Reliable communication in mine environments for autonomous vehicles

(1)

INOM

EXAMENSARBETE INFORMATIONS- OCH

KOMMUNIKATIONSTEKNIK, AVANCERAD NIVÅ, 30 HP

,

STOCKHOLM SVERIGE 2016

Reliable communication in mine

environments for autonomous

vehicles

ALESSANDRO TOMASI

KTH

(2)

(3)

KTH Royal Institute of Technology

School of Information and Communication

Technology

Dept. of Communication systems

Wireless@KTH

Degree project in Electrical Engineering

(4)

Jan Dellrud

(5)

Abstract

Automation in the mining industry has the potential to increase safety and produc-tivity while improving working conditions. Ore transportation within the mine is a repetitive task which is well suited to be replaced by an autonomous mining vehicle operating around the clock.

Scania, a world leader in sustainable transport solutions is investigating this new concept of vehicle. The autonomous operation is enabled by several technologies installed on the vehicle, including a communication system object of this thesis. Connectivity among vehicles is required in order to coordinate paths and exchange mission critical information.

In this thesis, after identifying the challenges of wireless propagations in mines, the communication technology is chosen and possible antenna configurations and communication ranges are found.

Through numerical link-budget simulations and subsequent range measurements, the potential communication range of this vehicle has been quantified. The results show the effectiveness of height diversity in extending the communication range. Lastly, the performance degradation caused by dust accumulated on the antennas is discussed.

(6)

(7)

Sammanfattning

Automatiseringen i gruvindustrin har potentialen att öka säkerheten och produk-tiviteten samtidigt som den förbättrar arbetsvillkoren. Malmtransporten inne i gruvan är en repetitiv uppgift som passar bra att bli utbytt av en autonom gruvtransport som är i drift dygnet runt.

Scania, en av de världsledande inom hållbara tranportlösningar, undersöker just nu denna typ av fordon. Den självstyrande driften aktiveras genom att ett fler-tal teknologier insfler-talleras på fordonet, inklusive ett kommunikationssystem som är ämne för denna avhandling. Anslutningen mellan fordonen är nödvändig för att kunna samordna banor och ge information i för uppdraget kritiska lägen.

I denna avhandling, efter att ha identifierat svårigheterna med en trådlös utbredning i gruvor, är kommunikationsteknologin vald och möjliga antennkonfigurationer och kommunikationsräckvidder funna.

(8)

(9)

Acknowledgment

First and foremost, I want to express my gratitude to my academic supervisor, Mats Nilson, and my industrial supervisor, Jan Dellrud. Over the last six months your guid-ance has been really helpful to overcome the challenges of this project.

A thesis like this would not have been possible without some experience in the mining sector. I would like to sincerely thank Jon Fangel for introducing me to the challenges of the mining industry and connecting me with Atlas Copco.

Thanks are also due to Wesley Santos and Ola Pettersson for their valuable support in radio controlled mining equipment and for sharing their experience with networking infrastructure deployed in mines.

I am deeply thankful for the software and hardware support received during my CST simulations by Christos Kolitsidas and Assistant Professor Oscar Quevedo Teruel.

Heartfelt thanks to my colleague Luca Manara, for embarking with me on this adventurous project. Our collaboration has been an invaluable asset in carrying out our master theses.

Last, but not least, a warm thank you, in Italian, goes to my family.

Un ringraziamento profondo e speciale è dedicato alla mia famiglia, per avermi incoraggiato ad intraprendere la mia carriera universitaria e per il sincero supporto dimostratomi negli anni. Grazie di cuore!!!

Södertälje, July 12, 2016

(10)

(11)

List of Figures

2.1. Rendering of autonomous mining vehicle c 2012 Allan Macdonald . 8 2.2. Phases of open-pit mine operation c 2013 IEEE . . . 9 2.3. Shape of the first Fresnel zone. . . 11 2.4. Dimension of first Fresnel zone at its widest point for 2.4GHz (left)

and 5.9GHz (right) . . . 12 2.5. Cross section of an open-pit mine, whose typical dimensions are shown

in table 2.1. When the truck, represented by the red box, and the infrastructure are placed on different levels of the mine, there might be coverage of the first Fresnel zone, thus resulting in NLOS conditions. [1] 17 2.6. Open-pit mines wireless communication systems c 2013 IEEE . . . . 18 2.7. A sample mesh router placement for an open-pit mine [2] . . . 18 3.1. Channel paths in a 2x2 MIMO configuration . . . 21 3.2. Representation of mutual coupling phenomena . . . 22 3.3. Representation of scattering parameter for a two port network . . . 23 3.4. Effect of correlation (ρ) on dual diversity with SC for two branches

with equal mean SNR (Γ1 = Γ2). . . 25

3.5. Effect of correlation (ρ) on dual diversity with SC for two branches with double mean SNR (Γ₁= 2Γ₂). . . 25 3.6. Effect of correlation (ρ) on dual diversity with SC for two branches

with highly different mean SNR (Γ₁ = 10Γ₂). . . 26 3.7. Effect of correlation on communication uptime . . . 26 3.8. Value of diversity gain achievable for different number of received

copies of signal (M). . . 26 4.1. Representation of truck platooning c www.eutruckplatooning.com . 32 5.1. Communication link model used in link-budget simulations. . . 37 5.2. Illustration of the two ray propagation model. . . 41 5.3. Mine environment used for determining empirical channel model

c

2008 IEEE . . . 43 5.4. RSS over distance for a SIMO 1x2 RX configuration with different

TX antenna height, 1.08m on the left and 2.72m on the right. . . 44 5.5. Combined RSS using SC on the same antennas configuration of figure

(16)

5.6. Diversity gain as function of position on the same antennas

configura-tion of figure 5.4 . . . 45

5.7. RSS according to ETSI ITS standard at 5.9GHz (left) and 2.4GHz(right). 47 5.8. Pathgain of underground channel model for 2.4GHz and 5.9GHz. . . 48

5.9. RSS resulted from a single realization in underground tunnel for 5.9GHz (left) and 2.4GHz(right). . . 48

5.10. Averaged RSS over 100 realizations of the pathloss model for under-ground tunnel for 5.9GHz (left) and 2.4GHz(right). . . 49

6.1. Screenshot of autonomous truck prototype. . . 51

6.2. Screenshot of autonomous truck simulation model. . . 52

6.3. Simplified model of a tunnel where the truck is placed for the simula-tions with the asymptotic solver. . . 54

6.4. Truck configuration with two patch antennas mounted at the front of the vehicle. . . 55

6.5. Truck configuration with side mounted and front mounted patch antennas. . . 56

6.6. Truck configuration with a front mounted patch and a chassis mounted dipole. . . 56

6.7. Initial hit points of ray-tracing simulations for a patch antenna mounted at the front of the vehicle. . . 57

6.8. Initial hit points of ray-tracing simulations for a patch antenna mounted at the side of the vehicle. . . 57

7.1. Illustration of points where range measurements were taken. . . 60

7.2. Antenna positions at the front of the truck . . . 60

7.3. RX configuration for range measurements. . . 62

7.4. Parked trailers in the measurement location behind the yellow truck. 62 7.5. Details of the production facility near the measurement location. . 62

