Content uploaded by Nadeem Javaid
Author content
All content in this area was uploaded by Nadeem Javaid on Nov 09, 2016
Content may be subject to copyright.
Optimizing Energy Consumption with Sink
Mobility Management in Underwater Wireless
Sensor Networks
Author
Engr. Waseem Raza
14F-MS-TE-02
Supervisor
Engr. Farzana Arshad
Assistant Professor
Department of Telecom Engineering
University of Engineering and Technology
Taxila
Co-Supervisor
Dr. Nadeem Javaid
Associate Professor
Department of Computer Science
COMSATS Institute of Information Technology
Islamabad
DEPARTMENT OF TELECOM ENGINEERING
FACULTY OF TELECOM & INFORMATION ENGINEERING
UNIVERSITY OF ENGINEERING AND TECHNOLOGY
TAXILA
May 2016
Optimizing Energy Consumption with Sink
Mobility Management in Underwater Wireless
Sensor Networks
Author
Engr. Waseem Raza
14F-MS-TE-02
A thesis submitted in partial fulfillment of the requirements for the degree of
M.Sc. Telecom Engineering
Supervisor
Engr. Farzana Arshad
Assistant Professor
Telecom Engineering Department
External Examiner Signature: ——————–
Thesis Supervisor Signature: ——————–
Co-Supervisor
Dr. Nadeem Javaid
Thesis Co-Supervisor Signature: ——————–
Associate Professor
Department of Computer Science
COMSATS Institute of Information Technology
Islamabad
DEPARTMENT OF TELECOM ENGINEERING
FACULTY OF TELECOM & INFORMATION ENGINEERING
UNIVERSITY OF ENGINEERING AND TECHNOLOGY
TAXILA
May 2016
ABSTRACT
Optimizing Energy Consumption with Sink Mobility Management in Underwater
Wireless Sensor Networks
Waseem Raza
14F-MS-TE-02
In this work we propose a novel routing strategy to cater the energy consumption
and delay sensitivity issues in deep underwater wireless sensor networks. This strat-
egy is named as Event Segregation based Delay sensitive Routing; in which sensed
events are segregated on the basis of their criticality and then forwarded to their
respective destinations. Forwarding decisions are made on the basis of forwarding
functions which depend on different routing metrics like: Signal Quality Index, Lo-
calization free Signal to Noise Ratio, Energy Cost Function and Depth Dependent
Function. We have comprehensively investigated the ranges of previously defined
forwarding functions. The problems of incomparable values of forwarding func-
tions in different depth regions of network cause uneven delays in forwarding pro-
cess. Hence forwarding functions are redefined to ensure their comparable values
in different depth regions. Packet forwarding strategy is based on event segregation
approach which forwards one third of the generated events (delay sensitive) to sur-
face sinks and two third events (normal events) are forwarded to mobile sinks. Mo-
tion of mobile sinks is influenced by the relative distribution of normal nodes. key
words— Energy efficiency, Delay-sensitive Routing, Forwarding Function, Hold-
ing Time, Event Segregation, Network Lifetime, Throughput
ii
UNDERTAKING
I certify that research work titled “Optimizing Energy Consumption with Sink
Mobility Management in Underwater Wireless Sensor Networks” is my own work.
The work has not been presented elsewhere for assessment. Where material has
been used from other sources it has been properly acknowledged / referred.
——————-
Waseem Raza
14F-MS-TE-02
iii
ACKNOWLEDGEMENT
I am extremely thankful to Almighty Allah, with out His countless blessing it would
not be possible for me to complete this work. I seek His endless blessing for future.
After that I am very thankful to my family especially to my father whose persistent
support and confidence in me has always been a great source of inspiration and
motivation throughout my educational career. He has always been a great person;
uneducated but full of rural wisdom and vision to look deeper into the future.
I am grateful to Dr Nadeem Javaid from Computer Science Department COM-
SATS Islamabad for His collaboration and technical support throughout this re-
search work. He has always been supportive and motivating. In the end I am
thankful to my supervisor Engr. Farzana Arshad for her considerations and devo-
tions for this work. Without her it would be quite impossible to go through all the
phases of research work. Her motivation and blessing has been a constant source of
encouragement and strength for me.
I am also thankful; to all the respective members of PGS at Telecom Engineering
Department UET Taxila, especially to our prestigious Chairman Dr Yasir Amin for
the support and matchless facilities provided for fulltime master scholars at TED
UET Taxila.
iv
This work is dedicated to:
Innocent victims of terrorism from Pakistan and
all over the world;
APS and Bacha Khan University Martyrs and
their families;
especially to;
Dr. Syed Hamid Hussain Shaheed.
v
TABLE OF CONTENTS
The title of my thesis i
Abstract iii
Undertaking iv
Acknowledgement v
Dedication vi
Table of Contents viii
List of Figures ix
List of Tables x
Chapter 1 Introduction and Background 1
1.1 StatementofProblem......................... 4
1.2 Research Challenges in UWSNs . . . . . . . . . . . . . . . . . . . 6
1.2.1 Capricious Underwater Environment . . . . . . . . . . . . . 6
1.2.2 LowDataRates........................ 6
1.2.3 Damage to Equipments . . . . . . . . . . . . . . . . . . . . 6
1.2.4 Deployment and Maintenance Cost . . . . . . . . . . . . . 7
1.3 Classification of UWSNs . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 One Dimensional UWSNs . . . . . . . . . . . . . . . . . . 7
1.3.2 Two Dimensional UWSNs . . . . . . . . . . . . . . . . . . 7
1.3.3 Three Dimensional UWSNs . . . . . . . . . . . . . . . . . 8
1.3.4 Four Dimensional UWSNs . . . . . . . . . . . . . . . . . . 8
1.4 Applications of UWSNs . . . . . . . . . . . . . . . . . . . . . . . 9
vi
1.4.1 Monitoring and Exploration Application . . . . . . . . . . . 9
1.4.2 Military Applications . . . . . . . . . . . . . . . . . . . . . 10
1.4.3 Assisted Navigation for Deep Sea Exploration . . . . . . . 11
1.4.4 Disaster Monitoring and Prevention . . . . . . . . . . . . . 12
1.5 A Typical UWSN and Solutions . . . . . . . . . . . . . . . . . . . 13
1.5.1 Subnero Solutions . . . . . . . . . . . . . . . . . . . . . . 13
1.5.2 DSPComm Solutions . . . . . . . . . . . . . . . . . . . . . 15
Chapter 2 Literature Review 17
2.1 Ad-hoc Network Protocols . . . . . . . . . . . . . . . . . . . . . . 17
2.1.1 Proactive Routing Protocols . . . . . . . . . . . . . . . . . 18
2.1.2 Reactive Routing Protocols . . . . . . . . . . . . . . . . . . 19
2.2 Underwater Routing Protocols . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Localization Based Routing Protocols . . . . . . . . . . . . 20
2.2.2 Localization Free Routing Protocols . . . . . . . . . . . . . 23
Chapter 3 Problem Statement and Motivation 28
3.1 Forwarding Function Analysis . . . . . . . . . . . . . . . . . . . . 28
3.2 Motivation for Event Segregation . . . . . . . . . . . . . . . . . . . 34
Chapter 4 ESDR: Event Segregation based Delay sensitive Routing 36
4.1 Network Initialization and Node Deployment . . . . . . . . . . . . 37
4.2 Statistical Event Generation and Segregation . . . . . . . . . . . . . 38
4.3 Packets Forwarding Strategy . . . . . . . . . . . . . . . . . . . . . 40
4.3.1 Updating Depth Threshold . . . . . . . . . . . . . . . . . . 42
4.4 TheChannelModel.......................... 44
4.4.1 Thorp Attenuation Formula . . . . . . . . . . . . . . . . . 44
4.4.2 Monterey-Miami Parabolic Equation . . . . . . . . . . . . . 46
4.5 Improvements in Terms of Forwarding Function . . . . . . . . . . . 48
4.6 Improvements in Terms of Holding Time . . . . . . . . . . . . . . . 50
vii
4.7 MobilityofSink............................ 50
4.7.1 Synchronized and Uniform Mobility of AUV . . . . . . . . 51
4.7.2 Adaptive Mobility of Courier Nodes . . . . . . . . . . . . . 54
Chapter 5 Performance Evaluation 56
5.1 SimulationSetup ........................... 56
5.2 Performance Evaluation Metrics . . . . . . . . . . . . . . . . . . . 57
5.2.1 Network Lifetime . . . . . . . . . . . . . . . . . . . . . . . 58
5.2.2 Average end to end delay . . . . . . . . . . . . . . . . . . . 61
5.2.3 Throughput.......................... 63
5.2.4 Transmission Loss . . . . . . . . . . . . . . . . . . . . . . 64
5.2.5 PathLoss........................... 65
Chapter 6 Conclusion and Future Work 68
viii
List of Figures
1.1 Underwater Communication Scenario . . . . . . . . . . . . . . . . 14
1.2 Schematic Diagram of a Node . . . . . . . . . . . . . . . . . . . . 14
4.1 Network Diagram in ESDR . . . . . . . . . . . . . . . . . . . . . . 38
4.2 Packet Format in ESDR . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 Eligible Neighbours with Range and Depth Threshold . . . . . . . . 43
4.4 Synchronized mobility of AUV . . . . . . . . . . . . . . . . . . . . 52
4.5 Adaptive mobility of AUV/CN . . . . . . . . . . . . . . . . . . . . 54
5.1 Lifetime................................ 60
5.2 Stability Period, half-life and life-time . . . . . . . . . . . . . . . . 60
5.3 Average end to end delay: Comparison . . . . . . . . . . . . . . . . 62
5.4 Delay comparison in ESDR . . . . . . . . . . . . . . . . . . . . . . 63
5.5 Throughput .............................. 64
5.6 Packet Delivery Ratio . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 TransmissionLoss .......................... 66
5.8 PathLoss ............................... 67
ix
List of Tables
3.1 Parameters values used in previous schemes . . . . . . . . . . . . . 30
3.2 Ranges w.r.t depth threshold . . . . . . . . . . . . . . . . . . . . . 31
3.3 Forwarding Functions Ranges . . . . . . . . . . . . . . . . . . . . 33
4.1 Centers of elliptical paths . . . . . . . . . . . . . . . . . . . . . . . 53
5.1 Parameters values used in simulations . . . . . . . . . . . . . . . . 58
x
CHAPTER 1
Introduction and Background
Man has always been curious about nature asking questions and willing to find
more about the things influencing on his life. Curiosity leads towards exploration
and observation of the natural phenomena which is the first step of the scientific
process. It provides an intellectual fodder for the scientific discoveries. It’s about
gaining knowledge and expanding our horizon and putting the questions that have
not yet been asked. Curiosity and exploration have been the motivation for some
great thinkers and scientist especially; of last three to four centuries, and it has led
towards some greatest discoveries of mankind. As in the word of Einstein [1]:
"The important thing is not to stop questioning. Curiosity has its
own reason for existing".
