Citation
Submarine detection using underwater acoustic sensor networks

Material Information

Title:
Submarine detection using underwater acoustic sensor networks
Creator:
Tellisi, Gaith A
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file. : ;

Subjects

Subjects / Keywords:
Submarines (Ships) -- Recognition ( lcsh )
Detectors -- Technological innovations ( lcsh )
Detectors -- Technological innovations ( fast )
Submarines (Ships) -- Recognition ( fast )
Genre:
non-fiction ( marcgt )

Notes

Abstract:
ABSTRACT It is well known that submarines can hide successfully from traditional sonar systems. For this reason, we propose a new technique implemented via the utilization of mobile robots comprising Distributed Underwater Sensor Networks (DUWSNs). In the design and operation of DUWSNs, the arising challenging issues are: (1) the establishing of reliable communication links among the sensor nodes; (2) the maximization of the life time of the network; (3) the sustaining of sufficient data rate within the network; (4) the architecture of the DUWSN and the routing policies deployed; (5) the effectiveness of the DUWSN, regarding submarine detection and identification. This thesis contributes towards an understanding of, and solution to, the impact of the latency on DUWSNs in underwater acoustic environments. Moreover, a new underwater K-Cells transmission protocol is implemented to overcome and exploit the large propagation delay inherent to acoustic underwater networks. We develop a novel DUWSNs that consists of mobile elementary nodes (ENs), a backbone network of cluster-heads (CHs) and a Base Station (BS). We provide a DUWSN node clustering technique that attains high data rate and low energy consumption; the DUWSN utilizes the audio signal emitted by the submarine's engine to detect its presence. We also present some simulation results using MATLAB
Thesis:
Thesis (M.S.)--University of Colorado Denver. Electrical engineering
Bibliography:
Includes bibliographic references.
General Note:
Department of Electrical Engineering
Statement of Responsibility:
by Gaith A. Tellisi.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
862826393 ( OCLC )
ocn862826393

Downloads

This item has the following downloads:


Full Text
SUBMARINE DETECTION USING UNDERWATER ACOUSTIC SENSOR NETWORKS
by
Gaith A. Tellisi
B.S., A1 Fateh University, 2004
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Electrical Engineering
2012


This thesis for the Master of Science
degree by
Gaith A. Tellisi
has been approved
by
Titsa Papantoni
Yiming J. Deng
Fernando Mancilla-David
Oct 19, 2012
II


Tellisi, Gaith A. (Master of Science, Communications Engineering, Electrical Engineering)
Submarine Detection using Underwater Acoustic Sensor Networks
Thesis directed by Professor Titsa Papantoni
ABSTRACT
It is well known that submarines can hide successfully from traditional sonar systems.
For this reason, we propose a new technique implemented via the utilization of mobile robots
comprising Distributed Underwater Sensor Networks (DUWSNs). In the design and operation of
DUWSNs, the arising challenging issues are: (1) the establishing of reliable communication links
among the sensor nodes; (2) the maximization of the life time of the network; (3) the sustaining
of sufficient data rate within the network; (4) the architecture of the DUWSN and the routing
policies deployed; (5) the effectiveness of the DUWSN, regarding submarine detection and
identification. This thesis contributes towards an understanding of, and solution to, the impact of
the latency on DUWSNs in underwater acoustic environments. Moreover, a new underwater K-
Cells transmission protocol is implemented to overcome and exploit the large propagation delay
inherent to acoustic underwater networks. We develop a novel DUWSNs that consists of mobile
elementary nodes (ENs), a backbone network of cluster-heads (CHs) and a Base Station (BS). We
provide a DUWSN node clustering technique that attains high data rate and low energy
consumption; the DUWSN utilizes the audio signal emitted by the submarines engine to detect
its presence. We also present some simulation results using MATLAB.
The form and the content of this abstract are approved. I recommend its publication.
Approved by: Prof. Titsa Papantoni
III


DEDICATION
Thanks God for giving me the strength to finish this journey.
This thesis is dedicated to my passed away Dad and to my wonderful Mom who have supported
me all the way since the beginning of my studies;
To my brother Ehab, who has done more than enough to help me during my studies, and has been
a great source of motivation and support;
Finally, many thanks go to my sister Rasha, who protected my Mom during the hard time that my
country went through.
IV


ACKNOWLEDGEMENTS
I owe my gratitude to several individuals who have made this dissertation possible by either their
contribution or valuable assistance in the preparation and completion of this study.
First and foremost, I want to express my gratitude to my advisor Dr. Titsa Papantoni whose
motivation and encouragement I will never forget. It has been an honor to be one of her Masters
students. I appreciate all her contributions of time and ideas to make my research happen. I am
also grateful to Dr. Yiming J. Deng with whom I have interacted the most during the course of
my graduate studies for his encouragement and helpful advice.
Lastly, I would like to thank my family for all their love and encouragement.
V


TABLE OF CONTENTS
Chapter
1. Introduction and Motivation.................................................. 1
1.1 Thesiss Roadmap............................................................ 3
2. Background.................................................................. 4
2.1 Characteristics of Underwater Acoustic Channels............................ 4
2.2 Differences between M-UWSNs and ground Sensor Networks..................... 5
3. Mobile Distributed-Underwater Sensor Networks Structure..................... 6
3.1. Underwater Positioning..................................................... 6
3. 2 System Model and Problem Formalization..................................... 6
3.3 De sign Methodology for the M-DUWSN......................................... 8
4. Acoustic Channel in Underwater Wireless Sensor Network...................... 19
4.1. Related Work................................................................. 19
4.2. Underwater Acoustic Channel................................................ 20
4.2.1. Underwater Environment................................................... 20
4.2.2. Propagation Delay........................................................ 22
4.3. Transmission Protocol (Random Access Transmission Protocol)................ 24
4.3.1. The system model and the algorithms........................................ 25
4.3.2. Underwater Signal Propagation............................................ 30
4.4. Energy-Efficient and Delay Decreasing Routing Protocol..................... 31
4.5. Numerical Evaluation to Compute the Expected Delay for both WSNs and D-
MWSNs........................................................................... 37
5. Conclusions................................................................. 51
Bibliography.................................................................... 53
VI


LIST OF FIGURES
Figure
3.1. The structure of the Distributed Mobil Fish Network Underwater Sensor Networks
(DMFN-UWSN) and Submarine........................................................... 7
3.2. The general hierarchical structure of the Distributed Mobile Sensor Network -
Distributed Underwater Sensor Networks (DUWSNs)..................................... 8
3.3. The architecture of the Distributed Mobile Sensor Network Distributed Underwater
Sensor Networks (DUWSNs).............................................................. 10
3.4. The reorganization of all ENs around each CH................................... 11
3.5. Some example of some other cluster shapes...................................... 11
3.6. The architecture between odd and even layers................................... 12
3.7. The architecture of the reorganized Distributed Mobile Network Distributed
Underwater Sensor Networks (DUWSNs)................................................... 13
3.8. The architecture of the Distributed Mobile Network Distributed Underwater Sensor
Networks (DUWSNs)..................................................................... 14
3.9. The architecture of the reorganized Distributed Mobile Network Distributed
Underwater Sensor Networks (DUWSNs)................................................... 14
3.10. The architecture of a different reorganized Distributed Mobile Network -
Distributed Underwater Sensor Networks (DUWSNs)....................................... 15
3.11. The architecture of the reorganized (DUWSN) and passing by submarine.......... 16
3.12. Submarine Detection using the proposed SDCA-UVAD and routing the alarm to
the US................................................................................ 17
3.13. Submarine Detection using the proposed SDCA-UVAD and routing the alarm to
the US................................................................................ 17
3.14. The architecture of the reorganized (DUWSN) and passing by submarine (a) First
position (b) Second position (c) Third position....................................... 18
4.1: Standard salinity profde [41].................................................... 23
4.2: Standard Temperature Profde [42]................................................. 23
4.3: Delay between two sensors with distance 100m using Equations 1-4................. 24
4.4: The k-cells algorithm............................................................ 26
4.5: The operation of the 2-cell algorithm (Imaginary Stack device)................... 29
VII


4.6: The 2-cell algorithm (Imaginary Stack device & Arrival Time)................... 29
4.7: The architecture of the M-DUWSNs............................................... 31
4.8: The shortest path direction depends on the BSs location........................... 33
4.9: The shortest path direction toward the BS is chosen................................ 34
4.10: Compare the delay between the two cases: (a) The shortest path direction toward
the BS is chosen (b) a random path toward the BS is chosen to forward the alarm..... 34
4.11: Compare the delay between the two cases: (a) The shortest path direction toward
the BS is chosen (b) a random path toward the BS is chosen to forward the alarm..... 35
4.12: Update the M-DUWSN and Replace the expired ENs.................................... 36
4.13: Update the M-DUWSN and Replace the expired CHs.................................... 36
4.14: Generating a random number of Poisson arrivals.................................... 37
4.15: The system keeps sensing until two consecutive non-collision slots detected, and a
Awindow of............................................................................. 38
4.16: Expected Delays for 3-cell Vs 4-cell algorithm. Admission Delay 100 Slots........ 41
4.17: Expected Delays for 3-cell Vs 4-cell algorithm. Admission Delay 200 Slot......... 41
4.18: Expected Delays for 3-cell Vs 4-cell. Admission Delay 400 Slots.................. 42
4.19: Expected Delays for 3-cell algorithm. Admission Delay =100, 200 and 400 Slots.... 42
4.20: Expected Delays for 4-cell algorithm. Admission Delay =100, 200 and 400 Slot. 43
4.21: underwater sensor networks-Expected Delays for 3-cell Vs 4-cell algorithm
Admission Delay =100 Slots.............................................................. 43
4.22: UWSNs- Expected Delays for 3-cell Vs 4-cell algorithm. Admission Delay =200
Slots................................................................................... 44
4.23: UWSNs- Expected Delays for 3-cell Vs 4-cell algorithm. Admission Delay =400
Slots................................................................................... 44
4.24: UWSNs- Expected Delays for 3-cell algorithm. Admission Delay = 100, 200 and
400 Slots............................................................................... 45
4.25: UWSNs- Expected Delays for 4-cell algorithm. Admission Delay =100, 200 and
400 Slots............................................................................... 45
4.26: UWSNs Vs WSN- Expected Delays for 3-cell algorithm. Admission Delay =100,
200 and 400 Slots....................................................................... 46
4.27: Expected Delays for 6-cell Vs 7-cell algorithm. Admission Delay =200 slots.... 46
4.28: Expected Delays for 6-cell algorithm. Admission Delay =400 Slots.................. 47
4.29: UWSNs-Expected Delays for 6-cell Vs 7-cell algorithm. Admission Delay =200
VIII


47
Slots.................................................................
4.30: UWSNs-Expected Delays for 6-cell Vs 7-cell algorithm. Admission Delay =400
Slots..................................................................... 48
4.31: UWSNs Vs WSN-Expected Delays for 6-cell Vs 7-cell algorithm. Admission
Delay =200 Slots.......................................................... 48
4.32: UWSNs Vs WSN-Expected Delays for 6-cell Vs 7-cell algorithm. Admission
Delay =400 Slots.......................................................... 49
Figure 4.33: Average Access Delay for 7-Cell (Admission Delay =400 and Lambda
=0.4),ATL S-ALOHA, and S-ALOHA protocols.................................. 50
IX


LIST OF TABLES
Table
1 Comparison of the acoustic, radio wave and optical communication in sea water
[31].......................................................................... 21
2 2-Cell Algorithm........................................................... 28
3 Communication rules between ENs, CHs, and BS............................... 32
X


1: Introduction and Motivation
In the last few years, significant research effort has been dedicated in the area of
communications within Wireless Sensor Networks (WSNs). In this thesis, we focus on research
on Underwater Wireless Sensor Networks (UWSNs), since they are deployed for tasks such as
military monitoring, ocean data collection, disaster detection, oilfield monitoring, etc. Sensor
networks have been envisioned as powerful solutions for several applications, [1-3], such as: i)
Surveillance, including reconnaissance, targeting, and intrusion detection; ii) Environmental
monitoring, where pollution in near-shore oceans is an urgent issue and needs close watch; iii)
Underwater explorations, where such explorations are difficult for human beings due to the high
water pressure, unpredictable underwater activities and vast size of unknown area. UWSNs may
also help in the exploration of underwater minerals and oilfields, as well as in determining
routines for laying undersea cables, etc; iv) Disaster prevention via continuous underwater
monitoring.
For viability of their deployment, the UWSNs must be able to coordinate their operation
by exchanging information. Numerous research activities are currently focusing on the
development of network protocols for wireless sensor networks, where new efficient and reliable
data such communication technologies are demanded by UWSNs, due to the unique and
challenging characteristics of the underwater acoustic communication channel, including limited
bandwidth capacity and variable delays.
The major requirements in the design of reliable Mobile Underwater Sensor Networks
(M-UWSNs) systems include: (1) Uimited power, since underwater systems are usually designed
for relatively long-term monitoring. (2) Sensor localization, for effective monitoring. (3)
Reduced delays, to secure effective communication.
In this thesis, a single interesting application has been chosen on behalf of UWSN
deployment. In particular, the selected application corresponds to military target tracking,
namely to submarine detection. Usually, modem submarines are protected by the newly
developed Uow Probability Detection technology UPD [9], To reduce a submarines acoustic
signature, the hull of the submarine is shelled with rubber anti-sonar protection tiles that reduce
acoustic detection. The UPD technology hides the submarine's acoustic signature to thwart active
1


SONAR probing and also reduces the intra-submarine noise to foil passive SONAR listening. The
submarine's noise level is comparable to ocean's background noise, thus legible acoustic signature
may only be collected within a very short distance from the submarine. Considering this kind of
counter-measure, M-UWSNs become an ideal solution for submarine detection. A number of
underwater robots and a large amount of underwater sensor nodes can be air-dropped to the
venue. In real time, each sensor node monitors local underwater activities and reports sensed data
via multi-hop acoustic routes to a distant command center. The new feature of this type of M-
UWSN systems (as compared to previous M-UWSN systems for estuary monitoring) may be the
addition of a built-in Underwater Voice Activity Detector per each UWSN, to avoid false alarms.
This work focuses on mobile underwater sensor networks (M-UWSNs) architectures
consisting of a large number of low cost sensors that can move with water current and dispersion.
Those M-UWSNs must be organized efficiently, to transmit/receive data with minimal delay.
The underwater communication links connecting those underwater devices are also designed to
update effectively the control unit in the main station with current collected data. Such design
may also help in energy conservation, since then the induced reduction in collisions results in
increase of the sensors life span. Unlike traditional Wireless Sensor Networks (WSNs),
Underwater Sensor Networks (UWSNs) induce special design and resources constraints such as
limited amount of energy, short communication range, low bandwidth, as well as limited
processing and storage space in each network [2], UWSNs present a new research topic,
introducing several new problems, due to the unique properties of the underwater environment.
An underwater acoustic channel (chapter 3) is different from a ground-based radio channel in
many aspects, including [1]:
1) Bandwidth is extremely limited. The attenuation of acoustic signals increases with frequency
and range [4] [5] [7], Consequently, the feasible band is extremely small. For example, a
short range system operating over several tens of meters may have available bandwidth of a
hundred kHz; a medium-range system operating over several kilometers has a bandwidth in
the order of ten kHz; and a long-range system operating over several tens of kilometers is
limited to only a few kHz of bandwidth [8] [7],
2) The propagation delay is long. The transmission speed of acoustic signals in salty water is
around 1500 meter/s [6], which is five times of magnitude lower than the speed of
electromagnetic wave in free space. Correspondently, propagation delay in an underwater
2


channel becomes significant. This is one of the essential characteristics of underwater
channels and has profound implications on localization and time synchronization.
3) The channel characteristics vary with time and highly depend on the location of the
transmitter and receiver. The fluctuation nature of the channel causes signal distortions.
4) The bit error probability is much higher, while temporary loss of connectivity sometimes
occurs, due to the extreme characteristics of the channel. The practical deployment and
design of UWSNs face some special challenges: First, the cost of manufacturing, deployment,
maintenance and recovery of underwater equipments is much higher than that of their
ground-based counterparts. Second, energy saving/efficiency is a critical issue for UWSNs.
1.1 Thesiss Roadmap
The remainder of this dissertation is organized as follows:
In Chapter 2, some background on underwater sensor networks is presented. First, the
characteristics of underwater acoustic channels are reviewed. Then, the distinctions between M-
UWSNs and ground sensor networks are discussed.
Chapter 3 focuses on the structure of Mobile Distributed-Underwater Sensor Networks (M-
DUWSNs) and is organized as follows: Section 3.1 introduces the factors that could affect the
position of each mobile elementary node (EN). Section 3.2 presents the structure of the proposed
M-DUWSN. The proposed model and some results are demonstrated in Section 3.3.
Chapter 4 focuses on the Acoustic Channel in Underwater Wireless Sensor Networks (UWSNs).
The main objective in this chapter is to enhance the performance of the acoustic channel by
decreasing the propagation delay. This is attained via the deployment of a novel underwater
Random Access Algorithm. This chapter is organized as follows: Section 4.1 presents some
related work. In Section 4.2, the underwater acoustic channel is analyzed. Section 4.3 introduces
a novel transmission protocol for both WSNs and UWSNs. Section 4.4 focuses on energy-
efficiency and delay decrease via the deployment of a routing protocol. In Section 4.5,
simulation results are presented.
3


