supervisors Skogforsk and Andreas Barth for his support and help. It was appreciated, especially the help with ArcGIS and for his interesting suggestions.

I also appreciate the help and support from the other partners in this project. They pro- vided me with valuable information and help to advance in the project. Many thanks to Anders Hedlund from Ascom, Tommy Ljunggren from Telia, P˚ al Frenger from Ericsson and Jesper Simons and the crew from PTS.

ii

Contents

Abstract i

Acknowledgements ii

List of Figures v

List of Tables vii

1 Introduction 1

1.1 Swedish Forest Industry . . . . 1

1.2 Telecommunication in Rural Environments . . . . 2

1.3 Previous Work . . . . 3

1.4 Problem Description . . . . 4

2 Review of System Standards 5 2.1 The First Generation . . . . 5

2.2 The Second Generation . . . . 6

2.3 The Third Generation . . . . 6

2.4 The Fourth Generation . . . . 7

2.4.1 E-UTRA . . . . 8

2.4.2 OFDMA . . . . 8

2.5 The Fifth Generation . . . . 8

3 Theory 11 3.1 Radio Propagation and Coverage . . . 11

3.2 Path Loss Modeling . . . 12

3.2.1 Path Loss Models . . . 12

3.3 Link Budget . . . 15

3.4 Fading . . . 15

3.4.1 Fast Fading . . . 15

3.4.2 Slow Fading . . . 18

3.5 Interference . . . 18

3.5.1 Problem . . . 18

3.5.2 Solution . . . 18

3.6 Diversity . . . 19

iii

Contents iv

3.6.1 Space Diversity . . . 19

3.6.2 Polarization Diversity . . . 20

3.6.3 Comparison . . . 21

3.7 Theory Conclusion . . . 21

4 Methodology 23 4.1 Comparison of Existing Maps . . . 23

4.2 Verification of Real Performance . . . 26

4.2.1 Background . . . 26

4.2.2 Tems Investigation . . . 27

4.2.3 Bredbandskollen . . . 28

4.2.4 Calculation of the Propagation Loss . . . 29

5 Results 31 5.1 Comparing Existing Maps . . . 31

5.2 Verifying Real Performance . . . 38

5.2.1 Tems Investigation . . . 38

5.2.2 Bredbandskollen . . . 38

5.2.3 Calculation of Propagation Loss . . . 39

6 Recommendations 45 6.1 Recommendations for Verifying Coverage . . . 45

6.2 Recommendations for Improving Coverage . . . 45

7 Conclusion 47 8 Future Directions 49 A Equipment and software 51 A.1 Equipment . . . 51

A.1.1 Antennas . . . 51

B Equipment and software 55 B.0.2 Screenshots . . . 55

Bibliography 59

List of Figures

1.1 Stacked timber . . . . 2

2.1 First generation phone . . . . 5

2.2 Second generation phone . . . . 6

2.3 Third generation phone . . . . 7

2.4 OFDMA . . . . 8

3.1 A hexagon . . . 12

3.2 A multipath . . . 16

3.3 Rayleigh Distribution . . . 16

3.4 Raician Distribution . . . 17

3.5 Space diversity . . . 20

3.6 Polarization diversity . . . 20

4.1 Telia 3G . . . 24

4.2 Modelbuilder in ArcGIS . . . 25

4.3 Road . . . 26

4.4 Euipment setup . . . 27

4.5 Measurement equipment . . . 28

4.6 Bredbandskollen . . . 28

4.7 Location of road . . . 30

4.8 Path loss . . . 30

5.1 Coverage Telia . . . 32

5.2 Coverage Tele2 . . . 33

5.3 3s coverage . . . 34

5.4 Telenor coverage . . . 35

5.5 Net 1 coverage . . . 36

5.6 Diagram . . . 37

5.7 Tems . . . 38

5.8 Telia blank spots . . . 39

5.9 Map 900 . . . 40

5.10 Map 1800 . . . 40

5.11 Graph Vaxholm 900MHz Location 1 . . . 42

5.12 Graph Vaxholm 900 MHz Location 2 . . . 42

5.13 Graph Vaxholm 1800MHz Location 1 . . . 43

5.14 Graph Vaxholm 1800 MHz Location 2 . . . 43

A.1 Mini mag . . . 51

v

List of Figures vi

A.2 Specs mini mag . . . 52

A.3 Specs mini mag . . . 53

A.4 Specs mini mag . . . 54

B.1 Bredbandskollen . . . 55

B.2 Bredbandskollen Tele2 . . . 56

B.3 Bredbandskollen Telia . . . 57

List of Tables

5.1 Results from ArcGIS maps. . . . 37 5.2 The results from Bredbandskollen. . . . 39

vii

Dedicated to my family

ix

Chapter 1

Introduction

Sweden consists of 41 million hectares of land. Almost 23 million hectares are covered by woods [1]. This makes Sweden one of the bigger exporters when it comes to different types of wood and things related to it.

Just as all the other industries, this industry is becoming more and more automated and utilizes technology to make it more efficient. Since the forest industry is spread throughout the long country of Sweden, a major problem is mobile internet coverage.

This service is of great importance when it comes to transferring data to, and from the machines out in the woods. As the back end office can be very far away from the field workers, the wireless communication can enable the workers with maps and directions.

On the other way around, the machine operator can give valuable information to the back end office. That can be size and quantity of the harvested trees as well as time for pick up.

Regular phone service is also a crucial service, since it in addition to content in general also can be used to avoid accidents. It can also be serious faults on the machinery that have to be solved quickly; phone contact can enable the machine operator with support to solve the issue. Stoppages are very expensive for the forest companies.

1.1 Swedish Forest Industry

The Swedish forest industry is important for the Swedish welfare. It creates jobs in all parts of the country and is also a significant part of Swedish exports. The Industry has helped to build modern Sweden and will continue to play an important role in the future. Interesting amounts of different products produced by wood are [2]:

1

Chapter 1. Introduction 2

• 15.9 million m

of sawn wood comes from coniferous forest.

• 12 million tons of paper pulp, of which 7.6 million tons of pulp, 3.6 million tons of pulp, 0.3 million tons of sulphite pulp and 0.6 million tons of semi-chemical pulp

• 11.4 million tonnes of paper and paperboard, of which 2.0 million tons of newsprint, 2.1 million tonnes of wood-containing printing paper, 1.4 million tonnes of woodfree printing paper, 1.0 million tonnes of packaging paper, 1.9 million tons of corrugated materials and 2; 6 million tons of paperboard for packaging..