7.6. Measured and simulated RSS for TX placed at glass height (2.60m). 63 7.7. Measured and simulated RSS for TX placed at engine height (1.05m). 63 7.8. Radiation pattern of the antenna developed for the RELCOMMH project. . . 64

7.9. Radiation pattern of RELCOMMH antenna with a plastic layer in front. . . 64

7.10. Polar representation of the antenna developed for the RELCOMMH project. . . 65

7.11. Polar representation of RELCOMMH antenna with a plastic layer in front. . . 65

7.12. Measurement setup with near field analyzer . . . 65

7.13. Measurement setup with plastic layer. . . 65

7.14. Measurement setup with dry dust layer. . . 65

(17)

List of Figures

7.16. Radiation pattern of RELCOMMH antenna covered with 1cm of wet dust. . . 66 7.17. Polar representation of RELCOMMH antenna covered with 1cm of

dry dust. . . 66 7.18. Polar representation of RELCOMMH antenna covered with 1cm of

wet dust. . . 66

(18)

(19)

List of Tables

2.1. Size of biggest open-pit mines in the world according to [3], [4]. . . 10 2.2. Typical dimensions of the cross shape of open-pit mines, from [1] . . 12 4.1. Main differences between 802.11a and 802.11p . . . 28 4.2. SNR required for achieving a frame-error-rate (FER) of 10% in different

environments under different multi-antenna configurations. c 2012 Springer . . . 30 4.3. Summary of differences between highway platooning applications and

mining autonomous driving solutions. . . 34 4.4. Summary of antennas available during this master thesis. . . 35 5.1. Limits for total RF output power and Power Spectral Density at

highest power level. Source: ETSI . . . 39 5.2. Summary of channels allocated for ITS purposes . . . 39 5.3. Attenuation of a sample of commercial cables at different frequencies 40 5.4. Parameters used for underground link-budget calculations. α

rep-resents the pathloss exponent and σ_x the standard deviation of the random variable ∆σ. . . 43

5.5. Communication range for different antenna height for V2V and V2I at 2.4GHz. . . 46 5.6. Communication range for different antenna height for V2V and V2I

at 5.9GHz. . . 46 6.1. S-parameters calculated for two patches at the front of the truck. . . 55 6.2. S-parameters calculated for two patches, one at the front of the truck

and one mounted on the side of the vehicle. . . 56 6.3. S-parameters calculated for a patch at the front of the truck and a

(20)

(21)

Glossary

AMV Autonomous Mining Vehicle. 2–4, 7, 33, 52, 59, 70, 72

ASSET Advanced Safety and driver Support for Essential road Transport. 32 BER Bit Error Rate. 28–31, 39

BSS Basic Service Set. 28

CST Computer Simulation Technology. 4, 5, 34, 35, 51, 53 DSRC Dedicated Short Range Communication. 27

ECC Envelope Correlation Coefficient. 4, 22, 23, 54, 69 EIRP Equivalent Isotropically Radiated Power. 38, 47, 68

ETSI European Telecommunications Standards Institute. xiv, 2, 38, 39, 42, 47,

67–69

ETTE Enabling Technologies for Transport Efficiencies. 32, 33 FCC Federal Communication Commission. 27

FDTD Finite Difference Time Domain. 13 FEM Finite Element Method. 13

GO geometrical optics. 13, 14

GPIB Generic Purpose Interface Bus. 61 GTD Geometrical Theory of Diffraction. 13 ISM Industrial, Scientific, and Medical. 16, 19, 27

ITS Intelligent Transport System. xiv, xvii, 2, 3, 5, 16, 19, 27, 29, 31, 33, 34, 39, 47,

54, 69, 72

(22)

MAC Medium Access Control. 27, 28 MCS Modulation Coding Scheme. 31

MIMO Multiple Input Multiple Output. 5, 21, 22, 29, 30 MISO Multiple Input Single Output. 29, 30

MoM Method of Moments. 4, 13, 53 MTU Maximum Transmission Unit. 30, 31 MWS MicroWave Studio. 53

NLOS Non Line of Sight. xiii, 9, 17, 30, 31, 33, 42, 43, 46, 47, 67, 68, 70 OFDM Orthogonal Frequency Division Multiplexing. 28, 38, 70

PE Parabolic Equation. 15 PHY physical. 27–29

PSD Power Spectral Density. 38

RELCOMMH Reliable Communication for Heavy duty vehicles. xiv, xv, 33, 64–66 RF Radio Frequency. 29, 39, 53

RMS Root Mean Square. 59

RSS Received Signal Strength. xiii, xiv, 13, 24, 37, 43–45, 47, 59, 61, 63, 67–70, 83 RSU Road side unit. 2, 8, 29, 31, 39, 40, 45, 46, 68, 70

RX receiver. xiii, xiv, 2, 10–12, 14, 21, 29, 30, 38–40, 43, 44, 60–62, 68, 71 SA Spectrum Analyser. xvii, 61

SARTRE Safe Road Trains for the Environment. 32 SBR Shooting and Bouncing Ray. 4, 5, 13, 19, 53, 69 SC Selective Combining. xiii, 22, 24–26, 44

SHF Super High Frequencies. 10, 11, 62

SIMO Single Input Multiple Output. xiii, 30, 41, 44 SINR Signal Interference Noise Ratio. 31

SISO Single Input Single Output. 29, 30

(23)

Glossary

SP Sveriges Tekniska Forskningsinstitut. 3, 32

TX transmitter. xiii, xiv, 2, 10, 11, 14, 21, 29, 30, 40, 43–46, 60, 61, 63 UHF Ultra High Frequency. 10, 11, 14, 62

V2I Vehicle to Infrastructure. xvii, 3, 29, 31, 34, 45, 46, 67–69, 72 V2V Vehicle to Vehicle. xvii, 3, 29–31, 33, 34, 44–46, 60, 68, 69 VHF Very High Frequency. 14

VPE Vector Parabolic Equation. 13

WAVE Wireless Access Vehicular Environments. 2, 27, 28 WBSS WAVE Basic Service Set. 28

(24)

(25)

1

Chapter 1.

Introduction

1.1. Mining

Several materials are present in the Earth’s crust. Ever since the beginning of civilization men have been interested in them. It is no coincidence that mining is one of the oldest human activity reported by history.

The mined materials, ranging from metals to coal, have been fundamental in driv-ing and sustaindriv-ing technological development. In the past, mindriv-ing was one of the toughest jobs due to its physically intense nature and dangerous working condi-tions.

Mining techniques evolved dramatically over the centuries. The use of machinery was introduced in order to replace human workers. From simple surface mining, mines became deeper and wider in a race to increase the amount of ore extracted. Several disastrous accidents have been witnessed throughout the years where thousands of workers lost their lives. These casualties have highlighted the importance of reducing human presence in mines in order to increase safety.

Nowadays mineral demands are rising [5] and there is a strong interest from mine operators worldwide to increase the amount of ore mined, reduce extraction costs and improve mine efficiency [6]. These challenges have to be met with an eye on safety, which is a primary concern in an industry so frowned upon in the past for its inhuman conditions.