2: Background
2.1 Characteristics of Underwater Acoustic Channels
Underwater acoustic channels are temporally variable due to the nature of the
transmission medium and the physical properties of the environments [10], Radio and optical
waves are considered as candidates for wireless communication in underwater environments.
However, neither of these latter waves is appropriate for M-UWSNs. The first reason is that the
radio waves suffer from high attenuation in salty water. Radio Frequency (RF) signals can
propagate at a long distance through conductive water only at extra low frequencies, 30-300 Hz,
which requires large antennae and high transmission power [4], Thus, the high attenuation of
radio waves in water makes them infeasible for M-UWSNs. Second, optical signals usually suffer
from the short-range and line-of-sight problems. Moreover, optical signals are severely impaired
by scattering when used in underwater, and the transmission of optical signals requires high
precision in communication. In short, it is widely accepted that acoustic channels are the most
appropriate physical communication links for underwater environments.
As compared to radio channels, underwater acoustic channels have the following unique
features. (1) Long propagation delays: The propagation speed of sound in water is about 1500m/s,
which is 5 times of magnitude lower than that of radio, 3x108 m/sec. Moreover, the propagation
speed of sound in water is affected by many properties of the water medium such as temperature
and salinity. Such low propagation speed of acoustic signals results in a very long propagation
even in a short distance. (2) Low accessible bandwidth: The available bandwidth of underwater
acoustic channels is very limited due to absorption and most acoustic systems operate below 30
kHz, which is extremely low compared with radio networks. Moreover, the available bandwidth
of acoustic channels depends on both transmission range and frequency. As a result, the
achievable bit-rate in acoustic channels is also determined by both transmission range and
frequency. (3) High channel error rates: The quality of acoustic channels is affected by some
factors such as signal energy loss and noise. Energy loss comes from signal attenuation and
geometry spreading, where signal attenuation is caused by the absorption of acoustic energy
(which increases with distance and frequency), and geometry spreading is caused by the
expansion of the wave fronts. Noise, man-made noise or ambient noise, clearly disturbs the
channels, multi-path possibly introduces inter symbol interference (ISI). All these factors
contribute to the high error rates of acoustic channels.
4


In short, underwater acoustic channels are featured with large propagation delays, limited
available bandwidth and high error rates. Furthermore, the bandwidth of underwater acoustic
channels is determined by both the communication range and frequency of acoustic signals. The
bigger the communication range, the lower the bandwidth of underwater acoustic channels.
2.2 Differences between M-UWSNs and ground Sensor Networks
The main differences between the M-UWSNs and ground sensor networks include the
following [9]:
Communication Method: as explained before, RF channels do not work well in
underwater environments. Instead, acoustic channels are the practical communication
method for M-UWSNs. The unique features of underwater acoustic channels: long
propagation delays, low communication bandwidth and high error rates, make the
protocols proposed for ground sensor networks unsuitable for M-UWSNs.
Cost: while ground sensor networks are expected to become inexpensive, underwater
sensors are expensive devices. This is mainly due to the small relative number of
suppliers.
Deployment and environments: An M-UWSN is usually deployed in 3-dimensional
space, whereas most ground sensor networks work in 2-dimensional space. The 3-
dimensional deployment makes many protocols (such as localization and routing)
designed for ground sensor networks unsuitable for M-UWSNs. Furthermore, the
physical environments of M-UWSNs are much harsher than those of terrestrial sensor
networks.
Power: the required power for the acoustic underwater communications is higher that in
ground radio communications because of the different physical-layer technology, the
longer distance, and the more complex signal processing techniques implemented at the
receivers.
Memory: while ground sensors have very limited storage capacity, underwater sensors
require some data cashing ability, since the underwater channel may be intermittent.
5


3: Mobile Distributed-Underwater Sensor Networks Structure
Underwater Wireless Sensor Networks (UWSNs) represent a promising approach to a
wide range of potential applications, both civilian and military [11-13], In particular, UWSNs are
becoming increasingly important in oceanographic applications as well as in monitoring efforts of
the undersea world: UWSNs facilitate unmanned underwater explorations and large-scale
underwater monitoring with reduced implementation cost and increased operational frequency
[14], However, the lifetime of a UWSN is largely restricted by the energy constraints of the
sensor nodes and the harsh conditions of the underwater acoustic transmission channels. In this
chapter, we develop a mobile distributed UWSN clustered architecture, named Mobile
Distributed Underwater Wireless Sensor Network (M-DUWSN), that prolongs the life time of the
network, by increasing the efficiency of the per sensor energy utilization. We first consider the
problem of determining the network mobility management scheme that conserves energy and also
allows for easy sensor replacement. The chapter is organized as follows. Section 3.1 introduces
the factors that could affect the position of each mobile Elementary Node (EN). Section 3.2
presents the structure of the proposed M- DUWSN. The proposed model and some results are
demonstrated in Section 3.3.
3.1. Underwater Positioning
Many ocean observation and surveillance applications require accurate position
knowledge of the mobile elementary nodes (ENs). Those ENs communicate with each other by
messages exchange, to attain position estimation with subsequent possible position updating, as
may be required. There are several factors that might affect the spatial distribution of the
underwater network such as current, temperature, depth, salinity, pressure, as well as the
deployment of a bad communications system. Therefore, the underwater channel differs from
node to node.
3.2. System Model and Problem Formalization
The general operational architecture of the M-DUWSNs is depicted in Figure. 3.1, where
sensors are symbolized by fish and clusters are symbolized by fish swarms. The objective of the
DUWSN is the detection of adversarial submarines. The architecture of the DUWSN is
comprised of clusters of Elementary Nodes (ENs), where each cluster contains a Cluster Head
(CH) and where CHs are connected with a Base Station (BS) via a backbone network. The BS is
6


basically a main-frame computer located at a carrier. The issues arising in the proposed M-
DUWSNs architecture are: (1) the establishing of reliable communication links among the sensor
nodes; (2) the maximization of the life time of the network; (3) the sustaining of sufficient data
rates within the network; (4) the routing policies deployed in the backbone network; (5) the
effective submarine detection and identification.
Kl
1 Iu>ur1tratl, _ - - '
Figure 3.1. The structure of the Distributed Mobil Fish Network Underwater Sensor Networks
(DMFN-UWSN) and Submarine
The ENs, CHs and BS perform the following functions: (1) The ENs are grouped into
distinct clusters, where each cluster contains a single CH. Each EN collects local data and
transmits them to its local CH. Some constructions of robot ENs are discussed in [18]-[21], (2)
Each CH collects the data sent by its local ENs and processes them, using an operation
determined by the network signal processing objectives. The CH then processes the compounded
processed data, utilizing an operation that is detennined by the network signal processing
objectives, and transmits the outcome to selected neighboring CHs and/or the BS. The CHs have
processing capabilities, energy and life-spans that are much higher than those of the ENs, where
their life-spans and energy may still be limited. (3) BS fuses data transmitted to it by neighboring
CHs, utilizing an operation that is determined by the network signal processing objectives. The
BS has practically unlimited life- span [17], (4) All sensor nodes (ENs) and cluster heads are
distributed randomly underwater per unit-size region (Ocean Depth X Spreading Length). (5)
7


Each EN and CH can move freely and are assumed to be unmanned robots. (6) Each mobile CH
has a specific communication range that includes all ENs in the area of interest. The CHs are
organized in layers by ocean depth, where each CH can only communicate with the CHs in the
layer directly above its own. As an only exception, the CHs located in the highest layer (those
closest to the ocean service) can communicate with each other.
3.3 Design Methodology for the M-DUWSN
The system incorporates many sensors where each sensor has multitasks such as: (1)
collecting data; (2) transmitting; (3) receiving; (4) processing (5) moving. Usually, both
transmitting and moving are the most expensive in terms of energy consumption. Efficient
techniques for multi-access and data forwarding play a significant role in reducing energy
consumption. Also, a mechanism may be needed to dynamically control the mode of the sensors
(switching between sleeping mode, wake-up mode, and working mode), to possibly conserve
more energy, hi addition, when sensors run out of battery power, they should be able to pop up to
the water surface for recharging or replacing, [14],
Figure 3.2. The general hierarchical structure of the Distributed Mobile Sensor Network -
Distributed Underwater Sensor Networks (DUWSNs)
8


In mobile DUWSNs, sensor localization is needed because most of the underwater
networks sensors are mobile with the current. To find the locations of mobile sensors in aquatic
environments is very challenging. At the same time, the improvement of localization accuracy
needs to be considered, while such accuracy could be significantly affected by poor acoustic
channel quality and node mobility. Figures 3.1 and 3.2 exhibit the localization process.
The steps needed for the construction of the general hierarchical structure are as follows:
(1) The number of layers is calculated based on the depth distance between the oceans surface
and its bottom.
(2) The cluster heads (CHs) are inserted in each layer as shown in Figure. 3.3. where each odd
layer has (M) cluster heads and where each even layer has (M-l) cluster heads; their specific
locations per layer are assigned via the use of equations (3.1) and (3.2).
Odd Laver:
CHodd(n) = CHodd(n 1) + Ax/2 (3.1)
Even Laver:
CHEven(n) = CHEven(n 1) + Ax (3.2)
(3) Insert the mobile sensor nodes (ENs) randomly as shown in Figure. 3.3. The sensor nodes
and cluster heads are marked by black and yellow circles, respectively. Each EN knows the
location of itself and its neighbour and is assigned a unique ID in the network. Data and alarms
are collected by underwater local sensor nodes and relayed by some cluster heads. They are
finally fused at the base station (BS) which is equipped with both acoustic and Radio Frequency
(RF) modems that can transmit data to the on-shore command centre by radio or satellite.
9


A
Layer# 1 *0

Layer #2 *
Cluster *

0 A o o
0 . 0 o
ft
0 ft 0 *0
ft ft %
0 0 0
< ft
ft ft
ft
0 0 0

Figure 3.3. The architecture of the Distributed Mobile Sensor Network Distributed Underwater
Sensor Networks (DUWSNs)
(4) Reorganize the whole sensor population, so that each cluster head (CH) is surrounded by
eight sensor nodes (ENs) in diameter d, as shown in Figure. 3.4, using Equations (3.3) and (3.4).
Each one of the inner and outer circles has four ENs, while consecutive sensors in each circle are
90 degrees apart.
SN(xi>yi) = [cH(ri) +^cos(0;) ; CH(n) + ^sin^)] (3.3)
SN(xi,yi) = [cH(n) +yCs(9i) ; CH(n) + ysin(0j)]
(3.4)


Cluster
, MfiiJ
Sensor Nodes
(Fidt)
Figure 3.4. The reorganization of all ENs around each CH
Those sensors are randomly distributed, then reorganized as shown in Figure 3.3, i.e. the
number of sensors and their locations in each cluster could be changed based on the objective of
the network, and those could be also randomly distributed as shown in Figure 3.4.
Figure 3.5. Example of some other cluster shapes
(5) Each cluster head (CH) can only communicate with its two closest cluster heads in the layer
above its own (Cross-section between them as shown in Figure. 3.6). This implies that there is no
connection between cluster heads that are located in the same layer (no cross-section).
11


Sensor Nodes
Figure 3.6. The architecture between odd and even layers
(6) The general re-structure of the proposed model is shown in Figure. 3.6, where the number of
CHs and ENs vary, depending on the number of layers (Depth) and desired spreading length.
12


Figure 3.7. The architecture of the reorganized Distributed Mobile Network Distributed
UnderWater Sensor Networks (DUWSNs)
To validate the effectiveness of our proposed scheme, Matlab is used to simulate the proposed
model and test its mobility management strategy. In the simulation, there are more than 103-256
mobile sensor nodes and 13-32 mobile cluster heads, first distributed randomly, as shown in
Figure. 3.7. All sensors are then reorganized using Equations (1.1)-(1.4), as shown in Figure. 3.8
and 3.9.
13


Mobile Distributed Fishes UnrterWaler Sensor Networks
Spreading Lenghi
Figure 3.8. The architecture of the Distributed Mobile Network Distributed Underwater Sensor
Networks (DUWSNs)
Submarine Deletion using MOF-UWSNs
Spreading Lenghi
Figure 3.9. The architecture of the reorganized Distributed Mobile Network Distributed
Underwater Sensor Networks (DUWSNs)
14


Submarine Detection using MDF-UWSNs
Figure 3.10. The architecture of a different reorganized Distributed Mobile Network Distributed
Underwater Sensor Networks (DUWSNs)
(7) Assign a random path to the underwater submarine. The submarine is detected as soon as the
submarines engine sound starts reaching the ENs, as shown in Figure. 3.10.
(8) As soon as some EN detects a suspicious sound, it sends an alarm to the CH and then to the
Base station (BS), using an assigned path (See chapter 5).
15


Figure 3.11. The architecture of the reorganized (DUWSN) and passing by submarine
Our objective is to detect the presence and trajectory of a submarine using the proposed
M-DUWSN. As mention earlier, an underwater voice activity detector is embedded in each
sensor, to detect the submarines engine sound. Figure 3.11 exhibits the scenario where a
submarine moves from the first to the second position. The clusters located in the vicinity of
those positions will detect the sound of the engine, while pertinent alanns will be immediately
transmitted to the BS using the routing protocol that will be explained in Chapter 4. Figure 3.12
exhibits the same detecting process via a different route. The simulation results using MATLAB
are shown in Figure 3.13.
16


Figure 3.12 Submarine Detection using the proposed SDCA-UVAD and routing the alarm to the
BS

Figure 3.13 Submarine Detection using the proposed SDCA-UVAD and routing the alarm to the
BS
17


Subniacim* r^lwi-lKin nsitny WDF-U'Ai7;Mi
Spreadirv;] LoogM
Stfcmarme Dstaclion using MIX IWSMs
5pt*ai3ir*g LMght
Submarine CW*r|ion 119*03 MCF-iJwSNs
Fig 3.14. The architecture of the reorganized (DUWSN) and passing by submarine:
(a) First position (b) Second position (c) Third position
18


4: Acoustic Channel in Underwater Wireless Sensor Networks
Underwater acoustic wireless sensor networks (UWSNs) consist of a variable number of
ENs that have the ability to communicate via an acoustic signal that has been sent through an
underwater channel. The design of an acoustic underwater channel could be somehow
challenging because of many constrains and limitations in underwater environments, such as
large propagation delays, limited bandwidth and high bit error rate. The main objective of this
chapter is to enhance the performance of the acoustic channel by decreasing the propagation
delay in such a large underwater network, using a novel underwater Random Access Algorithm.
In this chapter, Section 4.1 presents some related work. In Section 4.2, the underwater acoustic
channel is analyzed. Section 4.3 introduces a novel transmission protocol for both WSNs and
UWSNs. Section 4.4 focuses on energy-efficiency and delay decrease using some routing
protocol, while simulation results are presented in Section 4.5.
4.1. Related Work
Object tracking and monitoring in underwater are currently implemented by distributed
underwater wireless sensor networks whose architectures, operations and performance demands
are dictated by the tracking and monitoring objectives, and are also constrained by the
characteristics and limitations of the environment. Within the statistical inference domain, the
tracking and monitoring objectives are classified as either hypothesis testing (detection) or
parameter estimation or estimation of the acting data generating process, and the pertinent
performance criteria include decision/estimation accuracy and convergence rate, where
detection/estimation accuracy is generally monotonically increasing with the number of
observation data processed, [23], When time constraints are imposed on high accuracy
detection/estimation, the consequence is increase in the required overall data rates. At the same
time, in the distributed wireless networks considered, observation data may be collected and
processed by life-limited nodes, whose life-span is a function of the data rates they process, [24]-
[34],
Several underwater communication models for acoustic underwater channels were
proposed recently, focusing on the acoustic channel from different prospectives; for example,
salinity, temperature, depth, error, delay, energy, path loss and the deployed communication
technique. These distinctive features of UWSNs pose many new challenges for delivering
19


information from ENs to CHs in underwater sensor networks scenarios. In [35] a MAC layer
protocol was developed to attempt to deal with the long propagation delays. The protocol actually
sends synchronization messages containing a transmission time schedule for each node to attempt
to avoid collisions and delays. In [36] a protocol is presented to allow nodes to use a low-power
wakeup mode to conserve energy when idle. [37] presents an analysis of the effect of the
bandwidth-distance relationship on routing decisions based on hop length for energy-efficient
routing. However, in this chapter a new event random access protocol is developed to enhance the
communication link between ENs, CHs, and BS to attain energy efficiency, reliable data transfer
and decrease expected delays.
4.2. Underwater Acoustic Channel
4.2.1. Underwater Environment
The transmission medium in any underwater communication channel may be: (1) Sound
for acoustic communication; (2) Electromagnetic waves for radio wave communication; and (3)
Light for optical communication. As mentioned in Chapter 2, neither radio waves nor optical
communication perform well in deep underwater environments, while acoustic communication is
then widely used, due to the low attenuation of sound in water,(see Table 4.1). Although radio
waves propagate faster, in comparison to acoustic waves, they strongly attenuate, especially in
salt water, and propagate at long distances through conductive seawater, only at extra low
frequencies (30-300Hz) requiring large antenna and high transmitter power [3 8]-[40], However
[41], electromagnetic wave are preferred in shallow waters, subject to certain specific conditions.
Optical communications have a greater advantage in data rates exceeding 1 Giga bps [38],
However, when used in underwater environments, the light is rapidly absorbed and the
communication is profoundly affected by its scattering. Consequently, transmission of optical
signals requires high precision in pointing the narrow laser beams [39], Table 1 shows the
different characteristic of the acoustic, radio and optical propagation in seawater.
20


Table 1: Comparison of the acoustic, radio wave and optical communication in seawater [31]
Acoustic Radio Optical
Speed Propagation 1500m/s 3xl0A8 m/s 3xl0A8 m/s
Power Loss >0.1 dB/m/Hz -28 dB/km/lOOMHz Turbidity
Bandwidth -kHz -Mhz ~10-150Mhz
Antenna Size -0.1m ~0.5m -0.1m
Frequency Band -kHz -Mhz ~10A14-10A15 Hz
Transmission Range ~50m-5km ~lm-100m lm-lOOm
Underwater acoustic communication is a technique for sending and receiving messages
underwater using the propagation of sound. Typical frequencies associated with underwater
acoustics are between 10 Hz and 1 MHz. The slow propagation speed of sound in underwater is
1500m/s which is five orders of magnitude lower than that for radio and optical waves [42], The
features of the acoustic channel are large propagation delay (0.67s/km) [41], large delay variance,
limited bandwidth and high bit error rate [40] [42].
The major challenges in designing underwater acoustic networks are [43-46]:
Low propagation speed which results in large propagation delays.
Limited battery power that usually cannot be recharged requiring the deployment of
energy efficient communication protocols.
Underwater sensors are prone to failures because of fouling and corrosion.
High bit error and temporary losses of connectivity (shadow zone) will be experienced.
The available bandwidth is limited and depends on the transmission distance.
The underwater channel is severally impaired, especially due to multipath and fading
problems.
21


