Localization Techniques in Embedded

Sensor Networks

THE UNIVERSITY OF
NEW SOUTH WALES

Prasant Kumar Misra

Student ID: z3229880

A thesis research proposal submitted in partial fulfillment of the

requirements for the degree of
Doctor of Philosophy
School of Computer Science & Engineering
University of New South Wales

June 15, 2009

Supervisor : Prof. Sanjay Jha

Network Research Lab (NRL)
CSE, UNSW, Sydney, Australia.

Co-supervisor : Diet Ostry

CSIRO, Sydney, Australia.
 Abstract

Wireless sensor network (WSN) provide the potential for the extension of the wireless network
in areas that could have never been imagined. One such potential sensor application research
is Underground Mine Location Sensing System based on Wireless Sensor Networks. Under-
ground mines are extensive labyrinths. The mine tunnels are long and narrow. They are usually
a few kilometers in length but only a few meters in width. They employ hundreds of mining
personnel working at any point of time under extreme conditions. It is a recognizable fact that
the underground mining operations involve a high level of risk. Hundreds of miners die from
mining accidents every year. To ensure the safety of workers and perform co-ordination of
tasks, a communication and location sensing system is one of the more important infrastruc-
tures that needs to be deployed. This system is expected to deliver satisfactory performance in
terms of communicating and performing location estimation in routine and rescue operations.
The current state-of-the art communication technology used in underground mines are chal-
lenged by various limitations. Communication and tracking systems in underground mines is
an area that has not been actively researched as contemporary surface based systems. There are
few existing system and there is limited information regarding the actual implementation of the
tracking systems. This implies that location sensing is an unsolved problem in mines and there
is a need for technological improvement in order to cater to the challenging underground min-
ing conditions. This research seeks to design and develop a sophisticated underground mine
location sensing system using wireless sensor networks for tracking personnel and machinery
inside the mine that is able to mitigate the limitations faced by present systems. The lack of
dependence on fixed infrastructure and robust operation will aid in locating trapped personnel
in the event of an emergency. The proposed methodology is based on designing a new proto-
type sensor mote that would be able to withstand the harsh mining environment. Implementing
the solution at the low level would not only provide the flexibility of better signal control and
modulation but would also empower the system with better signal processing algorithms for
effective noise filtering. The existing solutions have not been able to cater to this challenge
because of weak solutions implemented at the lower protocol stack levels and then performing
extreme customization at the higher levels which are not worth the investment. The solution
will be tested and evaluated in a real world underground mine. The research problem provides
an opportunity for cross-disciplinary research in computer science and mining engineering.
Contents

1 Introduction 6
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Aim of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 Research Proposal Organization . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 Underground Mines : An Insight 16

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.1 Underground Channel Properties . . . . . . . . . . . . . . . . . . . . . 19
2.2.2 Underground Communication Techniques . . . . . . . . . . . . . . . . 20
2.2.2.1 Through-the-Wire (TTW) . . . . . . . . . . . . . . . . . . . 20
2.2.2.2 Through-the-Air (TTA) . . . . . . . . . . . . . . . . . . . . 22
2.2.2.3 Through-the-Earth (TTE) . . . . . . . . . . . . . . . . . . . 23
2.2.3 Existing Underground Communication Systems . . . . . . . . . . . . . 23
2.3 A Survey of Underground Mine Tracking Systems . . . . . . . . . . . . . . . . 25
2.3.1 The Location and Monitoring System for Personal Safety (LAMPS) . . 28
2.3.2 The Nexsys Real-time risk management system for underground mines 28
2.3.3 The Emergency two-way communications for underground miners . . . 28
2.3.4 Wireless location technologies for tracking and data collection . . . . . 29
2.3.5 The Trapped Miner Locating System . . . . . . . . . . . . . . . . . . 29
2.3.6 The Electromagnetic Location System for Trapped miners . . . . . . . 29
2.3.7 The Personal Emergency Device (PED) . . . . . . . . . . . . . . . . . 29

1
 2.3.8 The TRACKER Tagging . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.3.9 The TeleMag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.3.10 The Delta Electromagnetic (DeltaEM) Gradiometer Beacon Tracking

System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

2.3.11 The RadCAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.3.12 The Subterranean Wireless Communication System (SWECS) . . . . . 31

2.3.13 The Digital Radio for Underground Miners (DRUM) . . . . . . . . . . 32

2.3.14 The Canary 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.3.15 The TramGuard Proximity Warning System . . . . . . . . . . . . . . . 32

2.3.16 The TramGuard Miner Track System . . . . . . . . . . . . . . . . . . 32

2.3.17 The BreadCrumb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.3.18 The AXON Transceivers . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.3.19 The Mine Net Tracking System . . . . . . . . . . . . . . . . . . . . . 33

2.3.20 The BeckerTag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.3.21 The RFID: Multi-read Radio Frequency Identification and Tracking . . 34

2.3.22 The Wi-Fi based Real-time Tracking . . . . . . . . . . . . . . . . . . . 34

2.3.23 The Mine-NET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.3.24 The Mine Rescue System . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.3.25 The IPMine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

2.3.26 The MineAX Tracking and Tagging . . . . . . . . . . . . . . . . . . . 35

2.3.27 The Smart Tag Resource Tracking System . . . . . . . . . . . . . . . . 35

2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3 Methodology 37

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.2 Proposed Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.3 Approach and Justification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3.4 Proposed system overview and architecture . . . . . . . . . . . . . . . . . . . 42

3.4.1 General Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . 42

2
 3.4.2 Wireless sensor network design challenges . . . . . . . . . . . . . . . 43
3.4.3 System Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4.4 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4.5 System Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4.5.1 Hardware Platform . . . . . . . . . . . . . . . . . . . . . . 45
3.4.5.2 Software Platform . . . . . . . . . . . . . . . . . . . . . . . 47
3.4.6 Network Topology and Protocols . . . . . . . . . . . . . . . . . . . . 48
3.4.7 On Board Data Processing . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.8 Localization Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4.9 Auto-Localization of the anchor nodes . . . . . . . . . . . . . . . . . . 49
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

4 Research Plan 52
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2 Research program timetable . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Tentative Thesis Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

5 Preliminary Work 55
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2 Concurrent Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2.1 Improving the accuracy and distance ranging of ultrasound-based Cricket
localization systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2.2 Chirp generation using the US sensors . . . . . . . . . . . . . . . . . . 59
5.2.2.1 General waveform generation technique in ATmega 128 mi-
crocontroller . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2.2.2 Procedure for the generation of the chirp signal . . . . . . . 61
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6 Future Work and Conclusion 63

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3
List of Tables

2.1 Frequency Band Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.2 Underground Mine Tracking Systems . . . . . . . . . . . . . . . . . . . . . . 27

3.1 System Core Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.2 Radio Physical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.1 Research program timetable . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.1 Selected frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

4
List of Figures

1.1 Continental view of mining activities [1]. . . . . . . . . . . . . . . . . . . . . 7

1.2 Surface Mining [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.3 Underground Mining [3] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.4 View of a mine tunnel which are usually long and narrow [4]. . . . . . . . . . . 9

1.5 Mine disasters: A comparsion between China and the rest of the world. . . . . 10

1.6 Explosion in a mine [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.7 Roof fall inside a mine [6]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.8 Summary of the mining problems. . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1 Walls of the mine have been strengthned with wooden grids [7]. . . . . . . . . 17

2.2 Water inside the mine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.3 Muddy conditions inside the mine [3] . . . . . . . . . . . . . . . . . . . . . . 18

2.4 Underground communication techniques. . . . . . . . . . . . . . . . . . . . . 20

2.5 Communication achieved through Leaky Feeder cables . . . . . . . . . . . . . 21

2.6 RFID Tag based communication . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.7 Node based tracking systems : A Wireless Mesh . . . . . . . . . . . . . . . . . 26

3.1 Chirp based ranging technique . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.2 Communication within each level inside the mine . . . . . . . . . . . . . . . . 47

3.3 Communication within various levels inside the mine . . . . . . . . . . . . . . 48

3.4 Hardware implementation of the sensor mote . . . . . . . . . . . . . . . . . . 49

5.1 Modified Cricket setup with the 3-US transducer array . . . . . . . . . . . . . 58

5
Chapter 1

Introduction

In recent years, the rapid development in miniaturization, low power wireless communication,
micro sensor and microprocessor hardware, small-scale energy supplies in conjunction with a
significant progress in distributed signal processing, ad-hoc networks protocols and pervasive
computing have given Wireless Sensor Networks (WSNs) new technological possibilities. As
the Internet has revolutionized people’s lives readily through the exchange of diverse forms of
information among large number of users, WSNs may, in the near future, be equally significant
by providing information on the physical phenomena of interest and ultimately being able to
detect and control these phenomenon or enable the construction of more accurate models of the
physical world.
A Wireless Sensor Network is composed of sensor nodes that are autonomous devices
with integrated sensing and wireless communication facilities. They are deployed either inside
the phenomena or very close to it. Sensor nodes are fitted with an on-board processor. In-
stead of sending the raw data to the nodes responsible for the data fusion, they use processing
abilities to locally carry out simple computation and transmit only the required and partially
processed data. The other features that make them attractive are that they have a low cost, use

6
low power, have a small size and short transmission range and that their positions need not be
predetermined.
Individually, these resource constrained devices appear to be of little value. However, de-
ploying these sensors on a large scale across an area of interest can be more effective. Placing
the sensors in hostile or inaccessible regions may allow for data collection which was previ-
ously impossible. Hence, they offer excellent remote monitoring capabilities. Such a network
can easily collect sensor data and transmit them by radio and the accuracy of the information
can be greatly improved via collaboration of a group of sensor nodes. Sensor nodes monitor
the activities of a set of objects in a sensing region, and report their observations to a client,
called a sink node, through the wireless medium. In the network, sensor nodes can share and
process their observations, as a result of which meaningful information is available at the sink
node. Users can retrieve this information to ascertain the status of the monitored region. Wire-
less sensor networks can provide the means to link the physical world to the digital world. The
flexibility offered by this technology will soon cross over into myriads of domains. In the fu-
ture, applications of wireless sensor networks will appear in areas that could never have been
imagined. One such potential sensor application research is an Underground Mine Location
Sensing System based on Wireless Sensor Networks.
The mining industry has played an important role in the development and sustenance of
civilization. It can be dated as back as the iron and bronze age to the age of industrial revolution
and is one of the major driving forces in building the infrastructure of the present information
age. Apart from Antarctica where a treaty was signed for banning mining activities [8], mining
takes place in all the continents of the world (Fig. 1.1). The global mining industry has been
dominated by countries that include USA, Canada, Australia, South Africa and Chile.

Figure 1.1: Continental view of mining activities [1].

Mining is the process of extracting the mineral as economically as possible [9]. The

7
mining activity has been broadly divided into 2 major categories:

1. Surface mining (Fig. 1.2)

2. Underground mining (Fig. 1.3)

Figure 1.2: Surface Mining [2]

Figure 1.3: Underground Mining [3]

Surface mining method is employed when the minerals are found closer to the surface
[10]. This method is suitable when the surface deposits covering the mineral is relatively less
or when the material is deemed unsuitable for tunneling. Underground mining method is used
when the mineral deposits are found deep below the surface of the earth [10].

8
1.1 Motivation

Underground mines are extensive labyrinths. The mine tunnels are long and narrow (Fig. 1.4).
They are usually a few kilometers in length but only a few meters in width. They employ
hundreds of mining personnel working at any point of time under extreme conditions. It is a
intuitive fact that the underground mining operations involve a high level of risk as compared
to most other industries. To better understand this danger, [11] presented an analogy between
mining operations and a manufacturing system. In a manufacturing system, all the processes
take place using raw materials inside a controlled factory environment. On the other hand, in
mining, the ore reserves or minable deposits are fixed in the ground and are available in variable
chunks over a long distance. Hence, the overall mining process is mobile and the mining
machines have to be shifted as the mining operation progresses consequently encountering a
constantly changing environment. The combination of the ever changing ground conditions,
together with a dynamic mining system generates a variable profile of risks.