(26)

1.2. Intelligent Transport System

In 2009, the European Telecommunications Standards Institute (ETSI) allocated the frequency band from 5 855 MHz to 5 925 MHz for Intelligent Transport Sys-tem ITS applications. The new standard [7] specifies the 802.11p protocol, also called Wireless Access for Vehicular Environments WAVE. WAVE allows vehicles to benefit from information exchanges with other vehicles and road side units (RSU).

Vehicular communication poses great challenges from a telecommunication point of view. Transmitter (TX) and receiver (RX) are moving, giving rise to considerable Doppler shifts, while the road environment gives rise to multipath propagation, reflections, scattering and diffraction phenomena. Fleet and freight management, enhanced road safety, traffic control, and location based services are just some of the possible scenario enabled by the ITS [8].

While social barriers are a significant obstacle to the diffusion of autonomous vehicles on public roads, the benefits of this technology could be applied in controlled environments, such as mines, to meet the industry requirements.

Scania is a world leader in heavy duty vehicles, providing sustainable transport solution to several industries. Even though a line of trucks specific for mine oper-ation is already available, to further enter the mining market, the feasibility of an autonomous mining vehicle (AMV) is under investigation at Scania. Autonomous driving is made possible by the installation of several sensors on the vehicle, which together with a path planning algorithm and a communication system allow the vehicle to operate safely without human guidance. The final prototype is expected to be capable of continuous operation without human intervention, thus helping the mining sector reduce costs and enhance productivity.

The main task allocated to this truck is rock transportation from the excavation point to the processing facility of the mine. The business feasibility of this type of operation has been subject to several studies in the past. Commercialization strategies have been analysed in [9] while a case study, related to the Boliden Aitik mine in the north of Sweden, was carried out in [10].

Thanks to Scania’s commitment to innovation through research and development, the communication system of this concept of vehicle will be analysed in this work.

1.3. Problem statement

(27)

1.4. Research scope

project. Despite the presence of an electromagnetic compatibility department and a few antenna experts, there is a lack of electromagnetic modelling and simulations knowledge inside the company.

Within Scania, antenna placement problems are tackled from a practical perspective, consisting mostly of a trial and error approach and extensive real-world measurements. For problems that require simulations, such as data age estimation or antenna design for truck platooning application, Scania outsources the research to Sveriges Tekniska Forskningsinstitut (SP).

With ITS, connectivity becomes fundamental for the normal operation of a vehicle. As a consequence Scania is seeking experts to bring the lacking know-how and skills inside the organization, in order to evaluate how to best approach the design of mission critical wireless systems.

1.4. Research scope

The aim of this Master thesis is to ensure reliable wireless communication between autonomous trucks operating within mines. This consists in providing both vehicle to vehicle (V2V) and vehicle to infrastructure (V2I) communication, as the AMV needs to be manageable from a remote control room and be able to coordinate its operations with other vehicles.

The research will focus on the physical layer of wireless networks. Two versions of the 802.11 protocol family will be considered, operating at 2.4GHz and 5.9GHz respectively.

In particular, different antenna configurations and positions will be analysed with attention on their performance properties. The positions investigated are proposed in another master thesis and in close collaboration with the chassis designers of the innovative truck concept. The advantages and requirements of multi-antenna techniques will be evaluated and assessed in this report.

As the AMV project is still at an early pre-development, the entirety of this work is focused on the vehicle. This product focus is justified by the interest of Scania to have a vehicle adaptable to all mines and by a high site dependance of the infrastructure required for V2I. Therefore, throughout my work the infrastructure plays a marginal role, where assumptions related to it are made exclusively to proceed with the goal of my thesis.

1.5. Research questions

The thesis answers the following research questions:

(28)

• What is the expected communication range for the AMV in mine environments? • What are the antenna positions most suited for exploiting receive spatial

diversity?

• Which environmental factors affect signal propagation in mines?

1.6. Research method

The research started with a literature study focused on mine topologies, wireless vehicular communication, wireless propagation in mine environments and multi-antenna systems.

Within the background research, I carried out several interviews with vehicular com-munication and electromagnetic compatibility experts at Scania. I also contacted the researchers who helped REP, the system pre-development department, in designing their platooning communication system. Thanks to the collaboration with the head of Scania mining, I seeked consultation with Atlas Copco, a market leader in remote controlled mining equipment.

Following the state-of-the-art analysis, link-budget simulations, with closed form analytical channel models, were carried out in MatLab for both underground and open-pit mines. Given the recent standardization of vehicular communication and the wide presence of traditional Wi-Fi in mines, two carrier frequencies were considered, namely 2.4GHz and 5.9GHz. This first part of the thesis identified a rough estimate of the communication range for the autonomous truck. Moreover, it highlighted how multi-antenna techniques could extend this communication range.

The second main part of the investigation consisted in the evaluation of the mutual coupling among the multiple antennas deployed on the truck. After evaluating the radiation patterns obtained with different antenna configurations, scattering parameters and envelope correlation coefficient (ECC) were calculated using the Computer Simulation Technology CST software. CST’s Microwave Studio electro-magnetic simulation suite solves the electric and electro-magnetic fields equations with an integral solver based on the method of moments MoM, which is suitable for electrically large structures. Moreover, the Shooting and Bouncing ray SBR method available within CST was used in order to verify ray propagation in a simplified tunnel model.

The antenna configuration for the front of the truck was assembled on a representative truck model and radio measurements were made in an open-air field to validate the analytical results.

(29)

1.7. Thesis structure

Lastly, the environmental factors typical of mines impacting radio propagation were briefly researched. The impact of a layer of dust on the antennas radia-tion pattern was indicatively measured using a near-field scanner (EMScan RFx-pert).

1.7. Thesis structure

This thesis consists of eight chapters whose content is detailed in this section.

In the first chapter, an introduction to the ITS, mining industry and its challenges is given. The research questions are formulated and the research method is de-tailed.

Chapter two provides background information related to the typical mine environ-ments and their specific challenges in terms of wireless communication. The existing communication infrastructure deployed in mines is explained.

Chapter three focuses on the theoretical aspects behind multi-antenna techniques and diversity gain. The concept of correlation in multi-antenna systems is introduced and its impact on the communication system is highlighted.

In chapter four, the previous related works in ITS are discussed and existing research related to MIMO and tunnel propagation is analysed.

In chapter five link budget simulations are carried out. The cases discussed cover the entirety of possible scenarios for both open-pit and underground mines.

Chapter six discusses the physical layer simulations carried out with the aid of CST Microwave studio. Antenna configurations are analysed in this chapter with emphasis on the mutual coupling between the antennas. Moreover, SBR simulations are detailed in this part of the report.

Measurements carried out are described in chapter seven. This includes both range measurements to validate the choice of the analytical model and indicative measurements highlighting the effect of dust on the radiation pattern of anten-nas.

At the end of the thesis, obtained results and limitations are discussed in chapter eight, whose content also debates aspects related to ethics, sustainability and possible future works building on this work.

(30)

(31)

2

Chapter 2.

Background

This chapter introduces the concept of an autonomous truck for mining operation. Subsequently, environments typical of mines are discussed and their challenges in terms of radio propagations are explained. The chapter is concluded by a literature study on wireless propagation in tunnels and open-pit mines and the analysis of the existing networking infrastructure deployed in mines.