4.2.2. Propagation Delay
The sound speed in seawater depends on water properties such as temperature, salinity
and pressure (or depth). The speed of sound in seawater increases with the increase of any of the
above three parameters. This may lead to large variations in propagation speed, even for equal
distances. The underwater propagation speed has been modeled in [43] as follows:
Cx = 14929 + 3(T- 10) 6 X 103(T- 10)2 4 X 102(T 18)2 + 1.2(S- 35)
D
- 1(T2(T- 18)(S 35) + (4.1)
61
C2 = 1449.2 + 4.67 + 0.05572 + 0.000297s + (1.340 1(T27)(5 35) + 1.6 X 10~2D (4.2)
C3 = 1448.96 + 4.5917 + 0.0530472 + 0.000237473 + 1.340(5 35) + 0.0163Z) + 1.675 X 10~7D2
- 0.01027(5 35) 7.139 X 10~13TD3 (4.3)
C4 = 1449 + 4.67 + 0.05572 + 0.0037s + (1.39 0.0127)(5 35) + 0.017D (4.4)
C5 = 1449.2 + 4.67 + 0.05572 + 0.002097s + (1.34 0.0107)(5 35) + 0.06D (4.5)
Since pressure is a function of depth, it is customary to express the sound speed (c) in m/s
as an empirical function of temperature (T) in Celsius, salinity (S) in parts per thousand and depth
(D) in meters. Figures 4.1 and 4.2 show the standard salinity and temperature profile in the ocean.
Sound speed may be calculated using one of the formulas above. Therefore, many calculations
were performed based on the assumption that the salinity is constant, or nearly so. This
assumption is justified by the observation that the typical range of salinities in the open ocean is
usually small and that their impact on the speed of sound is practically negligible. In coastal
areas, and near rivers or ice, however, the assumption is not generally valid [48], There are
several underwater sound speed equations (1-5); we may use Equation 4.5, for example. It is
important to note that, unlike in ground radio links, distance alone is not sufficient for
determining delays. Some underwater short links could induce longer delays, as compared to long
wireless links.
22


Standard Unde***ei Salinity Profile
Figure 4.1: Standard salinity profde [41]
Standard Undeiwalar Temperature PreSle
Figure 4.2: Standard Temperature Profde [42]
The slow propagation speed of sound in seawater contributes to large propagation delays,
regarding system performance. The propagation delay may be calculated by the following
equation:
d

c
(4.6)
23


; where t is the propagation delay (in second), d is the distance between two nodes (in
meters) and c is the sound of speed (in m/s). We will talk about the effect of this propagation
delay on our K-cells algorithm later. Figure 4.3 shows how the delay between two sensors in
fixed distance from each other (100m) changes, for varying depths, where each one of Equations
(4.1) to (4.5) is used in the calculation of the speed of sound.
fjdlsy for Different Ctepllvs using 1>is 4-£quwion*
Figure 4.3: Delay between two sensors with distance 100m using Equations 1-4
4.3. Transmission Protocol (Random Access Transmission Protocol)
Random Access Transmission Protocols are implemented in this chapter, first for the
wireless on ground surface case, and then for the underwater case. We present and analyze a class
of limited sensing random access algorithms with powerful properties for both the on ground
surface Wireless Sensor Networks (WSNs) and the Underwater Wireless Sensor Networks
(UWSNs). We note that the algorithm has been previously implemented in on ground surface
wireless mobile environments. In this chapter, we will apply it to the underwater sensor networks
environment and compare then its performance with that of the on ground surface case. We start
by considering packet networks, and independent users whose identities are initially unknown to
the system and who transmit through a single common channel. The deployed transmission
protocol must then necessarily belong to the class of Random Access Algorithms (RAAs),[50],
24


Such scenarios arise, for example, in the Ethernet environment, as well as in the signaling stage
of packet radio and cellular telephony. The class of RAAs includes three subclasses, depending
on the level of channel sensing required by the users in the protocol operations. The Minimal
Sensing (MS) subclass requires that each user sense the transmission channel only at times when
he transmits. The Limited Sensing (LS) subclass requires that each user sense the transmission
channel continuously from the time he generates a packet to the time that this packet is
successfully transmitted; this subclass contains a number of algorithms, [51]-[53] with various
characteristics discussed in the paragraph below. The Full Sensing (FS) subclass requires that
each user know the overall feedback history of the transmission channel from the time when the
system starts operating, [54]; this subclass contains algorithms, [50], [54], whose operations are
of just academic interest, since the level of channel sensing they require is clearly non-
implementable.
The limit Poisson user model consists of infinitely many independent Bernoulli users and
it represents a worst case scenario for the study of RAAs, within a large class of user models, as
proven in [55],
In this thesis, we present a class of LS RAAs that attain the same throughput and also
possess similar resistance to channel errors as the algorithm in [53], The algorithms exhibit
varying delay behavior for low rates and possess advantageous operational characteristics. We
will focus on a single member in the class. The collision resolution operations of the latter
algorithm can be depicted by a k-cell stack; we thus name the algorithm, the k-cell algorithm.
4.3.1. The system model and the algorithms:
We assume packet-transmitting users, a slotted channel, binary collision-versus-non-
collision (C-NC) feedback after each slot, zero propagation delays, and initially absence of
feedback errors. We also assume that collided packets are fully destroyed and retransmission is
then necessary. Time is measured in slot units; slot t occupies the time interval [t, t+1) and xt
denotes the feedback that corresponds to slot t; xt = C and xt = NC then represent collision and
non-collision in slot, respectively.
25


I All the Backlog
! will be
CRI has ended at T transmitted on
T 1 this slot
1 J 1 | FCFS 1 1 L.. 1 i r Real Time
1 Data have 1 1 1 1 1 : Updated i
been P | ^-Window-* i !
successfully I 1 d III
transmitted | 1 , i 1 j 1 1 1
1 1 tt !- tf 1 111 1 1 Arrival Time
y/////// 1 II i 1
1 m-11 m 1 i
Figure 4.4: The k-cells algorithm
Each algorithm in the class is independently and asynchronously implemented by the
users. Indeed, in the limited sensing environment, it is required that each user monitor the
channel feedback only from the time he generates a packet to the time that this packet is
successfully transmitted. Therefore, the users' knowledge of the channel feedback history is
asynchronous. The objective in this case is to prevent new arrivals from interfering with some
collision resolution in progress. This is possible if each user can decide whether or not a collision
resolution is in progress within a finite number of slots from the time he generates a new packet.
The possibility of such decision can only be induced by the operational characteristics of the
algorithm. As we will explain below, each algorithm in the class possesses the appropriate
operational characteristics for such decisions.
Each algorithm in the class utilizes a window of size A as a operational parameter and
induces a sequence of consecutive Collision Resolution Intervals (CRIs). The window length A is
subject to optimal selection for throughput maximization. Each CRI corresponds to the
successfiil transmission of all packet arrivals within an arrival interval of lengthA. The length of
the CRI is determined by the number of users in the window A and the algorithmic steps of the
collision resolution process, Figure 4.4. The placement of the A-size window on the arrival
26


access is determined asynchronously by the users. We will first describe the collision resolution
process induced by the algorithm. Then, we will explain the process which determines the
placement of the A-size window per CRI.
The algorithmic class contains algorithms whose collision resolution process can be
depicted by a stack with finite number of cells. Let us consider this algorithm in the class which
can be described by a K-cell stack. Then, in the implementation of the collision resolution
process, each user utilizes a counter whose values lie in the set of integers, [ 1,2,... K]. We denote
by rt the counter value of some user at time t. The K different possible values of the counter
place the user in one of the K cells of a K-cell stack. When his counter value is 1, the user
transmits; he withholds at K-l different stages otherwise. When a CRI begins, all users in the A-
size window set their counters at 1; thus, they all transmit within the first slot of the CRI. If the
window contains at most one packet, the first slot of the CRI is a non-collision slot and the CRI
lasts one slot. If the window contains at least two packets, instead, the CRI starts with a collision
which is resolved within the duration of the CRI via the following rules:
The user transmits in slot t if and only if rt= 1. A packet is successfully transmitted in t
if and only if rt= 1 and xt = NC. The counter values transition in time as follows:
If xt 1 = NC and rt 1 = j; j = 2,3, ..,K, then rt = j 1
If xt 1 = C and rt 1 = j; j = 2,3,... K, then rt = j
If xt-l = C and rt 1 = 1, then, rt
1 w. p. 1/K
2; w. p. 1/K
3; w. p. 1/K
From the above rules, it can be seen that a CRI starts with a collision slot ends with K
consecutive non-collision slots, an event which cannot occur at any other instant during the CRI.
Thus, the observation of K consecutive non-collision slots signals the certain end of a CRI to all
users in the system; it either signifies the end of a CRI that started with a collision or the end of a
sequence of K consecutive length-one CRIs. Therefore, a user who arrives in the system without
any knowledge of the channel feedback history can synchronize with the system upon the
observation of the first K-tuple of consecutive non-collision slots. This observation leads to the
asynchronous by the users generating of the size-A window placement on the arrival axis.
Specifically, if a CRI ends with slot t, the window of the next CRI is selected with its right most
27


edge K-l slots to the left of slot t and it contains those packets whose updates fall in the interval
(t- K+l-A t-K+1). The updates \ tk r of a packet are generated as follows: Let t0 be the slot
within which a packet is generated. Then define t to be equal to t0. Starting with slot t, the
corresponding user senses continuously the channel feedbacks. He does so passively, until he
observes the first K-tuple of consecutive NC slots, ending with slot ft.. If te(ti -K+l A, ft -K+
1), the user participates in the CRI that starts with slot ft + 1. Otherwise, he updates his arrival
instant to t1 = t + A and waits passively until the end of the latter CRI, ending with slot t2. If t1 e
(t2-K+l-A, t2 -K+ 1), the user participates in the CRI which starts with slot t2; otherwise, he
updates his arrival instant by A again and repeats the above process. In general, if H tn r"n> 1
denotes the sequence of consecutive CRI endings since the first K-tuple of consecutive NC slots,
the packet participates in the kth CRI if tk l e(tk -K+l-A, tk -K+l) and tV(tni -K+ 1 -A, tn -K+l)
; for all n < k-2.
2-cell Algorithm Case: The 2-cell case will be explained with steps below:
If at least two packets attempt transmission within the same slot, a collision occurs and
such an event is initially the only cause for faulty transmissions. A collision results in complete
loss of the information carried by the collided packets; thus, retransmission of such packets is
necessary. As we said before that the stack has endless capacity because we have infinite number
of users, which means that CRI does not have a specific ending unless the users are synchronized
with system from the begimiing, they know die history of the system'. What happen in the CRI is
as follows:
Table 2: 2-Cell algorithm
Case number Outcome of the slot Counter value What will happen
Number 1 xt = NC ct = i Then, he is gone
Number 2 xt = NC Ct > 1 Ct+i = 1
Number 3 xt = C Ct > 1 ct+i = ct + 1
Number 4 xt = C Ct = l (2; w. p 0.5
28


Imaginary Device
Stack
Exam mg
Interaval
No Collision
Transmitted
Figure 4.5: The operation of the 2-cell algorithm (Imaginary Stack device)
From these rules it can be seen that a CRI that starts with a collision slot ends with two
consecutive non-collision slots. Furthermore, two consecutive NC slots cannot occur at any other
instant during a CRI, Figures 3.5 and 3.6. Thus, the observation of two consecutive NC slots
signals the end of such a CRI to all the users in the system.
lmiiginur>
^Siack
Rvamino
Intc'iava!
^ Trans milled
Figure 4.6: The 2-cell algorithm (Imaginary Stack device & Arrival Time)
29


4.3.2. Underwater Signal Propagation
As mentioned above, radio and optical communications do not perform well in deep
underwater environments. Thus, acoustic communication is then widely chosen, due to the low
attenuation of sound in water. Typical frequencies associated with underwater acoustics are
between 10 Hz and 1 MHz. To compute the underwater delay, we need to find the sound speed
and the distance between the sensors. The sound speed in the seawater depends on the water
properties such as temperature, salinity and pressure (or depth). The speed of sound in the
seawater increases with the increase of any of these three parameters.
After selecting the underwater parameters such as temperature (T) in Celsius (from
Figure 4.2), salinity (S) in parts per thousand (from Figure 4.1) and depth (D) in meters, we may
calculate the sound speed and subsequently the underwater propagation delay via Equation 4.6
(page 38, Figure 4.3), based on the selected distance between the sensors (in our program, the
distance is between 100m). The propagation delay has a direct effect on the operation and the
induced performance characteristics of the deployed random access transmission algorithm. The
propagation delay must be considered in the operation of the algorithm: In particular, it should be
added to the delay of observed channel outcomes when propagation delay is absent. Thus, since
feedbacks are observed shifted in time by the amount of the propagation delay, the latter is added
immediately to received channel feedbacks, in our simulations. The propagation delay value
could be added either as a constant or as a random variable, hi this work, the propagation delay
value is considered a constant, where its delay effect on the algorithmic operations may lead to
deadlocks in the limit. The equation below shows how the underwater expected delay is
simulated when the propagation delay is incorporated. The simulation results are shown in
Figure 4.29, 4.30, 4.31 and 4.32.
Underwater Expected Delay
= The time from the instant when the Events Ocurrs to the time we actually see it
Underwater Expected Delay = [TimeTransmited TimeArrivai\
+ Underwater Proagation Delay (equation 4.6) (4.7)
30


4.4. Energy-Efficient and Delay Decreasing Routing Protocol:
In order to perform underwater communication effectively via deployed M-DUWSNs,
we propose an energy-efficient and delay decreasing routing protocol. This means that, in
addition to the delay reduction induced by the k-cells algorithm, the deployment of the routing
protocol will further enhance performance: (1) Energy savings may be attained by minimizing
the number of data transmissions through the ENs and CHs. (2) Delay reduction may be attained
via shortest path routing across the CHs. (3) The lifetime extension of underwater sensor
networks is also accomplished by using the Layer by Layer replacement method.
In this section, the communication architecture of M-DUWSNs is recalled from Chapter
3, as shown in figure 4.6. This communication architecture is dictated by the network objectives:
(1) Gathering information continuously; (2) Monitoring and detecting objects effectively; (3)
Attaining energy-Efficiency and lifetime extension; (4) Guaranteeing reliable communication
links. We thus selected one of the communication architectures used in Chapter 3, as shown in
Figure 4.7.
Figure 4.7: The architecture of the M-DUWSNs
The ENs, CHs and BS perform the following functions: The ENs are grouped into
distinct clusters, where each cluster contains a single CH, where each EN collects local data and
transmits them to its local CH. Each CH collects the data sent by its local ENs and processes
them, using an operation detemiined by the network signal processing objectives. The CH then
processes the compounded processed data, utilizing an operation that is determined by the
network signal processing objectives, and transmits the outcome to selected neighboring CHs
31


and/or the BS. The CHs have processing capabilities, energy and life-spans that are much higher
than those of the ENs, where their life-spans and energy may still be limited. BS fuses data
transmitted to it by neighboring CHs, utilizing an operation that is determined by the network
signal processing objectives. The BS has practically unlimited life- span. All sensor nodes (ENs)
and cluster heads are distributed randomly underwater per unit-size region (Ocean Depth X
Spreading Length). Each EN and CH can move freely and are assumed to be unmanned robots.
Each mobile CH has a specific communication range that includes all ENs in the area of interest.
The CHs are organized in layers by ocean depth, where each CH can only communicate with the
CHs in the layer directly above its own. As an only exception, the CHs located in the highest
layer (those closest to the ocean service) can communicate with each other and the BS on the
service. The ENs and CHs are first randomly distributed and then reorganized in clusters and
layers as shown in Figure 4.8. Hence, each layer can handle as many ENs and CHs as required,
and the number of layers could be calculated based on the water depth and the distance between
the layers needed to accomplish reliable communication link.
All ENs and CHs are equipped with Voice Activity Detectors, to detect the presence of
various objects or any unusual activity in the area, and also possess underwater acoustic
transmitters and receivers. In our design, let assume that the water depth is about 100m, and that
the distance between the layers is 20m, as shown in Figure 3.5. The communication path to route
the information from ENs that are located at the bottom layer to the service layer where the BS is
located should be considered in the design, since it affects the overall transmission delay. The
communication rules pertinent to layering are included in Table 4.3.
Table 3: Communication rules between ENs, CHs, and BS
Ocean Service Where the BS is located to comiect the M-DUWSN with the Wireless Network
Layer #1 Is located on the service, where all CHs can communicate with each other and the BS.
Layer#2 (1) Each EN can communicate only with its CH within the cluster
(2) Each CH can communicate only with its two nearest CHs in the
above and the lower layer. (2 lower layer +2 higher layer = 4
Layer#N CHs)
32


Table 3: (Cont.)
(1) Find die location of the BS (Center, Right, or Left Side)
(2) The path direction to transmit information is programmed based
on the position of the BS as shown in Figure 5
Shortest Path to (3) As mentioned above, each CH can communicate only with its
the Service two nearest CHs in the above and the lower layer. (2 lower layer +2 higher layer = 4 CHs), but the CH from the above layer is selected based on this direction. For example, if the base station is located on the left side of the network, the chosen CH is the one on the left side. See figure 4.
Center
l.efi Side |
| Right Side
To decrease the delay while simultaneously saving in transmission energy, it is important
to configure the path between the layers and maintain it consistently. Figures 4.8 and 4.9 show
how the path could be configured. Each CH communicates with the CH located on the left side on
the above layer if the assumed path is the Left Side Path (LFP), and communicates with the right
CH if the assumed path is the Right Side Path (RSP).
33