Figure 1.4: View of a mine tunnel which are usually long and narrow [4].

Hundreds of miners die from mining accidents every year. The Beaconsfield mine col-
lapse [12] that occurred on the 25 April, 2006 in Beaconsfield, Tasmania is the latest in the
record of Australian mining disasters. It was a triggered by a seismic activity that resulted in
the rock fall at the mine and it claimed the life of a miner. The Sago Mine disaster was a coal
mine explosion on January 2, 2006, in the Sago Mine in Sago, West Virginia, USA near the
Upshur County seat of Buckhannon. The blast and ensuing aftermath trapped 13 miners for
nearly two days out of which only one of them survived [13]. China is the worst effected by
mining accidents. China currently accounts for the largest number of coal-mining fatalities,
accounting for about 80% of the world’s total mine accidents, although it produces only 35%

9
of the world’s coal [14] (Fig. 1.5). The Shandong coal mine flood [15] was an accident that oc-
cured on August 28, 2007 in Xintai, Shandong, China when heavy rains caused a river to burst a
levee creating flood in two mine shafts killing 181 miners. The Shanxi mine blast [16] occured
in a mine in Gujiao city near Taiyuan, the capital of Shanxi province of China on 21 Feburary
2009. This accident recorded a death toll of 74 miners with 114 injured people who were be-
ing treated for carbon-monoxide poisoning. The Sunjiawan mine disaster [17] took place on
February 14, 2005 in Fuxin city in the Liaoning province of China. It was the result of a gas
explosion that was triggered by an earthquake. It registered a record death of 214 miners. The
Shanxi mudslide [18] was another mine disaster that occured 8 September, 2008 when a mine
wall collapsed after torrential rains and it claimed the life of 11 miners. A fatal gas explosion
took place at the Nanshan Colliery [19] in Lingshi County, Jinzhong, Shanxi Province, China
on 13 November 2006. 24 miners were killed in this accident. The list of mining accidents and
disasters is endless. A comprehensive list of mining accidents and diasters have been compiled
in [20] while [21] gives a historical data of mine disasters in the US.

Figure 1.5: Mine disasters: A comparsion between China and the rest of the world.

A brief overview of the various problems and challenges faced in an underground mine
includes the requirement for monitoring gas concentrations [22, 23] which mainly include
methane which can lead to mine explosions when they exceed a threshold level (Fig. 1.6),
underground structure monitoring [24, 25, 26] as the unstable nature of geological construction
in mines makes an underground tunnel prone to structural changes (Fig. 1.7), determining the
location of miners [27, 28] and providing a way out of the mine in case of disaster as well
as monitoring dust concentration [29] and the risk of water irruption [30]. Consequently to
ensure the safety of workers and to perform the co-ordination of tasks, a communication and
location sensing system is one of the indespensible piece of infrastructure that needs to be de-
ployed. It would be expected to deliver satisfactory performance in terms of communicating

10
and performing estimation of location in routine and rescue operations.

Figure 1.6: Explosion in a mine [5].

In order to enhance rescue efforts in the event of a mine accident, the availability of the
following information would be extremely useful:

1. The location of trapped miners.

2. The two way communication between the trapped miners and the surface control station.

3. The environment conditions along the rescue path to the trapped miners that include
temperature and concentration of methane or other undesirable gases.

The current state-of-the art communication technology used in underground mines are
challenged by various limitations which prevent the above requirements from being satisfacto-
rily addressed. Communication and tracking systems in underground mines is an area that has
not been as actively researched as contemporary surface based systems. There are few existing
systems and there is limited information regarding the actual implementation of the tracking
systems presented in the previous section. This implies that location sensing is an unsolved
problem in mines and there is a need for technological improvement in order to cater to the
challenging underground mining conditions. Currently available tracking systems only register
when a person passes a certain location which presents a very limited application scenario.
The above observation reiterate the need for a cutting edge communication solution.

11
 Figure 1.7: Roof fall inside a mine [6].

While we have defined the problem space, it is clear that addressing all the challenges as-
sociated (Fig. 1.8) with this problem requires a scope that is beyond a PhD research thesis.
In this thesis, we will focus on designing a novel Underground Location Sensing System that
would not only provide a desirable location estimation of the miners and keep a track of their
moving locations but also be reliable and robust in delivering its routine functions in the event
of mine disasters and accidents.

Figure 1.8: Summary of the mining problems.

A portable wireless system would be the best option in mines because it offers the best
resistance to damage from roofs falls inside the mines, fires and explosions. Besides that, the
system can be carried by miners and it does not require pre-existing infrastructure in terms of
pre-installed antennas. Wireless sensor networks (WSNs) are the most plausible portable wire-
less systems. Using WSNs, the moving miners and machinery can estimate their location by

12
co-operating with nearby objects by sharing the sensor data in order to minimize the overall
location error. With a moderate upgrade of the processor used in the sensor nodes, fast and effi-
cient object tracking can be performed. In addition, it can be enhanced to monitor gas and dust
concentration inside the mines and the stability of underground structures with the addition of
a few extra sensors.
This thesis will also consider the layer upon which such a solution should be developed.
A solution developed on the network or application layer of the protocol stack would be excel-
lent for delivery of the data to the control station using optimum network capacities as well as
presenting the information in various interactive formats for the monitoring personnel to better
gauge the situation. However, the prime challenge for developing a location sensing system
for underground mines is that the communication signal should be able to sustain its signal
characteristics under the extremely hostile mining environment. We believe that the location
sensing scheme should be implemented as low in the protocol stack as possible, especially at
the physical and link layers. Designing and developing a solution at the low level would not
only provide the flexibility of better signal control and modulation but would also empower
the system with better signal processing algorithms for effective noise filtering. The existing
solutions have not been able to cater to this challenge because of weak solutions implemented
at the lower protocol stack levels and then performing extreme customization at the higher lev-
els which are not worth the investment. Therefore, the fine tuning and standardization of the
system from the lowest protocol stack level and then building better and effective solutions on
every level of the protocol stack would result in an improved location sensing system.
In addition, we believe that the proposed system should not only be functional as a sepa-
rate entity but should be able to seamlessly integrate with the existing wired or wireless com-
munication infrastructure in the mines. As the latest tracking systems being developed for the
mining industry are based on digital data networks [31], the proposed location sensing system
should have the ability to integrate itself with the conceivable solutions in the future. In order to
achieve this cross functionality, we propose to build it using light weight TCP/IP stack [32, 33].
Additionally, upon further customization and development, it can be made to operate over the
Internet as well. Though the Internet was initially designed and developed for low speed net-
works, current Internet technology is highly flexible in adopting and running over diversified
link layer technologies.
Nevertheless, implementing the solution at the lower layer for example the PHYSICAL
layer would require the solution to be changed for different communication technologies. We
believe that the level of interoperability should be customized to such an extent that the pro-
posed system would be able to swap over multiple link layer technologies. For example, the

13
sensor mote should be autonomously capable of communication with other motes using its
own low power radio. However, in the process of movement, if it comes in contact with any
other high power radio such as the Wi-Fi/Wireless mesh access points, then the sensor mote
should be able to communicate with these high-end radios through soft handoffs by changing
its end-points without any data loss. In order to achieve this level of cross-functionality, there
would be a requirement to design and develop a novel communication stack which would ex-
ceed the scope of the research. The limitation in research scope will be circumvented by using
an adaptive communication stack known as Rime [34] which is presently the part of the Contiki
operating system [35] designed for memory constrained networked embedded systems.

14
1.2 Aim of the Thesis

The thesis aims to design and develop a sophisticated underground mine location sensing sys-
tem using wireless sensor networks for tracking personnel and machinery inside the mine that
are able to mitigate the limitations faced by the present systems. In the event of a calamity, real-
time delivery of sensed data will improve the response time and the lack of dependence on fixed
infrastructure and robust operations will aid in locating trapped personnel. The system design
will be guided by empirical models of the characterization of wireless communication in harsh
mining conditions. The solution will be tested and evaluated in a real-world underground mine.
The research problem provides an opportunity for cross-disciplinary research in computer sci-
ence and mining engineering. Besides that, it aims to pioneer research and development with
regards to building the next generation location sensing systems in harsh environmental condi-
tions. Though the immediate motivation in designing this system is the underground mines but
the fundamental aim is to provide guidance in developing futuristic systems for surveillance in
hazardous consitions by benchmarking the system robustness.

1.3 Research Proposal Organization

Chapter 2 presents a literature review that explains the relation of my research with the signif-
icant literature and recent/current research on this problem. The following Chapter 3 outlines
the methodology that would be applied in solving the research problem. The research plan
has been listed in Chapter 4. The concurrent work that has been done in this regard has been
explained in Chapter 5. Chapter 6 outlines the scope of future work in the respective field of
research and concludes the thesis research proposal.

15
Chapter 2

Underground Mines : An Insight

2.1 Introduction

In order to develop an engineering and scientific foundation, understanding the underground

channel characteristics along with the challenges faced by wired and wireless communication
solutions is necessary. The purpose of this chapter is twofold. Firstly, it aims to provide an ori-
entation to the readers regarding the environmental extremeties present inside the underground
mines. Secondly, it presents a review that explains the relation of the research with respect to
recent/current research on this problem which includes the description of the underground mine
(UG) communication techniques along with existing communication systems and a survey of
the UG mine tracking systems. This would facilitate in the better understanding for the design
of an improved location sensing system.

16
2.2 Background

The underground mining environment is remarkably different from the conditions present on
the surface. Underground mines are structurally non-uniform. They contain many crosscuts, es-
cape ways, first-aid stations and blockages. Most of the hallways have railroads on the ground.
The walls are rough and the ground surface is uneven and may have small amounts of accumu-
lated water (Fig. 2.2,2.3). Some parts of the wall and ceilings are strengthened with wooden
grids (Fig. 2.1) and metal [36]. As the thesis focus is on the communication inside the mines,
the general conditions that are noteworthy can be stated as follows:

Figure 2.1: Walls of the mine have been strengthned with wooden grids [7].

Figure 2.2: Water inside the mine

• The dynamic change in underground topology : The walls of the mines may shift daily
as a result of the cutting of the mineral faces.

17
 Figure 2.3: Muddy conditions inside the mine [3]

• The unstable nature of geological construction : A mineral face consists of safe and
collapse zones. In the safe zones, there are hydraulic supporters to avoid collapses. In
collapse zones, there are no supporters and they can easily collapse either when the zone
becomes larger or in the event of mine quakes resulting in structural changes [37].

• Limited Line-of-Sight (LOS): This arises from the presence of pillars and undulations
following the mineral seam. These underground structures get carved and come into
existence during the mineral extraction process.

• Low loss dielectric medium : At certain frequencies, the mine tunnel acts a low-loss
dielectric [38] resulting in the degradation of the communication system.

• Ionized air : The air gets ionized as a result of fires inside the mine. The self ignition of
coal seams results from an exothermic reaction of coal and oxygen. If the concentration
of oxygen is more than 3%, then the oxidation heat is released from the coal and gives
way to fires [39].

• Humid and warm conditions : The relative humidity is greater than 90% and the temper-
ature is approximately 28 degrees [39].

• Gaseous environment : The main component of the gases that effuse with the extraction
of coal from the coal seams is Methane. When the concentration of Methane exceeds the
threshold value, it leads to gas blasts/coal-dust explosion [37]. Hence, there is continuous
ventilation to decrease the build-up of the gas. However, in case of a disaster, the power
supply to the mines is often cut leading to the compromise of the ventilation system [40].

18
Besides these natural conditions present inside mines which pose challenges to electronic com-
munication, every mine has a unique environment. Consequently, communication solutions
need to be engineered that cater to these varying conditions. The design and implementation of
any communication solution needs to be carefully engineered as the slightest of issues can lead
to a mine accident that could result in the loss of human life.