Figure 1.1: Cut timber that is ready for transport.

These digits are from 2012 and some of them are preliminary. As seen above, the use of wood is widespread and it is something that almost all humans today deal with in their everyday life. It can be used for everything from furniture to pizza boxes.

1.2 Telecommunication in Rural Environments

The rural coverage is at the moment dominated by the use of GSM 900 and CDMA2000 that utilizes the 450 MHz band. The latter one is a 3G standard. These lower frequency bands have great characteristics that enable coverage with much less base stations. As a comparison, using the same number of base stations with CDMA2000 in the 450 MHz band can cover an area that is 12 times larger than the area covered using 2100 MHz or 1800 MHz [3]. There is only one operator that offers CDMA2000 using the 450 MHz band in Sweden today and that is Net1.

In the transition of moving from analogue to digital TV broadcasting the 800 MHz band

suddenly became available. This was auctioned out by The Swedish Post and Telecom

Chapter 1. Introduction 3

Authority (PTS) [4]. Since this frequency band has good properties for rural coverage, the telecom operators could utilize it to supply rural households and companies with a broadband connection. Even though a lot of new customers have been provided with broadband connections there are still large parts of Sweden that are not covered.

Another solution would be to use LTE which offers higher data rates at the cell boarders with a very low latency. It has always been very tempting for the operators to use LTE in the lower frequency bands and some have started to change portions of the GSM 900 MHz into LTE. The same thing is done with the just mentioned 800 MHz band.

1.3 Previous Work

Not a lot of effort has been put on studying coverage for rural environments. The main interest in general is to solve the interference and capacity problems that occur in urban areas.

• In [5] a method that evolves from using an 80 MHz system to using the 450 MHz is studied. The 80 MHz band worked well for communication used by machines in the forest, but higher data rates are required today.

• In master thesis [6], the author investigates if LTE gives better coverage than GSM.

• In [7], operator diversity in forest and rural areas is examined. The author looks if utilizing all available operators will give a better result than only using the one with best overall coverage.

• Another thesis, [8] handled propagation in a coniferous forest using the 900 MHz band as a function of different mobile antenna heights.

• In [9] PTS have made checks in 10 random places in Sweden to see if the operators coverage maps are true. This was done since customers complain on these maps.

The customers are unsatisfied because of the optimistic claims of high download speeds given by the operators. This report aimed at telling if the maps are telling the truth.

However, I have not found a study handling solutions for improving 5G coverage in forest

areas. The term 5G is in this thesis used as a name for future system technology and

solutions for mobile internet. This might be improved versions of the existing systems

GSM/3G/LTE.

Chapter 1. Introduction 4

1.4 Problem Description

This thesis is part of the larger 5G Off Road collaboration between Telia, Ericsson, As- com, PTS, Skogforsk, LRF and Wireless@KTH. They worked towards an enhancement of coverage in rural areas.This topic has recently got some attention in different newspa- pers such as ATL [10] and Elektronik Tidningen [11]. PTS has also been very active in this manner. Even though a lot of work have been done by the operators and almost all households and companies in Sweden have access to a fast 4G connection at their postal adress. There are still many spots on the map that are not covered [12]. This is a big problem for the forest industry since they are spread all over the country. Furthermore, they do not work at specific places but rather change places all the time making it hard for the telecom operators to provide them with coverage. It is understandable that the operators do not want to deploy base stations and provide coverage where there are no population and where the network only will be used for a short period of time. However some compromise will have to be provided in order to make both parts satisfied, i.e. the operators and the forest industry.

It is not only the forest industry that has interest in this topic. A smaller part of Swedens population that live in remote areas and organizations that have tourism on remote places also have great interest in providing a Q&S for their users. People today want to be able to make Facebook and twitter updates as well as uploading and sending pictures from their vacation to friends and family. So there is no place on the map that do not need coverage.

What makes this master thesis different from previous work is that:

1. It will try to give a clearer picture of how much percentage that actually is not covered by the operators in practice. This is of great interest for the forest industry since they do not know how much of Swedens area that actually is covered in practice. Some previous research have been done where different individuals have been asked about their experience of the service. This is usually not that precise since personal feelings often overemphasis the answer.

2. It will propose solutions on how to improve service availability in the future, and give

methods on how to verify the coverage.

Chapter 2

Review of System Standards

2.1 The First Generation

The first cellular networks were introduced in the pre 1980s. They were analogue systems which mainly supported speech. The first running system was launched in Japan by Nippon Telegraph and Telephone (NTT). The year was 1979 and it was called the automobile telephone [13]. In Europe the first standard, The Nordic Mobile Telephone (NMT), was implemented in 1981 and utilized the 450 MHz and later the 900 MHz band. In US the AMPS was introduced around the year 1980.

The cellular systems were later on divided into generations making this the first gen- eration (1G). The main disadvantage with these 1G systems was the lack of frequency spectrum efficiency. This made it unsuitable for the systems to operate in the same fre- quencies near each other. Another problem was that roaming not could be implemented between different standards.

Figure 2.1: A first generation phone [14].

5

Chapter 2. Review of System Standards 6

2.2 The Second Generation

The next generation of cellular systems, known as Global System for Mobile communi- cations (GSM) was implemented in the early 1990s. The main things that differed it from the 1G systems were that it used a digital technology and that it also had to offer a common standard to create a single market for mobile phones [13]. New features were narrowband circuit swithed data service, Short Message Service (SMS) and better voice quality. 2G utilized the frequency band in a more efficient way. It came to be a very popular system which until today is used and expands all over the world.

The 2G system later on evolved and new features such as General packet radio service (GPRS) and Enhanced Data rates for GSM Evolution (EDGE) were introduced. These features enable packet data transport which is more robust than the previously used method. The maximum data rates that could be reached with these standards were 236 kbit/s with 4 timeslots in downlink.

Figure 2.2: A 2G phone, Nokia 3310 from 2000 which sold 126 million units [15].

2.3 The Third Generation

The next revolution in cellular systems is today known as 3G. International Telecom- munication Union (ITU) worked during 1986 to 2000 on defining a new standard that offered higher data rates. The final systems came to be of two types, namely the Wide- band Code Division Multiple Access (W-CDMA) or UMTS implemented in Europe and Multi-Carrier Code Division Multiple Access (MC-CDMA) implemented in America.