The imperceptible aspiration to explore and challenge the environs of what we know
has provided welfare to our society for centuries. Interest and investigation are
imperative motivation to the human soul. This motivation had led human explorers
to the farthest and unreachable areas on the earth. This interest has driven human
investigation to most noteworthy mountains, coldest landmasses and even into the
deeper space.
But there is one more unexplored area; much closer to our homes and host of largest
and diverse ecosystem; The Ocean. Study by World Register of Marine Species
(WoRMS) has shown that the aquatic environs may be home to as many as a million
species of animals and plants [2] but only about a quarter of them have actually
been properly discovered and recognized.
Ocean has been swaying on human race since ancient times. It has disseminated
people on different continents and have connected civilizations with each other for
1
different purposes. The seas give advantageous transport routes as around 90% of
all trade between nations is conveyed by ships. It provides oxygen to atmosphere
and act as sink for carbon dioxide, thus have helped to negate human-caused global
warming and climate change. Nearly half the CO2produced by human activities in
the last 200 years has dissolved into the ocean. More than half of the world popula-
tion live within 150km of coastline. Ocean is the source of their food, business and
employment. Besides providing sustenance items and livelihood ocean plays vital
role in weather systems, economy and defense system of a country. Ocean covers
72% of earth surface, that is 139 million square miles or 139 with 19 zeros after it
[3]. It is not just very wide rather it is 4000mdeep as well. That is 10 Empire State
buildings heaped on top of each other!
It is quite strange that such an important influencer on human civilization has not
been explored to its full potential. Conventional exploring mechanism involves
sending/deploying a sampler, to the bottom of the ocean or into the monitoring
area of interest, which collects required information and then returns on-shore data
center. There is lack of real time monitoring especially required in surveillance and
seismic monitoring applications. There is also no mechanism of retrieving if there
is failure or mis-configuration. These above mentioned limitations empower Under
Water Sensor Network (UWSN) to become an appropriate candidate for longtime
and reliable monitoring and data gathering applications [4].
UWSN is a fusion of wireless technology with extremely small micro mechanical
sensors having smart sensing, intelligent computing, and communication capabil-
ities. UWSN is a network of autonomous sensor nodes which are spatially dis-
tributed in underwater to sense the water-related properties such as quality, tempera-
ture, and pressure. The sensor nodes, stationary or mobile, are connected wirelessly
via communication modules to transfer various events of interest [5,6,7]. These
nodes are not aware of their exact position, however their depth can be found using
depth (pressure) sensors. Generally, in UWSN acoustic transceivers are used for
2
communication. Acoustic signal propagation speed depends on depth, temperature
and salinity of the water. Generally these signal propagate at about 1500m/s which
is five orders of magnitude lesser than that of radio links in terrestrial wireless net-
works. The acoustic waves are low frequency waves which offer small bandwidth
but have long wavelengths. Thus, acoustic waves can travel long distances and are
used for relaying information over kilometers. Hence these are some inherent char-
acteristic of acoustic communication like, limited bandwidth (<100 KHz), larger
delays, time varying multi path which may cause larger doppler shift.
The packets generated by each node are transmitted to surface sink using broad-
casting and multi-hop communication [8]. In this communication, sending node
includes its depth and routing information in data packet and broadcast that packet
to all the nodes within its transmission range. However only shallower nodes i.e.
node with lesser depth than current node depth receive that broadcast. Now only
those shallower nodes which are farther than depth threshold become eligible for
forwarding. All the eligible nodes, hold the packet for their respective holding time
which is inversely related with forwarding function. If these packet holding nodes
receive the copy of same packet with-in their holding time they discard the packet
and assume that some more eligible node has forwarded the packet. However if
holding time is elapsed and copy is not received, node considers itself as the best
forwarder and broadcasts that packet. This process goes on till the packet is reached
to surface sinks. In this way each time a node with maximum value of forwarding
function, gets minimum value of holding time and becomes next forwarder. This
communication process involves certain terms which are needed to be explained.
Depth Threshold: Depth threshold is a global parameter to control the number of
nodes involving in forwarding process. Greater the values of depth threshold
lesser the nodes involved in forwarding process. Value of depth threshold is
always less than transmission range.
Forwarding Functions: Forwarding function is a routing metric which can be de-
3
pendent on different parameters of nodes, like absolute depth, depth differ-
ence, residual energy, transmission range and depth threshold. Different for-
warding functions have been defined in literature and included in details in
next sections.
Holding Time: Holding time is assigned to each node to avoid redundant trans-
missions and collision. It is the time during which a node hold the packet and
avoids any transmission.
1.1 Statement of Problem
The problem statement is summarized here in the following three points are is given
in details in chapter 3.
•Underwater environment is very diverse in nature [9]. A wide range of living
and non-living activities take place in the same oceanic area. Generally, this
diversity increases as we go deeper into the oceans. There can be many di-
verse applications in which severity of the sensed value of all nodes may not
be similar. The data generated by some nodes may be much important than
that of the other nodes. Hence the costly deployment of UWSN cannot be
dedicated for sensing only one type of occurring events. This sternness can
be of many types like, in terms of delay criticality or in terms of size of data
generated. These diverse applications require appropriate routing for each
type of sensed event. So it is necessary to provide an underwater network so-
lution which consider diverse events occurrence and appropriate response to
that diverse characteristic (different delay criticality associated with different
occurring events).
•Mobile sinks are deployed in underwater for date collection applications but
these precious resources can be exploited with much more potential to en-
hance the network lifetime of energy constraint nodes.
4
•The multi-hop communication involves forwarding to shallower (lesser depth)
nodes. This forwarding is done on the basis of forwarding function. The
forwarding function values for different nodes must be different enough, but
comparable to ensure optimum forwarding process. Analysis on the ranges of
previously defined forwarding functions require the redefinition of forward-
ing function.
Our contribution in this work is enlisted in following steps.
•We have considered a diverse environment with three types of events occur-
ring in sensing field. These events are named as critical, very critical and
normal. Critical and very critical events are delay sensitive in nature so for-
warding to surface sink is planned for these types of event using respective
forwarding functions in different depth regions. This forwarding to surface
sink is accomplished using multi-hop communication involving upper nodes
as intermediate forwarder.
•Forwarding function are defined differently for different depth regions. How-
ever values of different forwarding functions must be comparable with each
other. This is necessary because in-comparable values of forwarding func-
tion causes, uneven delays in the regions where smaller values are calculated;
and collision and redundant transmission in the regions where larger values
of forwarding functions are observed. Analysis on the range of previously
defined forwarding function reveals the incomparable ranges of different for-
warding function. Hence the forwarding functions are redefined to eradicate
this issue.