Figure 4.9: The shortest path direction toward the BS is chosen
To evaluate the performance of the routing protocol, we select two cases that include
distance, sound speed, temperature, salinity (hence, Equations 4.1-4.5). The two cases are
exhibited in Figure 4.10. In the first case, the shortest path toward the BS is 111.8304m (see
chapter 3 to know the parameters of the MDUWSN structure). In the second case, the random
path is 140.4330m.
Figure 4.10: Compare the delay between the two cases: (a) The shortest path direction toward the
BS is chosen (b) a random path toward the BS is chosen to forward the alarm
Figure 4.11 exhibits the possible improvement in delay reduction, when the shortest path
toward the BS is chosen, instead of the random path. The performance is approximately -22.26%.
34


Delay for Different Deplhe using 1he 3-Eguaiions
Figure 4.11: Compare the delay between the two cases: (a) The shortest path direction toward the
BS is chosen (b) a random path toward the BS is chosen to forward the alarm
Considering the extension of the network lifetime, in this section we introduce a simple
methodology for EN and CH replacement, when any of the latter expire. Our methodology is
described below, where ENs and CHs are organized in layers according to their closeness to the
BS and its service function:
1- Old ENs are replaced by new ENs at the higher service layer, while the old ENs move to
the layer below. The same process is repeated across consecutive layers, as exhibited by
Figure 4.12.
2- This step applies to CHs, where the replacement process across CHs is the same as that of
the ENs, as exhibited by Figure 4.13.
35


New ENs
Expired
ENs
Figure 4.12: Update the M-DUWSN and Replace the expired ENs
Figure 4.13: Update the M-DUWSN and Replace the expired CHs
36


4.5. Numerical Evaluation to Compute the Expected Delay for both WSNs and M-
DUWSNs
The main goal of our simulations is to examine the delay for the deployed RAA in
two cases: the on-ground and the underwater wireless sensor networks. To compute the delay, we
need to first generate a number of random Poisson arrivals as shown in Figure 4.14. Then, those
arrivals are transmitted by our K-cell algorithm. The algorithm transmits arrivals occurred within
a window of size A, while each new arrival first senses transmissions and outcomes, until the first
group of K consecutive non-collision slots occurs, at which point this arrival participates in the
next collision resolution interval (CRI).
Number oF Arrivals
Figure 4.14: Generating a random number of poisson arrivals
Figure 4.15 shows the results of applying the 2 cells algorithm. The first part of each
figure is Poisson random arrivals, the second part exhibits the process of sensing the two
consecutive NCs and updating the A window, and the third part exhibits the transmitted arrivals.
Each time, the figures in the middle and right side are changed to show how many arrivals still
remain in the updated A window and how many arrivals are transmitted successfully.
37


Magnfeida
Figure 4.15: The system keeps sensing until two consecutive non-collision slots detected, and a
A window of the next CRI is selected with its right most edge m-1 slot
38


1
Stats Trartsmrtletf Successful^
cfArmralt
Under Process Window Random Access AJggnlhm
Number of An rals L*ndeo Process Windows Random Access AJgwishrn Slots Transmuted Successful?
Figure 4.15: The system keeps sensing until two consecutive non-collision slots detected, and a
A window of the next CRI is selected with its right most edge m-1 slot
39


Number erf Atrrwda Under Process- Window Random Access AigariShrp Slols Transmlted Successfcdy
Tima Stofi
Number ofArrvl
Tim? Slo1
Unde* Proc*?** Window P^nckm Access A^r/hm
t
03
0.8
0?
, 06
;<5
: a.A
08
o .;
t
Tims SI01
Slfl* TransmiMe*^ SuctessMy
A
Is.
15 5 16 166
Trf> Slid
Figure 4.15: The system keeps sensing until two consecutive non-collision slots detected, and a
A window of the next CRI is selected with its right most edge m-1 slot
In this section, we compare the 3-cell algorithm with the 4-cell algorithm for both on-
ground and underwater M-DUWSNs, when all algorithms operate in environments imposing
admission delay constraints. We simulated the algorithms in the presence of the limit Poisson
user model for admission delay constraints equal to 100, 200 and 400 slots and we then computed
the average delays. Our results for the 3 and 4-cell algorithms, for both on-ground (WSNs) and
underwater networks (UWSNs) are exhibited in Figures 4.16 4.26. From these figures, we
40


observe that, for 100, 200, 400 slots admission delay constraints, the 3-cell algorithm does
somewhat better than the 4-cell in terms of expected delays of those packets that are successfully
transmitted (50,000 Slots and different rates from 0.1 to 0.43). Moreover, the expected delays for
underwater networks are higher than those for on-ground networks.
Expects Dels* for 3-Cell Vs 4-Cell
Admission Delays 100
Figure 4.16: Expected Delays for 3-cell Vs 4-cell algorithm Admission Delay 100 Slots
Expected Dels} for 3-Cell Vs 4-Cell
Admsssion Delay=200
Figure 4.17: Expected Delays for 3-cell Vs 4-cell algorithm Admission Delay 200 Slots
41


Expected Delay for 3-CeJI Vs 4-Cell
Figure 4.18: Expected Delays for 3-cell Vs 4-cell Admission Delay 400 Slots
Enpected Delay for 3-Cell
42


Expected Delay for 4-Ce
Figure 4.20: Expected Delays for 4-cell algorithm Admission Delay =100, 200 and 400 Slots
UWSN Expected Delay for 3-Cell Vs 4-Ce!l
Figure 4.21: underwater sensor networks-Expected Delays for 3-cell Vs 4-cell algorithm.
Admission Delay =100 Slots
43


LWSN Expected Delay for 3-Cs* Vs 4-Cell
Admission Delay=200
Figure 4.22:UWSNs- Expected Delays for 3-cell Vs 4-cell algorithm
Admission Delay =200 Slots
LWSN Expected Delay for 3-Cell Vs 4-Cell
,, in4 Adrni&SMm Delay=4lX!
Figure 4.23:UWSNs- Expected Delays for 3-cell Vs 4-cell algorithm Admission Delay =400
Slots
44


Figure 4.24:UWSNs- Expected Delays for 3-cell algorithm Admission Delay = 100, 200 and 400
Slots
UWSN Expected Dfltay for + Cell
Figure 4.25:UWSNs- Expected Delays for 4-cell algorithm Admission Delay =100, 200 and 400
Slots
45


UVVStJ Vs WSN Expected Dwy for 3-Cell
Figure 4.26: UWSNs Vs WSN- Expected Delays for 3-cell algorithm Admission Delay =100, 200
and 400 Slots
Expected Detay far 6-Cell and 7-Cell
Adrms5*ofi Defsy=2CG
Figure 4.27: Expected Delays for 6-cell Vs 7-cell algorithm Admission Delay =200 Slots
We now compare the 6-cell algorithm with the 7-cell algorithm for higher number of
arrivals (20,000 to 80,000) for both WSNs and UWSNs, when all algorithms operate in
environments imposing admission delay constraints. The number of used slots is about 200,000
slots for different lambda rates (0.29 to 0.43). We simulated the algorithms in the presence of the
limit Poisson user model for admission delay constraints equal to 200 and 400 slots and then we
compute the expected delays. Our results for 6 and 7-cell algorithm for WSNs and UWSNs are
46


exhibited in figures 4.27 4.32. From these figures, we observe that 7-cell algorithm perform
much better than the 6-cell algorithm for both WSNs and UWSNs for both 200 and 400 slots
admission delay constraints. Also, the 7-cell and 6-Cell performs better when the admission delay
equals to 400 than 200 slots especially when we transmit high number of arrivals.
Expected Delay lor 6-Call and 7-CeO
, m4 Admission 0alay=40Q
Figure 4.28: Expected Delays for 6-cell algorithm Admission Delay =400 Slots
LfWSN E*peded Delay for 6-Cell arid 7-Cel I
x -jO4 Admission Delay=2O0
Figure 4.29:UWSNs-Expected Delays for 6-cell Vs 7-cell algorithm Admission Delay =200 Slots
47


10
UWSN Expected Delay for 6-Cell and 7-Cell
X to* Admission Dsl3y=d0
Ratio (Lambda)
Figure 4.30:UWSNs-Expected Delays for 6-cell Vs 7-cell algorithm Admission Delay =400 Slots
WSN Vs WSN Expected Delay for 6-Cell and 7-Cell
, m* Admission Delay=2D0
Figure 4.31:UWSNs Vs WSN-Expected Delays for 6-cell Vs 7-cell algorithm Admission Delay
=200 Slots
48


Figure 4.32:UWSNs Vs WSN-Expected Delays for 6-cell Vs 7-cell algorithm Admission Delay
=400
hi this part, a comparison between the 7-cell algorithm and the S-Aloha, [58], is
presented. The expected delay for 1000 transmitted arrivals is computed for the 7-cell algorithm
and compared to that induced by the S-Aloha protocol. The used lambda rate in this case is equal
to 0.4 and the admission delay constraint equals 400 slots. It is also assumed each slot is equal to
20 Micro Sec, [56-58], because most of the references use the per second rather than the per slot
time units. We thus converted the y-axis to (Sec) instead of (Slot) in our comparative evaluation.
Our results in Figure 4.33 show that the proposed K-cell model outperforms the S-Aloha protocol
[58],
49


Expe-ctes Delay using k-calts AJg&ftthm Vs S-ALQHA
Figure 4.33: Average Access Delay for 7-Cell (Admission Delay =400 and Lambda=0.4), ATL
S-ALOHA, and S-ALOHA protocols.
50


5: Conclusions
This thesis focuses on the impact of latency on Mobile Underwater Sensor Networks
(MUWSNs) operating in the underwater acoustic environment. The deployment of a novel
MUWSN that consists of mobile Elementary Nodes (ENs), a backbone network of Cluster-Heads
(CHs), and a Base Station (BS) has been proposed. We also provide an MUWSN node clustering
technique that attains high data rate and low energy consumption; the MUWSN utilizes the audio
signal emitted by the submarines engine to detect its presence.
During the implantations of the MUWSN, sensor localization is required. Thus, the
improvement of localization accuracy has been considered, while such accuracy could be
significantly affected by poor acoustic channel quality and node current-caused mobility. To
overcome problems such as: interference, power savings, data rate constraints and propagation
delay, we reorganized our network in clusters where each cluster has a CH and many ENs. In
general, the network design may be dictated by the objective of its application, in conjunction
with the constraints induced by the deployed transmission protocols and the environment (Power,
transmission rates, sensors life span,...etc.).
We presented a limited feedback sensing Random Access Algorithm (RAA), as the data
transmission protocol in each cluster. The algorithm utilizes binary (C-NC) feedback and a
slotted broadcast channel, while each sensor is required to sense the channel continuously from
the time it generates a packet to the time when this packet is successfully transmitted. We also
presented the analysis of the algorithm in the presence of the Poisson used model. We examined
the delays induced by the proposed RAA in two cases: when deployed on-ground; and when
deployed by underwater wireless sensor networks, where large propagation delays are inherent to
acoustic such networks. In both cases, we modeled the sensors limited life span by imposed
admission delay constraints in data transmissions. In each case, system evaluation was
represented by simulations, where such results were used to compare on-ground versus
underwater performance.
Finally, a perfonnance comparison between the proposed algorithm and the S-Aloha
algorithm was presented. The comparison encompassed expected delays and rejection rates as
the performance metrics. The results showed that as the rate of the limit Poisson traffic and the
51


imposed limit on the admission delay increase, the S-Aloha-type protocol collapses, while the
introduced underwater protocol maintains high standards.
52


BIBLIOGRAPHY
[1] Z. Jiang Underwater Acoustic Networks International journal of Intellgent Control and
Systems, Vol. 13, NO. 3, Sep 2008, 152-161
[2] J. Yick, B. Mukheijee, and D. Ghosal Wireless sensor network survey, Department of
Computer Science, University of California, Davis, CA 95616, United States
[3] J. Partan, J. Kurose, and B. N. Levine, A Survey of Practical Issues in Underwater Networks,
International Conference on Mobile Computing and Networking, Proc. of the 1st ACM
international workshop on Underwater networks, pp. 17 24, Los Angeles, CA, USA, Sept.,
2006.
[4] L. Berkhovskikh and Y. Lysanov, Fundamentals of Ocean Acoustics. New York: Springer,
1982.
[5] A. Quazi and W. Konrad, Underwater acoustic communications, IEEE Commun. Mag., pp.
24-29, Mar. 1982.
[6] R. J. Urick, Principles of Underwater Sound, 3rd Edition, McGraw-Hill Publishing
Company, New York, NY, 1983.
[7] R. Coates, Underwater Acoustic Systems, New York: Wiley, 1989.
[8] J. A. Catipovic, Performance Limitations in underwater acoustic
telemetry, IEEE J. Oceanic Eng., 15(3):205-216, Jul. 1990.
[9] E. M. Sozer, M. Stojanovic, and J. G. Proakis. Undersea Acoustic Networks. IEEE Journal of
Oceanic Engineering, OE-25(l):72-83, January 2000.
[10] http://www.cse.uconn.edu/~jcui/UWSN_papers/UCONN_CSE_UbiNet-TR05-02.pdf
[11] S. Zhou and P. Willett, Submarine Location Estimation Via a Network of Detection-Only
Sensors. IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 6, JUNE 2007
[12] P. Wang, C. Li, J. Zheng Distributed Minimum-Cost Clustering Protocol for Underwater
Sensor Networks (UWSNs). ICC 2007 proceedings.
[13] C. Wen-Yu, L. Jing-Biao UUV Mobility Management for Underwater Acoustic Sensor
Networks
[14] J. Cui, J. Kong, M. Gerla S. Zhou The Challenges of Building Scalable Mobile
Underwater Wireless Sensor Networks for Aquatic Applications
[15] T. and H. Lin, Research on Submarine Counter-detection Probability by Threat Surface
Ship Sonar. IEEE.
[16] A. C. KELLER, Submarine Detection by Sonar. IEEE.
[17] A.T. Burrell and P. Papantoni-Kazakos, Performance Monitoring in Sensor Networks,
IEEE.
53


[18] Carp Fish Robot, http://www.youtube.com/watch?v=e09oseiCTdk.
[19] Recycled Robot Fish Teach Japanese Children About Sealife, http://inhabitat.com/recycled-
robot-fish-teach-japanese-children-about-sealife/
[20] Video: Mini fuel cell powering a robot fish, http://www.crunchgear.eom/2010/03/16/video-
mini-fuel-cell-powering-a-robot-fish/
[21] Robotic fish consume less power by using magnets instead motors,
http://www.robaid.com/bionics/robotic-fish-consume-less-power-by-using-magnets-instead-
motors.htm
[22] Vinay K. Ingle and John G. Proakis, Digital Signal Prosseing Using Matalb.
[23] D. Kazakos and P. Papantoni-Kazakos, Detection and Estimation, Computer Science Press,
New York, NY, 1990.
[24] I. F. Akyildiz, W.Su, Y. Sankkarasubramaniam and E Cayirci, Wireless Sensor Networks:
A Survey, Computer Networks (Elsevier), vol. 38, no. 4, 2002, pp. 393-422.
[25] S. Bandyopadhyay and E.J. Coyle, An Energy Efficient Hierarchical Clustering Algorithm
for Wireless Sensor Networks, IEEE INFOCOM 2003.
[26] M. Bhardwaj and A.P. Chandrakasan, Bounding the Lifetime of Sensor Networks via
Optimal Role Assignments, in Proc. IEEE INFOCOM, New York, Jun. 23-27, 2002, pp. 1587-
1596.
[27] D. Blough and S. Paolo, Investigating Upper Bounds on Network Life-Time Extension for
Cell-Based Energy Conservation Techniques in Stationary Ad Hoc Networks, in Proc. ACM
MobiCom, Atlanta, GA, Sep. 23-28, 2002, pp. 183-192.
[28] J.H. Chang and L. Tassiulas, Routing for Maximum System Lifetime in Wireless Ad Hoc
Networks, in Proc. 37th Annual Allerton Conference on Communications, Control and
Computing, Monticello, IL, Sep. 1999, vol. 1, pp. 22-31.
[29] J.H. Chang and L. Tassiulas, Energy Conserving Routing in Wireless Ad Hoc Networks,
in Proceedings of IEEE INFOCOM, Tel Aviv, Israel, Mar. 26-30, 2000, pp. 22-31.
[30] Y. T. Hou, Y. Shi and H.D. Sherali, On Node Lifetime Problem for Energy-Constrained
Wireless Sensor Networks, ACM/Springer Mobile Netw.Applicat., vol. 10, no. 6, Dec. 2005, pp.
865-878.
[31] Y.T. Hou, Y. Shi, J.H. Reed and K. Sohraby, Flow Routing for Variable Bit Rate Source
Nodes in Energy-Constrained Wireless Sensor Networks, in Proc. IEEE International
Conference on Communications, Seoul Korea, May 16-20, 2005, pp. 3057-3062.
[32] Y. T. Hou, Y. Shi and H.D. Sherali, Rate Allocation and Network Lifetime Problems for
Wireless Sensor Networks, IEEE Trans. Networking, vol. 16, no. 2, April 2008, pp. 321-334.
54