2.1. Autonomous mining vehicle

The autonomous truck, rendered in figure 2.1, is an entirely new vehicle concept. Unlike traditional trucks, the AMV does not have a cabin as the human driver is replaced by technology.

Mechanically, the AMV is quite similar to existing 8x4 mining trucks. A minor difference consists in a strengthened chassis designed in order to benefit from the lack of cabin. As a consequence, load carrying capacity of the vehicle is in-creased.

The main differences between the AMV and conventional trucks lies in its modus operandi. A human driver is no longer responsible for controlling the vehicle. Autonomous unsupervised operation is enabled by the state-of-the-art sensors and controllers installed on the truck.

The computer vision algorithm, responsible of providing sight to the AMV, is based on the fusion of data collected from different sensors. These include LIDARs, a radar like sensor emitting light instead of radio waves, time of flight cameras and automotive radars. These sensor map the terrain and the surrounding environment, enabling autonomous obstacle avoidance and path planning.

(32)

Antennas on the truck have the objective to ensure proper communication with other vehicles and RSU. Wireless communication is a key requirement in synchronizing mine and fleet operations.

Thanks to the heavy automation, the operational cycle of the truck is a strenuous twentytwo hours per day and given its strict maintenance regime, it is likely to be sold as a service to mine operators.

Figure 2.1.: Rendering of autonomous mining vehicle c 2012 Allan Macdonald

2.2. Environments typical of mines

In order to discuss the challenges of wireless propagation in mines, the typical propagation environments are analysed in this section. Different mining techniques gave rise to different typologies of mines [11], which can be classified in open-pit and underground.

Open-pit mines, developed by surface mining, are deep holes on the Earth’s crust. The material is removed from the bottom of the mine and then transported to the processing facility on top.

(33)

2.2. Environments typical of mines

Figure 2.2.: Phases of open-pit mine operation c 2013 IEEE

mining process leads to a continuous extension of the network of underground tunnels.

2.2.1. Open-pit mines

Open-pit mines are exposed to harsh atmospheric conditions depending on their geographic location.

Wide temperature ranges between winter and the warm season are present and full exposure to atmospheric precipitations characterize this environment.

Moreover, open-pit mines are dusty environments and mining equipment is often covered in dust or mud that can impact the radio signal propagation.

The biggest open-pit mines of the world cover a very wide area and can reach very profound depths, as shown in Table 2.1.

Open-pit mines might have numerous simultaneous operations happening at the same time. As a consequence, the mine is always expanding and changing shape during its different phases, as seen in figure 2.2.

2.2.2. Underground mines

Despite several different underground mining techniques exist, these mines are typically characterized by tunnels excavated in the ground in order to reach the ore deposit. These tunnels can have different cross section shapes and sizes. Tunnels can have steep slopes and curves that could cause non-line-of-sight (NLOS) conditions in wireless communication.

(34)

Table 2.1.: Size of biggest open-pit mines in the world according to [3], [4].

Material Depth [m] Area [km2]

Hibbing Minnesota (USA) Iron 180 25

Garzweiler (GER) Lignite 450 48

Bingham canyon (USA) Copper 1200 63

Chuquicamata (CHL) Copper 850 13

Escondia (CHL) Copper 645 10

Unlike open-pit mines, the temperature inside the shafts is homogeneous over the year with a high humidity rate. Air ventilation is poor and after blasting explosive gases, like methane, have to be ventilated before manned operations can begin. Moreover, tunnels are characterized by the presence of metal objects, such as rooftop cables, ventilation fans and other mining equipment.

Given the complex nature of the environment, underground wireless propagation has been studied thoroughly in the past.

2.3. General challenges of radio propagation in mines

The goal of this section is to give an overview of the wireless propagation phenomena that communication systems operating at ultra-high frequencies (UHF) and super-high frequencies (SHF) can experience. At the centre frequency of 802.11p (5.9 GHz), the wavelength is approximately five centimetres, while for traditional Wi-Fi 12cm.

The physical environment in which wireless communication takes place greatly affects radio link performance. During signal propagation, the electromagnetic waves can be affected by several physical phenomena:

• Reflections occur when a radio wave hits a smooth surface with dimensions much greater than its wavelength; These are caused by the metal body of the vehicle and other mining equipment present in the mine.

• Diffraction is caused by an obstacle with size greater than the wavelength between transmitter and receiver.

• Scattering is caused by obstacles of size smaller than the wavelength in the path of the signal;

(35)

2.3. General challenges of radio propagation in mines

2.3.1. Line-of-Sight requirements

Communication systems operating at UHF and SHF require a LOS condition to meet their design specifications. The LOS requirement is not limited to a virtual line connecting TX and RX, but it consists in a cylindrical ellipse beam wrapping the radio-link, as seen in figure 2.3.

Figure 2.3.: Shape of the first Fresnel zone.

Fresnel zones are important in assessing the differences in path length and phase shifts experienced by objects placed between TX and RX.

The radius of the first Fresnel zone is an important indicator of the LOS condition of a wireless link. F1 = s λd1d2 d1+ d2 (2.1)

In equation 2.1, d1 and d2 represent the distances, in meters, between radio

equip-ment and the obstacle and λ is the wavelength of the communication systems considered.

The widest radius of the first Fresnel zone, calculated at the middle between TX and RX, assuming d₁= d₂, is found by using the following equation:

r =

s

dλ

2 (2.2)

In equation 2.2, d represents the distance expressed in kilometers between TX and RX and λ is the wavelength.

Beyond 20% blockage of the first Fresnel zone, signal loss will become significant and the free space approximation can no longer be used [12]. The radio waves at the

(36)

Figure 2.4.: Dimension of first Fresnel zone at its widest point for 2.4GHz (left) and 5.9GHz (right)

Mineral Bench Height [m] Bench Width [m] Bench Slope [◦]

Copper 12 - 18 24 - 38 50 - 60

Iron 9 - 14 18 - 30 60 - 70

Non - metallic 12 - 30 18 - 45 50 - 60

Coal 15 - 23 15 - 30 60 - 70

Table 2.2.: Typical dimensions of the cross shape of open-pit mines, from [1]

RX include also reflected and diffracted components which impact negatively the performance of the system.

2.4. State-of-the-art of wireless tunnel propagation

Telecommunication in tunnels has been the focus of numerous research during the past years. Communication in underground mines facilitates the normal work flow and, in case of emergencies, it enhances rescue operations.

Earliest research dates back to 1978, where a team of researchers measured the attenuation constant of radio waves in tunnel. A high pass channel behaviour, similar of the one shown by circular waveguides, was discovered [13]. Since then, researchers worldwide set to study this challenging environment with different technologies operating at different frequencies.

Several methodologies can be used to model wireless propagation in tunnels, as detailed in [14]. These methods are either based on mode analysis, channel models or on optical techniques.