•As per event segregation approach normal events are forwarded to mobile
sinks. Hence mobility of these nodes is influenced by the distribution and
rate of data generation of normal nodes. We have proposed two different
mobility schemes for data gathering nodes.
5
1.2 Research Challenges in UWSNs
Despite all the attractive features of UWSN it faces lots of challenges due to dy-
namic underwater environment, changing network topology due to water currents,
shipping and aquatic life activities. The authors [10,11] have extensively elabo-
rated various research challenges and issues for underwater on different layers. A
brief review of major of challenges is given in details.
1.2.1 Capricious Underwater Environment
Due to the capricious underwater environment, it is extremely challenging to install
the network in underwater which works consistently and wirelessly. The present
technology allows reserved communication but it incurs substantial cost of deploy-
ment, maintenance, and device recovery to cope with unpredictable underwater cir-
cumstances [12,13].
1.2.2 Low Data Rates
Water absorb lots of RF energy hence RF communication is not workable in under-
water environments. That is why much of the communication is done via acoustic
channel. Acoustic communication is mainly in low frequency typically less than 30
KHz. This ensure longer distance communication typically in the range of several
kilometers. However low frequency limits available bandwidth and hence lower
data rates are observed in underwater communications [14]. Higher data applica-
tions require comparatively larger time to communicate the whole data.
1.2.3 Damage to Equipments
The instruments used in underwater devices are vulnerable to tedious subaquatic
challenges, for example, algae and salt accumulation on camera lens, decreasing
the effectiveness of sensors and other important devices.
6
1.2.4 Deployment and Maintenance Cost
Finally, the energy requirements and budget of UWSNs are high compared to higher
power and regular battery replenishing techniques are quite costly. Analysis on the
deployment schemes are given in details in literature [15,16].UWSNs have brought
important areas of research for researchers due to the immense challenges in every
aspect of underwater communication systems. However there is still lot of work to
be done with the passion of innovation and knowing more about the hidden fact of
oceans.
1.3 Classification of UWSNs
There has been many different criteria for classification of Underwater sensor net-
works but in this thesis these are classified on the basis of network architecture. This
network architecture classification provides important road map while deciding the
deployment for a certain application [17,18].
1.3.1 One Dimensional UWSNs
In one dimensional networks sensor node are autonomous stand alone entities. Each
sensor node is liable for detecting, processing, and communicating the information
to the distant base station. Node can either be a floating buoy or a deep diving de-
vice in underwater, communicating with acoustic, Radio Frequency (RF), or optical
links.
1.3.2 Two Dimensional UWSNs
Flat 2D-UWSN architecture refers to a network where a group of sensor nodes
(cluster) is deployed underwater. Each bunch holds a cluster head (also called an
anchor node). The clusters are defined as they are anchored at the submerged sur-
7
face. Each member of the cluster gathers the underwater data and transmits it to
the anchor client. The anchor node gathers the info/data from all its member nodes
and relays it to the surface buoy nodes. In 2D-UWSN, the communication is ex-
tended in two dimensions; that is, (i) each member of the cluster communicates
with its anchor node with horizontal communication link while (ii) the anchor node
communicates with the surface buoyant node with vertical communication link. In
2D-UWSN, acoustic, visual, and RF communication can be applied depending on
the type of application and nature of the underwater environment In 2D-UWSN,
acoustic communication is preferred for underwater anchor node and the surface
buoyant node due to the typically higher distance between them. The 2D-UWSN
can be employed for both time-critical and delay tolerant applications.
1.3.3 Three Dimensional UWSNs
In this type of network, the detectors are deployed underwater in the pattern of
clusters and are anchored at different depths. Referable to the deployment of the
sensors at variable heights, the communication between the sensors goes beyond
the two proportions. In that respect, there are three communication scenarios in
this architecture: (i) inter cluster communication of clients at different depths, (ii)
intra cluster (sensor-anchor node) communication, and (iii) anchor buoyant node
communication.
1.3.4 Four Dimensional UWSNs
UWSN is designed by the combination of fixed UWSN, that is,3D-UWSN and
mobile UWSNs. The mobile UWSN consists of remotely operative underwater
vehicles (ROVs) to collect data from the anchor nodes and relay the data to the
remote station. ROVs can be autonomous submersible robots, vehicles, ships, and
even submarines. Each underwater sensor node can be autonomous in relaying the
8
data directly to ROVs depending on how close that particular sensor node is to
the ROV. The communication scenario between ROV and underwater sensor node
depends on the distance and data between them and either acoustic or radio can be
used. As the transmission is to be directly relayed to ROV, the sensors which have
large data and are close to ROVs can use radio links while the sensors which have
small data to transmit or are far from ROV can use acoustics links.
1.4 Applications of UWSNs
UWSNs are quickly getting ubiquity for empowering progresses in the areas of
ocean monitoring and observatory systems, profound ocean surveillance, tracking
of different substances of the oceanic environment [18]. Submerged sensor systems
can have multifarious applications ranging from oil industry to aquaculture, from
exploration and monitoring to contamination control, environment recording, seek
and study missions etc. A comprehensive study about the potential applications of
UWSNs is given in [19,20].
1.4.1 Monitoring and Exploration Application
Underwater monitoring and exploration alludes to a sensor network deployed to
monitor the underwater environment. These applications are especially identified
with observing the physical environment. The application are further categorized
into following main groups and given in details.
Water Quality Monitoring
Water is valuable asset and is chief variable for survival of living things under or
over the water surface. That is why, it is imperative to monitor and screen the nature
of water. The underwater quality observing applications fluctuate from checking
water nature of waterways to, lakes, rivers and oceans [21].
Marine Life Monitoring
9
Marine life observing application manages the checking of environment of different
species living over the water or submerged. ACMENet [22] is a submerged acoustic
sensor network which is equipped for observing marine environment, like fishes
and any human or nonhuman movement inside of the scope territory. The creators
have added to a convention taking into account TDMA master slave mechanism for
information exchange over the sensor hubs.
Fish Clusters Monitoring
Fishes in underwater live in the form of cluster and storms. Study of special type
of fishes and their natural behaviour over the longer period of time can be realized
using underwater sensor network with mobile and fixed sonobuoy.
Underwater Exploration
There are substantial number of minerals present in underwater environment which
are required to be investigated. Immersed cables are installed, which carry some of
the most basic social necessities such as oil, gas pipelines, and fiber optic cables.
Therefore UWSN can be used to explore the precious resources and also to monitor
the underwater pipeline and transmission cables. Underwater explorations and ap-
plications are further classified into (i) underwater natural resource searches and (ii)
cable and pipeline observing applications for submerged oil and gas assessments
1.4.2 Military Applications
UWSNs are utilized to help military applications also. These frameworks take the
guide of various sensors conveyed for discovery of various parts of military appli-
cations. Distinctive sensors, for example, cameras, imaging sonar, and metal iden-
tifiers incorporated with AUV are utilized to help with finding submerged mines,
securing ports, and submarines and are likewise utilized for checking and recon-
naissance. These applications can prompt monetary answer for ensure maritime
powers.
Mine Detection
10
Since a sensor can sense physical parameters, it is an intellectual exercise to detect
hidden mines underwater. This can assist the military ships in a trouble-free voyage.
The mines are usually made up of ferrous materials and they can be differentiated
from underwater clutter based on the fact that the clutter is a nonmetal. Thus, using
metal detecting sensors can help find clutters underwater.
Surveillance
UWSNs are additionally utilized for observation purposes where the invasion of any
undesired element is taken note. This can be utilized for close and seaward observa-
tion purposes. This takes in an enormous extent of utilizations offered for seashore
reconnaissance, for example, detection of battle ships and arrival of logistics.
Submarines
Target location is another application of UWSNs that takes the utility of AUVs to
localize submarines. Schematic ways of submarine localization require heavy com-
putation. These UWSNs based systems have a tendency to provide cost-effective
solutions to the localization complications.
1.4.3 Assisted Navigation for Deep Sea Exploration
Submerged environment is exceedingly uneven, unexplored, irregular, and dim with
expanding profundity. In such environment there is need of help for exploring the
vessels, water crafts, dispatches, and significantly swimmer and voyagers. Explor-
ing collaborative innovations which are most regular over the surface of water are
not utilized submerged because of the adjustment in medium of television. Subse-
quently, in parliamentary strategy to find, scout, and explore, there is a need of col-
laborative route advancements, framework, and application which could be utilized
submerged . In this prospect, UWSN can be used to give helped route frameworks
and applications [23].
11
1.4.4 Disaster Monitoring and Prevention
Broadly, the natural cataclysms are inexorable. Among others, water based natural
disasters are more dangerous and created huge devastation to the world. In view of
that, disaster monitoring and preventive measures are really necessary. UWSN of-
fers a broad range of applications for management and retrieval of such calamities.