[33] P. Papantoni Kazakos and A.T. Burrell The Implementation of Dynamic Rate Allocation in
Sensor Networks Journal of Intelligent and Robotic Systems, 2010, Vol. 58, pp. 211-238.
[34] V. Rodoplu and T.H. Meng, Minimum Energy Mobile Wireless Networks, IEEE J. Sel.
Areas Commun., vol. 17, no. 8, Aug. 1999, pp. 1333-1344.
[35] V. Rodoplu and M. Park, "An energy-efficient MAC protocol for underwater wireless
acoustic networks," in MTS/IEEE OCEANS, 2005.
[36] A. F. Harris, M. Stojanovic, and M. Zorzi, "Why underwater acoustic nodes should sleep
with one eye open: Idle-time power management in underwater sensor networks," in ACM
International Workshop on UnderWater Networks (WUWNet), 2006.
[37] A. F. Harris and M. Zorzi, "Energy-efficient routing protocl design considerations for
underwater networks," in IEEE Conference on Sensor; Mesh and Ad Hoc Communications and
Networks (SECON), 2007.
[38] Lanbo Liu, Shengli Zhou, and Jun-Hong Cui, Prospects and Problems of Wireless
Communication for Underwater Sensor Networks. Wiley WCWC Special Issue on Underwater
Sensor Networks.
[39] Milica Stojanovic, Underwater Acoustic Communication, For the Wiley Encyclopedia of
Electrical and Electronics Engineering. Department of Electrical and Computer Engineering
Northeastern University Boston.
[40] N. N. Ismail, L. A. Hussein, S. H.S.Ariffin, AnalyzingThe Performance of Acoustic
Channel in Underwater Wireless Sensor Network(UWSN), IEEE 2010 Fourth Asia International
Conference on Mathematical/Analytical Modeling and Computer Simulation, Page(s): 550 555.
[41] Xianhui Che, Ian Wells, A Static Multihop Underwater Wireless Sensor Network Using RF
Electromagnetic Communications 29 IEEE International Conference on distributed computing
system
workshop, Swansea metropolitan university UK, 2009.
[42] Jun-Hong Cui, Jiejun Kong, Mario Gerla, Shengli Zhou, Challenges: Building Scalable and
Distributed Underwater Wireless Sensor Networks (UWSNS) For Aquatic Applications .2005.
[43] Dario Pompili, Tommaso Melodia, Ian F. Akyildiz, routing algorithm for delay-insensitive
and delay-sensitive application in underwater sensor networks, in Proc. of ACM Conference on
Mobile Computing and Networking (MobiCom), Los Angeles, CA, September 2006
[44] Mari Carmen Domingo, Rui Prior, Energy Analysis Of Routing Protocol For Underwater
Wireless Sensor Networks, 2007
[45] Ian F. Akyildiz, Dario Pompili, Tommaso Melodia, Challenges for Efficient Communication
in Underwater Acoustic Sensor Networks.ACM Sighed Review, Vol. 1, Number 2, July 2004
55


[46] Dario pompili, Tommaso Melodia, lan F,akyildiz, Deployment Analysis In Underwater
Acoustic Wireless Sensor Network International Conference on Mobile Computing and
Networking ,2006.
[47] Paul C.Etter, Underwater Acoustic Modeling Principle, Techniques And Applications.
Elsevier Science Publishers, 1991.
[48] http://www.windows.ucar.edu/tour/link=/earthAVater/salinity_depth.h tml
[49] http ://www. windows .ucar.edu/tour/link=/earthAVater/temp .html
[50] J.L. Massey, Collision resolution algorithms and random access communications, in
Multiuser Communication, G. Longo Ed, New York, Springer-Verlag.
[51] B.S. Tsybakov and V.A. Mikhailov, Free synchronous packet access in a broadcast
channel with feedback, Problemy Peredachi Informatsii, Vol. 14, No. 4, pp. 259-280, 1978.
[52] L. Georgiadis and P. Papantoni-Kazakos, A 0.487 throughput limited sensing algorithm,
IEEE Trans. Inf. Th., Vol. IT-33, No. 2, pp. 233-237, Mar. 1987.
[53] M. Paterakis and P. Papantoni-Kazakos, A simple window random access algorithm with
advantageous properties, IEEE Trans. Inf. Th., Vol. 35, pp. 1124-1130, Sept. 1989.
[54] J. I. Capetanakis, Tree Algorithms for Packet Broadcast Channel, IEEE Trans.
Information Theory, Vol. IT-25, pp. 505-515, Sept. 1979.
[55] M. Paterakis, L. Georgiadis and P. Papantoni-Kazakos, On the relation between the finite
and the infinite population models for a class of RAAs, IEEE Trans. Commun., Vol. COM-35,
pp. 1239-1250, Nov. 1987.
[56] Chatzimisios, P.; Boucouvalas, A.C.; Vitsas, V., Packet delay analysis of IEEE 802.11
MAC protocol, IET Journals, Volume: 39, Page(s): 1358 1359.
[57] Guolin Sun; Bo Hu, Delay Analysis of IEEE 802.11 DCF with Back-off Suspension,
Future Generation Communication and Networking (FGCN 2007),Volume: 1, Page(s): 148 -
151.
[58] Rivero-Angeles, M.E.; Lara-Rodriguez, D.; Cruz-Perez, F.A., Access delay analysis of
adaptive traffic load type protocols for S-ALOHA and CSMA in EDGE, Wireless
Communications and Networking, 2003. IEEE, Volume: 3, Page(s): 1722 1727 vol.3
56


Full Text

PAGE 1

SUBMARINE DETECTION USING UNDERWATER ACOUSTIC SENSOR NETWORK S by Gaith A. Tellis i B.S., Al Fateh University, 2004 A thesis submitted to the University of Colorado Denver in partial fulfillment of the requirements for the degree of Master of Science Electrical Engineering 2012

PAGE 2

II This thesis for the Master of Scienc e degree by Gaith A. Tellis i has been approved by Titsa Papanton i Yiming J. Den g Fernando Mancilla-Davi d Oct 19, 201 2

PAGE 3

III Tellisi, Gaith A. (Master of Science, Communications Eng i neering, Electrical Engineerin g) Submarine Detection using Underwater Acoustic Sensor Networks Thesis directed by Professor Titsa Papanton i ABSTRAC T It is well known that submarines can hide successfully from traditional sonar systems. For this reason, we propose a new technique implemented via the utilization of mobile robots comprising Distributed Underwater Sensor Networks (DUWSNs). I n the design and operation of DUWSNs, the arising challenging issues are: (1) the establishing of reliable communication links among the sensor nodes; (2) the maximization of the life time of the network; (3) the sustaining of sufficient data rate within t he network; (4) the architecture of the DUWSN and the routing policies deployed; (5) the effectiveness of the DUWSN, regarding submarine detection and identification. This thesis contributes towards an understanding of, and solution to, the impact of the l atency on DUWSNs in underwat e r acoustic environmen ts Moreove r a new underwater KCells transmission protocol is implemented to overcome and exploit the large propagation delay inherent to acoustic underwater networks. We develop a novel DUWSNs that con s ists of mobile elementary nodes (ENs), a backbon e network of clusterheads (CHs) and a Base Station (BS). We provide a DUWSN node clustering technique that attains high data rate and l o w energy consumption; the DUWS N utiliz es the audio signal emitted by the submarines engine to detect its presence. We also present some simulation results using MATLAB. The form and the content of this abstract are approved. I recommend its publication. Approved by: Prof. Titsa Papantoni

PAGE 4

IV DEDICATIO N Thanks God for giving me the strengt h to finish this journey.. This thesis is dedicated to my passed away Dad and to my wonderful Mom who have supported me all the way since the beginning of my studies ; To my brother Eha b, who has done more than enough to help me during my studies, and has been a great source of motivation and support; Finally, many thanks go to my sister Rash a, who protected my Mom during the hard time that my country went thro u gh

PAGE 5

V ACKNOWLEDGEMENT S I owe my gratitude to several individuals who have made this dissertation possible by either their contribution or valuable assistance in the preparation and completion of this study. First and foremost, I want to express my gratitude to my advisor Dr. Titsa Papantoni whose motivation and encouragement I will never forget. It has been an honor to be one of her M asters students. I appreciate all her contributions of tim e an d ideas to m a ke my research happen. I am also grateful to Dr. Yiming J. Deng with whom I have inte rac te d the most during the course of my graduate studies for his encouragement and helpful advice. Lastly, I would like to thank my family for all their love and encourag ement

PAGE 6

VI TABLE OF CONTENT S Chapter 1. Introduction and Motivatio n..... 1 1.1 Thesiss Roadmap ... ... 3 2. Background ... ....... 4 2.1 Characteristics of Underwater Acoustic Channels. 4 2.2 Differences between M UWSNs and ground Sensor Networks .. 5 3. Mobile Distribute d Underwater Sensor Networks Structur e ... 6 3.1. Underwater Positionin g ... ... ....... 6 3. 2 System Model and Problem Formalization ... 6 3.3 Design Methodology for the M DUWSN 8 4. Acoustic Channel in Underwater Wireless Sensor Network 19 4.1. Related Work 19 4.2. Underwater Acoustic Channel 2 0 4.2.1. Underwater Environment ... ....... 2 0 4.2.2. Propagation Delay . 22 4.3. Transmission Protocol (Random Access Transmission Protocol ) .. .. .... 2 4 4.3.1. The system model and the algorithm s ... ......... 25 4.3.2. Underwater Signal Propagation 3 0 4.4. Energy Efficient and Delay Decreasing Routing Protocol .. ..... 3 1 4.5. Numerical Evaluation to Compute the Expected Delay for both WSNs and D MWSNs ............................ 37 5. Conclusion s .. 5 1 Bibliography 53

PAGE 7

VI I LIST OF FIGURES Figur e 3.1. The structure of the Distributed Mobil Fish Network Underwater Sensor Networks (DMF N UWSN) and Submarine ... 7 3.2. The general hierarchical structure of the Distributed Mobile Sensor Network Distributed Underwater Sensor Networks (DUWSNs ) 8 3.3. The architecture of the Distributed Mobile Sensor Network Distributed Underwate r Sensor Networks (DUWSNs ) ... ................... .... 1 0 3.4. The reorganization of all ENs around each CH ..... 1 1 3.5. Some example of some other cluster shapes .. 1 1 3.6. The architecture between odd and even layers .. 1 2 3.7. The architecture of the reorganized Distributed Mobile Network Distribute d Underwater Sensor Networks (DUWSNs ) 1 3 3.8. The architecture of the Distributed Mobile Network Distributed Underwater Senso r Networks (DUWSNs ) ... .. .. 1 4 3.9. The architecture of the reorganized Distributed Mobile Network Distribute d Underwater Sensor Networks (DUWSNs ) .. 1 4 3.10. The architecture of a different reorganized Distributed Mobile Network Distributed Underwater Sensor Networks (DUWSNs) ..... 1 5 3.11. The architecture of the reorganized (DUWSN) and passing by submarine ... . 1 6 3.12. Submarine Detection using the proposed SDCA UVAD and routing the alarm to the BS .. 17 3.13. Submarine Detection using the proposed SDCA UVAD and routing the alarm to the BS .. 17 3.14. The architecture of the reorganized (DUWSN) and passing by submarine (a) First position (b) Second position (c) Third position .... 18 4.1: Standard salinity profile [41] . 2 3 4.2: Standard Temperature Profile [42 ] 2 3 4.3: Delay between two sensors with distance 100m using Equations 1 4 .. ... .. 2 4 4.4: The k cells algorith m ..... 26 4.5: The operation of the 2 cell algorithm (Imaginary Stack device ) .. ... .. 29

PAGE 8

VII I 4.6: The 2 cell algorithm (Imaginary Stack device & Arrival Time) .. ... ... 29 4.7: The architecture of the M DUWSNs . 3 1 4.8: The shortest path direction depends on the BSs location .. 33 4.9: The shortest pat h direction toward the BS is chose n . 34 4.10: Compare the delay between the two cases: (a) The shortest path direction towar d the BS is chosen (b) a random path toward the BS is chosen to forward the alarm 34 4.11: Compare the delay between the two cases: (a) The shortest path direction towar d the BS is chosen (b) a random path toward the BS is chosen to forward the alarm 35 4.12: Update the M DUWSN and Replace the expired ENs 36 4.13: Update the M DUWSN and Replace the expired CH s ... ..... 36 4.14: Generating a random number of Poisson arrivals... ...... ...... 37 4.15: The system keeps sensing until two consecutive no n collision slots detected, and a

PAGE 9

IX Slots . 47 4.30: UWSNs Expected Delays for 6 cell Vs 7 cell algorithm. Admission Delay =40 0 Slot s . 48 4.31: UWSNs Vs WSN Expected Delays for 6 cell Vs 7 cell algorithm. Admissio n Delay =200 Slots ... ... 48 4.32: UWSNs Vs WSN Expected Delays for 6 cell Vs 7 cell algorithm. Admissio n Delay =400 Slots . 49 Figure 4.33: Average Access Delay for 7 Cell (Admission Delay =400 and Lambda =0.4),ATL S ALOHA, and S ALOH A protocols. 50

PAGE 10

X LIST OF TABLES Tabl e 1 Comparison of the acoustic, radio wave and optical communication in sea water [31] .. ... 2 1 2 2 Cell Algorithm ..... ..... ........... ... 28 3 Communication rules between ENs, CHs, and BS .... ........ ........... .. 3 2

PAGE 11

1 1: Introduction and Motivation In the last few years, significant research effort has been dedicated in the area of communications within Wireless Sensor Networks (WSNs). In this thesis, we focus on research on Underwater Wireless Senso r Networks (UWSNs), since they are deployed fo r tasks such as military monitoring, ocean data collection, disaster detection, oilfield monitoring, etc. Sensor networks have been envisioned as powerful solutions for several applications, [ 13], such as: i) Surveillance, including reconnaissance, targeting, and intrusion detection; ii) Environmental monitoring, where pollution in nearshore oceans is an urgent issue and needs close watch; iii) Underwater explorations, where such explorations are difficult for human beings due to the high water pressure, unpredi c table underwater activities and vast size of unknown area. UWSNs may also help in the exploration of underwater minerals and oilfields, as well as in determining routines for laying undersea cables, etc; iv) Disaster prevention via continuous underwater monitoring. For viability of their deployment, the UWSNs must be able to coordinate their operation by exchanging information. Numerous research activities are currently focusing on the development of network protocols for wireless sensor networks, where n ew efficient and reliable data such communication technologies are demanded by UWSNs, due to the unique and challenging characteristics of the underwater acoustic communication channel, including limited bandwidth capacity and variable delays. The major requirements in the design of reliable Mobile Underwater Sensor Networks (MUWSNs) systems include: (1) Limited power, since underwater systems are usually designed for relatively lon gterm monitoring. (2) Sensor localization, for effecti v e monitoring. (3) Reduced delays, to secure effective communication. In this thesis, a single interesting application has been chosen on behalf of UWSN deployment. In particular, the selected application corresponds to military target tracking, name l y to submarine detection. Usually, modern submarines are protected by the newly developed Low Probability Detection technology LPD [9]. To reduce a submarines acoustic signature, the hull of the submarine is shelled with rubber ant isonar protection til e s that reduce acoustic detection. The LPD technology hides the submarine's acoustic signature to thwart active

PAGE 12

2 SONAR probing and also reduces the intra-submarine noise to foil passive SONAR listening. The submarine's noise level is comparable to ocean's background noise, thus legible acoustic signature may only be collected within a very short distance from the submarine. Considering this kind of countermeasure, MUWSNs become an ideal solution for submarine detection. A number of underwater robots and a large amount of underwater sensor nodes can be ai rdropped to the venue. In real time, each sensor node monitors local underwater activities and reports sensed data via multihop acoustic routes to a distant command center. The new feature of this type of MUWSN systems (as compared to previous MUWSN systems for estuary monitoring) may be the addition of a builtin Underwater Voice Activity Detector per each UWSN, to avoid false alarms. This work focuses on mobile underwater sensor networks ( MUWSNs) a rchitectures consist in g of a large number of low cost sensors that can move with water current and dispersion. Those MUWSNs must be organized efficiently, to transmit/receive data with minimal delay. The underwater communication links connectin g those underwater devices are also designed to update effectively the control unit in the main station with curren t collected data. Such desig n may also help in energy conservation, since the n th e induced reduction in collisions results in increase of the sensors life span. Unlike traditional Wireless Sensor Networks (WSNs), Underwater Sensor Networks (UWSNs) induce special design and resources constraints such as limited amount of energy, short communication range, low bandwidth, as well as limited processing and storage space in each network [2]. UWSNs present a new research topic, introducing several new problems, due to the unique properties of the underwater environment. An underwater acoustic channel (chapter 3) is different from a groundbased radio channel in many aspects, including [1] : 1) Bandwidth is extremely limited. The attenuation of acoustic signa ls increases with frequency and range [4] [5] [7]. Consequently, the feasible band is extremely small. For example, a short range system operating over several tens of meters may have available bandwidth of a hundred kHz; a mediumrange system operating over sev e ral kilometers has a bandwidth in the order of ten kHz; and a longrange system operating over several tens of kilometers is limited to only a few kHz of bandwidth [8] [7]. 2) The propagation delay is long. The transmission speed of acoustic signals in salty water is around 1500 meter/s [6], which is five times o f magnitude lower than the speed of electromagnetic wave in free space. Correspondently, propagation delay in an underwater

PAGE 13

3 channel becomes signific a nt. This is one of the essential characteristics of underwater channels and has profound implications on localization and time synchronization. 3) The channel characteristics vary with time and highly depend on the location of the transmitter and receiver. T he fluctuation nature of the channel causes signal distortions. 4) The bit error probability is much higher, while temporary loss of connectivity sometimes occurs, due to the extreme characteristics of the channel. The practical deployment and design of UWSNs face some special challenges: First, the cost of manufacturing, deployment, maintenance and recovery of underwater equipments is much higher than that of thei r groundbased counterpar ts Second, energy saving/efficiency is a critical issue for UWSNs. 1.1 Thesiss Roadma p The remainder of this dissertation is organized as follows: In Chapter 2 some background on underwater sensor networks is presented. First, the characteristics of underwater acoustic channels are reviewed. Then, the distinctions between MUWSNs and ground sensor networks are discussed Chapter 3 focuses on the structure of Mobile Distribute dUnderwater Sensor Networks ( MDUWSNs) and is organized as follows: Section 3.1 introduces the factors that could affect the position of each mobile elementary node (EN). Section 3.2 presents the structure of the proposed M-DUWSN. The proposed model and some results are demonstrated in Section 3.3. Chapter 4 focuses on the Acoustic Channel in Underwater Wireless Sensor Networks (UWSNs). The ma i n objective in this chapter is to enhance the performance of the acoustic channel by decreasing the propagation delay. This is attained via the deployment of a novel underwater Random Access Algorithm. This chapter is organized as follows: Section 4.1 pr e sents some related work. In Section 4.2, the underwater acoustic channel is analyzed. Section 4.3 introduces a novel transmission protocol for both WSNs and UWSNs. Section 4.4 focuses on energ yefficiency and delay decrease via the deployment of a routin g protocol. In Section 4.5, simulation results are presented

PAGE 14

4 2: Background 2.1 Characteristics of Underwater Acoustic Channel s Underwater acoustic channels are temporally variable due to the nature of the transmission medium and the physical properties of the environments [10]. Radio and optical waves are considered as candidates for wireless communication in underwater environments. However, neither of these latter waves is appropriate for MUWSNs. The first reason is that the radio waves suffer from high attenuation in salty water. Radio Frequenc y (RF) signals can propagate at a long distance through conductive water only at extra low frequencies, 3 0300 Hz, which requir es large antennae and high transmission power [4]. Thus, the hi g h attenuation of radio waves in water makes them infeasible for MUWSNs. Second, optical signals usually suffer from the shor trange and line-ofsight problems. Moreover, optical signals are severely impaired by scattering when used in underwater, and the transmission of optical signals requires high precision in communication. In short, it is widely accepted that acoustic channels are the most appropriate physical communication links for underwater environments. As c ompared to radio channels, underwater acoustic channels have the following unique features. (1) Long propagation delays: The propagation speed of sound in water is about 1500m/s, which is 5 times of magnitude lower than that of radio, m/sec. Moreover, the propagation speed of sound in water is affected by many properties of the water medium such as temperature and salinity. Such low propagation speed of acoustic signals results in a very long propagation even in a short distance. (2) Low accessible bandwidth: The available bandwidth of underwater acoustic channels is very limited due to absorption and most acoustic systems operate below 30 kHz, which is extremely low compared with radio networks. Moreover, the available bandwidth of ac oustic channels depends on both transmission range and frequency. As a result, the achievable bitrate in acoustic channels is also determined by both transmission range and frequency. (3) High channel error rates: The quality of acoustic channels is affec t ed by some factors such as sign a l energy loss and noise. Energy loss comes from signal attenuation and geometry spreading, where signal attenuation is caused by the absorption of acoustic energy (which increases with distance and frequency), and geometry s preading is caused by the expansion of the wave fronts. Noise, ma nmade noise or ambient noise, clearly disturbs the channels, mult ipath possibly introduces inter symbol interference (ISI). All these factors contribute to the high error rates of acoustic channels.