2.2.1 Underground Channel Properties

The major factors that impact on communication with Electromagnetic (EM) waves in under-
ground mines [41] can be summarized as follows:

• Extreme path loss : Lower frequencies experience less attenuation than higher frequen-
cies due to material absorption. The rate of attenuation would increase with the increase
in humidity as well. The path loss increases as the square of the distance travelled by the
wave.

• Reflection/refraction : As mentioned above, the tunnel acts a low-loss dielectric at certain

frequencies and leads to a waveguide effect. Waves that impinge on a wall of the tunnel
are partially refracted into the surrounding dielectric and partially reflected back into the
waveguide resulting in signal losses. The reflected waves may result in a completely new
pattern that may not be recognized as information by the receiver, but as noise.

• Multi-path fading : The random addition of multiple propagation paths causes fluctu-
ations in signal strength with position and frequency, and, if reflectors, transmitters or
receivers are moving, also in time.

• Reduced propagation velocity : Waves propagating through a dielectric medium would

experience a reduced propagation velocity compared to that of air. With the change in
underground temperature, the dielectric property of the medium will change. Hence,
an increase in undermine temperature would alter the dielectric properties and lead to
increase in signal attenuation.

• Noise : The performance of the communication system is highly dependent on the EM

noise in the environment. Clouds contain electrical charges that are evidenced as light-
ning strikes under stormy conditions. This flow of current gives rise to EM radio waves
with sufficient intensity to interfere with radio communications. A considerable amount

19
 of noise ends up in the extremely low frequencies (ELF), voice frequencies (VF) and very
low frequencies (VLF) frequency band, which have a negative impact on the receivers.
The noise caused by electric motors, power lines and appliances is in the frequency bands
which are most suitable for underground communications [42].

• Realistic waveguide effect : In an ideal waveguide effect, electromagnetic waves are

confined and guided by the mine tunnel, but in a realistic scenario, the reflective and
absorption losses along the path result in an increase in signal attenuation.

2.2.2 Underground Communication Techniques

There are 3 ways of communication inside mines [42]:

1. Through-the-Wire (TTW)

2. Through-the-Air (TTA)

3. Through-the-Earth (TTE)

Figure 2.4: Underground communication techniques.

2.2.2.1 Through-the-Wire (TTW)

The TTW communication technique utilizes the wired communication infrastructure present
in the mine where the equipment that is carried by the mining personnel has to be tethered to
a cable. The type of cables used are twisted pair, coax, CAT5 (specially constructed twisted
pair), trolley cable, fiber optic and leaky feeder [43]. The most popular cable among these is
the Leaky Feeder. It is designed to radiate over the entire length and hence derives the name

20
Leaky for this characteristic (Fig. 2.5). The increase in signal range is the result of lesser
attenuation rate by the cable compared to the free space propagation in the mine.

Figure 2.5: Communication achieved through Leaky Feeder cables

Although the performance of the TTW is moderately satisfactory for routine operations,
it is subject to failure and wear and tear under the conditions of roof falls, mine fires and explo-
sions, power failure, interference from other machinery and inadequate maintainability [43].
In order to improve the reliability of the existing systems, various cable protection schemes
have been applied. They include putting the cable in a conduit(armor cable), burying the cable,
feeding cables through borehole connection to main lines, loop-around and redundant cabling
where multiple cables feed the same portion of the system [43]. However, these methods are
not only expensive but also add to the maintainenance and complexity of the system. Moreover,
the borehole cable protection method has its own set of problems that include impedance of the
radio signal and being highly vulnerable towards water getting into the cables. The material
absorption, material scattering, splice and bending losses result in attenuation of the signals in
the wired medium.

21
2.2.2.2 Through-the-Air (TTA)

The TTA communication technique refers to radio communication used in mines. Both met-
alliferous and coal mines present an unique set of challenges for radio communication. The
undulating structures inside the mine are unfavorable for TTA communication as it requires a
clear line-of-sight for propagation. At frequencies in the range of 200-400 MHz, the rock and
the coal bounding in the coal mine acts as relatively low-loss dielectrics with dielectric con-
stants in the range of 5-10 [44]. Transmission takes the form of waveguide propagation in the
tunnel, since the wavelengths of the UHF waves are smaller than the tunnel dimensions. An EM
wave can propagate in any one of a number of allowed waveguide modes. All of these are lossy
modes as any part of the wave that impinges on a wall of the tunnel is partially refracted into
the surrounding dielectric and partially reflected back into the waveguide. The refracted part
propagates away from the waveguide and represents a power loss. However, the propagation
of some frequencies is enhanced by a waveguide effect due to the sandwiching of radio signals
between layers of varying electrical properties. Wall roughness and uneven tunnel cross section
leads to an increase of the longitudinal attenuation. With respect to coal mines, the electrical
properties of coal attenuate certain frequencies more than others and hence a small fraction of
the radio signals are able to propagate down the coal mine. Heavy mining machineries, trol-
leys, high voltage power cables operating inside the mines lead to signal interference and thus
affect the propagational behavior of the signals. Other challenges include ionized air, adverse
environment and mine dynamics.
The frequency selection has a major impact on the signal propagation [40].

• The mine acts as a waveguide to propagate signals at Very High Frequency (VHF) and
Ultra High Frequency (UHF). They have high bandwidth and the devices using them
would require small antennae. However, they suffer from attenuation, corner losses and
require a clear line of sight for propagation.

• Extremely Low Frequency (ELF), Very Low Frequency (VLF) and Low Frequency (LF)
suffer less attenuation but can experience electrical interference from motors and other
machinery. These systems have limited bandwidth, have significant distance limitations
and require large antennas for operation.

• Medium Frequency (MF) have less attenuation characteristics than VHF and UHF sig-
nals. It does not experience the high noise levels of lower frequencies. These systems
have limited bandwidth and require larger antennas than VHF/UHF systems.

22
 Table 2.1: Frequency Band Range
Abbreviation Name Frequency Band
ELF Extremely Low Frequency 30 - 300 Hz
VLF Very Low Frequency 3 - 30 Hz
LF Low Frequency 30 - 300 KHz
MF Medium Frequency 300 - 3000 KHz
VHF Very High Frequency 30 - 300 MHz
UHF Ultra High Frequency 300 - 3000 Hz

Research has shown that MF offer better usability in underground mine communication. Table
2.1 gives the frequency band for each frequency.

2.2.2.3 Through-the-Earth (TTE)

The operating frequency that are used by the conventional systems is unable to penetrate the
rock strata and becomes unoperational in the event of mine accidents. TTE communication
systems prove to be superior as they do not require pre-existing cables/open-air communication
technology and use ultra low frequency signals for communication [42]. The antenna is located
on the mine surface which provides coverage to various parts of the mine, thus reducing the
risk of damage in emergency. Some of the TTE communication systems researched by the US
Bureau of Mines are as follows and can be useful in certain critical situations.

• Ground conduction : This consists of transmitting and receiving signals through the
ground medium through ground stake connections.

• Seismic : The entire setup of the seismic communication system is carried on a truck
and accessories need to be carried by the miners for location detection. Trapped miners
generate a seismic signal by pounding on the mine surface such as the roof, floor or ribs.
Geophone sensors, installed on the surface or underground, detect these signals.

2.2.3 Existing Underground Communication Systems

The existing underground communication systems [38][42] can be broadly classified as gen-
eral communication and tracking systems. The general communication systems include the
following:

• Telephones : The basic operationality of the underground mine phones is similar to that of
the surface-type phones. The Private Branch Exchange (PBX) inside the mine is respon-
sible for the make/break of the call and these telephones are interconnected by multiple

23
 pair cables. Circuit breakers and lightning arrestors are responsible for protecting the
system from sudden energy surges by limiting the electrical energy to safe levels. These
telephone systems are easy to use but are vulnerable to damage from roof falls, mine fires
and explosions.

• Pager Phones : The pager phones are battery operated, party line telephones with pro-
vision for loudspeaker paging. It is cheap and simple but is noisy even in the usual
transmission mode.

• Trolley Phones : Trolley phones can be fixed or mobile (carried on locomotives). The mo-
bile units are subjected to constant vibration and suffer the temperature extremity along
with humid and dusty conditions. The transmission lines pass across various mining ma-
chineries which result in degradation of the communication quality. The main advantage
is that they provide communication to all the rail haulage vehicles using trolley cables
but are limited in coverage.

• Hoist Phones : It is a communication facility between the persons in the hoist cage
(used to raise and lower conveyances within the mine shaft) and the surface/underground
wherein a phone line directly connects the cage to the mine communication system.

• Walkie Talkie : It is a portable, bi-directional radio transceiver which has the appearance
of telephone handset with an antennae. It is basically a half duplex communication sys-
tem wherein only one person can talk at a time. They provide a wireless communication
system with an better coverage area but have to be used in conjunction with leaky feeder
cables and line amplifiers for signal transmission across corners and bents.

The tracking systems include the following:

• RFID Tag based communication : It consists of Radio Frequency Identification (RFID)

tags that is carried by the workers/machinery. As it passes the tag readers pre-positioned
at fixed locations throughout the mine, they are able to recognize the object by the coded
RF signal emitted. This information is sent to a central location for monitoring (Fig. 2.6).

24
 Figure 2.6: RFID Tag based communication

2.3 A Survey of Underground Mine Tracking Systems

The underground mine tracking systems can be classified into the following types based on the
tracking type [40].

• Zone/Proximity based systems : They are able to detect the presence of the object in a
particular region. The RFID based systems belong to this category. Resolution depends
on number of readers installed in a surveillance area.

• Node based systems : A radio device capable of communicating with other nodes is
carried by the miner. The location is determined by identifying the node with which the
miner was able to communicate. Resolution depends on the number of nodes and the
fidelity of the signal processing technique (Fig. 2.7).

• Autonomous systems : They determine the location independent of any deployed mine
infrastructure.

25
 Figure 2.7: Node based tracking systems : A Wireless Mesh

The latest tracking systems are based on digital data networks which include TCP/IP,
Ethernet, WiFi, Wireless Mesh Networks, VoIP, Cell phone technology.
There are many research agencies and manufacturers that have conducted extensive re-
search on developing effective tracking systems for underground mines. A detailed list is avail-
able at [45] [46] [47] [48] [49]. The U.S. Bureau of Mines, CSIRO (Australia), CSIR (South
Africa) are some of the noteworthy names in this regard. The U.S. Bureau of Mines had ex-
tensively focused its research on operational and post-disaster communication systems during
the period from 70s to 80s [50]. This lead to development of numerous mining communica-
tion products that include selectable pager phones, visual paging, multiplexed communication
system, UHF extended system, EM deep mine transmitter, adaptive transmitter systems, TTE
voice communication and many more. [51] gives a complete list of MSHA approved commu-
nication and tracking technologies. Designing and developing such systems for underground
mines should have certain de facto standards. [52] gives a list of the ideal requirements for
confined space communication systems. As per the US government regulations for mines,
electrical communication devices have to be approved by the Mine Safety and Health Admin-
istration (MSHA) as permissible. Permissibility can be achieved through Explosion Proof (XP)

26
and Intrinsically Safe (IS) designs.
The following subsections present a concise overview of the commercially available
tracking products available with the manufacturers and vendors. An assessment of perfor-
mance and limitations has not been described as the utilization of these products in mines is
presently not known and the compiled information is from the respective company websites
which feature mostly on the promised functionality. Nevertheless, it can serve as a good start-
ing point for research in this area. Table 2.2 categorizes the tracking systems discussed in the
following subsections based on the type of tracking.