2.4.1. Channel models

(37)

2.4. State-of-the-art of wireless tunnel propagation

Deterministic channel models are characterized by a closed form mathematical expression and a set of parameters depending on the specific environment. Numerical methods include Finite Difference Time Domain (FDTD), Method of Moments (MoM), Finite Element Method (FEM) and the use of a Vector Parabolic Equation (VPE). Tunnels can also be modelled deterministically as a waveguide by mode analysis. This approach is appropriate for tunnels without objects in the cross section and when it is of interest to analyse higher modes of propagation.

Empirical channel models, instead are based on measurements of the received signal strength (RSS) and subsequent curve fitting. Their accuracy depends entirely on the similarity between the environment in which the model was developed and the application where said model is used. Two-slope models are a typical example of an empirical method to calculate path loss of a LOS link. They include a linear zone, called the near region, where free space propagation dominates and a far region where the dominance of constructive interference approximates well to a waveguide model.

Moving vehicles in tunnels could be addressed by statistical channel models that try to predict attenuation and signal fluctuations by the use of statistical means. Such methods, for instance, compute probability density functions of different fading phenomena.

2.4.2. Optical methods

In optical methods, the radio wave is approximated by a ray. Thanks to optical theory, the different paths followed by the signal can be predicted. These methods are based on the assumption that the plain wave generated in far field partially penetrates the tunnel wall and it is partially reflected back.

Assuming the angle of incidence similar to the angle of reflection, geometrical optics (GO) method can be used to approximate pathloss as a superposition of the different signal copies coming from different paths. If obstacles are present, the model can be extended to consider diffraction by using the Huygens’s principle [15].

Optical methods include techniques such as shooting and bouncing rays (SBR), image method, ray density normalization and geometrical theory of diffraction (GTD).

Given the characteristics of optical methods, they are suitable for tunnels with constant cross section and smooth walls of a known material. The close match requirement with reality is a limitation in my investigation, as mining tunnels present different characteristics depending on the type of material and type of mining technique employed. However, SBR provides an intuitive graphical representation of

(38)

radio propagation which allows to verify the LOS condition among TX and RX for different antenna positions.

2.4.3. Relevant works in past literature

Theoretical modelling of electromagnetic propagation in tunnels has been studied extensively in the literature. In this section, a selection of past works utilizing different methods is listed.

In [16] a study based on mode propagation has been carried out. Propagation parameters have been obtained for lower modes and for different cross section shapes. A comparison with experimental measurements show good agreement with the closed form expression derived. At very high frequencies (VHF), however, tunnel curvature and wall roughness worsen the predicted attenuation of signals propagating at higher modes.

The effect of curvature has been further studied in [17] for UHF by using a waveguide approximation and GO. The model proposed was computationally light compared to purely GO methods and it proved good accuracy with subsequent measurements both for straight and curved tunnels.

In [18], a simple rectangular waveguide approximation is used and the results suggest the usage of frequencies greater than 2GHz and a base-station switch network structure.

A railway tunnel was modelled as an oversized rectangular periodically loaded waveguide in [19]. The surface impedance method was used in order to verify which modes can propagate. The presence of trains has been proven to affect field distribution and an attempt to study the influence of passengers on the train has been made.

Along similar lines, the influence of vehicles in tunnels occurs in two ways and it is computed in close form by using the uniform theory of diffraction in [20]. The in-mode

loss is caused by the plane waves hitting the front surface of the vehicle and being

reflected back. Mode coupling loss, instead, is due to the diffraction that happens when the plane wave hits the edge between two surfaces of the vehicle.

Sun and Akyildiz conducted a detailed analysis of underground wireless propagation

environments in [21]. They extended their research to consider room and pillar areas in underground mines. The models obtained with the multi-mode method have shown good correlation with measurements carried out later on in the study. In the near region, intense signal attenuations occur due to the presence of several active modes. In the far region, the received signal power is more predictable as the lowest order mode is dominating. The dominating mode depends on the tunnel size and the carrier frequency used. Power distribution between different modes depends on the position of the transmitter and receiver antenna.

(39)

2.4. State-of-the-art of wireless tunnel propagation

of polarization is dependent on the physical size of the tunnel, however it has been shown that for low and wide tunnels, horizontal polarization is suggested while for high and narrow tunnels vertical polarization performs better. Another important conclusion is the very little influence of air parameters, such as humidity, pressure, temperature. This is justified by the fact that these factors have no effect on air permittivity but affect its conductivity.

The electrical parameters of the tunnel walls can also be considered constant as they would not affect significantly the calculation. This claim is further supported by [22]. After examining the electrical characteristics of tunnels,the authors argue that the influence of conductivity on the characterization of tunnels can be neglected in most cases.

In the iron mine of Kiruna in the North of Sweden, measurements of radio interference, path loss and multipath propagation were carried out [23]. The signal present in the mine are GSM at 900MHz, the underground train communication operating at 459.8MHz and a strong interference, caused by passing trains, was observed at 2.5GHz. Measurements carried out show the rms delay spread of the received signals is low (100ns) even for LOS conditions indicating that MIMO techniques can not be fully exploited. A strong attenuation is observed between 2-2-5GHz caused by ground plane reflections of the two ray propagation model.

More measurements were conducted in a straight, two way arched tunnel with the goal of modelling the radio channel for frequencies between 2.8GHz and 5GHz [24]. The measurements conducted lead to an accurate analytical model which can be used to model MIMO channels in such conditions.

Several studies utilized PE to calculate the electric and magnetic field in tunnels. This method is computationally efficient and it can be used both for straight and curved tunnels, assuming that electromagnetic fields travel within 15◦ degrees of the axis of propagation. PE can be used to model the effect of lossy rough walls [25] [26] or to obtain deeper knowledge about propagation in tunnels with less computational needs [27] [28]. However, the mathematical complexity is high and the results produced are too detailed for this initial phase of the investigation at Scania.

An existing study using the ray tracing method in a rectangular tunnel highlights the importance of having a large height difference between transmitting and receiving antenna to increase the range of propagation [29].

Ray tracing was used also in [30], to model tunnels with varying cross section, typical of mine environments. Concave and convex profiles have been considered with the assumption that the cross section changes gradually along the tunnel axis. The proposed solution has been validated successfully with a measurement campaign. Another application of this optical method is found in [31], where ray tracing was used to understand how the antenna radiation pattern and its position affect the radiation inside tunnels. Results indicate how the field distribution is sensitive to the positioning of the antenna. As a consequence, the choice of the optimal antenna depends on

(40)

the desired position as a specific radiation pattern could reduce the standing waves compared to an omni-directional antenna for certain placements.

The effect of antenna position have been further researched in [32]. Placing a transmitter close to the walls of the tunnel results in more complex propagation with higher attenuation. The middle of the tunnel has been identified as the most suitable position to place a transmitting antenna.

2.4.3.1. Frequencies considered in the literature

A wide range of frequencies has been studied over the years. The first studies concentrated on the unlicensed spectrum around 900MHz, suitable for long range point-to-point LOS communications [27] [33] [34].

The Industrial, Scientific, and Medical ISM band at 2.4GHz was also of interest for underground wireless propagation in mesh and sensor networks.

Several measurement campaigns with the goal of modelling the underground radio channel have been conducted at ISM and ITS frequencies. It is the case of [35], [36] and [37].