PAGE 15

5 In short, underwater acoustic channels are featured with large propagation delays, limited available bandwidth and high error rates. Furthermore, the bandwidth of underwater acoustic channels is determined by both the communication range and fr e quency of acoustic signals. The bigger the communication range, the lower the bandwidth of underwater acoustic channels. 2.2 Differences between MUWSNs and ground Sensor Networks The main differences between the MUWSNs and ground sensor networks incl u de the following [9 ]: Communication Method: as explained before, RF channels do not work well in underwater environments. Instead, acoustic channels are the practical communication method for MUWSNs. The unique features of underwater acoustic channels: l ong propagation delays, low communication bandwidth and high error rates, make the protocols proposed for ground sensor networks unsuitable for MUWSNs. Cost: while ground sensor networks are expected to become inexpensive, underwater sensors are expensiv e devices. This is mainly due to the small relative number of suppliers. Deployment and e nvironments: An MUWSN is usually deployed in 3-dimensional space, whereas most ground sensor networks work in 2dimensional space. The 3dimensional deployment makes many protocols (such as localization and routing) designed for ground sensor networks unsuitable for MUWSNs. Furthermore, the physical environments of MUWSNs are much harsher than thos e of terrestrial sensor networks. Power: the required power for the a coustic underwa t er communications is higher tha t in ground radio communications because of the different physica llayer technology, the longer distance, and the more complex signal processing techniques implemented at the receivers. Memory: while ground s e nsors have very limited storage capacity, underwater sensors require some data cashing ability, since the underwater channel may be intermittent.

PAGE 16

6 3: Mobile Distribute dUnderwater Sensor Networks Structure Underwater Wireless Sensor Networks (UWSNs) represent a promising approach to a wide range of potential applications, both civilian and military [1 113]. In particular, UWSNs are becoming increasingly important in oceanographic applications as well as in monitoring efforts of the undersea world: UWSNs facilitate unmanned underwater explorations and larg escale underwater monitoring with reduced implementation cost and increased operational frequency [14]. However, the lifetime of a UWSN is largely restric t ed by the energy constraints of the sensor nodes and the harsh conditions of the underwater acoustic transmission channels. In this chapter, we develop a mobile distributed UWSN clustered architecture, named Mobile Distributed Underwater Wireless Sensor N e twork ( MDUWSN), that prolongs the life time of the network, by increasing the efficiency of the per sensor energy utilization. We first consider the problem of determining the network mobility management scheme that conserves energy and also allows for e asy sensor replacement. The chapter is organized as follows. Section 3.1 introduces the factors that could affect the position of each mobile Elementary Node (EN). Section 3.2 presents the structure of the proposed MDUWSN. The proposed model and some re s ults are demonstrated in Section 3.3. 3.1. Underwater Positionin g Many ocean observation and surveillance applications require accurate position knowledge of the mobile elementary nodes (ENs). Those ENs communicate with each other by messages exchange, to attain position estimation with subsequent possible position updating, as may be required. There are several factors that might affect the spatial distribution of the underwater network such as current, temperature, depth, salinity, pressure, as well a s the deployment of a bad communications system. Therefore, the underwater channel differs from node to node. 3.2. System Model and Problem Formalization The general operational architecture of the MDUWSNs is depicted in Figure. 3.1, where sensors are s ymbolized by fish and clusters a re symbolized by fish swarms. The objective of the DUWSN is the detec t ion of adversarial submarines. The architecture of the DUWSN is comprised of c l usters of Elementary Nodes (EN s ), where each cluster contains a Cluster Head (CH) and where CHs are connected with a Base Station (BS) via a backbone network. The BS is

PAGE 17

7 basically a mainframe computer located at a carrier. The issues arising in the proposed MDUWSNs architecture are: (1) the establishing of reliable communication links among the sensor nodes; (2) the maximization of the life time of the network; (3) the sustaining of sufficient data rates within the network; (4) the routing policies deployed in the backbone network; (5) the effective submarine detection and identification Figure 3.1. The structure of the Distributed Mobil Fish Network Underw ater Sensor Networks (DMF N-UWSN) and Submarine The ENs, CHs and BS perform the following functions: (1) The ENs are grouped i nto distinct clusters, where each cluster contains a single CH. Each EN collects local data and transmits them to its local CH. Some constructions of robot ENs are discussed in [18 ][2 1]. (2) Each CH collects the data sent by its local ENs and processes t h em, using an operation determined by the network signal processing objectives. The CH then processes the compounded processed data, utilizing an operation that is determined by the network signal processing objectives, and transmits the outcome to selected neighboring CHs and/or the BS. The CHs have processing capabilities, energy and lif espans that are much higher than those of the ENs, where their lif espans and energy may still be limited. (3) BS fuses data transmitted to it by neighboring CHs, utilizi ng an operation that is determined by the network signal processing objectives. The BS has practically unlimited lif espan [17 ]. (4) All sensor nodes (ENs) and cluster heads are distributed randomly underwater per uni tsize region (Ocean Depth X Spreading Length). (5)

PAGE 18

8 Each EN and CH can move freely and are assumed to be unmanned robots. (6) Each mobile CH has a specific communication range that includes all ENs in the area of interest. The CHs are organized in layers by ocean depth, where each CH can only c ommunicate with the CHs in the layer directly above its own. As an only exception, the CHs located in the highest layer (those closest to the ocean service) can communicate with each other. 3.3 D esig n M ethodology for th e MDUWS N The system incorporates many sensors where each sensor has multitasks such as: (1) collecting data; (2) transmitting; (3) receiving; (4) processing (5) moving. Usually, both transmitting and moving are the most expensive in terms of energy consumption. Ef f icient techniques for mult iaccess and data forwarding play a significant role in reducing energy consumption. Also, a mechanism may be needed to dynamically control the mode of the sensors (switching between sleeping mode, wak eup mode, and working mode) to possibly conserve more energy. In addition, when sensors run out of battery power, they should be able to pop up to the water surface for recharging or replacing, [14]. Figure 3.2. The general hierarchical structure of the Distributed Mobile Sen so r Network Distributed Underw ater Sensor Networks (DUWSNs )

PAGE 19

9 In mobile DUWSNs, sensor localization is needed because most of the underwater networks sensors are mobile with the current. To find the locations of mobile sensors in aquatic environments is very challenging. At the same time, the improvement of localization accuracy needs to be considered, while such accuracy could be significantly affected by poor acoustic channel quality and node mobility. Figures 3.1 and 3.2 exhibit the localization proce ss. The steps needed for the construction of the general hierarchical structure are as follows: (1) T he number of layers is calculated based on the depth distance between the oceans surface and its bottom. (2) The cluster heads (CHs) are inserted in ea c h layer as shown in Figure. 3.3. where each odd layer has (M) cluster heads and where each even layer has ( M1) cluster heads; their specific locations per layer are assigned via the use of equations (3.1) and (3.2) Odd Layer : Even Layer: (3 ) Insert the mobile sensor nodes (ENs) randomly as shown in Figure. 3.3. The sensor nodes and cluster heads are marked by black and yellow circles, respectively. Each EN knows the location of itse l f and its neighbour and is assigned a unique ID in the network. Data and alarms are collected by underwater local sensor nodes and relayed by some cluster heads. They are finally fused at the base station (BS) which is equipped with both acoustic and Radi o Frequency (RF) modems that can transmit data to the o n-shore command centre by radio or satellite.

PAGE 20

10 Figure 3.3. The architecture of the Distributed Mobile Sensor Network Distributed Underwater Sensor Networks (DUWSNs) (4) Reorganize the whole sensor population, so that each cluster head (CH) is surrounded by eight sensor nodes (ENs) in diameter d, as shown in Figure. 3.4, using Equations (3.3) and (3.4). Each one of the inner and outer circles has four ENs, while cons e cutive sensors in each circle are 90 degrees apart.

PAGE 21

11 Figure 3.4. The reorganization of all ENs around each CH Those sensors are randomly distributed, then reorganized as shown in Figure 3.3, i.e. the number of sensors and their locations in each cluster could be changed based on the objective of the network, and those could be also randomly distributed as shown in Figure 3.4. Figure 3.5. E xample of some other cluster shapes (5) Each cluster head (CH) can only communicate with its two closest cluster heads in the layer above its own (Cros ssection bet w een them as shown in Figure. 3.6 ). This implies that there is no connection between cluster heads that are located in the same layer (no crosssection)

PAGE 22

12 Figure 3.6. The architecture between odd and even layer s (6) The general r estructure of the proposed model is shown in Figure. 3.6, where the number of CHs and ENs vary, depending on the number of layers (Depth) and desired spreading length.

PAGE 23

13 Figure 3.7. The architecture of the reorganized Distributed Mobile Network Distributed UnderWater Sensor Networks (DUWSNs) To validate the effectiveness of our proposed scheme, Matlab is used to simulate the proposed model and test its mobility management strategy. In the simulation, there are more than 103-256 mobil e sensor nodes and 1 332 mobile cluster heads, first distributed randomly, as shown in Figure. 3.7. All sensors are then reorganized using Equations (1.1 )(1.4), as shown in Figure. 3.8 and 3.9.

PAGE 24

14 Figure 3.8. The architecture of the Distributed Mobile Network Distributed Underwater Sensor Networks (DUWSNs ) Figure 3.9. The architecture of the reorganized Distributed Mobile Network Distributed Underwater Sensor Networks (DUWSNs )

PAGE 25

15 Figure 3.10. The architecture of a different reorganized Distribu ted Mobile Network Distributed Underwater Sensor Networks (DUWSNs ) (7) Assign a random path to the underwater submarine. The submarine is detected as soon as the submarines engine sound starts reaching the ENs, as shown in Figure. 3.10. (8) As soon as some EN detects a suspicious sound, it sends an alarm to the CH and then to the Base station (BS), using an assigned path (See chapter 5)

PAGE 26

16 Figure 3.11. The architecture of the reorganized (DUWSN) and passing by submarin e Our objective is to detect the presence and trajectory of a submarine using the proposed MDUWSN. As mention earlier, an underwater voice activity detector is embedded in each sensor, to detect the submarines engine sound. Figure 3.11 exhibits the scenario where a submarine moves fro m the first to the second position. The clusters located in the vicinity of those positions will detect the sound of the engine, while pertinent alarms will be immediately transmitted to the BS using the routing protocol that will be explained in Chapter 4 Figure 3.12 exhibits the same detecting process via a different route. The simulation results using MATLAB are shown in Figure 3.13.

PAGE 27

17 Figure 3.12 Submarine Detection using the proposed SDCA-UVAD and routing the alarm to the BS Figure 3.13 Submarine Detection using the proposed SDCA-UVAD and routing the alarm to the BS

PAGE 28

18 Fig 3.14. The architecture of the reorganized (DUWSN) and passing by submarine: (a) First position (b) Second position (c) Third position

PAGE 29

19 4: Acoustic Channel in Underwater Wi r eless Sensor Networks Underwater acoustic wireless sensor networks (UWSNs) consist of a variable number of ENs that have the ability to communicate via an acoustic signal that has been sent through an underwater channel. The design of an acoustic underwater channel could be somehow challenging because of many constrains and limitations in underwater environments, such as large propagation delays, limited bandwidth and high bit error rate. The main objective of this chapter is to enhance the perfor m ance of the acoustic channel by decreasing the propagation delay in such a large underwater network, using a novel underwater Random Access Algorithm. In this chapter, Section 4.1 presents some related work. In Section 4.2, the underwater acoustic channel is analyzed. Section 4.3 introduces a novel transmission protocol for both WSNs and UWSNs. Section 4.4 focuses on energ yefficiency and delay decrease using some routing protocol, while simulation results are presented in Section 4.5. 4.1. Related Work Object tracking and monitoring in underwater are currently implemented by distributed underwater wireless sensor networks whose architectures, operations and performance demands are dictated by the tracking and monitoring objectives, and are also const rained by the characteristics and limitations of the environment. Within the statistical inference domain, the tracking and monitoring objectives are classified as either hypothesis testing (detection) or parameter estimation or estimation of the acting dat a generating process, and the pertinent performance criteria include decision/estimation accuracy and convergence rate, where detection/estimation accuracy is generally monotonically increasing with the number of observation data processed, [23]. When tim e constraints are imposed on high accuracy detection/estimation, the consequence is increase in the required overall data rates. At the same time, in the distributed wireless networks considered, observation data may be collected and processed by lif elimit ed nodes, whose lif espan is a function of the data rates they process, [24 ][34]. Several underwater communication models for acoustic underwater channels were proposed recently, focusing on the acoustic channel from different prospectives; for example, salinity, temperature, depth, error, delay, energy, path loss and the deployed comm u nication technique. These distinctive features of UWSNs pose many new challenges for delivering

PAGE 30

20 information from ENs to CHs in underwater sensor networks scenarios. In [35] a MAC layer protocol was developed to attempt to deal with the long propagation de lays. The protocol actually sends synchronization messages containing a transmission time schedule for each node to attempt to avoid collisions and delays. In [36] a protocol is presented to allow nodes to use a low-power wakeup mode to conserve energy whe n idle. [37] presents an analysis of the effect of the bandwidthdistance relationship on routing decisions based on hop length for energ yefficient routing. However, in this chapter a new event random access protocol is developed to enhance the communicati on link between ENs, CHs, and BS to attain energy efficiency, reliable data transfer and decrease expected delays. 4.2. Underwater Acoustic Channel 4.2.1. Underwater Environment The transmission medium in any underwater communication channel may be: ( 1) Sound for acoustic communication; (2) Electromagnetic waves for radio wave communication; and (3) Light for optical communication. As mentioned in Chapter 2, neither radio waves nor optical communication perform well in deep underwater environments, whi le acoustic communication is then widely used, due to the low attenuation of sound in water,(see Table 4.1). Although radio waves propagate faster, in comparison to acoustic waves, they strongly attenuate, especially in salt water, and propagate at long d istances through conductive seawater, only at extra low frequencies (3 0300Hz) requiring large antenna and high transmitter power [38 ][40]. However [41], electromagnetic wave are preferre d in shallow waters, subject to certain specific conditions. Optic a l communications have a greater advantage in data rates exceeding 1 Giga bps [38]. However, when used in underwater environments, the light is rapidly absorbed and the communication i s profoundly affected by its scattering. Consequently, transmission of op tical signals requires high precision in pointing the narrow laser beams [39]. Table 1 shows the different characteristic of the acoustic, radio and optical propagation in seawater

PAGE 31

21 T able 1: Comparison of the acoustic, radi o wave and optical communication in seawater [31] Acoustic Radi o Optica l Speed Propagation 1500m/s 3x10^8 m/s 3x10^8 m/s Power Los s >0.1 dB/m/Hz ~28 dB/km/100MHz Turbidit y Bandwidth ~kH z ~Mh z ~1 0 150Mhz Antenna Size ~0.1m ~0.5m ~0.1m Frequency Band ~kH z ~Mh z ~10^14 10^15 Hz Transmission Range ~50m 5k m ~1 m 100m 1m 100m Underwater acoustic communication is a technique for sending and receiving messages underwater using the propagation of sound. Typical frequencies associated with underwater acoustics are between 10 Hz and 1 MHz. The slow propagation speed of sound in underwater is 1500m/s which is five orders of magnitude lower than that for radio and optical waves [42]. The features of the acoustic channel are large propagation delay (0.67s / km) [41], large delay variance, limited bandwidth and high bit error rate [40] [42] The major challenges in designing underwater acoustic networks are [4 3-46]: Low propagation speed which results in large propagation delays. Limited battery power tha t usually cannot be recharged requiring the deployment of energy efficient communication protocols. Underwater sensors are prone to failures because of fouling and corrosion. High bit error and temporary losses of connectivity (shadow zone) will be experie nced The available bandwidth is limited and depends on the transmission distance. The underwater channel is severally impaired, especially due to multipath and fading problems.