Table 2.2: Underground Mine Tracking Systems

Tracking System Concerned Authority Tracking Category
Location and Monitoring System for Personal Safety CSIRO, Australia Zone/Proximity based
(LAMPS)
Nexsys Real-time risk management system CSIRO, Australia Zone/Proximity based
Emergency two-way communications CSIRO, Australia -
Wireless location technologies for tracking and data CSIRO, Australia -
collection
Trapped Miner Locating System CSIR, South Africa Zone/Proximity based
Electromagnetic Location System for Trapped miners Institute for Advanced Physics, University of Zone/Proximity based
Innsbruck , Austria
Personal Emergency Device (PED) Mine Site Technologies, Australia Zone/Proximity based
TRACKER Tagging Mine Site Technologies, Australia Zone/Proximity based
TeleMag Transtek, USA
Delta Electromagnetic (DeltaEM) Gradiometer Bea- Stolar, USA Zone/Proximity based
con Tracking System
RadCAT Stolar, USA Zone/Proximity based
Subterranean Wireless Communication System Kutta Consulting, USA Zone/Proximity based
(SWECS)
Digital Radio for Underground Miners (DRUM) Kutta Consulting, USA Zone/Proximity based
Canary 2 Los Alamos National laboratory, licensed to Zone/Proximity based
Vital Alert, Canada
TramGuard Proximity Warning System GeoSteering, USA Zone/Proximity based
TramGuard Miner Track System GeoSteering, USA Zone/Proximity based
BreadCrumb Rajant, USA Node based
AXON Transceivers Innovative Wireless Technologies, USA Node based
Mine Net Tracking System AMR, USA Zone/Proximity based
BeckerTag Becker Mining Systems, Germany Zone/Proximity based
RFID: Multi-read Radio Frequency Identification and Davis Derby, UK Zone/Proximity based
Tracking
Wi-Fi based Real-time Tracking Ekahau, USA Node based
Mine-NET Embigence, Germany Zone/Proximity based
Mine Rescue System Falcon, USA -
IPMine IPackets, Canada Zone/Proximity based
MineAX Tracking and Tagging Tunnel Radio, USA Zone/Proximity based
Smart Tag Resource Tracking System Varis, Canada Zone/Proximity based

27
2.3.1 The Location and Monitoring System for Personal Safety (LAMPS)

This system was proposed by CSIRO, Australia. The key idea of this system is to distribute the
traffic along a key number of redundant paths [53]. Hence, with the disruption of a path, the
communication would still be possible through other paths. It requires a network of wireless
beacons distributed throughout the mine. The system consists of equipping miners cap-lamp
with a transponder that transmits the miners ID, his location and vital health signs to the wire-
less beacon. A control and monitoring system on the surface gathers the communication traffic
from different paths and displays the location information. In case of emergency, it serves to
communicate escape route information for indication by the individual beacons. The personal
transponders have a receiver. In case, the personal transponders fail to maintain communication
with the network, two alarm conditions can be flagged. The underground staff can be alerted
when they are beyond the network coverage as well as when the surface control station loses
communication contact with the underground staff.

2.3.2 The Nexsys Real-time risk management system for underground

mines

CSIRO Australia, in collaboration with the Japan Coal Energy Centre (JCOAL), has also devel-
oped the Nexsys Real-time Risk Management System for underground mines [54]. The system
is comprised of:

• The Nexsys software package

• Electronic report-capturing system

• Suite of Ethernet-based, fiber optic and communications devices.

Nexsys can gather information directly from existing proprietary systems. It then integrates
and interprets the data in accordance with a pre-determined set of rules and can autonomously
initiates a response to breeches of these rules. Information and real-time risk profiles are con-
tinuously displayed on handheld wireless units such as pocket PDAs.

2.3.3 The Emergency two-way communications for underground miners

Again CSIRO Australia but this time with the support from the Australian Coal Association
Research Program has developed a two-way TTE communication [55] system. Using this

28
technology, miners will be able to ask for information, report their conditions and location and
guide rescuers in emergencies. Even though, it has been designed for general communication
purposes, it can serve the purpose of tracking.

2.3.4 Wireless location technologies for tracking and data collection

This is yet another type of tracking system that has been developed by CSIRO Australia.
Though this technology is not specifically targeted for mining environments, using radio fre-
quency tracking, it is suitable for locating objects in challenging radio environments such as
mines [56].

2.3.5 The Trapped Miner Locating System

CSIR, South Africa has developed this application. It consists of a uniquely coded miners tag
that can be worn with the belt and a portable search unit [57]. Inbuilt LEDs and a buzzer inform
that a miner has been detected by the searching unit. The push button on the belt enables the
miner to communicate with the searching team. The testing prototype was able to detect and
locate trapped miners at the distance of 30m through solid and fragmented rocks.

2.3.6 The Electromagnetic Location System for Trapped miners

The Institute for Advanced Physics, University of Innsbruck , Austria has developed an tracking
application for trapped miners [58]. It consisted of a transmitter (beacon) contained in the
miners cap lamp and a hand-held location receiver that could search for the trapped miners
location by tracking the transmitter. The transmitter emits an ELF field of which amplitude
and direction is measured in at least 3 points with a receiver. It is able to measure amplitude
and direction of the magnetic field with three orthogonal receiving antennas. The research
showed that the location of a trapped miner is possible with magnetic signals in the VLF-ELF
region. The theoretical basis and limits for the parameter frequency and distance depending
on the medium parameters have been proposed in [58]. Field tests at the Schewaz/Tirol mine
demonstrated a detection accuracy of 50 cm.

2.3.7 The Personal Emergency Device (PED)

PED has been developed by Mine Site Technologies, Australia. It is an ultra low frequency
TTE communication system used for paging, control, centralized blast initiation as well provide

29
service in case of emergency [59]. The system has the ability to send text messages to warn
or inform specific situations. It uses either a surface or underground antennae loop which
radiates a radio frequency signal enabling one way communication to the miners. It can track
the people/machinery with the help of tags worn by miners or attached to moving machinery.

2.3.8 The TRACKER Tagging

Mine Site Technologies, Australia has developed this product. It is an underground tracking
system that tracks active Tags. These tags are carried by the personnel or attached to the moving
vehicles and other equipments [59]. The system consists of 3 main components:

• RFID tags : Transmit unique ID.

• The Readers/Beacons : Wireless access points that receive the Tag IDs transmitted by
the Wi-Fi signal and then transmit this data to the monitoring systems through the mine
Ethernet or other type of digital network.

• The TRACKER software : Records all the Tag logins and known locations in real time.

2.3.9 The TeleMag

It has been developed by Transtek, USA. It is a TTE, real time, two-way voice communication
between the surface and underground [60]. It employs a single sideband modulated carrier
technique and a digital signal processor based tracking comb filter for attenuating harmonic-
induced noise, which improves the signal-to-noise ratio thus improving the range of the system.
It is not portable. ComCell and ResQCom are two other system designed to complement the
functionality of TeleMag. ResQCom is a portable system designed for mine rescue and other
emergency situations.

2.3.10 The Delta Electromagnetic (DeltaEM) Gradiometer Beacon Track-

ing System

Stolar, USA has developed this system. It is a system consisting of beacon transmitter and a
wave gradiometer (receiver) [61]. The receiver is portable and is used on the surface of the
mine to locate the beacon transmitter. A global positioning system (GPS) receiver and a radio

30
frequency (RF) modem are integrated into the gradiometer. The gradiometer sensor data are
time and position stamped with information from the GPS. The RF modem allows wireless
communication with the gradiometer receiver.

2.3.11 The RadCAT

This is another system that has been developed by Stolar, USA. The system provides a two-
way voice and text messaging communication among moving miners, mine rescue team and
between the surface monitoring center and underground personnel [61]. It also provides track-
ing capabilities for moving, trapped and barricaded miners. The system consists of:

• Tracking beacon.

• Multi mode cap lamp transceiver for voice and text messaging.

• Delta Tracker for locating trapped miners from the surface.

• Directional fox hunter antenna.

• F1/F1 repeater network

• Surface network computer..

• Personal data assistant.

2.3.12 The Subterranean Wireless Communication System (SWECS)

SWECS is being developed for the US Army Communications and Electronics Research, De-
velopment and Engineering Center (CERDEC)[62]. The portable system consists of:

• Digital Radio

• PDA type of display

• Small and portable antenna

31
It can provide two-way voice, text and image transfer through a stationary multi-node network
as well as TTE. The nodes will establish an ad-hoc mesh network and the portable units can
communicate with these stationary nodes as well as between peer and peer. It uses Software
Defined Radio. The system can also be configured such that in an emergency such as a roof fall
where multiple nodes fail, the node to node communication can be via a lower earth penetrating
frequency that can hop over disabled nodes to establish the link connectivity.

2.3.13 The Digital Radio for Underground Miners (DRUM)

The system is under funding from the National Institute for Occupational Safety and Health
(NIOSH) and has been developed by Kutta Consulting, USA. The system is an emergency
backup communication system in large mines and primary communication system in medium
and small mines [62]. The system uses the existing metallic mine infrastructure to propagate the
radio signals. It consists of a collection of portable radios and fixed repeaters that collaborate
to transmit, receive and forward digital data using medium frequency wireless antennae.

2.3.14 The Canary 2

The product was developed out of the Los Alamos National laboratory and licensed to Vital
Alert, Canada. It is a two-way, real-time prototype communication system that uses very low
frequency. It has designed a new underground radio that overcomes the barrier of RF signal
penetration through thick surfaces [63].

2.3.15 The TramGuard Proximity Warning System

It has been developed by GeoSteering, USA. It is a proximity-based tracking system [64]. It

functions by establishing a magnetic marker field around continuous mining machines which
are detected by personal alarm devices (PAD) worn by mining personnel. The PAD system
warns with an audible sound. A display mounted on the continuous mining machine visually
warns the miners to keep them outside the turning radius.

2.3.16 The TramGuard Miner Track System

This is another system developed by GeoSteering, USA. It is a tracking system that uses the
existing field generators and personal alarm devices (PAD) that were developed for use with

32
Geosteerings TramGuard proximity system [64]. Each miners PAD communicates via UHF
to field generators located at various points around the mine which then transmit the unique
ID numbers of each PAD that the system is in contact using VLF signals to surface mounted
receivers. By knowing the field generator that the PAD is in contact with, the location of the
PAD can be determined.

2.3.17 The BreadCrumb

The system is currently used by the military, police, SWAT fire fighters and other first responder
agencies in USA and has been developed by Rajant. It has been used to establish communi-
cation in areas that have experienced natural disasters. The system consists of portable nodes
that use the 802.1b Wi-Fi networking standard at 2.4GHz [65]. The nodes can be deployed as a
standalone network inside the mine or can be connected to other networks on the surface with
communication links outside the mine such as satellite modem, DSL, cable modem etc. The
basic unit is the SE model that provides two radios and an Ethernet port. The XL is a long
range model cable of 11 MBps communication at 6 miles LOS and even greater distances at
lower-speed of communication. The XLE includes an MPEG video encoder which can be used
to provide a video stream to the network. WE and ME are two smaller units that provide at
least one radio and an Ethernet port. WE is portable.

2.3.18 The AXON Transceivers

Innovative Wireless Technologies, USA has developed this product. It is a transceiver module
designed to work with the IEEE 802.15.4 specifications [66]. It is deployed as an ad-hoc mesh
network. The infrastructure consists of a set of fixed nodes that will be stationary throughout
the mine. The mining personnel carry hand-held units to facilitate the tracking process.

2.3.19 The Mine Net Tracking System

This system has been developed by AMR, USA. The system is composed of readers with three
antennas that are placed throughout the mine to define zones and sub zones [67]. The miners
wear a tag that is read by the reader. The tag has a button that is used to send coded messages
in emergency situations.

33
2.3.20 The BeckerTag

The system consists of Active Tags that transmit their unique identification codes every sec-
ond [68]. Locally connected Tag Readers via a RS485 bus interpret the code and update a
centralized database over Fiber Optic, RF transmissions or WiFi. The system can distinguish
between various Tags such as vehicles, personnel, assets or control tags. It has been developed
by Becker Mining Systems, Germany.

2.3.21 The RFID: Multi-read Radio Frequency Identification and Track-

ing

Developed by Davis Derby, UK, the system consists of RFID readers and antennae are strate-
gically installed throughout the mine [69]. Each person entering the mine is equipped with tags
configured with a unique ID number. The readers then automatically identify the object and
pass the information to the surface control centre via the telemetry link. In the event on an
accident, an alarm can be activated in the respective danger zone and the directional signals can
be illuminated.