In recent years, the interest is shifting towards millimetre wave, as shown in [38] and [39] where exploratory radio measurements at 18GHz were carried out. Following measurements at 60GHz [40], the capacity of the mm-Wave underground channel has been obtained in [41] for the abandoned gold mine in Val d’Or, Quebec in Canada. In the same location, a performance evaluation of a LOS link operating at 60GHz with different antenna configurations was investigated in [42] and a MIMO system for short range communication was characterized [43].

2.5. Open-pit mine propagation

Due to the vast geographical area covered and their changing shape, radio planning in open-pit mines is not a static process but a dynamic one. Deployed infrastructure should be self organizing and able to adapt to changing surroundings. Several tech-nologies, summarised in figure 2.6, have been adopted depending on the applications connectivity requirements.

Cellular networks provide a wide area coverage, however they operate on licensed spectrum in a centralized fashion and their deployment lacks flexibility and requires expensive equipment.

(41)

2.6. Existing networking infrastructure in mines

Furthermore, as pointed out in [45], they can be sectorized with the goal of increasing energy efficiency by shutting off unused areas, which for battery powered mobile nodes is a real benefit.

Wireless links in open-pit mines can be modelled by using the ground reflection channel model. This approximation considers two contributions to the received signal, namely the LOS component and the ground reflection. The obstruction of the first Fresnel zone becomes particularly relevant in this environment in the case of communication between two different levels of the mine. The LOS could be blocked by the cliff edges if the transmitter is not placed on the road edge, as shown in picture 2.5

Figure 2.5.: Cross section of an open-pit mine, whose typical dimensions are shown in table 2.1. When the truck, represented by the red box, and the infrastructure are placed on different levels of the mine, there might be coverage of the first Fresnel zone, thus resulting in NLOS conditions. [1]

2.6. Existing networking infrastructure in mines

Connectivity is helping the mining industry transition towards remote controlled and automated machinery. Communication systems are already present in mines because they provide several advantages both during normal operations and in emergency situations.

Open-pit mines have either proprietary cellular networks deployed, such as WiMax, or wireless mesh networks. In underground tunnels, WiFi access points are densely de-ployed, typically every 50-100m, thus making it the dominant technology.

(42)

Figure 2.6.: Open-pit mines wireless communication systems c 2013 IEEE

(43)

2.7. Choice of method

Given the diffusion of mining networks in the ISM band at 2.4GHz, my master thesis research is extended, in addition 5.9GHz systems, to that frequency in order to provide a solution for currently deployed networks as well as future coverage at ITS frequencies.

2.7. Choice of method

For my link-budget simulations, closed form channel models are the most appropriate method to use. Therefore, to model path loss in underground mines, an empirical channel model is used. While the other methods discussed earlier are very suitable for modelling wireless propagation in tunnels, they produce results too detailed and for several propagation modes. For these reasons and their inherent complexity, waveguide modelling and parabolic equation methods are not used. For open-pit mines, the two ray propagation model, a two slope and free space channel model with modified path-loss exponent are used.

Despite the high site specificity of optical methods, SBR in a simplified mine tun-nel model is simulated. This ray tracing approach identifies the behaviour of the propagating rays and the effect of objects present in the cross section of the tun-nel. Moreover, SBR produces intuitive graphical results tha could help identify best antenna characteristics, such as high directivity, for placement in front of the truck.

(44)

(45)

3

Chapter 3.

Multi-antennas system

Multi-antenna systems have several antennas both at the TX and the RX, which enable the use of different techniques benefiting vehicular communication. MIMO can be exploited is several ways:

• Diversity gain consists in receiving independent copies of the same signal on different antennas.

• Multiplexing gain consists in transmitting indipendent data streams in parallel channels, so throughput is increased.

• Smart antennas are antenna arrays capable of beamforming and dynamic configuration of their radiation pattern.

Figure 3.1.: Channel paths in a 2x2 MIMO configuration

In vehicular communication, robustness and reliability of the wireless link is of primary importance therefore diversity techniques are the main focus of this sec-tion.

(46)

copies which travelled on different paths and thus experienced uncorrelated fading. The diversity receiver can then recombine the signal from the multiple copies. Several linear combining strategies are reported in the literature:

• Selective combining (SC) consists in choosing the signal with the highest SNR.

• Equal gain combining consists in adding coherently, with respect to their received phase, the received copies of the signal.

• Maximum ratio combining adds coherently and with regards to the received SNR the multiple copies received.

In order to exploit diversity, space-time block coding is required at the source in order to transmit multiple copies of a data stream over several antennas.

De-correlation between the radiation patterns of the multiple antennas is required for the signals to experience independent fading and thus benefit from a diversity gain.

3.1. Mutual coupling

Mutual coupling is the interaction between the antennas of an array. When two closely spaced antennas radiate, they interact with each other, causing changes to the overall radiation pattern and to their input characteristics.

This unwanted interaction affects the radiation efficiency of the configuration: as shown in figure 3.2, the radiation of the element m translates into another current at the antenna n, whose effect is to un-match the impedance and cause more reflections. Mutual coupling, together with the envelope correlation coefficient (ECC) are important indicators used to quantify the performance advantages of a MIMO solution, especially antenna arrays.

(47)

3.2. Scattering parameters

The effects of mutual coupling can be minimised by increasing the spacing between the antennas, changing polarization or the orientation of the radiation patterns of the antennas.

3.2. Scattering parameters

At microwave frequencies, dealing with voltages and currents is troublesome as they behave like a travelling wave. Moreover, modelling the interaction between components in complex radio frequency network is quite complex. The use of S-parameters abstracts any type of network and allows it to be characterized just by measuring signals at its ports, where a port is defined as any point where a signal can be applied to the the system.

While S-parameters are complex quantities indicating attenuation and phase shifts between the ports considered, a simplification considering only the attenuation magni-tude is made in quantifying mutual coupling between matched antennas.

For an N port network, S-parameters are in the form of a N ∗ N matrix. Typical network components have two ports and the formulation of the related S-parameters can be computed as follow, where SN,M means the response at port N due to a

signal applied at port M:

Figure 3.3.: Representation of scattering parameter for a two port network

" b1 b2 # = " S11 S12 S21 S22 # ∗ " a1 a2 # (3.1)

In the scattering matrix, S₁₁ and S₂₂ represent the reflection coefficient of the antenna while S12 and S21 are transmission coefficients. A graphical representation

of reflection coefficient is the Smith chart, which displays the input impedance of the port.

The ECC, is an indicator of the correlation between two antennas’ radiation patterns which can be calculated either from the analytical expression of far-field patterns or more efficiently from S-parameters, as shown in 3.2:

(48)

ρe = |S∗ 11S12+ S21∗ S22|2 [1 − (|S11|2+ |S21|2)][1 − (|S22|2+ |S12|2)] (3.2)

3.3. Diversity techniques

Diversity schemes provide different copies of the transmitted signal at the branches of the receiver. In order for diversity techniques to work, the received copies must be decorrelated, which means they have to experience independent fading.

To achieve the condition of independent fading, many techniques are available. Antenna diversity methods use diversified antennas with different radiation pat-tern, polarization or orientation. Frequency diversity consists in transmitting at different frequencies the same copy of the signal. With time diversity, the same signal is sent repeatedly over different time instances. Spacial diversity, instead, is achieved by spacing physically the antennas at either the transmitter or the receiver.