These different CDMA techniques differs from the previously Time Division Multiple

Access (TDMA) used in GSM systems. CDMA originally came from the military and

Chapter 2. Review of System Standards 7

was employed during World War II. Instead of allocating a frequency to each user CDMA uses the entire spectrum that is available. Each user is encoded with digital sequence and by doing so more users can be connected and the spectrum is utilized in an efficient way, resulting in higher data rates. [16] Improvements were also done in 3G systems.

New additions to W-CDMA going under the names High-Speed Packet Access (HSPA) and HSPA+ were implemented. This enabled higher data rates of up to 21 Mbps using 64 Quadrature amplitude modulation (QAM) [17].

Figure 2.3: A 3G phone, with e616 NEC enabled video calls for the first time. [18]

2.4 The Fourth Generation

The next generation in mobile networks entered the market in the end of 2009 by Telia in Sweden. The biggest difference in this new system is that it is all packet switched.

It was made possible by the new radio interface Evolved Universal Terrestrial Radio Access (E-UTRA) and core network evolution System Architecture Evolution (SAE).

This ground breaking technology was mainly implementable due to the few restrictions on being compatible with the older generations, given by 3GPP. The motivating factors for LTE are given below [19] :

• The data rates provided by 3G had to be increased to satisfy the user demand.

• A transition from circuit switched to a packet switched system had to be imple- mented.

• The constant demand for cost reduction (CAPEX and OPEX).

• A system with low complexity was required .

• Paired and unpaired band operations had to be merged.

Chapter 2. Review of System Standards 8

Another very important feature in LTE is that it supports spectrum allocations between 1.4 MHz and 20 MHz [19]. This enables operators around the world that do not own spectrum in the higher frequencies to implement LTE in their networks.

2.4.1 E-UTRA

The air interface in LTE is referred to as E-UTRA. It is a standard that, in time, would replace the older Universal Mobile Telecommunications System (UMTS) and HSPA used in 3G systems. The E-UTRA used in LTE is unlike HSPA a totally new air interface system that is separate from W-CDMA. It solely runs with packet data and provides very high data rate at the same time as it provides a low latency. Orthogonal Frequency Division Multiple Access (OFDMA) is used in the downlink and Single Carrier Frequency Division Multiple Access (SC-FDMA) in the uplink [20].

2.4.2 OFDMA

In OFDMA the bandwidth is divided into many related orthogonal subcarriers which are shared between users. By doing so, the spectrum is efficiently used. However, this method requires very fast processors. A drawback with this method is that it requires a large amount of signal processing power which in turn increases the battery consumption.

This is not a problem in the downlink (DL), but would cause problems in the uplink (UL). Therefore another method that uses less power is used for the UL. As mentioned earlier in the report and also visible in the figure below, OFDMA is used for the DL and SC-FDMA for the UL. [20].

Figure 2.4: OFDMA and SC-FDMA [20].

2.5 The Fifth Generation

At the present time there is no finished standard or requirements for the 5G. However,

that does not stop the companies from claiming that they are offering 5G. It is clear

Chapter 2. Review of System Standards 9

that 5G has to offer higher bitrates and capacity than the current available 4G but exactly where the boarders are is not that clear as it was in the transitions of former generations. Nevertheless, it is most likely that the new systems will offer speeds in the range of Gbps and that it will use Multiple Input Multiple Output (MIMO) antennas, since some companies already have made promising tests with these features [21]. In this thesis the term 5G is used as a name for future system technology and solutions for mobile internet. In this case, this might be new technology as just mentioned or improved versions of the older generations GSM/3G/LTE.

The Swedish telecom operator Telia, which is included in this ”5G Off Road” project, presented their requirements for 5G in a newspaper. They were [22]:

• To define 5G as one unit so the customers know what they get.

• Higher speeds for larger areas.

• New possible use, such as the surf in 3G.

• Higher speeds and lower response time.

• Higher amount of transfered data per person in a small area (e.g. soccer arena).

• Cheaper cost.

However, an equal responsibility was put on the handset and device makers. As none of

the mentioned points could be fulfilled if no device could support it.

Chapter 3

Theory

The following chapter is meant to give an insight on important parts of radio commu- nication. It is mainly intended for readers that are not familiar with the area of radio communications.

3.1 Radio Propagation and Coverage

The electromagnetic waves that are used in radio communications are not visible to the human eye and very hard to predict. They bounce of objects, split into different parts and are being reflected. How the waves behave depends on the wave length. Usually the waves are not affected by objects that are smaller than the wave length. This is why different frequencies are suitable for different types of communication. [23]

The area that is provided with a signal from a base station is usually referred to as cell.

If the cell has a sufficient signal, i.e. the Quality of Service (QoS) is reached, it is said that it has coverage. Small areas within the cell that are not provided with a sufficient signal are referred to as white spots. In reality the cells are not regularly shaped due to different reasons such as reflection by different objects and terrain conditions. However, in order to plan and describe efficient cell plans hexagonal cells are usually used. In a perfect world, the signals from two adjecent base stations are equal at the cellboarders of the hexagonal cells (see figure 3.1). This is called equisignal level. The hexagonal shape is commonly used and can be found in many situations in nature. Examples of that are snowflakes, bee hives and some molecule shapes. It is a very effective reuse shape which minimizes the cell boarders [23].

11

Chapter 3. Theory 12

Figure 3.1: Hexagonel cell pattern.

3.2 Path Loss Modeling

As described in the previous section, electromagnetic waves are hard to predict. Nev- ertheless, they can be described by Maxwell equations. These equations can be used to calculate how the waves interact with different mediums. This method is very useful in radio communications since it shows how much energy that is lost when a signal is sent from a transmitter to a receiver or vice versa. This is called path loss. However in order to get an accurate result, the mediums and path conditions have to be precisely known.

This makes the method of using Maxwells equations in real scenarios very complex and not practical to use. Path loss models have been created in order to describe how the waves behave in different situations. The path loss is given by:

L

= P

G

G

P

(3.1)

Where the parametars are as follows:

L

= Path loss

P

= Transmitted power P

= Received power

G

= Transmit antenna gain G

= Receive antenna gain

3.2.1 Path Loss Models

There are mainly three path loss models that are suitable for different circumstances.

These models are mainly divided into three types, namely [24]:

Chapter 3. Theory 13

Empirical Models: These are models that strictly use observations and measurements.

They are not used to explain physical behavior of systems. They are instead used to predict the path loss.

Deterministic Models: This model utilizes the laws of electromagnetic waves to deter- mine the signal power at a specific position.

Stochastic Models: These models are least accurate. Thats because they use series of random variables to model the environment. On the other hand these models require the least information and use less processing power for the predictions.