More markedly, it links to the monitoring of events that intensify a disaster reper-
cussion. Along with inadequate resources for comprehensive monitoring of the vast
expanse of water (e.g., Sea), the exertion becomes even more challenging with oc-
casionally callous meteorological conditions. Thus, effective monitoring of marine
and aquatic dynamics is a substantial research challenge [24] . UWSN monitoring
strategies for disaster management and prevention can be developed into a broad
assortment of applications such as floods, underwater volcanic outbreaks, and un-
derwater quakes and their subsequent tsunamis, and oil leaks which lead to above
the water and underwater biological instabilities.
Flood Alerts
The aftermaths of a deluge and its expanded rate have pushed the researchers to
discover systems for convenient flood forewarnings. The forewarnings need not just
be situated in urban shores and henceforth call for remote arrangements. UWSN
creates arrangements of submerged sensor organizations with over-the-water hand-
off operators to gauge river vitals. These vitals assemble at the remote station and
investigated for flood indication.
Earthquake and Tsunami
Submerged earthquakes and volcanoes are common calamities and foundations for
jeopardizing living things. These normal fiascos can happen at whatever time and
anyplace over the surface of the earth and are yet all the more disturbing when they
occur submerged, contingent on the seismic and the land changes that occur under
the earth. In this way, it is vital to screen such conditions.
Oil Spills Monitoring 12
Man-made pollution is an important element to look at when talking about marine
life health. Aquatic life is extremely affected by oil spills, contamination and there-
fore UWSNs have contributed a way to find out the location and wideness of oil
spills in water which can accelerate the cleanup process.
1.5 A Typical UWSN and Solutions
In a typical underwater network consist of many different sensing, communicat-
ing and moving devices deployed for different purposes. One or more sinks are
deployed on the surface. These sinks are equipped with both acoustic modem RF
modem. RF modem is used for inter sink communication or communication be-
tween sink and on shore data centers. In some applications data centers can be far
away from the shore, in these scenarios surface sink can be capable to communicate
with satellite link to far away data center; virtually anywhere on the planet. Some
sinks can be employed for underwater data gathering applications. These devices
are termed as AUV, Mobile Sink,or Courier Node. In some applications nodes are
fixed with some vertical anchor. Position of the anchor nodes is exactly known and
these nodes collaborate for the localization of other unanchored nodes. Sensing and
Forwarding nodes can be sperate, or same node can be utilized for both purposes.
A typical underwater network scenario is shown in Figure 1.1. In this section some
current available solutions for underwater sensor networking are introduced and,
are explained in details. Schematic diagram of a typical underwater node is also
given in Figure 1.2. It consist of sensing unit, power supply unit, transcending
unit, and mobility unit.
1.5.1 Subnero Solutions
Subnero solution includes a software defined modem working at carrier frequency
of 27kHz, bandwidth of 18kHz and data rates 0.5 kbps for command links and
13
Figure 1.1: Underwater Communication Scenario
1
Figure 1.2: Schematic Diagram of a Node
2−10 kbps for data links. Error correction codes and modulation formats are soft-
ware defined, customise able with default settings as 1/3 rate convolutional and
PSK OFDM respectively. It also provides the communication range of about 2
14
to 3km, with 0.1m ranging precision for subaquatic networking applications. An-
other solutions of subnero is SWAN (Subnero Water Monitoring and Networks); a
duck shaped floating node equipped with in-situ water quality measuring probes,
navigation, sensing and realtime date communication capabilities. Since underwa-
ter environment lacks GPS facilities so for subaquatic navigation subnero provides
UPN (Underwater Positioning Network) in which deployed beacon nodes collabo-
rate with each other for the position estimation of a moving AUV. This provides the
facility of localization in underwater for mobile nodes and deployed nodes.
1.5.2 DSPComm Solutions
DSPComm is a leading company which provides solutions for underwater connec-
tivity 2. AquaComm is subaquatic wireless modem manufactured by DSPComm.
It is the best option for underwater communications when utmost consistency and
flexible design combination are crucial. The product is available in 100 bit/sec and
480 bit/sec varieties and in two arrangements. AquaNetwork is an improved Aqua-
Comm with networking resources in harsh underwater environments. AquaCase
is for casing requirements to avert equipment form severe underwater atmosphere.
AquaTrans is a hydrophone for underwater communication with omnidirectional
antenna and can be deployed up to 500m depth for communication purpose. In
order to facilitate the end users with data logging facilities DSPComm provides
AquaStore; ability to be connected with AquaComm modem and provide analog
inputs. It is also accessible in sleep mode and has ability to connect various analog
input sensors with modem.
Rest of the thesis is organized as follows: In chapter 2 we have provided an exten-
sive literature review about the routing protocols especially available for underwater
sensor networks. Problem statement and Motivation for this protocol is explained
in chapter 3. Event segregation approach, network initialization and node deploy-
2http://www.dspcomm.com/
15
ment, channel model, improvements in terms of forwarding function and holding
time and mobility of courier nodes are explained in sections given in chapter 4.
Chapter 5 contains simulation setup and performance evaluation with explanation
of routing metrics. Finally conclusion and future work is given in chapter 6.
16
CHAPTER 2
Literature Review
In this chapter we present a comprehensive literature review, and classification of
wireless network protocols. Mobile Ad hoc Networks (MANETs) are introduced
and their classification is explained in details. Moreover the imperfections in using
them in underwater wireless environment are also explained. Wireless networks
have gained much popularity due to some attractive features, especially the mobil-
ity of connected devices [25,26]. These networks are broadly categorized into two
types.
(1)Infrastructure based networks; in which a fix and central control station is avail-
able to provide connectivity and routing.
(2)Mobile Ad-hoc Networks also known as MANETs in which no centralized sta-
tion is available.
2.1 Ad-hoc Network Protocols
MANETs have quite complex and distributed systems in which mobile devices can
be connected and self configured into temporary and arbitrary network topology.
The surveys and reviews on some routing protocols in MANETs are given in liter-
ature [27,28]. Some key features of MANETs are listed below:
•Since there is absence of centralized infrastructure, network topology remains
varying and dynamic in nature, control and operation is done collaboratively
and distributively in ad hoc networks.
•A node in ad hoc networks acts as end host and as switching or routing device
at the same time.
•If two communicating nodes are far away a third node can be employed as an
17
intermediate node for multi hop communication process.
Routing protocols in MANETs are capable to consider the mobility of the nodes
and their role as routing and relay devices for multi hop communication. These
protocols are broadly divided into two categories: proactive and reactive [29,30,
31].
2.1.1 Proactive Routing Protocols
In proactive or global routing protocols-like OLSR (Optimized Link State Rout-
ing) [32] and DSDV (Dynamic Destination Sequenced Distance-Vector) [33]-nodes
routes are made initially irrespective their requirement to a certain node. Nodes
periodically share their routing table information with each other and routes are
updated and maintained by using some periodic route updating mechanisms. The
contrast between these conventions exist in the way the routing information is up-
graded and the kind of data kept at each routing table. Most global routing protocols
cannot be scaled very well because their updating procedure consumes a substan-
tial amount of network bandwidth. Some other routing protocols in this category is
given in following [34].
WRP: Wireless Routing Protocol
GSR: Global State Routing
FSR: Fisheye State Routing
STAR: Source Tree Adaptive Routing
DREAM: Relative Distance Micro-Discovery Ad Hoc Routing
CGSR: Cluster-head Gateway Switch Routing
HSR: Hierarchical State Routing
18
2.1.2 Reactive Routing Protocols
In reactive routing protocols,like AODV [35] and its variants like MR-AODV (Mul-
tiple Route AODV) [36] and DSR (Dynamic Source Routing) [37]; routes are es-
tablished on the basis of demand initiated from source. These protocols have some
attractive features over their proactive counterparts in terms of reduced broadcasting
overhead. However this route establishing mechanism would be slow and causes
larger delay due to low propagation speed of acoustic signals. So these protocols
cannot be used for underwater environment. These protocols are enlisted here and
are given in details in [34].
ROAM: Routing On-demand Acyclic Multi-path
TORA: Temporally Ordered Routing Algorithm
CBRP: Cluster Based Routing Protocol
FORP: Flow Oriented Routing Protocol
ARA: Ant-colony-based Routing Algorithm
Wireless sensor networks is the branch of ad hoc networks in which some deployed
nodes in far areas sense the environment, and respond to a central base or action
center. There have been lots of routing schemes in terrestrial wireless sensor net-
works in literature. These schemes are unable to be deployed in harsh underwater
environments.
2.2 Underwater Routing Protocols
There have been many proposed solution for routing in underwater scenarios. These
schemes are broadly divided into following two categories. Different aspect of these
schemes are given in details .