PAGE 32

22 4.2.2. Propagation Delay The sound speed in seawater depends on water properties such as temperature, salinity and pressure (or depth). The speed of sound in seawater increases with the increase of any of the above three parameters. This may lead to large variations in propagation speed, even for equal distances. The underw a ter propagation speed has been modeled in [43] as follows : Since pressure is a function of depth, it is customary to express the sound speed (c) in m/s as an empirical function of temperature (T) in Celsius, salinity (S) in parts per thousand and depth (D) in meters. Figures 4.1 and 4.2 show the standard salinity and temperature profile in the ocean. Sound speed may be calculated using one of the formulas above. Therefore, many calculations were performed based on the assumption that the salinity is constant, or nearly so. This assumption is justified by the obser v ation that the typical range of salinities in the open ocean is usually small and that their impact on the speed of sound is practically negligibl e In coastal areas, and near rivers or ice, however, the assumption is not generally valid [48]. There are s everal underwater sound speed equations ( 15); we may use Equation 4.5, for example. It is important to note that, unlike in ground radio links, distance alone is not sufficient for determining delays. Some underwater short links could induce longer delay s, as compared to long wireless links.

PAGE 33

23 Figure 4.1: Standard salinity profile [41] Figure 4.2: Standard Temperature Profile [42] The slow propagation speed of sound in seawater contributes to large propagation delays, regarding system performance. The propagation delay may be calculated by the following equation:

PAGE 34

24 ; where t is the propagation delay (in second), d is the distance betwe e n two nodes (in meters) and c is the sound of speed (in m/s). We will talk about the effect of this propagation delay on our Kcells algorithm later. Figure 4.3 shows how the delay between two sensors in fixed distance from each other (100m) changes, for v arying depths, where each one of Equations (4.1) to (4.5) is used in the calculation of the speed of sound. Figure 4.3: Delay between two sensors with distance 100m using Equations 1-4 4.3. Transmission Protocol (Random Access Transmission Protocol ) Random Access Transmission Protocols are implemented in this chapter, first for the wireless on ground surface case, and then for the underwater case. We present and analyze a class of limited sensing random access algorithms with powerful properties for both the on ground surface Wireless Sensor Networks (WSNs) and the Underwater Wireless Sensor Networks (UWSNs). We note that the algorithm has been previously implemented in on ground surface wireless mobile environments. In this chapter, we will apply i t to the underwater sensor networks environment and compare then its performance with that of the on ground surface case. We start by considering packet networks, and independent users whose identities are initially unknown to the system and who transmit through a single common channel. The deployed transmission protocol must then necessarily belong to the class of Random Access Algorithms (RAAs) 50

PAGE 35

25 Such scenarios arise, for example, in the Ethernet environment, as well as in the signaling stage of pa cket radio and cellular telephony. The class of RAAs includes three subclasses, depending on the level of channel sensing required by the users in the protocol operations. The Minimal Sensing (MS) subclass requires that each user sense the transmission c hannel only at times when he transmits. The Limited Sensing (LS) subclass requires that each user sense the transmission channel continuously from the time he generates a packet to the time that this packet is successfully transmitted; this subclass cont a ins a number of algorithms, 51 [5 3 with various characteristics discussed in the paragraph below. The Full Sensing (FS) subclass requires that each user know the overall feedback history of the transmission channel from the time when the system starts operating, 54 ; this subclass contains algorithms, 50 54 whose operations are of just academic interest, since the level of channel sensing they require is clearly no nimplementable The limit Poisson user model consists of infinitely many indep e ndent Bernoulli users and it represents a worst case scenario for the study of RAAs, within a large class of user models, as proven in 55 In this thesis, we present a class of LS RAAs that attain the same throughput and also possess simi l ar resistance to channel errors as the algorithm in 53 The algorithms exhibit varying delay behavior for low rates and possess advantageous operational characteristics. We will focus on a single member in the class. The collision resolution operatio n s of the latter algorithm can be depicted by a kcell stack; we thus name the algorithm, the kcell algorithm 4.3.1. The system model and the algorithms: We assume packettransmitting users, a slotted channel, binary collisionversu s-noncollision ( CNC) feedback after each slot, zero propagation delays, and initially absence of feedback errors. We also assume that collided packets are fully destroyed and retransmission is then necessary. Time is measured in slot units; slot t occupies the time inte r val t, t+1) and xt denotes the feedback that corresponds to slot t; xt = C and xt = NC then represent collision and noncollision in slot, respectively

PAGE 36

26 Figure 4.4: The kcells algorithm Each algorithm in the class is independently and asynchronously implemented by the users. Indeed, in the limited sensing environment, it is required that each user monitor the channel feedback only from the time he generates a packet to the time that this packet is successfully transmitted. Therefore, the users knowledge of the channel feedback history is asynchronous. The objective in this case is to prevent new arrivals from interfering with some collision resolution in progress. This is possible if each user can decide whether or not a collision r esolution is in progress within a finite number of slots from the time he generates a new packet. The possibility of such decision can only be induced by the operational characteristics of the algorithm. As we will explain below, each algorithm in the c l ass possesses the appropriate operational characteristics for such decisions. Each algorithm in the class utilizes a window of size as a operational parameter and induces a sequence of consecutive Collision Resolution Intervals (CRIs). The window leng th is subject to optimal selection for throughput maximization. Each CRI corresponds to the successful transmission of all packet arrivals within an arrival interval of length The length of the CRI is determined by the number of users in the window and the algorithmic steps of the collision resolution process, Figure 4.4. The placement of the size window on the arrival

PAGE 37

27 access is determined asynchronously by the users. We will first describe the collision resolution process induced by the algori t hm. Then, we will explain the process which determines the placement of the -size window per CRI. The algorithmic class contains algorithms whose collision resolution process can be depicted by a stack with finite number of cells. Let us consider this algorithm in the class which can be described by a Kcell stack. Then, in the implementation of the collision resolution process, each user utilizes a counter whose values lie in the set of integers, 1,2, K We denote by rt the counter value of some user at time t. The K different possible values of the counter place the user in one of the K cells o f a Kcell stack. When his counter value is 1, the user transmits; he withholds at K1 different stages otherwise. When a CRI begins, all users in the size window set their counters at 1; thus, they all transmit within the first slot of the CRI. If the window contains at most one packet, the first slot of the CRI is a noncollision slot and the CRI lasts one slot. If the window contains at least two packets, instead, the CRI starts with a collision which is resolved within the duration of the CRI via t he following rules : The user transmits in slot t if and only if A packet is successfully transmitted in t if and only if and The counter values transition in time as follows: If and then If and then If and then, From the above rules, it can be seen that a CRI starts with a collision slot ends with K consecutive noncollision slots, an event which cannot occur at any other instant during the CRI. Thus, the observation of K consecutive noncollision slots signals the certain end of a CRI to all users in the system; it either signifies the e nd of a CRI that started with a collision or the end of a sequence of K consecutive lengt hone CRIs. Therefore, a user who arrives in the system without any knowledge of the channel feedback history can synchronize with the system upon the observation of the first Ktuple of consecutive noncollision slots. This observation leads to the asynchronous by the users generating of the siz ewindow placement on the arrival axis. Specifically, if a CRI ends with slot t, the window of the next CRI is selected w ith its right most

PAGE 38

28 edge K1 slots to the left of slot t and it contains those packets whose updates fall in the interval (tK+ 1, tK+1). The updates tk of a packet are generated as follows: Let t0 be the slot within which a packet is generated. T h en define t0 to be equal to t0. Starting with slot t0, the corresponding user senses continuously the channel feedbacks. He does so passively, until he observes the first Ktuple of consecutive NC slots, ending with slot t1.. If t0 (t1 K+1 t1 K+ 1), the user participates in the CRI that starts with slot t1 + 1. Otherwise, he updates his arrival instant to t1 = t0 + and waits passively until the end of the latter CRI, ending with slot t2. If t1 (t2 K+ 1, t2 K+ 1), the user participate s in the CRI which starts with slot t2; otherwise, he updates his arrival instant by again and repeats the above process. In general, if tn n 1 denotes the sequence of consecutive CRI endings since the first Ktuple of consecutive NC slots, the packet participates in the kth CRI if tk-1 (tk K+ 1, tk K+1) and tn (tn+ 1 K+ 1 tn K+1) ; for all n k-2. 2cell Algorithm Case: The 2cell case will be explained with steps below : If at least two packets attempt transmission within the same slot, a c o llision occurs and such an event is initially the only cause for faulty transmissions. A collision results in complete loss of the information carried by the collided packets; thus, retransmission of such packets is necessary. As we said before that the st ack has endless capacity because we have infinite number of users, which means that CRI does not have a specific ending unless the users are synchronized with system from the beginning, they know the history of the system. What happen in the CRI is as f o llows : Tabl e 2: 2-C el l a lgorith m Case numbe r Outcome of the slot Counter value What will happe n Number 1

PAGE 39

29 Figure 4.5: The operation of the 2cell algorithm (Imaginary Stack device) From these rules it can be seen that a CRI that starts with a collision slot ends with two consecutive noncollision slots. Furthermore, two consecutive NC slots cannot occur a t any other instant during a CR I Figures 3.5 and 3.6. Thus, the observation of two consecutive NC slots signals the end of such a CRI to all the users in the system Figure 4.6: The 2cell algorithm (Imaginary Stack device & Arrival Time)

PAGE 40

30 4.3.2. Underwater Signal Propagation As mentioned above, radio and optical communications do not perform well in deep underwater environment s Thus, acoustic communication is then widely chosen, due to the low attenuation of sound in water. Typical fre q uencies associated with underwater acoustics are between 10 Hz and 1 MHz. To compute the underwater delay, we need to find the sound speed and the distance between the sensors. The sound speed in the seawater depends on the water properties such as temper a ture, salinity and pressure (or depth). The speed of sound in the seawater increases with the increase of any of these three parameters. After selecting the underwater parameters such as temperature (T) in Celsius (from Figure 4.2), salinity (S) in par t s per thousand (from Figure 4.1) and depth (D) in meters, we may calculate the sound speed and subsequently the underwater propagation delay via Equation 4.6 (page 38, Figure 4.3), based on the selected distance between the sensors (in our program, the dis tance is between 100m). The propagation delay has a direct effect on the operation and the induced performance characteristics of the deployed random access transmission algorithm. The propagation delay must be considered in the operation of the algorithm : In particular, it should be added to the delay of observed channel outcomes when propagation delay is absent. Thus, since feedbacks are observed shifted in time by the amount of the propagation delay, the latter is added immediately to received channel f e edbacks, in our simulations. The propagation delay value could be added either as a constant or as a random variable. In this work, the propagation delay value is considered a constant, where its delay effect on the algorithmic operations may lead to dead l ocks in the limit. The equation below shows how the underwater expected delay is simulated when the propagation delay is incorporated. The simulation results are shown in Figure 4.29, 4.30, 4.31 and 4.32. (4.7)

PAGE 41

31 4.4. Energy-Efficient and Delay Decreasing Routing Protocol: In order to perform underwater communication effectively via deployed MDUWSNs, we propose an energ yefficient and delay decreasing routin g protocol. This means that, in addition to the delay reduction induced by the kcells algorithm, the deployment of the routing protocol will further enhance performance: (1 ) Energy savings may be attained by minimizing the number of data transmissions through the ENs and CHs. (2) Delay reduction may be attained via shortest path routing across the CHs. (3) The lifetime extension of underwater sensor networks is also accompli s hed by using the Layer by Layer replacement method In this section, the communication architecture of MDUWSNs is recalled from Chapter 3, as shown in figure 4.6. This communication architecture is dictated by the network objectives: (1) Gathering infor m ation continuously; (2) Monitoring and detecting objects effectively; (3) Attaining energ yEfficiency and lifetime extension; (4) Guaranteeing reliable communication links. We thus selected one of the communication architectures use d in Chapter 3, as show n in Figure 4.7. Figure 4.7: The architecture of the MDUWSNs The ENs, CHs and BS perform the following functions: The ENs are grouped into distinct clusters, where each cluster contains a single CH, where each EN collects local data and transmits th e m to its local CH. Each CH collects the data sent by its local ENs and processes them, using an operation determined by the network signal processing objectives. The CH then processes the compounded processed data, utilizing an operation that is determine d by the network signal processing objectives, and transmits the outcome to selected neighboring CHs

PAGE 42

32 and/or the BS. The CHs have processing capabilities, energy and lif espans that are much higher than those of the ENs, where their lifespans and energy ma y still be limited. BS fuses data transmitted to it by neighboring CHs, utilizing an operation that is determined by the network signal processing objectives. The BS has practically unlimited lif espan. All sensor nodes (ENs) and cluster heads are distribu ted randomly underwater per uni tsize region (Ocean Depth X Spreading Length). Each EN and CH can move freely and are assumed to be unmanned robots. Each mobile CH has a specific communication range that includes all ENs in the area of interest. The CHs a r e organized in layers by ocean depth, where each CH can only communicate with the CHs in the layer directly above its own. As an only exception, the CHs located in the highest layer (those closest to the ocean service) can communicate with each other and t he BS on the service. The ENs and CHs are first randomly distributed and then reorganized in clusters and layers as shown in Figure 4.8. Hence, each layer can handle as many ENs and CHs as required, and the number of layers could be calculated based on t he water depth and the distance between the layers needed to accomplish reliable communication link. All ENs and CHs are equipped with Voice Activity Detectors, to detect the presence of various objects or any unusual activity in the area, and also possess underwater acoustic transmitters and receivers. In our design, let assume that the water depth is about 100m, and that the distance between the layers is 20m, as shown in Figure 3.5. The communication path to route the information from ENs that a r e located at the bottom layer to the service layer where the BS is located should be considered in the design, since it affects the overall transmission delay. The communication rules pertinent to layering are included in Table 4.3 Tabl e 3: Communication rules betwee n ENs, CHs, an d BS Ocean Service Where the BS is located to connect the M DUWSN with the Wireless Networ k Layer # 1 Is located on the service, where all CHs can communicate with each other and the BS. Layer# 2 Layer# N (1) Each EN can communicate only with its CH within the cluste r (2) Each CH can communicate only with its two nearest CHs in the above and the lower layer. (2 lower layer +2 higher layer = 4 CHs )

PAGE 43

33 Tabl e 3 : ( Co nt.) Shortest Path to the Servic e ( 1 ) The path direction to transmit information is programmed based on the position of the BS as shown in Figure 5 (3) As mentioned above, each CH can communicate only with its two nearest CHs in the above and the lower layer. (2 lower layer +2 higher layer = 4 CHs), but the CH from the above layer is selected based on this direction. For example, if the base station is located on the left side of the network, the chosen CH is the one on the left side. See figure 4 Figure 4.8: The shortest path direction depends on the BSs location To decrease the delay while simultaneously saving in transmission energy, it is important to configure the path between the layers and maintain it consistentl y Figures 4.8 and 4.9 sh ow how the path could be configured. Each CH communicates with the CH located on the left side on the above layer if the assumed path is the Left Side Path (LFP), and communicates with the right CH if the assumed path is the Right Side Path (RSP).

PAGE 44

34 Figu r e 4.9: The shortest path direction toward the BS is chosen To evaluate the performance of the routing protocol, we select two cases that include distance, sound speed, temperature, salinity (hence, Equations 4.14.5). The two cases are exhibited in Fig u re 4.10. In the first case, the shortest path toward the BS is 111.8304m (see chapter 3 to know the parameters of the MDUWSN structure). In the second case, the random path is 140.4330m Figure 4.10: Compare the delay between the two cases: (a) The shortest path direction toward the BS is chosen (b) a random path toward the BS is chosen to forward the alarm Figure 4.11 exhibits the possible improvement in delay reduction, when the shortest path toward the BS is chosen, instead of the random path. The performance is approximately ~22.26%.

PAGE 45

35 Figure 4.11: Compare the delay between the two cases: (a) The shortest path direction toward the BS is chosen (b) a random path toward the BS is chosen to forward the alarm Considering the extension of the network lifetime, in this section we introduce a simple methodology for EN and CH replacement, when any of the latter expire. Our methodology is described below, where ENs and CHs are organized in layers according to their closeness to the BS and its serv ice function: 1Old ENs are replaced by new ENs at the higher service layer, while the old ENs move to the layer below. The same process is repeated across consecutive layers, as exhibited by Figure 4.12. 2This step applies to CHs, where the replacement pro c ess across CHs is the same as that of the ENs, as exhibited by Figure 4.13.