2.3.22 The Wi-Fi based Real-time Tracking

Ekahau’s(USA) real-time automated system continually monitors the location of assets or per-
sonnel on a campus area and informs the authorized users via the corporate network [70]. It
consists of tags, reference devices, data network, server software and end-user application soft-
ware. It uses Wi-Fi (802.11a/b/g/n) standard access points as the reference devices for tag
location and as the data network.

2.3.23 The Mine-NET

Based on the RFID tag technology [71], Mine-NET consists of a system integrated with the
underground WLAN network infrastructure where an RFID reader is integrated into a WLAN
access point location. It has been developed by Embigence, Germany.

2.3.24 The Mine Rescue System

This system developed by Falcon, USA is packaged in a lightweight battery powered backpack
that can be initiated by underground personnel in less than a minute [72]. It then broadcasts a

34
signal that can be identified by surface sensory units. It can be used either for vertical rescue
(Surface to Mine Rescue) or horizontal rescue (Mine to Mine Rescue). Both field transportable
and permanent installations of the system can be implemented. Permanent sensor installations
can monitor 7x24 underground mine beacons for immediate response and rescue.

2.3.25 The IPMine

Developed by IPackets Canada, IPMine is a a real-time, 2-way wireless communications and

tracking solution that uses 802.11 interface between wireless clients and base stations [73].
When miners enter a new zone, the information from their mobile device is intercepted by
Zone Access Points (ZAPs) and relayed back to the control centre. The flashing of a miners
headlamp or LEDs on the device is controlled and set at different speeds to indicate normal,
priority, or emergency messages.

2.3.26 The MineAX Tracking and Tagging

This supports multiple modes of data transport including wireless readers via leaky feeder, Eth-
ernet and hardwire [74]. In wireless mode, the tag readers can be located anywhere within the
leaky feeder radio system coverage area thus providing customized coverage for each unique
installation. It has been developed by Tunnel Radio, USA.

2.3.27 The Smart Tag Resource Tracking System

Consisting of active RFID tags, networked RFID readers and a PC/Server running Smart Tag
software [75], this system allows readers to connect to an Ethernet network using twisted-pair,
fiber optics or Leaky Feeder. It was developed by Varis, Canada.

2.4 Summary

This chapter has given an orientation regarding the working environment and conditions present
inside the mine. Besides that, it has examined the current state-of-the-art tracking systems
that have been developed for underground mines. The current/under-development tracking
devices are based on the radio technology. As discussed in the chapter, radio communication
inside the mines is impaired by various unfavourable environmental conditions. Though the

35
tracking devices under development are focussing on different types of radio technology, yet
the efficiency of these systems cannot be commented unless tested in a real-world mine. As
a result, combining the radio technology with other type of sensor devices would prove to be
beneficial in overcoming the current limitations.

36
Chapter 3

Methodology

3.1 Introduction

This chapter will outline the methodology proposed to address the problem stated in this thesis.
The various approaches have been stated and have been supported with a vivid justification for
the respective plan of action. A logical breakdown of the problem has been presented for better
understanding. The later part of the chapter discusses the proposed system architecture in terms
of system platform (hardware/software), network topology, protocols, localization and signal
processing algorithms.

3.2 Proposed Solution

We propose an Ultrasound based Wireless Sensor Network(WSN) system with Chirp based
spread sprectrum technique for designing a location sensing system for underground mines.
There are a number of open questions that need to be considered in order to achieve this

37
goal. The following are some of the prime questions that have been answered in the following
sections:

• What are the different types of sensors required to build this system ?

• Are the proposed sensors avaliable in a form that can be used directly or does it require
custom tailoring and modification for the scope of the problem?

• How can these sensors be powered for a long period of time without replacement?

• Are the sensors strong enough to withstand the damage that would be caused by mine
accidents?

• What protective measures would be applied in order to protect these sensors from per-
manent damage or faulty operations in case of accidents inside the mines or in normal
routine operations?

• What are the types of information required from the system for performing its designated
task?

• How, in what form, when and where should the information be retrieved?

• How should the information be stored/exchanged?

• How can we know the health of the system at any point of time and especially after the
occurance of a mine accident?

• How can the system be repaired for erroneous behaviour? Is human intervention required
or can it be done efficiently from a remote control station?

• How should the data received from the system consisting of distributed sensors be con-
verted into meaning information for best possible decisions under a given circumstance?

• Is it a sensible solution?

3.3 Approach and Justification

A portable wireless system would be the best option in mines because they offer the best resis-
tance to damage from roofs falls, fires and explosions. Besides that, the system can be carried
by miners and they do not require pre-existing infrastructure in terms of pre-installed anten-
nas. Using WSNs, the objects can estimate their location by co-operating with nearby objects

38
by sharing the sensor data in order to minimize the overall location error. With a moderate
upgradation of the processor used in the sensor nodes, fast and efficient object tracking can be
performed. In addition, it can be enhanced to monitor gas and dust concentration inside the
mines and stability of underground structures with the addition of a few extra sensors.
The motivation for using ultrasonic sensors is due to their unique propagational charac-
teristics. The propagation of the sound waves from the transmitter to the receiver is the key
to the ultrasound distance estimation. The speed of the ultrasound signals in air at room tem-
perature is about 340m/s. However, this would be seriously effected in underground mining
environment where the high humidity and temperature along with moderate dielectric effect of
the medium would result in decreased propagational velocity of the sound waves. The bliss
of relatively slow speed of acoustic signals with respect to radio waves turns out to be a boon
for location estimation purposes. The radio/light signals travel at very high speeds as a result
of which even a small timing error or lag would result in astronomically large ranging errors.
While, on the otherhand, these measurement error are under acceptable limits for the slow
traveling acoustic signal. Even though, a millisecond error for ultrasound signals would get
converted into an error of 30cm or more, yet it can be controlled by designing effective algo-
rithms as compared to radio signal where a mere microsecond error would translate to an error
of 300m. As a result of this, the electronic complexity of the ultrasound based systems are rel-
atively low where the need for high-end clocks and synchronization circuits can be substituted
with moderate precision clocks leading to an appreciable decrease in the manufacturing cost
of such devices. Additionally, the extra lag in ultrasonic signal propagation gives more time
for position calculations which can substantially improve the system precision and accuracy in
terms of more processing time or considerable time to recover from disoriented system state.
Although distance measurements for ultrasonic systems is based on the line-of-sight mea-
surements, yet it can be useful for estimation in places that may be around an acute turn or
corner of the room (in our case the non-uniform structure inside the mine containing crosscuts,
escape-ways, blockages) due to the diffraction property of the sound waves, though with re-
duced accuracy. The ultrasound waves cannot travel through walls (in our case mine walls) and
hence all the reflected waves would be contained in the same place. Effective filtering mecha-
nism have to be adopted to eminently retrieve the direct line-of-sight signal amidst the whole
array of direct and indirect reflected waves. Nevertheless, it can be useful in simultaneous in-
dependent measurements at different places in the application space which may be useful for
better estimation purposes.
The range of audible frequencies for human hearing is from 20Hz to 20Khz. Therefore,
any frequency above 20Khz would be considered to be in the ultrasound inaudible domain.

39
It may be exciting at the first glance that there is a very broad range of frequencies to apply
in the system as there is no upper bound on the frequency range. However, this excitement
soon sublimes down when [76] mentions that the upperbound is restricted by the absorption
of ultrasound in air and the rate of absorption increases sharply with the frequency increase.
The Sound Pressure Level (SPL) of a 170Khz ultrasonic signal suffers attenuation at the rate of
upto 10dB/m in air while the a 40Khz US signal is able to withstand the signal attenuation rate
to around 1dB/m. Besides this, all waves irrespective of their frequencies suffer signal attenu-
ation in accordance to the inverse square law where the path loss increases as the square of the
distance traveled by the wave. This leads to an addition 6dB decrease in the SPL of the ultra-
sound signal with the doubling of the distance from the source. Hence ultrasound transducers
commonly operate in the frequency range of 40-75kHz wherein distance of upto 10m between
the transmitter(source) and the receiver(destination) can be measured with better confidence
intervals.
Despite all of the facts, an inherent problem with ultrasonic sensing systems is that the
performance is severely affected by ultrasonic noise in the form of interference and reflection
of the signals. In case of mines, such noise would be produced by electric motors, power lines,
mining appliances etc. Distance measurements that are done using reflected signals would
be longer than those using direct path signals and hence special effort in the form of intelli-
gent filtering and correlating algorithms have to be designed. Broadband and correlation (that
use pulse waveform) based ultrasonic systems prove to be better than narrowband (that use
continuous waveform) systems in dealing with reflected signals. Direct pulses emitted by the
transmitter reach the receiver first as compared to reflected pulses that take a longer time to
travel the transmitter-receiver path. As a result, the first signal (or peak) that is available after
the correlation operation is the direct path signal while the remaining peaks in the profile are
a result of the refection of the signal and hence can be ignored. Thus broadband and correla-
tion based systems are effective than narrowband systems as they are productive in separating
the reflected signals from the direct path signal. However, there can occasionally be situations
when the direct path signals is blocked. In such a case, the only pulses that reach the receiver
are the result of the phenomenon of reflection. In such a scenario, if the first peak of the signal
profile can be erroneously taken to be the direct path signal and distance calculation performed
using this value would result in incorrect distance estimation. Hence, the signal processing al-
gorithm should be intelligent to reject such erroneous data resulting from the reflected signals.
The prime challenge for developing a location sensing system for underground mines is
that the communication signal should be able to sustain their signal characteristics under the
extremely hostile mining environment. Guided by this application scenario, the solution would

40
be implemented at the physical and link layers of the protocol stack. This would not only
provide the flexibility of better signal control and modulation but also empowering the system
with better signal processing algorithms for effective noise filtering. Therefore, fine tuning and
standardization of the system from the lowest protocol stack level and then building on better
and effective solutions on every level of the protocol stack would result in an improved location
sensing system.
Broadband US transducers and chirp based spread spectrum signaling techniques would
be used to overcome the limitations of the currently available acoustic based sensing systems.
Spread spectrum communication technique is a method in which the signal generated at a par-
ticular bandwidth is spread in the frequency domain resulting in a signal of higher bandwidth.
Besides other advantages, the major advantages it would offer to a undermine based location
sensing system are better noise immunity, selective addressing capabilities and multiple access
communication. Chirp based spread spectrum technique would be employed to encode the
identity information of the transmitter (in our case, the moving miner). The basic methodology
of this technique is to use wideband frequency modulated chirp pulses in conjunction with the
carrier signal during the pulse interval. A chirp is a sinusoidal signal whose frequency increases
or decreases over a certain period of time. As the chirp based approach uses the entire band-
width to send the signal, hence it is extremely robust to noise in the channel. This transmission
techniques makes the transmission resistant to multi-path fading even when functioning at low
power levels. Unlike, other spread spectrum techniques that make use of the pseudo random
numbers in the transmitted signal to distinguish it from channel noise, it makes use of the na-
ture of the chirp pulse for signal and noise differentiation purposes.
The proposed system would not only be functional as separate entity but would also be
able to seamlessly integrate with the existing wired or wireless communication infrastructure
present in the mines. As the latest tracking systems that are being developed for the mining
industry as based on digital data networks, hence the proposed location sensing system should
have the ability to integrate itself with the conceived futuristic solutions. In order to achieve this
cross functionality, the system platform would be build it using light weight TCP/IP stack. Ad-
ditionally, upon further customization and serious development, it can be made to operate over
the Internet. Though the Internet was initially designed and developed for low speed networks,
the current Internet technology is highly flexible in adopting and running over diversified link
layer technologies.
Implementing the solution at the physical layer would require the solution to be changed
for different communication technologies. The level of interoperability would be customized
to such an extent that the proposed system should be able to swap over multiple link layer

41
technologies. For example, the sensor mote should be autonomously capable of communica-
tion with other motes using its own low power radio. However, in the process of movement,
if it comes in contact with any other high power radio such as the Wi-Fi/Wireless mesh access
points, then the sensor mote should be able to communicate with these high-end radios through
soft handoffs by changing its end-points without any data loss. An adaptive communication
stack known as Rime which is presently the part of the Contiki operating system designed for
memory constrained networked embedded systems to achieve this functionality.