19
2.2.1 Localization Based Routing Protocols
In these protocols position information of each node is calculated with some loca-
tion finding algorithm. There have been many localization algorithms in literature
[38,39,40,41,42]. A modified version of DSR for underwater environment named
as LASR (Location Aware Source Routing) is proposed in [43]. LASR is on de-
mand routing protocol to cope with high latency, high probability of packet loss. It
takes into account location and link quality metric during routing process. AODV-
BI [44], modified version of AODV for underwater environment is proposed to
realize on demand routing in underwater environment with decreased overhead of
control messages.
Vector Based Forwarding (VBF) was proposed in [45], which aims to increase net-
work lifetime taking into account mobility of nodes. A hypothetical vector, which
makes a virtual pipe of radius ’R’ from source to destination is assumed. Nodes with
in the range of vector are selected as candidates for forwarding process. Instead of
having a single virtual vector (vector) in Hop-by-Hop VBF [46]-an improvement in
VBF-many hop to hop vectors (pipes) are introduced, in which each node can adap-
tively select better routing pipe. This overcome the low packet delivery issues in
VBF specially in sparse networks. Establishment and maintenance of Hop-by-Hop
vectors on each hop is an extra overhead especially in dense deployment scenarios.
A distributed self adaptation algorithm to reduce traffic load is utilized. In this al-
gorithm all candidate nodes coordinate to calculate desirableness factor, and some
most "desirable nodes" are directed to forward the packet.
EVBF (Efficient-VBF) [47] is an extension and improvement in VBF, in which
"desirable-ness" factor αand Adaptation Time Tadaptation are utilized in forwarding
decision and selecting most suitable forwarder. This scheme take into account and
workable for both sink based queries and source generated queries. Another variant
of this scheme is LE-VBF (Lifetime-Extended VBF) [48] in which desirableness is
redefined employing position and energy. Routing pipe radius threshold has very
20
much effect on the efficiency of scheme. VBF also requires localization information
and extra equipment to measure Angle of Arrival (AoA) which is an overhead and
causes extra energy consumption. Nodes closer to the axis of routing pipe become
hot spot and their faster energy depletion causes void regions.
In order to overcome the void area problems faced in VBF, and HH-VBF, Vector
Based Void Avoidance (VBVA) is proposed in [49]. VBVA has capability to detect
a void and bypass it on demand and hence does not need to keep topology infor-
mation. Void avoidance is realized using two mechanisms: vector-shift and back
pressure. Void is detected if a node finds no neighbour with "advance" greater than
its own "advance". Vector shift method allows nodes out of routing pipe to forward
the packet and hence a shifted vector is formed. If a node cannot find neighbours
to shift; it broadcasts a Back Pressure packet to reroute the packet with better path
which has no void. However void detection process is only capable to detect void on
edge of the network and has no solution if void occurs in middle of networks due
to unexpected deaths of intermediate nodes. Moreover backward flooding would
cause extra over head especially in sparse networks.
A power control based cross layer approach named as Focused Beam Routing
(FBR) is proposed in [50]. In FBR sender is allowed to select its transmit power
level from P1to Pn. Next forwarders are selected from a predefined cone using RTS
(request to send) and CTS (clear to send) frames, and source Sto destination Ddis-
tance DS. Sender node makes cone closing end with cone angle θ, and length of
cone is equal to transmission range. Next hop forwarders are selected from ±θ/2
and nodes within this range reply with CTS packet. But cone angle is independent
of link quality which causes decrease in throughput especially in sparse networks.
CTS packets from different node candidate forwarders may collide due to lack of
collision avoidance mechanism for these types of packets especially in denser net-
works.
SBR-DLP (Sector-based Routing with Destination Location Prediction) [51] is also
21
localization based approach in which fix located destination nodes are not required
and there is relaxation in terms of precise localized information. It has mechanism
for predicting the location of destination node. Unlike FBR in which a node locates
next forwarder from a cone, sender in this scheme may selects next forwarder from
whole communicate circle. Localization information requirements and preplanned
sink mobility awareness to all source node causes extra overhead.
Depth Control Routing (DCR) [52] is another geographic routing protocol which
overcome the void area issue by adjusting the depth of disconnected nodes using a
centralized algorithm. However criteria to determine disconnected nodes requires
global vision of network which increase control overhead. Moreover newer position
of these nodes are calculated on some central station and than needed to be informed
to each node. Moving to newer depth itself requires mechanical coordination among
nodes.
Link quality based control and directional flooding protocol is proposed in [53]
named as DFR (Directional Flooding based Routing protocol). Link Quality is
measured in terms of Expected Transmission Count (ETX) and flooding rather than
established paths is employed during routing process. Flooding zone is locally
formed based on link quality of neighbour nodes. If link quality is poor it involves
more nodes in forwarding process otherwise few sufficient nodes are employed. It
involves two type of angles: Current Angle of node i CAiand Reference Angle of
previous sender k RAk, in forwarding process. It also provide the mechanism for
adaptive change and improvement in RAkby each forwarding node. Geographi-
cally controlled flooding zone and directional (towards sink) forwarding based on
link quality ensures better packet delivery. But this scheme requires localization
information which is itself a challenge in UWASN. Link quality information get-
ting process for each node increases control packet broadcasting and causes extra
energy consumption. DFR does not have efficient mechanism for redundant packet
avoidance.
22
In Depth Adaptive Routing Protocol (DARP) [54] authors considered the fact that
speed of sound varies with depth and proposed shortest delay path for deep sea
(about 8000m) underwater environments. Relative Distance Based Forwarding
(RDBF) is proposed in [55]. In RDBF fitness factor is normalized by transmis-
sion range to get the value in range of [0 1]. Holding time is also dependent on
transmission range and speed of sound in acoustic environment.
Multi Path Routing (MPT) proposed in [56] ensure balance between energy con-
sumption and end to end delay for time critical underwater applications. Its a cross
layer approach which employs: (1)power control on physical layer according to
channel status and network conditions, (2)Packet combining mechanism on sink
for multiple copies of corrupted packets received from different paths. MPT per-
forms well in dense deployments scenarios while suffers void regions in sparse
networks
All the above mentioned localization based protocol requires some collaborative
mechanism for finding the location of deployed nodes. Localization process in
harsh underwater environment is itself a research challenge. It also requires some
certain number of node to act as Anchor Nodes (AN). Anchor nodes are the nodes
which are attached with some support and their location is known. Unknown loca-
tion nodes communicate with more than one anchor nodes and try to estimate their
own location. Error in localization can leads to erroneous communication. Hence
its preferred to have localization free routing protocol for UWSN.
2.2.2 Localization Free Routing Protocols
This category consists of protocols which are either completely free of any local-
ization information (Work on hop count base) or they may require partial localiza-
tion information (DBR requires only depth). A localization free routing protocol
(named as DBR:Depth Based Routing) for underwater sensor network is proposed
in [57]. In DBR exact only depth information is enough for routing decisions and
23
exact location information of node is not required. In this protocol a receiving node
from the set of potential forwarders utilizes its depth and depth threshold (a global
parameter) to make the forwarding decision in packet forwarding process.
Forwarding decision is done individually by each receiving node on the basis of
receiving time of a certain packet and holding time of that node. Holding time is
assigned to each node in such a way that node with lesser depth has less holding
time and more probability to be forwarder. Other nodes are prohibited to forward
the packet by assigning holding time values more than the optimal forwarder value.
Two Queues "priority queue Q1" and "Packet History Buffer queue Q2" and holding
time are collectively utilized to suppress the redundant packet forwarding. DBR
performs well to enhance the stability period but when nodes start dying there is no
mechanism to rescue and prolong the instability period (time period between the
death of first and last node). Throughput of DBR in sparse networks is relatively
low, which may require to implement multiple sinks. Additionally if two nodes
have same depth then there is no mechanism to suppress the transmission redundant
packet.
Depth Based Multi-hop Routing(DBMR) [58] is a variant of DBR which strives
to reduce flooding and redundant packet transmission issues in DBR. In DBMR
multi-hop communication is realized during forwarding process. WDFAD-DBR
(Weighting Depth and Forwarding Area Division DBR) is proposed in [59]. Void
area occurrence is reduced by selecting next forwarder considering the depth of cur-
rent node and expected next forwarding node. Weighted sum of depth difference
of two hops have role to predict the holding time. There is mechanism to change
the forwarding region adaptively according to node density and channel condition.
However neighbour node depth prediction error would lead to increase control over-
head.
Another depth based routing scheme is hydraulic pressure based anycast routing
called: HydroCast, [60]. In HydroCast firstly an opportunistic routing mechanism
24
is proposed which reduces co-channel interference and ensures maximum delivery.