Free Space

The simplest path loss model is used in situations where no obstacles occur between the transmitter and receiver. It is called the free space propagation model and it simply divides the transmitted power equally over a spherical area with the radius r. It is given by the following relationship [23]:

L

= 4π ∗ r

λ (3.2)

Plane Earth Model

This model is suitable for common propagation situations. It is called the plane earth model and it is based on a multipath propagation. It means that the signal that is transmitted from the transmitter is divided on the way and appears as several signals at the receiver. It is assumed that two waves are received. By examining the difference in phase it is possible to see if the resulting wave is attenuated or amplified. In order to use the equation for this model, some assumptions have to be made. Namely that the distance is much larger than the heights of the transmitter and receiver antennas and that the ground is perfectly reflecting the waves. This is given in the following equation:

L

= r

G

G

h

h

(3.3)

h

= Height of the transmitter, h

= Height of the receiver (3.4)

In order to simplify calculations, this equation is transformed to decibels and the mul-

tiplications and divisions becomes additions and subtractions:

Chapter 3. Theory 14

L

= 40log10r − 20log10hT x − 20log10hRx + 10log10GT x + 10log10GRx (3.5)

Okumura Model

This is an empirical model which was developed in Japan. It was formed by making a wide range of measurements using many different frequencies. Frequencies between 15-1920 MHz where used, but according to [25], up to 3000 MHz were tested. Since this model was developed for macrocells in the range of 1-100 km and antenna heights between 30-1000 m, it is very useful for many different situations.

Hata Model

This model is also known as the Okumura-Hata model, as it was developed from the Okumura model. It was published by Masaharu Hata 1980 [26], and is the most widely used model for prediction in urban areas. Hata used the information from the field strength curves in the Okumura model and created new path loss equations. Several propagation studies were carried out in the Tokyo area, both inside the site and around, which lead to this curves that illustrate how the field strength varies with the distance.

Major drawbacks with the Hata model are the limitation in path length and the limited upper frequency. Some modified models that compensate for these limitations are given below.

Cost 231 Extension of the Hata Model

This model, which is known as, the cost 231-Hata model compensates for the limitation in frequency range. It has the following specifications [27]:

• It can handle frequencies between 1500-2000 MHz

• Antenna heights for mobile stations at 1-10m

• Base station antenna height between 30-200m

• Link distance of 1-20 km

and is given by [28]:

L

= 46.3 + 33.9log(f ) − 13.82log(Hb) − a + [44.9 − 6.55log(Hb)]log(d) + C

(3.6)

Chapter 3. Theory 15

Where the constant C

is 0 dB for average sized cities and suburban areas and 3 dB for dense populated areas with a lot of buildings.

3.3 Link Budget

The radio link between a base station and a handheld Mobile Station (MS) is best described by a link budget. It is the initial step when a radio network is being planned.

The link budget expresses the gains and losses between the transmitter and the receiver.

It is usually expressed as signal energy-to-noise power spectral density in digital signaling [23]. The following equation is retrieved (in dB):

E

N

= EIRP − L

+ G

− F − kT

− R (3.7) Where:

E

= Energy per bit

N

= Noise spectral density L

= Basic path loss

R = Bit rate F = Noise factor

k = Boltzman’s constant T

= Room temperature

EIRP = Effective Isotropic Radiated Power

This equation enables the system designer to swap values in order to get the desired performance from the system.

3.4 Fading

Another major problem in radio communications is fading. It happens when a signal is transmitted in a wireless medium and reflections are causing a ”standing wave” pattern.

That is a wave which remains in a constant position. This results in a variation in signal level which causes problems in the communication. Fading is usually divided into two types. These two types will be described below.

3.4.1 Fast Fading

When a signal is transmitted it can either reach the receiver directly in a so called line

of sight case, or it can reach it in numerous replicas called multipath. In figure 3.2 an

Chapter 3. Theory 16

example with one light of sight wave and three reflected waves is visible. The latter case is most common, but in reality the signal is most often scattered, diffracted, reflected and absorbed by different objects on the way [23]. This results in many incoming waves at the receiver which have different amplitudes and angles. For this reason the signal level varies quickly, hence the name fast fading.

Figure 3.2: Multipath propagation.[29]

Fast fading is usually modeled by distributions, commonly distributions that are used is Rayleigh and Rician distributions.

• Rayleigh Distribution: Is appropriate to use when there is no dominating line of sight propagation between the transmitter and receiver. It assumes that the signal in the channel varies randomly and will fade according to this distribution.

The Rayleigh distribution is normally used in dense populated areas that have a lot of buildings. The probability density function (PDF) for different values sigma is illustrated below.

Figure 3.3: PDF for the Rayleigh distribution.[30]

Chapter 3. Theory 17

• Rician Distribution: Is in contrast to Rayleigh fading normally used when the line of sight signal is dominating. It is a stochastic model which is used when the signal arrives at the receiver by multipath and at least one path is varying, as in figure 3.2. This distribution is visible in the figure below.

Figure 3.4: PDF for the Rician distribution. Where σ is distance between the refer- ence point and the center of the bivariate distribution and v is the scale. [31]

Chapter 3. Theory 18 3.4.2 Slow Fading

This type of fading is caused by the objects and terrain in the way of the signal. It varies over the distance and the signal strength will be different depending on this distance. It is typically caused by large obstacles; hence it is also called shadow fading. It is often modeled by a log-normal distribution [32].

3.5 Interference

3.5.1 Problem

All waves that are transmitted in cellular networks are interfering with each other, especially when they are using the same frequencies or carrier waves. This is very problematic and causes major issues when it comes to providing a satisfying wireless service. The fact that operators use different generations of cellular networks at the same time makes the problem more complex. This is due to the close frequencies the different generations utilize. As an example, LTE can be operated in 800 MHz, 900 MHz, 1800 MHz, 2100 MHz and 2600 MHz bands [33]. Theses frequencies are very close to frequencies used by 3G and 2G systems. The increased demand in wireless communication makes the spectrum more crowded which leads to even more interference.

3.5.2 Solution

Guard band

To solve the interference problem, the most straight forward thing to do is to implement a guard band between two network systems or carriers. It is a simple solution which leaves a part of the frequency band between frequencies unused. However, since spectrum is a very precious resource which is hard to get by, this method may waste valuable frequencies.