The approach proposed to address the problem will be divided into the following parts:

1. Investigate the underground mining conditions with respect to communication inside the
mine.

2. Design the architecture for the location sensing system that would include the hardware
and software architecture.

3. Develop an signal processing algorithm to perform chirp processing to achieve best-

possible distance estimation.

4. Investigate the distance estimation accuracy of the proposed system in a simulation study.

5. Investigate the distance estimation accuracy of the proposed system in a testbed experi-
mental study in real-world underground mine.

6. Develop a tracking algorithm to continuously track the position of the mobile sensor
motes that would be worn by the person.

7. Investigate the tracking accuracy of the proposed system both in simulation and experi-
mental study.

3.4 Proposed system overview and architecture

The following section discusses the general system requirements for the prototype location
sensing system followed by the system functionality and architecture.

3.4.1 General Requirements

The general requirements for the Location Sensing system for underground mines is as follows:

42
 1. The ranging distance between the transmitter and the receiver sensor nodes should be
atleast 10m.

2. The nodes should be able to survive the hostile underground mine environment.

3. The ADC sampling rate should be between 80KHz and 96KHz and resolution should not
be less than 16 bits.

4. The loss in data transmission should be less than 1% of the total data transmitted.

5. The system should be able to perform continuous tracking.

6. The system should be able to track the position of atmost 4 persons at a time.

7. The system should provide different levels of information details such as fine grained
location, coarse grained location, information about its previous location when the device
moved and out of range information when the devices are far apart such that they cannot
communicate.

3.4.2 Wireless sensor network design challenges

As the proposed wireless sensor network is being designed for the harsh mining environment,
there are a number of design challenges that need to be considered.
High-resolution signal sampling and collection : As the system would be operated in a
highly noisy enviornment, the high resolution signal sampling would facilitate in better signal
reproduction at the receiver end. A good signal reconstruction is one of the prime requirements
as it has to be correlated with a previously stored signal for transmitter indentification. As a
result, it would reduce the chances of false positive for identification purpose.

Timing accuracy : To facilitate in the accuracy of location sensing and continuous track-
ing of the proposed system, the various sensor nodes has to be correctly syncronized with re-
spect to a common clock. On the contrary, data fusion methods that are independent of time
synchronization requirements have to be employed.

Reliability under harsh operating conditions : The system should be able to provide
reliable data collection and interpretation under the mining conditions that are not conducive
for communication systems. It should delay tolerant to compensate for failed nodes or slow
communication among nodes. In order to circumvent these problems, local buffering and data
redundancy techniques would be applied.

43
 Node protection : As the objective of the system is to provide graceful service under
normal as well as extreme conditions inside the mine, the sensor nodes have to be protected
against damage. This challenge would be addressed by encompassing the node inside a pro-
tective casing that is fire and dust resistant. However, as our system would be equipped with
ultrasound sensors, there should be an opening for the US transducers. Providing an outlet
for the transducers would reduce the sensor lifespan as it would be subjected to dust inside
the mine and result in its malfunctioning. We propose to use a low density material (such as
wooden streaks) to cover these outlets. Our choice is based on the fact that the high frequency
US sensors would be able to penetrate the low density material and perform regular opera-
tion along with guranteed protection. Dessicants (silica gel) would be put inside the protective
casing to prevent the sensor node from succumbing to the high humidity inside the mines.

Long lasting battery power : The setup of the entire system would require effort in the
placement of the sensor nodes on mine walls. The battery for powering these sensor motes
should be such that they do not require replacement before a year.

3.4.3 System Functionality

The proposed sensor motes would have the flexibility to be configured as transmitters or re-
ceivers. It would use the time difference of arrival (TDOA) between the RF and the US signals
for distance estimation. Though both RF and US are being used, each has its own functional
role. The US would be used for distance estimation purposes while all other communica-
tion would be carried out through the radio. Broadband US transducers and chirp based spread
spectrum signaling techniques would be used for better signal transmission and reception by the
transmitters and receivers respectively. The broadband ultrasonic transducers on the transmitter
would emit a chirp using spread spectrum methodology. This broadband US signal would be
received by the US receiver that would correlate this signal with an existing one stored with it.
An spread spectrum signal processing technique would be applied at the matching filter with
standard correlation methods to identify the receiver and extract the information from the sig-
nal. The identity information would be encoded in the chirp. Hence, in the presence of multiple
transmitters, each unique chirp generated by the transmitters would be correctly identified at
the receiving end.

3.4.4 System Architecture

The system architecture has been divided into 3 tiers.

44
 1. Tier 1: Individual tunnel stage operation.

2. Tier 2: Individual level stage operation.

3. Tier 3: Overall system operation.

Figure 3.1 depicts tier 1 of the design where the wireless sensor network consisting of sensor
motes would collaborate with each other to estimate the respective locations of the moving
miners in a mine tunnel. Each miner would carry a sensor mote that would emit an uniquely
encoded chirp signal. All such signals would be received by the static sensor nodes that are in
range with the transmitter node. The static sensor nodes would perform chirp processing and
sent back the distance estimate to the respective mobile node being carried by the miner. On
receiving 3 such range estimates, the mobile node would perform a trilateration to estimate its
location. Figure 3.1 shows a scenario where 3 static sensor nodes have been attached to the
walls of the mine. The mobile nodes that are carried by the miners emits unique chirp signals.
The two different chirp signals are received by the sensor node 2 where further processing and
distance estimation is done with intelligent signal processing algorithms.
Figure 3.2 shows the tier 2 of the design. Wireless access points/routers would be in-
stalled at the beginning of each tunnel so that all the location informations are transferred back
from the sensor nodes to the access points. Tier 3 of the design has been described in figure
3.3. All the data from the bottom-most level would be routed to the next higher level until
it reaches the final base/control station on the surface where a sophisticated software would
display movement details of the miners on a pre-configured time basis.

3.4.5 System Platform

3.4.5.1 Hardware Platform

The hardware platform to be used for this work would be based on a new sensor mote that
would be designed and developed. This new class of sensor board would aim to improve the
architecture of the existing Cricket motes as well as encompass better functionality than any
other motes till date. We plan to revamp the design of the Cricket Indoor Location System
[77]. The Cricket system was chosen because it is the latest in the arsenal of ultrasonic based
location systems which was quite popular in the wireless sensor networks community. The
new class of sensor motes would address all of the outlined drawbacks and additionally pro-
vide the feature of expandability by providing interfacing units to add additional sensors or
sensor interface boards or high-end DSP circuit boards. Broadband ultrasonic transducers with

45
 Figure 3.1: Chirp based ranging technique

ultrasound frequency of 40Khz would be used in conjunction with a moderately sophisticated

micro-controller and DSP daughter boards. The DSP circuit boards would have a sampling fre-
quency in the range from 80-96Khz. The transmitter mote that to be worn by the miners would
have an interface to charge itself from the battery pack that is carried by the mining personnel.
The hardware implementation would have the following modules:

1. Microcontroller : A 16 bit microcontroller would be used for faster processing capabili-

ties.

2. RF transceiver : A CC1000 RF transceiver would be used with an operating frequency

of 455 MHz. This band has less interference compared to the 900 MHz band.

3. Ultrasonic transmitter : The US transmitter would be a broadband transmitter and would

emit a chirp signal.

4. Ultrasonic receiver : The US receiver would be a broadband receiver.

5. Expansion slot : This would facilitate in enhancing the sensor mote by adding extra
sensors to the system.

46
 Figure 3.2: Communication within each level inside the mine

6. USB interface : A USB or compact flash interface would be used to faster data transfer

7. Temperature sensor : The ambient temperature would be measured with the temperature
sensor to compensate for the speed of sound variation with change in temperature.

A brief requirement study of the proposed system core features and radio physical properties
has been given in table 3.1 and 3.2.

3.4.5.2 Software Platform

The software would consist of the Contiki operating system. It is multi-tasking OS for memory
constrained networked embedded systems. It has been designed for micro-controllers with
limited amount of memory. A standard Contiki configuration uses 2KB of RAM and 40KB of
ROM. It contains 2 communication stacks : 1) µIP 2) Rime. µIP is a light weight TCP/IP stack
designed for communication over the Internet. Rime is a light weight communication stack
designed for low power radios and would give the flexibility to swap between different MAC
layers.

47
 Figure 3.3: Communication within various levels inside the mine

3.4.6 Network Topology and Protocols

The network infrastructure would consist of static nodes and mobile nodes. The static nodes
would be placed at known location inside the area of interest and would server as anchor points.
The mobile nodes would be worn by the moving person. The mobile nodes would acts as the
transmitter and send a chirp pulse to the fixed anchor nodes that would act as the receiver.
As we have adopted a correlation based signal processing technique, hence there is a
need to store the some chirp pulses in the receiver so that when it receives the signal from the
transmitter, it can perform the matching operation. Hence, there would be a limit on the number
of transmitter that can be localized/tracked at the same time. Therefore, there is a need to design
an effective MAC protocol that can sense this phenomenon and refrain from transmitting when
the receiver has reached the maximum limit for concurrent localization of transmitters. This
would also contribute in power conservation by the transmitter.

48
 Figure 3.4: Hardware implementation of the sensor mote

3.4.7 On Board Data Processing

The On-board data processing would include chirp processing wherein the receiver has to iden-
tify the transmitter by perform a correlation of the received signal with the stored chirp pattern.
This would result in a generate a profile of signal peaks.

3.4.8 Localization Algorithm

The localization algorithm would extend existing Monte-Carlo localization algorithm and com-
bine both location information and distance estimation between the transmitter and the receiver
for tracking the mobile node.

3.4.9 Auto-Localization of the anchor nodes

It is troublesome to visit each anchor node and assign it coordinates in a large deployment
scenario. Hence, there is a need anchor free auto localization algorithms that can automatically
measure the inter node distance without human intervention.

3.5 Summary

This chapter presented the methodology that would be taken to the address the problem state-
ment outlined in the thesis. We have presented a novel idea of using the radio technology in

49
 Table 3.1: System Core Features
Feature Description
Microcontroller MSP430
Architecture 16Bit
Speed 8 MHz
Program Memory 48 kB
Data Memory 10 kB
Configuration Memory 4 kB EEPROM
Storage Memory 1024 kB SRAM
ADC Resolution 16Bit
OnBoard Sensors 2
UI Components 3 LEDs 1 Button

Table 3.2: Radio Physical Properties

Feature Description
Radio Chipcon CC1000
Standard ISM
Frequency Band 315-916 MHz
Data Rate 38.4 kbps
Setup Time less than 50 msec
TX Power +/-10 dBm
Sensitivity -101 dBm
Modulation FSK
Internal Antenna Wire
External Antenna MMCX conn.
Range 40 m
Channels 4

50
conjunction with broadband ultrasound sensors and performing chirp processing in order to
increase the ranging distance and the accuracy of the location sensing device. The proposed
solution would be imlemented on a new prototype sensing system whose system architecture
and requirement details has also been discussed. The related challenges that would arise in the
process of designing the wireless sensor network and their respective solutions have also been
addressed.

51
Chapter 4

Research Plan

4.1 Introduction

The main objective of this chapter is to give a vivid outline of the research plan and define a
template for the final thesis in the form of tentative thesis chapter outline.

4.2 Research program timetable

This section sets out the research plan that we would be undertaking, with emphasis for the
next 12 months of the candidature. Table 4.1 shows the research plan. In the midst of the
research, we hope to produce at least one peer-reviewed publication on the new sensor mote
platform, one peer-reviewed publication on the signal processing algorithm and at least one
peer-reviewed publication on the system implementation and evaluation.