Secondly a simple depth based void area avoidance mechanism is proposed. In
hydroCast void nodes maintain the recovery route to a node who’s depth is lower
than its own depth. Then the void node packet is rerouted out of void using greedy
forwarding approach.
Opportunistic directional flooding is realized in Void Aware Pressure Routing (VAPR)
[61]. VAPR operation of two parts: enhance beaconing and opportunistic direc-
tional forwarding. Additional information such as sender depth, hop count to sonobuoy,
sequence number and forwarding direction, is added to each node’s beacon in en-
hance beaconing phase. VAPR employs local greedy directional forwarding and
utilizes a factor called surface reachability as forwarding metric. A Q-learning
based cross layer approach is proposed in [62].
Instead of hop count or depth Q-learning routing path selection process includes
traffic load, latency and queue length. Q learning in distributed nodes ensures bal-
ance energy consumption and load balancing. Path Unaware Layered Routing Pro-
tocol (PULRP) with uniform deployment of nodes and with non uniform deploy-
ment of nodes is proposed in [63]. It consists of two phases, in first phase called
layering phase, Spherical layers of radius Rrwith sink as layer 0 are formed and
layering number is increased by each node. A node in layer l is to find forwarder
from layer l−1 on the basis of transmission range and layering radius. Communica-
tion between nodes is done between potential relay nodes from each layer. Energy
optimized-PULRP [64] is an enhancement in which so called on the fly routing is
considered to select relay node. E-PULRP incorporate energy in routing decisions,
which was not consider in previous versions.
An improvement in DBR named as EEDBR is proposed by A Wahid in [65]. Un-
like DBR in this scheme routing decision is done by sending node using residual
energy and depth difference with neighbours. holding time is based on residual en-
ergy of the nodes. Higher energy nodes correspondingly have lower holding time
25
and vice versa. EEDBR performs better than DBR in terms of network lifetime
and throughout. But as next forwarder can be of lower depth and it involves rela-
tively more number of nodes in relaying process than DBR, which increases energy
consumption and delay.
Hop-by-Hop Dynamic Addressing Based H2-DAB [66,67] routing protocol is an-
other localization free protocol. In H2-DAB nodes are given different size addresses
directly depending on depth of nodes. Deeper nodes have large size addresses and
vice versa. CRP (Connectivity based Routing Protocol) protocol is proposed in
[68] in which next forwarder node is directly related to CI (Connectivity Index) of
neighbour node. where CI is defined as "number of neighbouring nodes closer to
sink from a forwarding node".
Hop count based Cross layer approach CARP (Channel Aware Routing Protocol)
is proposed in [69,70]. On the basis of recent history of successful transmissions
a particular node is selected a next forwarder. CARP is hop count based proto-
col with varying power levels. An energy efficient routing protocol for internet of
underwater things and an improvement in CARP is presented in [71], named as
Enhanced-CARP. E-CARP consider the reusability of sensed data in certain for-
warding applications. PING −PONG strategy is simplified in selecting the most
appropriate relay node at each point, during the steady network topology.
Reputation based Channel Aware Routing (R-CARP) [72] is another improvement,
which considers BLS, a short digital signature algorithm, to reduce communication
overhead and secure the routing protocol. R-CARP is a secure and reliable com-
munication protocol which contrast the malicious node behaviour. Another local-
ization free scheme AMCTD (Adaptive Mobility of Courier nodes in Threshold-
optimized DBR) is proposed in [73]. In this scheme forwarding of packet is im-
proved and based on weight, which is dependent on depth, residual energy and
some priority values assigned to them. CN are vertically moveable nodes with no
energy constraints on them.
26
An improvement in AMCTD is proposed in [74] iAMCTD (improved AMCTD)
in which different routing metrics are defined depending on the depth of region.
These metrics include Localization free-SNR (LSNR),Signal Quality Index (SQI)
and Energy Cost Function (ECF) and Depth Dependent Function (DDF). Moreover
hard and soft threshold based simulation scenario is adopted in which a sensed
event’s value greater than hard threshold is reported to sink. Moreover authors
employed Courier Nodes (CN) in [73,74] for data gathering tasks. Here mobility
of nodes start when certain fraction of total nodes have died. which is not efficient
and full use of deployed resources. Additionally holding time calculation process
does not guarantee optimum value of holding time ensuring packet suppression and
minimum end to end delay.
27
CHAPTER 3
Problem Statement and Motivation
This chapter highlights some problems in previous schemes and explains the moti-
vation for our proposed scheme especially about event segregation approach.
•Initially, we present a comprehensive analysis on the ranges of forwarding
functions defined in previous schemes. We have identified problems in those
forwarding functions and redefined them in next section. These redefined
forwarding functions causes much better and systematic forwarding.
•In second part of this section we explain broadcasting overhead issue in deep
sea acoustic networks and motivation for event segregation approach.
3.1 Forwarding Function Analysis
Some more aspects of communication in underwater are neighbor selection, for-
warding process (function) and holding time calculation. Holding time is the time
for a certain node during which it avoids any transmission and waits for the more
appropriate forwarder to transmit the packet. Its value for each node is related to
forwarding function of the underlying scheme. Difference of values of holding
times for two successive nodes should be sufficiently large enough to avoid any
collision. On the other hand very large value of holding time causes delay during
transmission process. This delay may be negligible on a single node but in multi
hop communication accumulated sum of all delays eventually causes larger consid-
erable delay. So an optimum value of holding time is required to ensure redundant
packet suppression and minimum delay.
In the following paragraph we have considered the neighbour selection process,
forwarding function and its impact on holding time. DBR and other depth based
28
protocols select neighbors on the basis of depth difference ∆dbetween previous
node depth dpcurrent node depth dc. In this selection process nodes with lesser
depth have more chances to become next forwarder even with less remaining en-
ergy. This eventually causes the premature death of relatively lower depth nodes
and creates void regions. Holding time in DBR is also function of depth. Depth of
broadcasting node can have range from at least TRto Dmax which directly affects
the holding time calculation. So this fails to explain the upper and lower bounds
on possible holding time values while considering the delay and packet suppression
tradeoff.
In EEDBR a sender node maintains depth, residual energy and ID information of
its neighbours. Neighbours with smaller depth are selected and their list of IDs is
included in data packet sent by sender. Receiving nodes hold the packet for holding
time which is based on residual energy Er. This may create problems as two nodes
with same residual energy would have same holding time and could transmit at the
same time. Moreover EEDBR involves more number of hops in forwarding process
and eventually brings increased delay and enhanced energy consumption.
Weight based forwarding function in AMCTD is dependent on both remaining en-
ergy and depth difference and is given in following equation:
W1=pv×Er
Dmax −Di
(3.1a)
where Erand Diare the residual energy and depth of ith node respectively, pvis
priority value which is a system parameter. Dmax is the maximum depth of under-
water network. Weight calculation is updated after 2% death of total nodes using
following equation.
W2=pv×Di
Er
(3.1b)
29
Weight calculation is updated after 80% death of total nodes by equation 3.1c.
W3=Er
pv×Di
(3.1c)
In order to predict the forwarding behaviour in network and its impact on holding
time of nodes it is convenient to have knowledge of domain and range of forwarding
function. We have done comprehensive analysis on the respective forwarding func-
tions of each scheme considering the upper and lower bounds of affecting metrics.
This analysis is explained as, if deepest layer of network is considered as reference
level, hi=Dmax −Digives the information about the height of node. A node during
its lifetime can have energy between 0 and Eo. Minimum and maximum values of
weight based forwarding functions are calculated using set of equations given in
3.2.
min(W1) = min(Er)
max(hi)max(W1) = max(Er)
min(hi)
min(W2) = min(Di)
max(Er)max(W2) = max(Di)
min(Er)
min(W3) = min(Er)
max(Di)max(W3) = max(Er)
min(Di)
(3.2)
For a unit value of pvall the forwarding function Wicalculated for AMCTD using
Table 3.1: Parameters values used in previous schemes
Parameter Name Value
Initial Energy (Eo) 70J
Volume of the Network 500 ×500 ×500×
Data Packet 64B
Control Packet 8B
Number of nodes |N|225
Number of courier nodes |C|4
Maximum Depth (Dmax) 500m
Depth Margin 1(D1) 150m
Depth Margin 2(D2) 350m
Depth threshold 1 (dth1) 60m
Depth threshold 2 (dth2) 40m
Depth threshold 3 (dth3) 20m
Transmission Range (RT) 100m
equation 3.1 has range in interval [0,∞). Holding time values depending on this
forwarding function would vary abruptly and would not meet the optimum delay
and reduced redundant transmission requirements. In AMCTD depth threshold dth
30
is also updated after certain percentage of deaths, but weight is independent of dth
and transmission range RTof nodes. Anyhow if there comes a lower bound on Er
defined as min(Er≥1)and Didefined as min(Di≥1)it causes a significant change
on the upper bound of all the respective forwarding function Wias in Table 3.3.