The focus of this work is receive vertical diversity, where two antennas are mounted at the front of the truck at different heights. This configuration, allows the receiver to mitigate the deep fades in RSS typical of the two ray propagation model.

3.4. Diversity gain

The impact of the correlation between two Rayleigh signals was quantified for the first time in 1956 [46]. Through statistical analysis of the received signal level failing below a threshold, the curves of figure 3.7 are found. The x-axis represent the correlation factor among two Rayleigh fading variables, while the y-axis the percentage of time each signal is below the threshold level. Each curve represents the same level of performance in terms of percentage of time where the received signals are below the desired threshold. For example, the 5% curve refers to received signals falling under the desired threshold for 5% of the time. In this case, communication would be lost (0.22%) of the time for uncorrelated signals (k = 0), and approximately 4% for perfectly correlated signals (k = 1).

In another work focusing on dual receive diversity with SC, graphs in the form of dual diversity gain for different correlation coefficients are displayed in [47] for different SNR ratios among the branches.

(49)

3.4. Diversity gain

level below 0.95 each outage rate below 1% will manifest a constant diversity gain of 4.6dB compared to the uncorrelated fading (ρ = 0). The same reasoning holds for figure 3.6, where the stronger branch has a double mean SNR compared to the weaker one. In figure 3.6, instead, the branches differ by a power factor of 10 and diversity gain is constant for correlations below 0.8.

The authors report that the effect of correlation on dual diversity is the reduction of the effective mean SNR on the branches, by the following relationship:

Γkef f = Γk

q

1 − |ρ2_| _(3.3)

From equation 3.4 it is possible to compute that ρ has to be over 0.85 in order to experience a reduction of a factor of two (3dB) in SNR levels.

To compare this result, diversity gain for two antennas at the receiver is estimated to be between 10 and 20dB from figure 3.8, also derived in [47].

Figure 3.4.: Effect of correlation (ρ) on dual di-versity with SC for two branches with equal mean SNR (Γ1= Γ2).

Figure 3.5.: Effect of correlation (ρ) on dual diversity with SC for two branches with double mean SNR (Γ1= 2Γ2).

(50)

Figure 3.6.: Effect of correlation (ρ) on dual di-versity with SC for two branches with highly different mean SNR (Γ1 =

10Γ2).

Figure 3.7.: Effect of correlation on commu-nication uptime

(51)

4

Chapter 4.

ITS application in mine

environments

The focus on this chapter is to extend the background research by investigating the state-of-the-art of ITS applications. The performance of 802.11p has been researched in the literature, with attention on multi-antenna techniques and tunnel propagation. Furthermore, the most relevant ITS research projects have been reported.

Afterwards, the importance of LOS is discussed and the unique challenges of ITS applications in mines are detailed.

The chapter ends with a list of antennas available during the investigation, both in terms of computational models and physical prototypes.

4.1. Wireless Access for Vehicular Environments

In 2004, the standardization body of the IEEE 802.11 specifications set to work on the definition of a standard for vehicular communication. This followed the Federal Communication Commission FCC standardization of the Dedicated Short Range Communication DSRC in 1999.

DSRC was allocated seven channels, centred around 5.9GHz, for applications related to vehicular communication, such as cooperative adaptive cruise control, collision avoidance and electronic toll collection.

The IEEE 802.11 is a family of protocols that specifies the physical (PHY) and medium access control (MAC) layers of wireless systems operating in the industrial, scientific and medical ISM band. The 802.11p version, also known as WAVE, is an adaptation of the previous 802.11a specification to suit the characteristics of ITS application.

(52)

controls the transport of packets from source to destination. The destination can be a single node or all nodes present in a geographical area. As a consequence, safety critical and traffic planning messages can be broadcasted to the specific affected location.

In order to adapt 802.11 to vehicular channels, several changes at the MAC layer have been devised in order to simplify and reduce the overhead in medium access procedures [48]. The major modification has been the introduction of “WAVE mode” to allow data exchange between nodes without requirements on previous basic service set BSS associations. WAVE mode is the process that allows fast connection setup between the mobile node and the WAVE basic service set (WBSS). 802.11p has the possibility to use a wildcard BSSID, where nodes already connected to a WBSS can transmit to other nodes, in a peer to peer fashion, safety critical messages.

At PHY, instead the concept was to minimize changes from the 802.11a version operating at 5.0GHz. Both these 802.11 versions, whose differences are summarized in table 4.1, utilise Orthogonal Frequency Division Multiplexing OFDM, however channels are 10MHz wide instead of 20MHz in the p amendment. This modification has the effect of halving the maximum data rate, and consequently the carrier spacing, as the OFDM time parameters are doubled. This modification renders the physical layer more robust against the higher delay spread present in vehicular environments, where the normal 802.11a physical layer configuration could otherwise suffer from inter-symbol interference in worst cases.

The main challenges at PHY are related to providing robust connectivity under tight latency requirements over multipath channels. The vehicular channel can change rapidly over a symbol period as it is affected by high Doppler shifts and fast Rayleigh fading[49].

The improvements of WAVE have been quantified in terms of BER performance over fading channels in [50] where it is found that 802.11p shows higher performance than 802.11a for channels with a high delay spread.

fc BWOF DM rbmax

802.11a 5.0 GHz 20MHz 54 Mbps

802.11p 5.9 GHz 10MHz 27 Mbps

Table 4.1.: Main differences between 802.11a and 802.11p

4.2. Existing research on 802.11p

(53)

4.2. Existing research on 802.11p

environment, traffic and density of road intersections.

Compared to cellular networks, TX and RX are positioned at the same height, as in V2V, or with a very low height difference, such as in V2I. The metal body of the vehicles causes shadowing and reflections, while smaller obstacles, such as lamp posts and road signs scatter the radio signal. The radio channel in these conditions is multipath, where the main path varies according to traffic conditions and environmental factors, such as the geometry of surrounding buildings [51]. Fading models for buildings at traffic junctions, based on real world measurements, are already existing in the literature [52] [53].

These environmental factors of the vehicular channel make it highly location depen-dant. As a consequence, empirical channel models derived from measurements are the most prominent in the literature. A channel model for V2I communication has been refined using parameters generated from measurements in [54].

A typical characteristic of the vehicular channel is the high relative speed between moving vehicles. Attention to the effects of high Doppler spread was given in [55] and in [56], where receive diversity with two antennas is found to improve packet error rate.

Overall performance, in terms of packet error rate, delay and throughput, of the 802.11p protocol has been evaluated in [57] and [58].

A physical layer evaluation showed differences in the SNR required based on the antenna height of the RSU. A recommendation from [59] is to place antennas higher than the height of vehicles to enhance the reliability of the link. Furthermore, in the same paper, the influence of packet length on BER has been highlighted. Bigger packets require a longer transmission time, thus increasing the effects of the multipath channel and degrading performance.

4.2.1. Multiple Input Multiple Output

The physical layer of 802.11p currently does not support multiple antennas, never-theless this technology has been considered and proved appropriate for ITS applica-tions.