Co-site interference

When the operators built up their networks with previous generations of cellular systems,

huge amounts were invested. So when the fourth and fifth generations of systems were to

be deployed a lot of investments could be saved if they could co-exist and share sites with

the previous systems. The problem that arises with this is the co-site interference. A lot

of solutions have been developed, e.g. co-site neighboring channel interference solutions

[33]. An example of that is the UMTS and GSM which run on 5 MHz respective 200

Chapter 3. Theory 19

KHz bands and do not require a guard band. The same method can be used to control interference when implementing newer generations of networks.

Co-channel interference

To avoid interference from adjacent cells in mobile networks same frequencies are not used in the neighboring cells, except for CDMA systems which uses codes to separate users in the same channel. However, useful frequencies are very hard to get so they have to be re-used as close as possible. The distance that describes this is called the reuse distance and is given by [23]:

D

= R

√

3K where K is the clustersize (3.8)

But these frequencies can sometimes still reach other cells that use the same frequencies, causing co-channel interference. To avoid this phenomena an isolating zone that is using another frequency is put between the cells. Another solution is that the GSM900 networks probably will decrees, moving to newer networks in the future and that will leave room for the newer generations to use its frequency.

3.6 Diversity

Using diversity is an effective way of minimizing the effect of Rayleigh fading. The technique is mostly used on the receiving side, at the base station. The idea is to transmit several copies of the same signal. The receiver can then receive one good copy or combine several signals into one satisfying copy. There are two types of diversity used in mobile communications, space diversity and polarization diversity. Space diversity was the dominating method for a long time, but polarization diversity has increased dramatically and is the dominating method in mobile networks today.

3.6.1 Space Diversity

As indicated by the name, two antennas are placed close with a distance D from each

other as illustrated in the figure below. The two antennas send the same signal to the

receiver, and the distance D needs to be carefully chosen in order to ensure independent

signals. There are to methods that can be used on the receiver side, the simple method

lets the receiver chose the strongest signal, this is called selection diversity. The more

complex but ideal method that combines the signals is called maximum ratio combining

[23].

Chapter 3. Theory 20

Figure 3.5: Space diversity.[34]

3.6.2 Polarization Diversity

This method combines pairs of antennas with orthogonal polarization. This method can protect a system from mismatches in polarization that otherwise would result in fading by pairing two opposite polarizations.

The main advantage with polarization diversity is that it requires fewer antenna units.

Deployment of a single antenna box is cheaper and at the same time more environment friendly than double antennas. It also minimizes the wind load on the tower, which can be a big problem on very high towers.

Figure 3.6: Polarization diversity.[34]

Chapter 3. Theory 21 3.6.3 Comparison

As mentioned above, the polarization diversity requires fewer antennas and it offers less wind load on the tower. But in general, it is safe to say that space diversity always performs better. The difference between the two techniques is not very noticeable in urban areas where the cells are small, so it is more preferable to use polarized diver- sity. However, in rural areas where the cells are very large the number of sites can be dramatically decreased by using space diversity.

3.7 Theory Conclusion

As mentioned in the beginning of this chapter, it was intended for readers that were not familiar with radio communication. Especially the first section that handled ra- dio propagation and coverage. These mechanisms are relevant for all types of wireless telecommunication, including this thesis.

Since rural environments with large cells are handled in this thesis the Hata model and its extension to higher frequency was selected as the most appropriate to use as a path loss model. It is an empirical model with a broad range of input parameters. The model was used in the results section to calculate the propagation loss. In addition, link budget was also used to find the propagation loss.

Since this thesis handled rural environments the most fading objects are large and not moving. This is due to the terrain, where large trees and hills might be fading. However, the fast fading had to be taken into account as well due to the small wavelength used in GSM 900 MHz and 1800 MHz. In order to compensate for the changes in amplitude, the device was move back and forth approximately 1m in the stationary measurements.

On the other hand, interference was not considered; it mostly causes problems in dense populated areas. In rural environments there are few base stations covering large areas with lower capacity demands so the interference was assumed to be insignificant and thus not considered.

Diversity is utilized in rural environments. By using space diversity it is possible to

cover a larger area than with polarization diversity in rural environments. This is due

to the lack of multipaths that is needed for the polarization diversity to perform well

[35]. So space diversity makes it possible to reduce the number of sites, which are very

expensive. Although diversity may enhance coverage by some dB, this has not been

included in the link budget calculations.

Chapter 4

Methodology

The main aim of this report was to give a clearer view on how many percent of Sweden that were covered with a mobile network connection. From the forest industry workers point of view. The aim was also to suggest on how to make the coverage better in a 5 to 10 year perspective. In order to solve the first point, a map of Sweden with points illustrating where Skogfors’s partners were active was compared to the mobile operator’s coverage maps provided by PTS. However, it is very hard to predict radio propagation using simulation tools and sometimes the operators overestimate the coverage. To solve the last item, field measurements were carried out to verify the coverage map and to propose a solution to how the maps can be verified in the future. This chapter will give a detailed description on how the maps were compared and how the performance was tested.

4.1 Comparison of Existing Maps

To illustrate the mobile internet coverage situation for the Swedish forest industry, their workplaces were compared with the areas where mobile operators have stated that they have coverage. Skogforsk’s map showed GPS coordinates on places where they had finished harvesting trees. There was 43727 points in total. In reality the places consist of unregularly shaped areas, but to make it easier to compare with the coverage maps, points from the center of these areas were used.

Skogforsk’s map was given in so called shape format (.shp) and the operator’s maps were provided by PTS in tab format (.tab). Skogforsk’s map was from 2012 and PTS maps were from the end of 2012. The operators were only asked to provide maps with places where they have coverage and which type of standard they were using (e.g. GSM, 3G

23

Chapter 4. Methodology 24

or LTE). In total there was 12 maps provided by five operators, namely; Telia, 3, Tele2, Telenor and Net1.

First step was to convert the maps into the same format. An open source software called Quantum GIS was used to do this. The maps were then processed in Esri’s ArcGIS software. GIS stands for Geographic Information System and lets one integrate hardware, software and data for managing, analyzing and displaying different forms of geographic information [36].

The method was to create different layers, one layer with the points and one layer with the coverage maps and then put them on top of each other to see where they overlay, see figure 4.1. A problem that came up was that the maps were in different geographical coordinate systems. This resulted in the fact that the maps were showed next to each other instead of on top of each other. To solve this, all maps were projected into the official Swedish referencesystem called SWEREF99 [37].

Figure 4.1: To the left a coverage map example and to the right harvestry places.