Since forwarding function of AMCTD are independent of depth threshold so max-
imum values are same regardless of dth. Dimensional analysis of each forwarding
function describes that W1and W3are dimensionally equivalent (=Newton) while
W3is dimensionally reciprocal of W1and W3.
Table 3.2: Ranges w.r.t depth threshold
Properties Minimum values Maximum Value
dth =20 dth =40 dt h =60
Depth Difference (li)20m40m60m100m
Absolute Depth (Di)100m100m100m500m
Absorption Coefficient (Abc) 0.1990dB 0.3980dB 0.5969dB 0.9949dB
Scattering Coefficient (Sc) −25.485dB −20.969dB −18.328dB −15dB
(Scm) 19.5154dB 24.0309dB 26.6723dB 30dB
Total Attenuation (At) −25.286dB −20.571dB −17.731dB −14.005dB
19.7144dB 24.4289dB 27.2692dB 30.9949dB
Attenuation A(l,f) 0.003 0.0088 0.0169 0.0398
93.6m−1277.3m−1533.2m−11245.4m−1
Noise N(f)20.925dBreµPa or 1.12 ×10−5Pa
3.29 ×10−89.75 ×10−81.87 ×10−74.42 ×10−7
Attenuation Noise Product A(l,f)×N(f)1×10−33.1×10−35.9×10−314.1×10−3
Forwarding function in iAMCTD is defined differently for different depth regions.
SQI based forwarding function is defined in equation 3.3a.
SQI =LSN R ×Er
li
(3.3a)
In this equation liis depth difference between two successive nodes. LSNR is
Localization free Signal to Noise Ratio given by equation 3.3b.
LSNR =P
t
A(l,f)×N(f)(3.3b)
where P
tis transmitted power and A() ×N() is Attenuation Noise factor, which is
product of path loss and ambient noise. A(l,f) is the sum of absorption coefficient
(Abc) and scattering coefficient (Sc) whose minimum and maximum values for dif-
31
ferent depth thresholds are given in Table 3.2.
SQI based forwarding decisions are made if depth of node is less than D1which
is top shallow region of underwater environment. This region involves more ship-
ping activities. We have considered the statistical fact that variance is the mea-
sure of randomness in a set of observations. Careful analysis of SQI reveals that
the range of output from this function is in the order of 105as shown in Table
3.2. Such values must not be used directly as forwarding function as holding
time calculated from these values would have high value of variance. So these
values are needed to be linearized before being used as forwarding function and
holding time. This is explained as follows: If f=30kHz and depth difference l
has range in the interval [dth =60,R=100] absorption coefficient given in equa-
tion 4.6c has range [0.5969dB 0.9949dB]. A(l,f) eventually comes out with in
the range of [533.2 1245.4]. For f=30kHz noise N(f) from equation 4.7 comes
out to be 1.12 ×10−5using this, Attenuation Noise product range is calculated
and finally LSNR is in range of [4.52 ×1061.06 ×107] making SQI in range of
[0 1.01 ×105]. This analysis is for dth =60mbut when dth is modified to 40 and
20 value of forwarding function for different nodes would be with larger values of
variance. Direct use of values in this range is impractical and is needed to properly
linearized. In our solution linearized version of SQI is proposed to be used as for-
warding function. ECF based forwarding function is utilized when depth of node is
between D1=150 and D2=350. ECF is defined as:
ECF =pv×Er
Di
(3.3c)
where Diis the absolute depth of sensor node which can be between D1and D2.
Value of Eris between 0 and 70Jand this results ECF values between [0 70] which
is very different from the range in SQI based forwarding function. ECF is also
modified to include transmission range and Dth. If node is in deepest region of
32
networks with depth greater than D2DDF based forwarding function is employed
using following equation.
DDF =SQI ×li
Di
(3.3d)
min(ECF ) = min(Er)
max(Di)
max(ECF ) = max(Er)
min(Di)
min(SQI) = min(LSNR)×min(Er)
max(li)
max(SQI) = max(LSNR)×max(Er)
min(li)
min(DDF) = min(SQI)×min(li)
max(Di)
max(DDF) = max(SQI)×max(li)
min(Di)
(3.4)
Range of DDF is also given in Table 3.3. For a certain value of depth threshold dth
Table 3.3: Forwarding Functions Ranges
Scheme Forwarding Function Minimum Value Maximum Values
dth =20 dth =40 dt h =60
W10.0020 70 70 70
AMCTD W20.0143 500 500 500
W30.0020 70 70 70
LSNR 4.52 ×1066.07 ×1072.05 ×1071.06 ×107
LSNR 142.98 1.92 ×103648.46 337.18
SQI 0 3.03 ×1051.33 ×1051.01 ×105
iAMCTD ECF 0 70 70 70
DDF 0 3.03×1041.33 ×1041.01 ×104
maximum and minimum values of DDF are roughly one tenth of the corresponding
SQI function. Maximum and minimum values of forwarding function used in iAM-
CTD are calculated using set of equations given in 3.4. Holding time calculated
for AMCTD and iAMCTD is given in following equations:
HT= (1−FF)×Htmax (3.5)
Holding time calculation depends on two factors Htmax and forwarding function FF.
There is no appreciable explanation of Htmax which is termed as system parameter
in previous literature. Whereas values for different forwarding functions can have
33
range very much different from each other and these values are not even comparable
with each other. Holding time calculated with these forwarding functions has very
abrupt fluctuations for different depth nodes which causes increased delay in some
regions and redundant transmission and collision in other depth regions. So there is
need to have a relation providing comparable values of holding times. At least one
must be able to predict the range of values to be assigned as holding time.
3.2 Motivation for Event Segregation
Much of contemporary work in underwater sensor networks strives to have an ef-
fective routing mechanism ensuring decreased energy consumption and increased
network lifetime. Each node strives to find a set of best possible nodes for the suc-
cessful delivery of packet to final destination i.e. on surface sink. This situation
(forwarding to surface sink) would be appropriate if data generating nodes are only
a fraction of all deployed nodes. However in some applications in which all the de-
ployed nodes are equally probable to sense and generate data packets, this approach
would cause increased broadcasting overhead. Each node has its own data plus it
may has to forward the packets of other deeper nodes. This broadcasting overhead
issue becomes prominent in deep underwater environment which involves compara-
tively greater number of hops in forwarding process. In this scenario node would be
wasting its resources by causing congestion and increasing probability of collision
of packets. Hence it is no more a wise option to broadcast all the generated data to
on surface sink. There is a need to decrease the broadcasting load on intermediate
nodes to ensure optimized energy consumption. This motivates for the formulation
of event segregation approach explained in details in the next section.
One possible solution to decrease broadcasting load in some recent work [74] is to
deploy certain number of underwater sinks (AUV or CN) which can go to the ap-
propriate location for data collecting jobs. But it requires proper tour planning and
34
coordination between nodes and sinks. Nodes may have to keep data for sometime
till the sink arrival requiring extra memory. It would also cause increased delay
and more susceptibility of congestion, collision and increased packet loss. It also
requires a decisive parameter for a node so that it can be decided whether to for-
ward to CNs or broadcast to shallower nodes. This broadcasting choice is needed
to be dependent on some decisive criteria. Event segregation approach provides an
important criteria for this decision making step.
Moreover conventional underwater schemes are generally planned for single appli-
cation in which all sensing nodes are assigned the task of sensing a particular type
of event and reporting to the floating sink. Occurring events have same occurrence
probability in sensing field and require similar delay sensitivity for all nodes. These
schemes provide no solution for diversified applications in which multiple events,
with different tolerable delay requirements, can occur with different probability of
occurrence. Moreover generally UWSN are considered delay tolerant in which ma-
jor constraint is energy consumption while it is assumed that there is relaxation in
terms of delay. This is over simplified assumption as some events although with
very less probability of occurrence may require to be reported much earlier than
normal events. To the best of our knowledge there is no solution in literature which
considers the diverse events occurrence and their energy efficient delivery to sink.
So there should be a mechanism of relatively faster delivery of time critical events
to sinks. Event segregation approach considers this delay sensitivity associated with
every node and plans more efficient approach for packet delivery.
35
CHAPTER 4
ESDR: Event Segregation based Delay sensitive
Routing
In this chapter our proposed scheme Event Segregation based Delay sensitive Rout-
ing is introduced and salient features of ESDR are explained. Directed Graph (di-
graph) theory is employed to serve as model for network topology in ESDR. De-
ployed nodes map to vertices and links between nodes represent edges of digraph.
In digraph G= (N,A)vertices Nare connected via edges or arcs A. Sets N,
Eand Care defined as.