A MIMO extension of the 802.11 PHY was developed and benchmarked in [60]. It is shown that multi-antenna techniques yield considerable performance advantages over SISO at the cost of higher transceiver complexity. Similar conclusions based on BER performance are reached in [61].

Antenna Selection Diversity is a technique that limits the complexity of MIMO transceivers. Out of the antennas present, only the ones with the highest SNR are connected to the available electronic radio frequency (RF) chains.

In [62], BER for MIMO, MISO and SISO systems have been computed. The result

(54)

has indicated how MISO could be a sensible approach to investigate diversity in vehicular communication; this solution mitigates the tradeoff between performance advantages obtained by full MIMO and the lack of signalling between TX and RX typical of SISO systems.

Receive diversity was also considered an ideal trade off in [63]. Several antennas configuration have been tested under seven propagation environments. MIMO 4x4 and SIMO 1x4 yielded BER performance improvements in all scenarios over SISO, as shown in table 4.2. However, the theoretical advantage of the former solution over the latter is not manifested in reality. The MIMO 4x4 configuration can not exploit fully diversity due to poor channel estimation techniques and the degrading effects of varying Rician fading in the vehicular channel, therefore SIMO 1x4 is the best performing solution.

Channel SISO SIMO

1x2 SIMO 1x4 MIMO 2x2 MIMO 4x4 VTV-expressway oncoming 13.17 10.05 5.81 10.68 11.91 RTV-urban canyon 8.78 6.44 3.05 6.83 4.26 RTV-expressway 14.71 10.30 5.75 10.65 8.62

VTV_urban canyon oncoming 9.16 6.60 3.23 6.95 5.08

RTV-suburban street 11.08 7.65 4.12 7.84 6.10

VTV-express sam dir. with wall 11.80 6.32 3.52 6.30 6.79

Table 4.2.: SNR required for achieving a frame-error-rate (FER) of 10% in different environments under different multi-antenna configurations. c 2012 Springer

The potential of RX diversity has been further validated in [64] for different combining techniques over a SISO configuration. The influence of packet size is also mentioned as larger maximum transmission unit (MTU) lead to higher error rates.

4.2.2. Tunnel propagation

Wireless propagation in tunnels has been researched focused both on traditional WiFi and 802.11p.

Wi-Fi based V2V has been evaluated in a tunnel in Vuokatti, Finland. The tunnel where this study was carried out has a perfect hemisphere shape with smooth walls. Its higher point is 18m tall and the tunnel extends for 1.2km, with a bend in the middle. Maximum communication ranges were found to be 150m for a single data stream and 100m for two data stream [65]. More importantly, no major performance degradation was witnessed when switching from LOS to NLOS conditions.

(55)

4.3. Previous ITS research projects

of WPA-PSK and OpenHIP is 24% at 300m. Throughput performances are therefore acceptable even with the security overhead, up to a distance of 300m in NLOS conditions.

Extensive measurements of V2I in tunnel environments have been conducted in [67] in order to assess frame success ratio of the connection. Different traffic densities have been found to impact the reception of data, especially inside the tunnel. The worst case is when a large vehicle, such as a truck, precedes a car: in these conditions the signal is shadowed. Moreover, to improve goodput the use of higher modulation coding scheme (MCS) is suggested over increasing the MTU.

The time varying parameters of V2V channel in tunnel have been studied in [68]. The main paths are caused by the reflections on the tunnel walls and the ventilation system. Random scattering due to vehicles was also observed. Moreover, it is observed that Doppler spreads and stationarity time follow a log normal distribu-tion.

4.2.3. Previous simulation studies

Given the several factors influencing wireless communication in urban environ-ments and the need for accurate performance estimation under a high density of nodes, network simulations are highly relevant in the assessment of 802.11p net-works.

A propagation model, accounting both for LOS, obstructed LOS and NLOS conditions was implemented in [69]. Subsequent comparisons with real world measurements and simulations in OMNet++ highlight the importance of mapping accurately the buildings surrounding the signal path. Inaccurate building modeling, and other small obstacles obstructing the LOS can influence significantly the quality of the received signal.

Another advantage of simulations is their scalability. Communication range, BER and SINR have been estimated from large scale simulations utilizing a Nakagami-m channel model in [70]. The critical communication range, for a transmission power of 23dBm and a received power threshold of -89dBm, is found to be around 400 metres.

The relationship between throughput and density of RSU has been simulated in [71], following previous radio measurements at the same location.

4.3. Previous ITS research projects

ITS has already been the focus of several European research projects, which are briefly described in this section.

(56)

Safe Road Trains for the Environment (SARTRE) [72] investigated platooning applications on public highways, with the goal to increase fuel efficiency, comfort and safety on public roads. A truck platoon, pictured in figure 4.1 consists in a set of vehicles travelling at a fixed small distance to each other. These vehicles are equipped with state-of-the-art driving support systems which enable them to behave in unison. This transport configuration reduces aerodynamic drag which in turn can decrease fuel consumption and lead to a lower transportation cost.

Figure 4.1.: Representation of truck platooning c www.eutruckplatooning.com

Advanced Safety and driver Support for Essential road Transport (ASSET) [73] pro-poses several applications with road side infrastructure directed at vehicle tracking and traffic control. These concepts included thermal vehicle imaging and vehicle identification, tracking and tracing.

The CityMobil project [74], aimed at improving the organization of urban transport, showcased autonomous vehicles in several European cities. The goal of CityMo-bil was to identify, test and evaluate new moCityMo-bility solutions for passenger trans-port.

In the kingdom of Sweden, several initiatives are particularly close to the topic of this master thesis: iQMatic is a research project being carried out by Swedish universities in collaboration with industries in the transportation sector. The scope of the project is the development of a fully autonomous heavy vehicle for goods transportation and other industrial application. The iQMatic test truck is remotely programmed and it is then able to autonomously navigate to the destination, avoinding obstacles on its way and making independent decisions on the path to follow [75].

Enabling Technologies for Transport Efficiencies (ETTE) set to research and develop key technologies enabling vehicular communication. Within ETTE, a vehicular traffic router and several antennas were designed for ITS application by SP [76].

Reliable communication in mine environments for autonomous vehicles

Reliable communication in mine

environments for autonomous

vehicles

ALESSANDRO TOMASI

KTH Royal Institute of Technology

School of Information and Communication

Technology

Dept. of Communication systems

Wireless@KTH

Acknowledgment

Contents

List of Figures

List of Tables

Glossary

1

Introduction

1.1. Mining

1.2. Intelligent Transport System

1.3. Problem statement

1.4. Research scope

1.5. Research questions

1.6. Research method

1.7. Thesis structure

2

Background

2.1. Autonomous mining vehicle

2.2. Environments typical of mines

2.3. General challenges of radio propagation in mines

2.4. State-of-the-art of wireless tunnel propagation

2.5. Open-pit mine propagation

2.6. Existing networking infrastructure in mines

2.7. Choice of method

3

Multi-antennas system

3.1. Mutual coupling

3.2. Scattering parameters

3.3. Diversity techniques

3.4. Diversity gain

4

ITS application in mine

environments

4.1. Wireless Access for Vehicular Environments

4.2. Existing research on 802.11p

4.3. Previous ITS research projects