Chapter 4. Methodology 25

In order to execute several functions in ArcGIS, one can use the so called model builder.

This model can among many things change projection, make new layers or add a new field to the table that shows all the values. An example, the model used in this thesis is given below.

Figure 4.2: The Modelbuilder in ArcGIS.

The top part, denoted 1, handled the first stage. The upper part of 1 was for the Skogforsk map and the lower part of 1 was for the operator maps. This stage fixed the problem caused by different reference systems in the operator maps and created a feature layer for the Skogforsk map. The function define projection overwrote the coordinate system information stored with a dataset. Then the spatial data was projected to the new coordinate system. Next function selected features from one layer based on a spatial relationship to features in the other layer. On the lower part, denoted 2, the combined maps were saved into one layer. Further, a new field was added and the values were calculated, displaying a 1 for coverage and a 0 for no coverage.

There were five mobile internet providers and three different standards that they use,

namely GSM, 3G and LTE. The just mentioned procedure was repeated for all operators

and standards resulting in a total of 12 maps. The data was presented in tables and

staple diagrams showing the total points and how many of the points that were covered

in percent.

Chapter 4. Methodology 26

4.2 Verification of Real Performance

4.2.1 Background

In order to see if the coverage maps were telling the truth and to give a clearer insight on how the theory really works, field measurements were taken. The results would also help to suggest a solution on how to verify if there really was coverage in a larger and statistically correct way. The field measurements were made on a remote road close to Vaxholm, Stockholm.

A notice here is that the measurements and verification for real performance in principle only handled GSM 900 MHz and 1800 MHz band, although som weak 3G was available.

There was no LTE coverage at the measurement location.

Two methods were used, the first was using professional Ascom Tems equipment and software. The second method was using an Note II Android phone with its internal antenna and a free application.

The specific place, S¨ oderby, was chosen after examining the operators coverage maps and finding a blank spot in this area. Since the maps provided by PTS were approximately 6 months old, Telia’s homepage was used to get an updated map that confirmed that this area was without coverage. This map is visible in figure 5.8. The road where measurements were taken on was a small gravel road which can be seen in figure 4.3.

An interesting notice related to this topic was that trees had been cut and stacked in this area. So the test environment was very close to the real world scenario which can be faced by workers in the harvest machines.

Figure 4.3: The road where measurements were taken left, and woodstack right.

Chapter 4. Methodology 27 4.2.2 Tems Investigation

The software used with the equipment listed in Appendix A is Ascom Tems investi- gation. It is the leading tool for troubleshooting, optimizing and maintaining wireless networks. Tems investigation can provide important features such as continuous scan- ning of GPS coordinates, received signal level, channel number and base station identity codes (BSIC). The values of these features can then be exported into tables, lists or Esri shape files.

Main focus was put on received signal level, which indicates if acceptable communication is possible between the mobile unit and the base station. This value was exported into an Esri shape file, which showed the values of the received signal level on a map. The values were represented by colored dots and the scale is visible at the top left corner. A limit for acceptable phone service is normally -104 dBm for GSM 900 and -102 dBm for GSM 1800. The BSIC and absolute radio-frequency channel number (ARFCN) could also give useful information since they could be used to find the exact location of the transmitting base station.

The measurements were taken from a car. Antennas were put on top of the roof and the car was driven back and forth on the designated road. Three antennas were mounted on the top of the car but only two were used, as the far left one in figure 4.4 was a 4G antenna and no such signals were available there. An illustration of the setup is given in figure 4.4. The used antennas were one GPS antenna and one Smarteq Minimag Antenna. The last one is a magnetic mount antenna that covers the 900 MHz and 1800 MHz frequency bands. Its specifications can be found in Appendix A.

Figure 4.4: Equipment setup.

Scans were made on Telia’s GSM 900 MHz and 1800MHz bands and in 3G 2100 MHz

band in different ways. Firstly explicitly GSM or 3G was chosen. Secondly GSM and 3G

Chapter 4. Methodology 28

mode was activated, where the phone switched mode automatically according to signal strength. Since the GPS function on the Tems Z750i did not work simultaneous as tests were taken, the PCTEL EX was used to log the GPS coordinates. It is visible in figure 4.5.

Figure 4.5: The measurement equipment, antennas to the left and cell phone and GPS device to the right.

4.2.3 Bredbandskollen

To test the environment with something closer to reality tests using Bredbandskollens coverage app were carried out. No external equipment was used, only a Android smart- phone; Note II running the application. This application measures the speed for uplink and downlink as well as the response time to the server. The measurement was done to the nearest internet exchange point and the application was available for Android and IPhone. A Screenshot of the interface is visible in figure 4.6.

Figure 4.6: Startpage of the bredbandskollen application.

Chapter 4. Methodology 29 4.2.4 Calculation of the Propagation Loss

Since the BSIC and the ARFCN of the transmitting base station was retrieved using Tems Investigation, its exact location could be found.

Together with information about the base station, such as height, frequency, transmit power, and distance to the mobile unit, the projection loss of the received signal could be calculated.

When the value was retrieved it could be compared to a standard Hata propagation path loss graph from [38]. This graph is seen in figure 4.8. Hence, better understanding of the situation could be gotten as the results would show what terrain the pathloss corresponded to. The procedure was done for two base stations located on opposite sides of the two measured spots. The spots were located at the two ends of the road in figure 4.7. Where the blue line corresponds to the designated road. The frequencies that were used were 900 MHz and 1800 MHz. The tower heights and transmitted power varied between the two base stations.

Three sector antennas were used on the sites. Two antennas covered the road from one site while only one antenna was pointed at the roads direction from the other site. These types of antennas normally have about 16 dBi gain for 900 MHz and 18 dBi for 1800 MHz. A typical such antenna is 2m long and its polarization is dual linear, slant (± 45

) [39]. This makes it possible to use MIMO.

One of the tower heights was lower than the height used in [38]. This means that the

path loss would be higher and the lines in the graph would move up in figure 4.8. In

order to find how much they would change, the propagation loss formulas in [38] were

used. It was enough to calculate the path loss for 1 km and 10 km and then draw a line

between them.

Chapter 4. Methodology 30

Figure 4.7: The location of the measured road.

Figure 4.8: Graph with path loss as a function of distance and terrain (900MHz).[38]

Chapter 5

Results

This chapter is divided into two parts. The first part compares Skogforsk’s maps to the operator’s coverage maps. These results will be given for each separate system, frequency band and operator. The reason for that was to give Skogforsk a detailed knowledge of the coverage and a chance to pick the one that suited them best, according to the areas and network standards that they used.