52
 Table 4.1: Research program timetable
Timeline Work
September 2009 Design the architecture for the location sensing system that
would include the hardware and software architecture
September 2009 Develop a signal processing algorithm to perform chirp pro-
cessing to achieve best-possible distance estimation.
December 2009 Investigate the distance estimation accuracy of the proposed
system in a simulation study.
March 2010 Develop, built and configure the sensor board.
May 2010 Investigate the distance estimation accuracy of the proposed
system in a testbed experimental study in real-world under-
ground mine.
October 2010 Develop a tracking algorithm to continuously track the po-
sition of the mobile sensor motes.
December 2010 Investigate the tracking accuracy of the proposed system
both in simulation and experimental study.
March 2011 Auto-Localization of the anchor nodes.
Mid 2011 Write up dissertation.

4.3 Tentative Thesis Chapter Outline

An attempt to define the final thesis template in the form of tentative thesis chapter outline has
been presented in this section.

• Chapter 1 : Introduction

• Chapter 2 : Underground Mines : An Insight

• Chapter 3 : Location Management in Wireless Communication Systems

• Chapter 4 : Ultrasonic Tracking Systems

• Chapter 5 : Simulation Testbed for Chirp Processing

• Chapter 6 : A Prototype Location Sensing System

• Chapter 7 : Location Sensing in Real-time Real-world underground mines

• Chapter 8 : High Performance Tracking in Real-time Real-world underground mines

53
 • Chapter 9 : An Improved Algorithm for Auto-Localization of Anchor Nodes in a WSN

• Chapter 10 : Lessons Learnt from the Development and Deployment of the Prototype
Locations Sensing System

• Chapter 11 : Conclusion and Future Work

4.4 Summary

The chapter discusses the research plan and defines a template for the final thesis in the form
of tentative thesis chapter outline.

54
Chapter 5

Preliminary Work

5.1 Introduction

The main aim of this chapter is to present the preliminary work that has been carried out with
respect to the designing of the prototype location sensing system.

5.2 Concurrent Work

Apart from doing a thorough literature review of the underground mining conditions and study-
ing the technical details of the location sensing system (both underground and surface based
indoor/outdoor ), we had being involved in understanding the limitations of the Cricket indoor
localization system. It was chosen as a starting point as it is the latest of the ultrasound based
localization systems and is quite popular in the wireless sensor networks community with re-
spect to indoor tracking. As the proposed system would also be based on ultrasound sensing,
hence it would provide valuable insight about the practical challenges and limitations faced by

55
these sensors. Apart from serving as a good road map, it would be used a initial platform for
testing out the proposed enhancements with the respect to the robustness of the signal process-
ing algorithms. We hope to eventually integrate the results into the thesis. We will highlight
the relevant aspects of the work in this section.

5.2.1 Improving the accuracy and distance ranging of ultrasound-based

Cricket localization systems

On scrutinizing the Cricket system’s design and methodology, the following drawbacks were
noted:

1. The ultrasound transmission is highly directional. This fact has also been highlighted by
[78].

2. It uses a threshold based approach to detect the arrival of the ultrasonic signal. The
listener assumes the arrival of a US signal when the output of the US amplifiers goes
beyond 65 mV.

3. The gain used in the receiver circuitry has been fixed at 127 and does not make use of the
gain tuning circuitry.

The fact that the ultrasound transmission from the transducers are highly directional has been
highlighted in a number of instance in the literature. This drawback would result in the loss
of robustness of the proposed system which is being designed with respect to the underground
mining conditions. In order to circumvent this limitations, we have two ways of improving the
directionality and making the transducers omni-directional which include altering the transmit-
ter or the receiver setup. A thorough analysis was conducted inorder to take a decision regarding
the change of either the transmitter or the receiver circuitry. At a first glance of the system, we
were lured to go for a change in the transmitter which was driven by our understanding that if
the US signals were spread equally in all direction from the emitting source, it would result in
omni-directionality. However, if the US receivers are not positioned at the required places, then
the transmitters would be sending signals which would not be useful at the expense of the extra
power being consumed for achieving this task. Besides that, the reflections from the surfaces
would result in signal scattering and the proper placement of the receivers would not also help
in such scenarios. Since, the bottleneck of the system is the graceful ability of the US receivers
to effectively detect an US signal that can arrive after multiple reflections, hence a better alter-
native would be to transform the receiver setup in order to achieve omni-directionality.

56
 With this idea in mind, we altered the receiver setup and devised an array of 3 US trans-
ducers with a dodecahedron arrangement (Fig. 5.1) that was plugged in the place of the existing
receiver. An important design decision was the angle at which the 3 transducers must be aligned
with each other so as to give the maximum signal reception. The data sheet the US transducers
claims the beam is typically about 110 degrees wide (+/- 55 degrees ) at the half voltage points
(-6dB). The dodecahedron arrangement of the transducers was decided because the faces make
an angle of about 63 degrees to each other, close enough to 55 degrees, so adjacent transducers
would add their patterns to give a roughly constant response in the plane joining them. The
current transducers are about 35 degree away from vertical ( i.e. looking straight up) when the
array is on a horizontal surface, which means that in a vertical plane the response of one trans-
ducer between the -6dB points is 35 +/- 55 degrees, i.e from about 90 degrees (i.e horizontal
) to +20 degrees over vertical, i.e the current transducer array should have a roughly constant
response in azimuth, but a stronger response looking straight up. The height difference between
the transmitter and receiver has an effect on the best transducer angle. Given a maximum range
R and height difference h, the receiver transducer elevation angle for maximum response (ig-
noring reflections etc.) at the maximum range is arcsine ( h / R ). For example, a range of
10.5m and h of 3m gives an elevation of 17 degrees ( 73 degrees from vertical), but this would
not necessarily be optimal for shorter horizontal ranges. The gain of the receiver circuitry was
fixed at 127 in the original cricket configuration which can scale up to a maximum value of
255. As a result of this, the cricket mote was being set to half its maximum range. To overcome
this drawback, the gain was set to the maximum value of 255 to enhance the cricket’s ranging
ability. As the cricket mote have a temperature sensor, it can also add to the error in distance
estimation. However, the temperature correction is only a numerical scale factor in the com-
puted range. We were more interested in improving the range rather than distance error. As
the temperature correction would not be beneficial for improving the maximal physical range,
it was overlooked.

Ranging experiments were performed with and without the modified setup of the cricket
motes. The comparative results were not exciting because the modified setup did not prove to
be beneficial in enhancing the distance estimates. In order to study this behavior of the modi-
fied setup that included the array of 3 US transducers, an experiment was performed to check
the performance of the receiver transducers with respect to the strength of the reception of the
Ultrasound pulse by the receiver circuitry. An experimental module was coded for the transmit-
ter cricket mote to send only Ultrasound pulses at a rate of 10 pulses/second. An oscilloscope
was used for tracing the US pulse reception pattern. The experimental was conducted with two
different setups:

57
 Figure 5.1: Modified Cricket setup with the 3-US transducer array

1. Original single transducer on the receiver.

2. Array of 3 transducers on the receiver.

The following were the findings from the experimental setup 1:

1. The pulse was detected by the receiver circuit. The receiver transducer showed the max-
imum reception strength when it was placed in the Line of Sight of the transmitter mote.

2. A variation of receiver strength was recorded as the receiver transducer face was gradu-
ally turned away from the Line-of-Sight (LOS) path to the path making a 90 degrees with
respect to the LOS thereby ranging from the maximum receive strength to the minimum
receive strength.

58
 3. The pulse detected by the receiver was quite promising in this setup.

The following were the findings from the experimental setup 2:

1. The pulse detected by the array of 3 transducers was not very satisfactory as the reception
strength in the LOS path was weaker as compared to the Experimental Setup 1.

2. The detected pulse pattern detoriated drastically when the face of the 3 transducer array
was moved away from the LOS and became extremely weak as compared to the Setup 1.

The following were the inferences from the experiment:

1. The 3 transducer array was leading to the signal cancellation effect as the distance be-
tween the each pair of transducers was relatively greater than the desired.

2. The 3 transducer array could not be rebuilt as there was no scope for bringing the trans-
ducers nearer than the present setup.

3. The signal generated by the transmitter is a Sine wave.

4. Even in conditions when the cricket is not able to get a distance estimate, the receiver is
able to capture the signals. But as it is below the specified threshold, the cricket rejects
the signals. Hence, the cricket is not utilizing the functionality of the gain control circuit
that should ideally be intelligent enough to adjust the gain(increase/decrease) depending
on whether the signal received by the receiver is weak enough to be detected.

The experiment suggested that the present hardware modification done would not be
able to give any satisfactory improvement in terms of range unless the cricket functionality is
changed with respect to the transmitter and receiver circuitry. Instead of using the narrowband
transducers, it was suggested to investigate the applicability of broadband transducers and chirp
based spread spectrum technique.

5.2.2 Chirp generation using the US sensors

A modification to the methodology was made to apply the chirp based spread spectrum tech-
nique rather than the narrowband US sine pulse. For the initial setup, it was decided to conduct
the chirp processing in a software environment. In order to simulate the technique, it was de-
cided to generate the chirp from the US transducers of the cricket, capture the US signals using
a broadband US microphone (receiver)and then direct these signals to the laptop using an ADC

59
converter. The following section describes the first part of the implementation which includes
generating the chirp waveform from the US transducer using ATmega 128 microcontroller. The
remaining work is still under progress.

5.2.2.1 General waveform generation technique in ATmega 128 microcontroller

The module responsible for the generation of the waveform is the 8-bit Timer/Counter2 module
of ATmega 128 microcontroller. The following are the registers associated with the functioning
of this module:

• TCNT - Timer/Counter

• OCR2 - Output Compare register

• TIFR - Timer Interrupt Flag register

• TIMSK - Timer Interrupt Mask register

• OC - Output Compare register

• TCCR2 - Timer/Counter clock register

The OC pin is set to either high or low. Values are set in the TCNT2 and the OCR2 registers at
the start of the waveform generation phase. In the case of the cricket motes, TCNT2 is set to
0 and OCR2 is set to 0x5B (decimal equivalent = 91). With the passing of every clock cycle,
the value in the OCR2 register is decremented. The TCNT2 and the OCR2 are continuously
compared by the comparator circuitry. Whenever,the value of TCNT2 equals OCR2 (which in
the case of the cricket happens when OCR2 = 0), a match is signaled by the comparator. This
triggers an Output Compare Interrupt. Simultaneously, the value of the OCR2 is reset to the
initial value of 0x5B at the next clock cycle. Hence, the OC pin remain in high/low value for
91 clock cycles and then toggles its state to low/high. The same process is repeated again.
The reason for the generation of the sine wave of 40 KHz by setting the value of 0x5B in
the OCR2 register has been explained as follows:
Clock frequency of the Atmega 128L microcontroller = 7.3728 Mhz
= 7.3728×106 Hz.
Hence, 1 clock cycle of the microcontroller = 1÷(7.3728 × 106 )

60
= 1.3563×10−7 seconds = 0.1356 microseconds.
182 (91+91) clock cycles of the microcontroller = 1.3563×10−7 ×182
= 2.4685 × 10−5 seconds = 24.685 microseconds.
Frequency of the wave generated in 182 clock cycles = 1÷2.4685 × 10−5
= 40510.43 Hz = 40.5 Khz.
This was the reason behind using 0x5D (=91) in the command outp(0x5D,OCR2).

5.2.2.2 Procedure for the generation of the chirp signal

We are interested in generating a linear chirp which is of the form:

freq = fstarting f requency + k : where k = chirp rate/frequency increase.
According to the current functionality of the code in the cricket motes, it allows only 16 cycles
of the wave to be generated after which the respective registers are cleared off. Hence, only 16
different frequencies are to be selected between 20 Khz and 40 Khz.
A matlab code was written to select the optimum value of k for which 16 different frequencies
can be generated between 20 Khz and 40 Khz. It was found that for k=1.3, 16 different fre-
quencies can be generated between the required range. They generated frequencies along with
their decimal and hex values are given in table 5.1.