PAGE 46

36 Figure 4.12: Update the MDUWSN and Replace the expired EN s Figure 4.13: Update the MDUWSN and Replace the expired CH s

PAGE 47

37 4.5. Numerical Evaluation to Compute the Expected Delay for both WSNs and MDUWSNs The main goal of our simulations is to examine the delay for the deployed RAA in two cases: the o nground and the underwater wireless sensor networks. To compute the delay, w e need to first generate a number of random Poisson arrivals as shown in Figure 4.14. Then, those arrivals are transmitted by our Kcell algorithm. The algorithm transmits arrivals occurred within a window of size while each new arrival first senses transmissions and outcomes, until the first group of K consecutive noncollision slots occurs, at which point this arrival participates in the next collision resolution interval (CRI) Figure 4.14: Generating a random number of poisson arrivals Figure 4.15 shows the results of applying the 2 cells algorithm. The first part of each figure is Poisson random arrivals, the second part exhibits the process of sensing the two consecutive NCs and updating the window, and the third part exhibits the t r ansmitted arrivals. Each time, the figures in the middle and right side are changed to show how many arrivals still remain in the updated window and how many arrivals are transmitted successfully

PAGE 48

38 Figure 4.15: The system keeps sensing until two consecutive noncollision slots detected, and a window of the next CRI is selected with its right most edge m-1 slot

PAGE 49

39 Figure 4.15: The system keeps sensing until two consecutive noncollision slots d e tected, and a window of the next CRI is selected with its right most edge m-1 slot

PAGE 50

40 Figure 4.15: The system keeps sensing until two consecutive noncollision slots detected, and a window of the next CRI is selected with its right most edge m-1 slo t In this section, we compare the 3cell algorithm with the 4cell algorithm for both o nground and underwater MDUWSNs when all algorithms operate in environments imposing admission delay constraints. We simulated the algorithms in the presence o f the limit Poisson user model for admission delay constraints equal to 100, 200 and 400 slots and we then computed the average delays. Our results for the 3 and 4-cell algorithms, for both on-ground (WSNs) and underwater networks (UWSNs) are exhibited in Figures 4.16 4.26. From these figures, we

PAGE 51

41 observe that, for 100, 200, 400 slots admission delay constraints, the 3-cell algorithm does somewhat better than the 4cell in terms of expected delays of those packets that are successfully transmitted (50,000 Slots and different rates from 0.1 to 0.43). Moreover, the expected delays for underwater networks are higher than those for on-ground networks. Figure 4.16: Expected Dela y s for 3cell Vs 4cell algorithm Admission Delay 100 Slots Figure 4.17: Expected Dela y s for 3cell Vs 4cell algorithm Admission Delay 200 Slots

PAGE 52

42 Figure 4.18: Expected Delays for 3cell Vs 4cel l Admission Delay 400 Slots Figure 4.19: Expected Delays for 3cell algorith m Admission Delay =100, 200 and 400 Slots

PAGE 53

43 Figure 4.20: Expected Delays for 4cell algorith m Admission Delay =100, 200 and 400 Slots Figure 4.21: underwater sensor networksExpected Delays for 3cell Vs 4cell algorith m. Admission Delay =100 Slots

PAGE 54

44 Figure 4.22:UWSNsExpected Delays for 3cell Vs 4cell algorith m Admission Delay =200 Slots Figure 4.23:UWSNsExpected Delays for 3cell Vs 4cell algorith m Admission Delay =400 Slot s

PAGE 55

45 Figure 4.24:UWSNsExpected Delays for 3cell algorith m Admission Delay = 100, 200 and 400 Slot s Figure 4.25:UWSNsExpected Delays for 4cell algorith m Admission Delay =100, 200 and 400 Slot s

PAGE 56

46 Figure 4.26: UWSNs Vs WS NExpected Delays for 3cell algorith m Admission Delay =100, 200 and 400 Slots Figure 4.27: Expected Delays for 6cell Vs 7cel l algorith m Admission Delay =200 Slots We now compare the 6cell algorithm with the 7cell algorithm for higher number of arrivals (20,000 to 80,000) for both WSNs and UWSNs, when all algorithms operate in environments imposing admission delay constrain t s. The number of used slots is about 200,000 slots for different lambda rates (0.29 to 0.43). We simulated the algorithms in the presence of the limit Poisson user model for admission delay constraints equal to 200 and 400 slots and then we compute the e x pected delays. Our results for 6 and 7cell algorithm for WSNs and UWSNs are

PAGE 57

47 exhibited in figures 4.27 4.32. From these figures, we observe that 7cell algorithm perform much better than the 6cell algorithm for both WSNs and UWSNs for both 200 and 400 slots admission delay constraint s. Also, the 7cell and 6Cell performs better when the admission delay equals to 400 than 200 slots especially when we transmit high number of arrivals. Figure 4.28: Expected Delays for 6cell algorith m Admission Delay =400 Slots Figure 4.29: UWSNsExpected Delays for 6cell Vs 7cell algorith m Admission Delay =200 Slots

PAGE 58

48 Figure 4.30:UWSNsExpected Delays for 6cell Vs 7cell algorith m Admission Delay =400 Slots Figure 4.31:UWSNs Vs WSNExpected Delays for 6cell Vs 7cell algorith m Admission Delay =200 Slots

PAGE 59

49 Figure 4.32:UWSNs Vs WSNExpected Delays for 6cell Vs 7cell algorith m Admission Delay =400 In this part, a comparison between the 7cell algorithm and the SAloha, [58], is presented. The expected delay for 1000 transmitted arrivals is computed for the 7cell algorithm and compared to that induced by the SAloha protocol. The used lambda rate in this case is equal to 0.4 and the admission delay constraint equals 400 slots. It is also assumed each slot i s equal to 20 Micro Sec, [5658], because most of the references use the per second rather than the per slot time units. We thus converted the yaxis to (Sec) instead of (Slot) in our comparative evaluation. Our results in Figure 4.33 show that the proposed Kcell model outperforms the SAloha protocol [58].

PAGE 60

50 Figure 4.33: Average Access Delay for 7-Cell (Admission Delay =400 and Lambda=0.4), ATL SALOHA, and S-ALOHA protocols.

PAGE 61

51 5: Conclusions This thesis focuses on the impact of latency on Mobile Underwater Sensor Networks ( MUWSN s ) operatin g in the underwater acoustic environment. The deployment of a novel MUWS N that consists of mobile E lementary N odes (ENs), a backbone network of C luste r-H eads (CHs), and a Base Station (B S ) has been proposed. We also provide an M UWSN node clustering technique that attains high data rate and l o w energy consumption; the MUWS N utiliz es the audio signal emitted by the submarines engine to detect its presence. Du r in g the implantations of the MUWS N sensor localization is required. Thus the improvement of localization accuracy has been considered, while such accuracy could be significantly affected by poor acoustic channel quality and node curren tcaused mobility To overcome problems suc h as: interference, power savings, data rate constraints an d propagation delay, we reorganized our network in clusters where each cluster has a CH and many E N s. In general, t he network design ma y be dictated by the objective of i t s application, in conjunction with the constraints induced by the deployed transmission protocols and the environment (Power, transmission rat es, sensors li fe span,etc. ). We presented a limited feedback sensing Random Access Algorithm (RAA), as the dat a transmission protocol in each cluster. The algorithm utilizes binary ( CNC) feedback and a slotted broadcast channe l while each sensor is required to sense the channel continuousl y from the time it generates a packet to the time when this packet is suc c essfully transmitted. We also presented the analysis of the algorithm in the presence o f the Poisson used model. We examined the delays induced by th e proposed RAA in two cases: when deploye d on-ground; and when deployed by underwater wireless sensor n e tworks wher e large propagation delay s are inherent to acoustic such networks In both cases, we modeled the sensors limited life span by impose d admission delay constraint s in data transmissions. In each case, system evaluation was represented by s imul a tio n s, where such results were used to compare o nground versus underwater performance. Finally, a performance comparison between the proposed algorithm and the SAloh a algorithm was presented. The comparison encompassed expected delays and rejection rates as the performance metrics. The results showed that as the rate of the limit Poi s son traffic and the

PAGE 62

52 imposed limit on the admission delay increase, the SAloh atype protocol collapses, whil e the introduced underwater protocol maintain s high standard s.

PAGE 63

53 BIBLIOGRAPHY [1] Z. Jiang Underwater Acoustic Networks International journal of Intellgent Control and Systems, Vol. 13, NO. 3, Sep 2008, 152-161 [2] J. Yick, B. Mukherjee, and D. Ghosal Wireless sensor network survey , Department of Computer Science, University of California, Davis, CA 95616, United States [3] J. Partan, J. Kurose, and B. N. Levine, A Survey of Practical Issues i n Underwater Networks, International Conference on Mobile Computing and Networking, Proc. of the 1st ACM international workshop on Underwater networks, pp. 17 24, Los Angeles, CA, USA, Sept., 2006. [4] L. Berkhovskikh and Y. Lysanov, Fundamentals of O c ean Acoustics. New York: Springer, 1982. [5] A. Quazi and W. Konrad, Underwater acoustic communications, IEEE Commun. Mag., pp. 2429, Mar. 1982. [6] R. J. Urick, Principles of Underwater Sound, 3rd Edition, McGra wHill Publishing Company, New York, N Y, 1983. [7] R. Coates, Underwater Acoustic Systems, New York: Wiley, 1989 [8] J. A. Catipovic, Performance Limitations in underwater acousti c telemetry, IEEE J. Oceanic Eng., 15(3):205216, Jul. 1990. [9] E. M. Sozer, M. Stojanovic, and J. G. Proaki s Undersea Acoustic Networks. IEEE Journal of Oceanic Engineering, OE-25(1):7283, January 2000. [10] http://www.cse.uconn.edu/~jcui/UWSN_papers/UCONN_CSE_UbiNe tTR0 5-02.pdf [11] S. Zhou and P. Willett, Submarine Location Estimation Via a Network of Dete c tio nOnly Sensors. IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 55, NO. 6, JUNE 200 7 [12] P. Wang, C. Li, J. Zheng Distributed MinimumCost Clustering Protocol for Underwater Sensor Networks (UWSNs). ICC 2007 proceedings. [13] C. We nYu, L. JingBiao UUV Mobility Management for Underwater Acoustic Sensor Networks [14] J. Cui, J. Kong, M. Gerla S. Zhou The Challenges of Building Scalable Mobile Underwater Wireless Sensor Networks for Aquatic Applications [15] T. and H. Lin, Research on Submarine Counterdetection Probability by Threat Surface Ship Sonar. IEEE. [16] A. C. KELLER, Submarine Detection by Sonar. IEEE. [17] A.T. Burrell and P. PapantoniKazakos, Performance Monitoring in Sensor Networks, IEEE.

PAGE 64

54 [18] Carp Fish Robot, http://www.youtube.com/watch?v=eO9oseiCTdk. [19] Recycled Robot Fish Teach Japanese Children About Sealife, http://inhabitat.com/recycledrobotfis hteac hjapanesechildre n-aboutsealife / [20] Video: Mini fuel cell powering a robot fish, http://www.crunchgear.com/2010/03/16/videomin i-fuelcel l-powering-a-robotfish / [21] Robotic fish consume less power by using magnets instead motors, http://www.robaid.com/bionics/roboticfis h-consumeless-power-by-usingmagnet sinsteadmotors.htm [22] V inay K. Ingle and John G. Proakis, Digital Signal Prosseing Using Matalb. [23] D. Kazakos and P. Papanton iKazakos, Detection and Estimation, Computer Science Press, New York, NY, 1990. [24] I. F. Akyildiz, W.Su, Y. Sankkarasubramaniam and E Cayirci, Wi r eless Sensor Networks: A Survey, Computer Networks (Elsevier), vol. 38, no. 4, 2002, pp. 393-422. [25] S. Bandyopadhyay and E.J. Coyle, An Energy Efficient Hierarchical Clustering Algorithm for Wireless Sensor Networks, IEEE INFOCOM 2003. [26] M. Bhar d waj and A.P. Chandrakasan, Bounding the Lifetime of Sensor Networks via Optimal Role Assignments, in Proc. IEEE INFOCOM, New York, Jun. 2327, 2002, pp. 15871596. [27] D. Blough and S. Paolo, Investigating Upper Bounds on Network Lif eTime Extension f or Cel lBased Energy Conservation Techniques in Stationary Ad Hoc Networks, in Proc. ACM MobiCom, Atlanta, GA, Sep. 23-28, 2002, pp. 183-192. [28] J.H. Chang and L. Tassiulas, Routing for Maximum System Lifetime in Wireless Ad Hoc Networks, in Proc. 37t h Annual Allerton Conference on Communications, Control and Computing, Monticello, IL, Sep. 1999, vol. 1, pp. 22-31. [29] J.H. Chang and L. Tassiulas, Energy Conserving Routing in Wireless Ad Hoc Networks, in Proceedings of IEEE INFOCOM, Tel Aviv, Israel Mar. 26-30, 2000, pp. 22-31. [30] Y. T. Hou, Y. Shi and H.D. Sherali, On Node Lifetime Problem for Energ y-Constrained Wireless Sensor Networks, ACM/Springer Mobile Netw.Applicat., vol. 10, no. 6, Dec. 2005, pp. 865-878. [31] Y.T. Hou, Y. Shi, J.H. Ree d and K. Sohraby, Flow Routing for Variable Bit Rate Source Nodes in Energ yConstrained Wireless Sensor Networks, in Proc. IEEE International Conference on Communications, Seoul Korea, May 16-20, 2005, pp. 3057-3062. [32] Y. T. Hou, Y. Shi and H.D. Sher a li, Rate Allocation and Network Lifetime Problems for Wireless Sensor Networks, IEEE Trans. Networking, vol. 16, no. 2, April 2008, pp. 321-334.

PAGE 65

55 [33] P. Papantoni Kazakos and A.T. Burrell The Implementation of Dynamic Rate Allocation in Sensor Networks Journal of Intelligent and Robotic Systems, 2010, Vol. 58, pp. 211-238. [34] V. Rodoplu and T.H. Meng, Minimum Energy Mobile Wireless Networks, IEEE J. Sel. Areas Commun., vol. 17, no. 8, Aug. 1999, pp. 1333-1344. [35] V. Rodoplu and M. Park, "An energ yefficient MAC protocol for underwater wireless acoustic networks," in MTS/IEEE OCEANS, 2005. [36] A. F. Harris, M. Stojanovic, and M. Zorzi, "Why underwater acoustic nodes should sleep with one eye open: Idl etime power management in underwater sensor ne t works," in ACM International Workshop on UnderWater Networks (WUWNet), 2006. [37] A. F. Harris and M. Zorzi, "Energ yefficient routing protocl design considerations for underwater networks," in IEEE Conference on Sensor; Mesh and Ad Hoc Communications and Networks (SECON), 2007 [38] Lanbo Liu, Shengli Zhou, and Ju nHong Cui, Prospects and Problems of Wireless Communication for Underwater Sensor Networks. Wiley WCWC Special Issue on Underwater Sensor Networks. [39] Milica Stojanovic, Underwater Acoustic Communication, For the Wiley Encyclopedia of Electrical and Electronics Engineering. Department of Electrical and Computer Engineering Northeastern University Boston [40] N. N. Ismail, L. A. Hussein, S. H.S.Ariffin, AnalyzingThe Performance of Acoustic Channel in Underwater Wireless Sensor Network(UWSN), IEEE 2010 Fourth Asia International Conference on Mathematical/Analytical Modeling and Computer Simulation, Page(s): 550 555. [41] Xianhui Che, Ian Wells, A Static Multihop Underwater Wireless Senso r Network Using RF Electromagnetic Communications 29 IEEE International Conference on distributed computing system workshop, Swansea metropolitan university UK, 2009. [42] JunHong Cui, Jiejun Kong, Mario Gerla, Shengli Zhou, Challenges: Building Scalabl e and Distributed Underwater Wireless Sensor Networks (UWSNS) For Aquatic Applications .2005 [43] Dario Pompili, Tommaso Melodia, Ian F. Akyildiz, routing algorithm for dela yinsensitive and dela ysensitive application in underwater sensor networks. in P r oc. of ACM Conference on Mobile Computing and Networking (MobiCom), Los Angeles, CA, September 2006 [44] Mari Carmen Domingo, Rui Prior, Energy Analysis Of Routing Protocol For Underwater Wireless Sensor Networks, 200 7 [45] Ian F. Akyildiz, Dario Pompili, Tommaso Melodia, Challenges for Efficient Communication in Underwater Acoustic Sensor Networks.ACM Sigbed Review, Vol. 1, Number 2, July 2004

PAGE 66

56 [46] Dario pompili, Tommaso Melodia, lan F,akyildiz, Deployment Analysis In Underwater Acoustic Wireless Sensor Network ,International Conference on Mobile Computing and Networking ,2006. [47] Paul C.Etter, Underwater Acoustic Modeling Principle, Techniques And Applications. Elsevier Science Publishers, 1991. [48] http://www.windows.ucar.edu/tour/link=/earth/Water /salinity_depth.h tml [49] http://www.windows.ucar.edu/tour/link=/earth/Water/temp.htm l 50 J.L. Massey, Collision resolution algorithms and random access communications, in Multiuser Communication, G. Longo Ed, New York, SpringerVerlag 51 B.S. Tsybakov and V.A. Mikhailov, Free synchronous packet access in a broadcast channel with feedback, Problemy Peredachi Informatsii, Vol. 14, No. 4, pp. 259-280, 1978. 52 L. Georgiadis and P. PapantoniKazakos, A 0.487 throughput limited sensi n g algorithm, IEEE Trans. Inf. Th., Vol. IT-33, No. 2, pp. 233237, Mar. 1987. 53 M. Paterakis and P. PapantoniKazakos, A simple window random access algorithm with advantageous properties, IEEE Trans. Inf. Th., Vol. 35, pp. 1124-1130, Sept. 1989. 54 J. I. Capetanakis, Tree Algorithms for Packet Broadcast Channel, IEEE Trans. Information Theory, Vol. I T-25, pp. 505-515, Sept. 1979. 55] M. Paterakis, L. Georgiadis and P. Papanton iKazakos, On the relation between the finite and the infini t e population models for a class of RAAs, IEEE Trans. Commun., Vol. CO M-35, pp. 1239-1250, Nov. 1987. [56] Chatzimisios, P.; Boucouvalas, A.C.; Vitsas, V., Packet delay analysis of IEEE 802.11 MAC protocol, IET Journals, Volume: 39, Page(s): 1358 1359. [57] Guolin Sun; Bo Hu, Delay Analysis of IEEE 802.11 DCF with Bac koff Suspension, Future Generation Communication and Networking (FGCN 2007),Volume: 1, Page(s): 148 151. [58] River oAngeles, M.E.; Lar aRodriguez, D.; CruzPerez, F.A., Access delay analysis of adaptive traffic load type protocols for SALOHA and CSMA in EDGE, Wireless Communications and Networking, 2003. IEEE, Volume: 3, Page(s): 1722 1727 vol.3