Secondly the chapter handled the verification of real performance which included in field measurements and calculations of the propagation loss.

5.1 Comparing Existing Maps

The red dots in the following figures correspond to Skogforsk’s identified workplaces. If the workplace was covered with mobile internet the red dot became green. The total workplaces that Skogforsk had in the maps were 43727. In reality the workplaces were unregularly shaped but GPS co-ordinates illustrated as dots from the center of gravity for these shapes were used to simplify the comparison.

The operators coverage maps were provided by PTS, and they were only showing areas where the operators stated that they had coverage and did not indicate how strong the signal was or the the bitrate to expect.

31

Chapter 5. Results 32

Telia

The figure shows Telia’s three different standards, GSM, 3G and LTE per 2012. Results showed that Telia covered:

• 97.63% with their GSM network (includes both 900 and 1800 MHz).

• 74.03% with their 3G network.

• 6.38% with their LTE network.

Figure 5.1: Telias coverage, from left: GSM, 3G and LTE.

Chapter 5. Results 33

Tele2

The figures below illustrate Tele2’s coverage maps compared to Skogforsk’s coverage maps. The standards that are visible are GSM, 3G, LTE 900 MHz and LTE 2600 MHz.

A remark here is that Tele2 and Telenor now share LTE and GSM networks under the name Net4Mobility [40]. Results showed that Tele2 covered:

• 76.64% of the points with their GSM network.

• 66.17% of the points with their 3G network.

• 48.36% of the points with their LTE 900 MHz network.

• 2.27% of the points with their LTE 2600 MHz network.

Figure 5.2: Tele2’s coverage, from left: GSM, 3G, LTE 900 MHz and LTE 2600 MHz.

Chapter 5. Results 34

3

The maps below shows the operator 3’s coverage maps compared to Skogforsk’s maps.

The standards offered by 3 were 3G and LTE (800 and 2600 MHz). A remark here is that 3 did not have a GSM network. Results have showed that 3 covered:

• 68.36% with their 3G network.

• 14.17% with their LTE network.

Figure 5.3: 3’s coverage from left: 3G and LTE.

Chapter 5. Results 35

Telenor

The pictures below shows Telenor’s coverage maps on top of Skogforsk’s map. The standards that are visible are GSM and 3G. As mentioned above, Telenor shared their LTE and GSM networks with Tele2. The results showed that Telenor covered:

• 47.49% of the points with their GSM Network.

• 64.76% of the points with their 3G network.

Figure 5.4: Telenor’s coverage from left: GSM and 3G.

Chapter 5. Results 36

Net1

The map below shows Net1’s coverage compared to Skogforsk’s workplaces. Net1 only offers 3G internet using CDMA2000 in the 450 MHz band. The results showed that Net1 covered:

• 98.01% of the points with their 3G network.

Figure 5.5: Net1 coverage for 3G.

Chapter 5. Results 37 Operator and type Number of covered points Number of uncovered points Percentage

Telia GSM 42689 1038 97.63%.

Telia 3G 32369 11358 74.03%.

Telia LTE 2789 40938 6.38%.

Tre 3G 29892 13835 68.36%.

Tre LTE 6195 37532 14.17%.

Telenor GSM 20768 22959 47.49%.

Telenor 3G 28316 15411 64.76%.

Tele2 GSM 33514 10213 76.64%.

Tele2 3G 28933 14794 66.17%.

Tele2 LTE 900 21148 22579 48.36%.

Tele2 LTE 2600 992 42735 2.27%.

Net1 3G 42855 872 98.01%.

Table 5.1: Results from ArcGIS maps.

Figure 5.6 shows how many of the 43727 work points that were covered to the left and how many percentage that corresponds to on top of each staple. Table 5.1 shows the values in numeric form.

Figure 5.6: Diagram showing the number of covered points to the left and the per- centage on top of each staple.

Chapter 5. Results 38

5.2 Verifying Real Performance

Previous section gave a clearer insight on how the situation looked like according to coverage maps done by the operators. This section on the other hand gives a more real- istic illustration of how the coverage actually was in field. Two methods were used, one professional with sophisticated tools and software usually used by the industry. While the other method could be done by anyone with an IPhone or Android smartphone. The section finishes of with calculations of the theoretical propagation loss as a comparison.

5.2.1 Tems Investigation

The scan in figure 5.7 shows how the coverage looked like when the phone in idle mode decided when to attach to and switch between GSM and 3G. The light green represent GSM, the dark green 3G and the red shows spots with insufficient signal level. This map can then be compared to the following methods presented beneath.

Figure 5.7: Map with received signal level for GSM 900 MHz.

5.2.2 Bredbandskollen

As can be seen on Telias coverage map and the Tems test map the blank spots that

lack coverage were situated between the two ends of the road. This was also obvious

in the Bredbandskollen test, since tests could not be made using Telia in between the

just mentioned points. An error message was then given as seen in figure 4.6. The

measurements were taken from left to right at four different spots seen in figure 5.8. At

Chapter 5. Results 39

the first round a Telia SIM card was used and in the second round a Tele2 card was used. The Telia card was not able to connect to the server at two of the spots (location 2 and 3), while Tele2 was able to connect at all spots. The values retrieved are given in Table 2, and all screenshots of the values from the interface are given in the Appendix C.

Figure 5.8: Telias Blankspots.

Telia Tele2

Location Speed UL/DL Response Time Location Speed DL/UL Response Time

1 0.033/0.013 Mbit/s 2085 ms 1 0.044/0.012 Mbit/s 2006 ms

2 x x 2 0.046/0.017 Mbit/s 204 ms

3 x x 3 0.084/0.014 Mbit/s 212 ms

4 8.769/1.281 Mbit/s 87 ms 4 0.172/0.090 Mbit/s 314 ms

Table 5.2: The results from Bredbandskollen.

5.2.3 Calculation of Propagation Loss

Maps

The received signal strength from the two Telia base stations is visible in figures 5.9 and 5.10. Different colors and shapes were used since the 900 and 1800 MHz frequency bands were scanned from two base stations. Figure 5.9 shows coverage for the 900 MHz band. In total coverage from three sector antennas were scanned in this case. Two of them were from one site and one from the other site.

Figure 5.10 shows the 1800 MHz case from two antennas. The color interval is visible in the top left corner and the values are in dBm. Input leveles sufficient for coverage were taken from [38]. As expected from the theory, the 1800 MHz band offers worse coverage.