The procedure for calculating the decimal equivalent values for the selected frequency is as
follows:
Let the required frequency to be generated = frequired Khz
= (frequired × 103 ) Hz
Time period of the wave = 1÷(frequired × 103 ) seconds
1.3563×10−7 secs is the time required to generate 1 clock cycle
1÷(frequired × 103 ) secs is required to generate (7373÷frequired ) clock cycles
Hence, the actual value to be given to the register = 7373÷(frequired ×2)
= 3686.5÷frequired

5.3 Summary

This chapter discussess the preliminary work that has been carried out with respect to the de-
signing of the prototype location sensing system. A dodecahedron arrangement of 3 ultrasonic
transducers was engineered for better ranging of the Cricket system. The experiment suggested
that the present hardware modification done would not be able to give any satisfactory improve-

61
 Table 5.1: Selected frequencies
Frequency Decimal Hex value
Value
20.0 184 B8
21.3 173 AD
22.6 163 A3
23.9 154 9A
25.2 146 92
26.5 139 8B
27.8 133 85
29.1 127 7F
30.4 121 79
21.7 116 74
33.0 112 70
34.3 107 6B
35.6 104 68
36.9 100 64
38.2 97 61
39.5 93 5D

ment in terms of range unless the cricket functionality is changed with respect to the transmitter
and receiver circuitry. Instead of using the narrowband transducers, it was suggested to inves-
tigate the applicability of broadband transducers and chirp based spread spectrum technique.
The task of generating a chirp signal from the broadband transducers was implemented.

62
Chapter 6

Future Work and Conclusion

Communication and tracking systems in underground mines is an area that has not been actively
researched as contemporary surface based systems. There are few existing system and there is
limited information regarding the actual implementation of such tracking systems. Currently
available tracking systems only register when the person passes a certain location. Designing
and developing autonomous tracking system (e.g. MEMS based Inertial navigation systems)
that are capable of real time continuous tracking need to be researched. Wireless underground
location systems using UltraWide Band (UWB) and Software Defined Radios (SDR) are other
promising research topic in order to counter the challenges posed by radio propagation in mines.
Additionally, the proposed underground location sensing system can be upgraded to monitor
gas and dust concentration inside the mines and stability of underground structures. Future
work would also include enhancing the system to function as a remote control for underground
equipments as well as perform the operation of a remote centralized blasting system.
In this thesis research proposal, the motivation and aim of the research has been pre-
sented inorder to convince the readers that there is a need for research for an undermine lo-
cation sensing system which is an important and significant problem. The specific objectives

63
of the research and the contribution of my work has been stated. A comprehensive literature
review has been drafted to place the proposed research topic in its relevant research context and
to demonstrate the awareness of significant relevant research. The theoretical approaches have
been explained to provide a conceptual framework for the thesis. It also explains the proposed
methods that have been selected to solve the thesis problem statement. The research program
timetable along with work-effort estimation has been outlined to provide the readers with a plan
for the entire duration of the thesis work. Last but not the least, it also puts forwards a tentative
thesis chapter outline.
The thesis can potentially serve as a beginning in entering the problem space of design-
ing effective location sensing systems for underground mines using the latest technological
advances. From the understanding and the results from the thesis, we hope to provide effec-
tive insight into the problem of location sensing which could lay the foundation for better and
improved solutions in the future both with respect to the underground mine as well as define
a guideline for building such systems from the scratch and apply it to the specific functional
environment.

64
Bibliography

[1] http://www.mbendi.com.

[2] http://en.wikipedia.org/wiki/File:Kalgoorlie_The_Big_Pit_DSC04498.JPG.

[3] https://www.cs.tcd.ie/˜jsacha/PeruBolivia2004/Potosi_Mines/?tsize=1024x768.

[4] http://www.nocere.com/structural-collapse/mine-shaft-kills-18-miners/.

[5] http://www.cdc.gov/niosh/nas/mining/whatis-history.htm.

[6] http://www.javno.com/en-world/hopes-fade-for-65-feared-drowned-in-tanzania-mine_1

[7] https://tiscali.co.uk/news/newswire.php/news/reuters/2008/02/17/world/china-warns-

[8] http://www.coolantarctica.com/Antarctica%20fact%20file/science/threats_mining_oil.

[9] http://www.rmg.se/RMG2005/pages/attachments/Mining_trends_atlas_Copco_200309._Tren

[10] http://www.greatmining.com.

[11] B. Ward. Personal communication. 2000.

[12] http://www.theage.com.au/news/national/earthquake-caused-mine-collapse/2006/04/26/

[13] http://www.wvminesafety.org/sagointerviews.htm.

[14] http://www.chinadaily.com.cn/english/doc/2004-11/13/content_391242.htm.

[15] http://articles.latimes.com/2007/aug/19/world/fg-chinamine19.

[16] http://www.nytimes.com/2009/02/23/world/asia/23miners.html?_r=1&hp.

[17] http://news.bbc.co.uk/2/hi/asia-pacific/4266179.stm.

[18] http://www.china.org.cn/government/local_governments/2008-09/12/content_16438962.h

[19] http://www.theaustralian.news.com.au/story/0,20867,20749177-1702,00.html.

[20] http://en.wikipedia.org/wiki/List_of_accidents_and_disasters_by_death_toll#Coal_mi

[21] http://www.msha.gov/MSHAINFO/FactSheets/MSHAFCT8.HTM.

[22] Li Fangmin, Han Ping, and Liu Xinhua. A Wireless Localization Method Used in Coal
Mine. 2007.

65
[23] O. Obst, X. R. Wang, and M. Prokopenko. Using Echo State Networks for Anomaly
Detection in Underground Coal Mines. 2008.

[24] Li Mo and Liu Yunhao. Underground Structure Monitoring with Wireless Sensor Net-
works. 2007.

[25] Mo Li and Yunhao Liu. Underground coal mine monitoring with wireless sensor net-
works. ACM Trans. Sen. Netw., 5(2):1–29, 2009. 1498916.

[26] Mo Li. Building Sensor Network Surveillance Systems: On the Applicability. PhD thesis,
Hong Kong University of Science and Technology, 2009.

[27] Ding En-Jie, Meng Qing-Yong, Gao Pei-Yan, and Zhou Qiang. Improved Pattern Match-
ing Localization of WSN in Coal Mine. 2007.

[28] Wang Xiaodong, Zhao Xiaoguang, Liang Zize, and Tan Min. Deploying a Wireless Sensor
Network on the Coal Mines. 2007.

[29] Cheng Xuezhen, Cao Maoyong, and Michael Collier. An on-line detection system for coal
mine dust. 2008.

[30] Chen Chin-Sheng, Cheng Chun-Feng, and Cai Yuan-Long. The application of pattern
recognition for the prediction of the coal mining water irruption. 1988.

[31] Prasant Misra, Diet Ostry, and Sanjay Jha. Underground mine communication and track-
ing systems : A survey. Technical Report UNSW-CSE-TR-0910, The University of New
South Wales, 2009.

[32] Adam Dunkels. Design and implementation of the lwip tcp/ip stack. Technical report,
Swedish Institute of Computer Science, 2001.

[33] Adam Dunkels. Programming Memory-Constrained Networked Embedded Systems. PhD

thesis, Swedish Institute of Computer Science, 2007.

[34] Adam Dunkels, Fredrik #214, sterlind, and Zhitao He. An adaptive communication
architecture for wireless sensor networks. ACM, 2007. 1322295 335-349.

[35] Adam Dunkels, Bjorn Gronvall, and Thiemo Voigt. Contiki - A Lightweight and Flexible
Operating System for Tiny Networked Sensors. IEEE Computer Society, 2004. 1034117
455-462.

66
[36] Serhan Yarkan and Huseyin Arslan. Statistical wireless channel propagation character-
istics in underground mines at 900mhz. In Military Communications Conference, 2007.
MILCOM 2007. IEEE, pages 1–7, 2007.

[37] Bai Meng, Zhao Xiaoguang, Hou Zeng-Guang, and Tan Min. A wireless sensor net-
work used in coal mines. In Networking, Sensing and Control, 2007 IEEE International
Conference on, pages 319–323, 2007.

[38] J. N. Murphy and H. E. Parkinson. Underground mine communications. Proceedings of

the IEEE, 66(1):26–50, 1978.

[39] Heping Xie and Tad S. Golosinsk. Mining Science and Technology ’99. Taylor and
Francis, 1999.

[40] http://www.cdc.gov/NIOSH/Mining/commtrack/commoverview.htm.

[41] Ian F. Akyildiz and Erich P. Stuntebeck. Wireless underground sensor net-
works: Research challenges. Ad Hoc Networks, 4(6):669–686, 2006. doi: DOI:
10.1016/j.adhoc.2006.04.003.

[42] Schiffbauer-WH Mowrey-GL. Preliminary assessment of communications systems for

underground mines [draft]. 2006.

[43] http://www.cdc.gov/NIOSH/Mining/pubs/pdfs/cmc.pdf.

[44] A. Emslie, R. Lagace, and P. Strong. Theory of the propagation of uhf radio waves in coal
mine tunnels. Antennas and Propagation, IEEE Transactions on, 23(2):192–205, 1975.

[45] http://www.wvminesafety.org/comtraclibrary.htm.

[46] http://www.wvminesafety.org/PDFs/communications/Additional%20Documents/Appendix_D_

[47] http://www.wvminesafety.org/PDFs/communications/Additional%20Documents/Appendix_E_

[48] http://www.wvminesafety.org/PDFs/communications/Additional%20Documents/Appendix_F_

[49] http://www.wvminesafety.org/PDFs/communications/Additional%20Documents/Communicati

[50] http://www.cdc.gov/NIOSH/mining/pubs/pdfs/aootb.pdf.

[51] http://www.msha.gov/techsupp/PEDLocating/MSHAApprovedPEDproducts.pdf.

[52] http://www.rmtltd.com/m_comm_technical_paper_2.htm.

67
[53] Location and Monitoring for Personal Safety (LAMPS), 1998.

[54] http://www.csiro.au/science/Underground-communications.html.

[55] http://www.csiro.au/science/EmergencyComms.html.

[56] http://www.csiro.au/science/PositionLocationSystem.html.

[57] http://www.mhsc.org.za/dmdocuments/Reports/GEN/gen502/gen502.pdf.

[58] Norbert H. Nessler. Electromagnetic location system for trapped miners. 1999.

[59] http://www.minesite.com.au/.

[60] http://www.transtekcorp.com/.

[61] http://www.stolarhorizon.com/.

[62] http://www.kuttaconsulting.com/.

[63] http://www.vitalalert.com/.

[64] http://www.geosteeringminingservices.com/.

[65] http://www.rajant.com/.

[66] http://www.iwtwireless.com/.

[67] http://www.americanmineresearch.com/.

[68] http://www.becker-mining.com.

[69] http://www.davisderby.com/.

[70] http://www.ekahau.com/.

[71] https://www.embigence.com/.

[72] http://www.falcongroupllc.com/.

[73] http://www.ipackets.com/.

[74] http://new.tunnelradio.com/.

[75] http://www.varismine.com/.

68
[76] Andrew Martin Robert Ward. Sensor-driven Computing. PhD thesis, University of Cam-
bridge, 1998.

[77] Nissanka Bodhi Priyantha. The Cricket Indoor Location System. PhD thesis, MAS-
SACHUSETTS INSTITUTE OF TECHNOLOGY, 2005.

[78] N. B. Priyantha, H. Balakrishnan, E. D. Demaine, and S. Teller. Mobile-assisted localiza-

tion in wireless sensor networks. In INFOCOM 2005. 24th Annual Joint Conference of
the IEEE Computer and Communications Societies. Proceedings IEEE, volume 1, pages
172–183 vol. 1, 2005.

69