1800 MHz is mostly used for capacity and not coverage.

Chapter 5. Results 40

Figure 5.9: The 900 MHz measurement.

Figure 5.10: The 1800 MHz measurement.

Chapter 5. Results 41

Graphs

The two graphs in figure 5.11 and 5.12 maps the actual propagation loss from the two sites. One graph shows the results for an antenna that had the height 100 m while the other one was located at 27 m.

The measured propagation loss calculated using the formulas in [38] is illustrated as the horizontal blue respectively red lines. These values were calculated using the sensitivity threshold, transmitter power and antenna gain. Following values were used:

• Thresholds used were -102 for GSM 1800 MHz and -104 for GSM 900 MHz.

• Transmit power for GSM 900 MHz 47 W (antenna at 100m) and 43 W (antenna at 27m) .

• Transmit power for GSM 1800 MHz 45 W (antenna at 100m) and 42 W (antenna at 27m).

For the base station with antenna at 100 m the propagation loss was found to be 152 dB and for the one at 27 m the loss was 148 dB.

As seen in the graphs, the two spots that were chosen showed a bad signal level. Even though the base station in the first case was located at a relatively close distance, it had worse path loss than an urban indoor environment at the same distance. The reason for that was most definitely because of the terrain varriations and deep forest that surrounded the measured area.

The other graph for the antenna at the height 27 m showed that the path loss was slightly

lower than the one at 100 m. The path loss was between the urban and suburban path

losses at this distance. This is shown in figure 5.10. There could be many reasons for

that. One reason could be that the wave path from that side mainly traveled over water

which makes the attenuation much smaller.

Chapter 5. Results 42

Figure 5.11: Path loss vs Cell Radius, BS height = 100 m (GSM 900).

Figure 5.12: Path loss vs Cell Radius, BS height = 27 m (GSM 900).

Chapter 5. Results 43

The graph in figure 5.13 corresponds to the same case as in figure 5.11 where the base station heigth was 100 m, however this time the 1800 MHz was used.

Figure 5.13: Path loss vs Cell Radius, BS height = 100 m (GSM 1800).

The graph in figure 5.14 corresponds to the same case as in figure 5.12 where the base station heigth was 27 m, however this time the 1800 MHz was used. As the antenna height is lower the path loss becomes higher.

Figure 5.14: Path loss vs Cell Radius, BS height = 27 m (GSM 1800).

Chapter 5. Results 44

The verification showed that Telia’s coverage map was very close to reality. If their

coverage map from their website in figure 5.8 is compared with the map retrieved from

Tems investigation in figure 5.7 it is inevitable that the white spots lack coverage. It is

seen that most red dots are situated in-between the two ends off the measured road.

Chapter 6

Recommendations

6.1 Recommendations for Verifying Coverage

The application Bredbandskollen or something similar could be a competitor to Ascoms Tems investigation if a low cost compromise is sought. An own application could also be developed if the available ones lacked some features of interest. However, there might be a lot of ready applications ready for use already. In [41] a interesting mobile coverage verification was carried out. It was stated that few vehicles visit as many places in the communities as the garbage trucks. So smartphones running an Android application called T¨ ackningskollen was put in the dash of the garbage trucks. One smartphone for each operator. The measurements start when the driver turns on the ignition to the truck and stops when he turns it off.

On the other hand, If a more detailed verification would be of interest Ascom offers a wide range of products.

6.2 Recommendations for Improving Coverage

As discussed earlier in this report, some types of compromises had to be made by the operators and the companies of interest. Some Suggestions are for improvments are:

• The operators could over a long period of time expand their coverage. However the companies might not want to wait for that long.

• Faster solutions might be to aim antennas in the direction of harvesting places to increase the signal strength.

45

Chapter 6. Recommendations 46

• Companies could invest in portable base stations in forms of trailers to move to wanted places or satellite connected base stations [42].

• Cell phones with a high sensitivity could be chosen to maximize the possibilities of communication. Research have shown that the choice of mobile can dramatically reduce the amount of dropped calls at the cell boarders [9].

• Most cell phones have a connection for external antennas such as the one used in

this thesis. If a antenna is mounted on the roof of a harvest machine, this can

increase the signal strength dramatically.

Chapter 7

Conclusion

The enhancement of rural mobile internet coverage was desired by many parts, especially the forest industry. This thesis confirmed the existing coverage from the forest industry point of view, were it stated which operator and system standard that provided best coverage. Further it proposed verification methods and solutions on how to improve coverage in the future.

As for the 5G standard, key features would be to expand the coverage and at the same time provide higher data speeds.

The study showed that the best coverage was provided by Net1 using a 3G standard.

The best GSM coverage was given by Telia, and the best LTE was provided by Tele2.

As for the verification of coverage, Ascom products offered the most complete solution with many methods to use. The Tems investigation method was proven to work as verification for coverage since it gave the same results as the available coverage map from Telia. The sensitivity threshold from the 3GPP specification was proven to be good coverage citeria for vehicle antennas. However, if a budget solution with less information would be sought, Android applications could be used.

47

Chapter 8

Future Directions

The limitation of time in this study restricted the field measurements and standards used. To get a more complete view of the situation more measurements should be done. As an example a bachelor thesis could be done where further measurements are taken. Since only measurements using 3G and GSM were carried out in this study, LTE measurements should also be studied.

The term 5G was also hard to define in this study since the industry was at variance concerning the term. It might be easier to define it when more work have been done and the industry move towards an unite goal.

49

Appendix A

Equipment and software

A.1 Equipment

• Dell Laptop running Windows XP Professional

• Ascom Tems Sony Ericsson Z750i; GSM & 3G with external and internal antenna

• Ascom PCTEL EX, LTE2600 UMTS and GPS function

• Note II running Android 4.2.1

• MiniMag, magnet mount antenna

A.1.1 Antennas

• MiniMag, magnet mount antenna

• 2.6meter cable, SMA-male (specifications below)

• Frequency range 824-960/1710-2690 MHz

• 2.15 dBi gain

Figure A.1: Mini Mag Antenna.

51

Appendix A. Appendix Title Here 52

Figure A.2: Antenna specification.

Appendix A. Appendix Title Here 53

Figure A.3: Antenna specification.

Appendix A. Appendix Title Here 54

Figure A.4: Antenna specification.

Appendix B

Equipment and software

B.0.2 Screenshots

Figure B.1: User interface of the application.

55

Appendix B. Bredbandskollen 56

Figure B.2: Screenshots from Tele2 measurements.