N={n|nis a deployed node}
E={n|nis an event sensing and data generating node}
B={n|nis a broadcasting/forwarding node}
C={n|nis a deployed courier node}
In event driven or threshold sensitive applications a set Esuch that E⊂Nis
selected as data generator and other nodes of set Bact as intermediate forwarder to
relay that packets to next hop nodes. Following relations satisfy this scenario:
B=N-E,|E|<|N|and |E|<|B|
There are some other (regular data gathering) applications in which all nodes are
equally responsible for generation and forwarding tasks. In such cases following
basic relations are satisfied:
E⊆N,B⊆Nand |E|=|N|
We have consider the later case in which all nodes can be both data sensing and
forwarding nodes. Summary of whole chapter is enlisted below:
•In first section network initialization and node deployment is explained.
•Motivation for event segregation and its positive impacts on network are ex-
36
plained in this section
•The channel model for underwater environment is explained in details. En-
ergy consumption, delay and losses are calculated using the equations given
in channel model.
•Improvements in terms of forwarding function calculation process and hold-
ing time calculation process is explained in subsequent sections.
•Mobility of courier node is also impacted by event segregation approach. Mo-
bility of deployed courier nodes in ESDR is explained in this section.
4.1 Network Initialization and Node Deployment
ESDR considers deep underwater environment which comprises of lm×bm×Dmaxm.
Initially Nnodes having same initial energy Eoare randomly deployed. Each node
is equipped with appropriate sensors and a communication module. Node uses
acoustic communication to send packets with each other. Each node is aware of its
depth which is calculated using pressure sensors. It is also assumed that each sens-
ing node on the basis of value of sensed attribute can determine its type and delay
criticality associated with it. This assumption can be realized by if different nodes
with capability of sensing different natural phenomena are deployed. It can also be
realized by predefining some threshold levels of being sensed parameter and then
conditioning their type by those threshold values. After sensing the event a node
can determine its type by simple comparison with threshold and as per its type it
decide whether to broadcast to shallower nodes or to patrolling sinks.
In order to analyze the scheme in sparse and denser environment node density is
varied however all the comparing schemes are implemented in same volume and
with same number of nodes. Four courier nodes are also placed on different depths
as shown in Figure 4.1. Courier nodes are deployed for data collection and have the
capability of vertical motion. Necessary parameters of each node in all comparing
37
schemes are given in Table 5.1.
Figure 4.1: Network Diagram in ESDR
4.2 Statistical Event Generation and Segregation
One of the unique aspects of ESDR is event segregation approach as it considers
delay sensitivity associated with different types of events and assures efficient deliv-
ery to appropriate destination. We have devised a statistical model which considers
different events occurring in sensing field. It is assumed that sensing node is able to
determine the type of a particular sensed event and delay criticality associated with
that event. This delay criticality value is mapped to a set of distinct and discrete
values of affordable delays. In order to simplify the analysis occurring events are
divided into three types: very critical, critical and normal/ordinary with the proba-
bility of occurrence pvc,pcr and pno respectively. Numerical values used for these
probabilities are given in Table 3.1. Nodes with critical and very critical events
38
are collectively termed as delay sensitive nodes and their data is termed as delay
sensitive data. The nodes of each type of events are members of the sets defined as:
Evc={x|x has sensed a very critical event}
Ecr={x|x has sensed a critical event}
Eno={x|x has sensed a normal event}
Cardinality of these set of node is as given as:
|Evc|=bpvc × |E|e (4.1a)
|Ecr|=bpcr × |E|e (4.1b)
|Eno|=bpno × |E|e (4.1c)
Where sign be represent the round off decimal values to nearest integer. These set
of nodes also satisfy the following relation.
|Evc|+|Ecr|+|Eno|6|E|=|N|(4.1d)
This means for a specific node of its lifetime say runits (rounds or seconds), b66.6×
retimes it would sense ordinary event, b22.2×retimes it would sense critical event
and b11.1×retimes it would be very critical event. This probability distribution
also ensures that at any time, b66.6× |E|e out of |E|event sensing nodes would be
ordinary nodes, b33.3× |E|e nodes would be very critical and b11.1× |E|e nodes
would sense very critical event. However in this model nodes are selected randomly
as critical, very critical and normal but it is ensured that the length of each set
of node is not changed. Packet format for ESDR is shown in Figure 4.2. One
important field is packet type which can be very critical, critical or ordinary. Events
Figure 4.2: Packet Format in ESDR
39
segregation benefits are threefold and enlisted below:
•This type of segregation not only allows to make UWSN workable for differ-
ent type of applications but it also provides more flexibility while planning
the forwarding and broadcasting of different type of events.
•It reduces the broadcasting overhead especially in deep underwater monitor-
ing applications packet forwarding strategy ensure that only delay sensitive
packets are broadcasted to surface sinks. Normal events are sent to courier
nodes.
•It provides the mechanism for adaptive mobility of courier nodes. Otherwise
courier nodes had no decisive parameters to plan their next stop in their so-
journ tour.
These positive impacts of event segregation become clear in next sections.
4.3 Packets Forwarding Strategy
In our proposed scheme it is envisioned to formulate broadcasting mechanism for
all types of nodes considering the delay sensitivity requirements of critical and very
critical nodes. We have planned the task dividing approach in which packets are
forwarded either to sink or to CN based on their type and criticality requirements.
Sojourn tour of CN can cause larger value of delay as data collected by CN is de-
livered to data center after its complete round trip. Hence it is ruled out to forward
delay sensitive packets to CN. On the other hand it is also inadequate to forward
all the generated packets to surface sink(s) especially in deep underwater environ-
ments or in regular data gathering applications. That is why it is planned that delay
sensitive nodes will forward the packet to surface sink using broadcasting and multi
hop communication approach. The algorithm 1 explains the forwarding or broad-
casting of packets to shallower nodes. If the type of packet; generated by the node
qor received from deeper node, is either critical or very critical than it is forwarded
40
using respective forwarding functions. These forwarding functions are improved
and depended on depths of nodes, explained in details in upcoming sections. If the
type of packet; generated by the node qor received from deeper node is normal it
is forwarded to CN using algorithm 2. Normal nodes try to forward or broad-
Algorithm 1 Broadcasting to Surface Sink
1: S←SensedE vent
2: q←EventSensingN ode
3: Dq←Deptho f qt hnod e
4: D10←NewU pperDepthLimit
5: D20←NewLowerDepthLimit
6: Sink f←SinkFound
7: FF←ForwardingFunction
8: SQI0←Im provedSignalQualityIndex
9: ECF 0←I m provedEnergyCostFunction
10: DDF0←ImprovedDepthDependentF unction
11: while Sink f=f alse do
12: if q∈Evc||q∈Ecr then
13: ForwardtoU pperNod es
14: if Dq≤D10then
15: FF=SQI0
16: end if
17: if D10≤Dq≤D20then
18: FF=ECF 0
19: end if
20: if Dq≥D20then
21: FF=DDF0
22: end if
23: else
24: if q∈Eno then
25: ForwardingtoCN
26: CNAlgorithm
27: end if
28: end if
29: Sink f=true
30: end while
cast their respective packets to CN. Hence CN mobility is planned and influenced
as per normal nodes distribution and data generation only. Broadcasting of normal
node’s packets to CN can also involve one hop communication. If a normal node
has packet to forward and it is not in the range of CN than it would check if any
41
of its neighbor is in range of CN then the packet is sent to that neighbor which re-
lays the packet to nearest CN. In order to accomplish this normal nodes share their
neighbor list with each other. Disjoint neighbors are updated by each node in the
list named as neighbours_f or_courier. Nodes with depth less than transmission
range directly send to surface sink irrespective of their type.
Algorithm 2 Forwarding to CN
1: S←SensedE vent
2: q←EventSensingN ode
3: c←NearestCourierNode
4: nc←Nod eNearerToCourierNode
5: Dq←Deptho f qt hnod e
6: Dc←Deptho f nearestCN
7: RT←TransmissionRange
8: CNf←CourierN ode f ound
9: F
C←ForwardingtoCourierNode
10: FN←ForwardingtoNeighborNode
11: while CNf=f al se do
12: if q∈Eno then
13: ForwardtoCN
14: if Dc−Dq≤RTthen
15: F
C
16: CNf=true
17: else
18: if Dc−Dq≥RTthen
19: ForwardtoNeighbor
20: if nc∈neighbours_f or_courier then
21: FN
22: CNf=true
23: end if
24: end if
25: end if
26: end if
27: CNf=true
28: end while
4.3.1 Updating Depth Threshold
Depth threshold is the radian distance from a certain node idepth in which no other
node can be neighbour of node i. Depth difference of node iand its all the eligible
42
Algorithm 3 Updating Depth Threshold
1: dead ←N umberO f DeadN odes
2: Nth1←First DeathT hereshold
3: Nth2←Second Deat hT hreshol d
4: dth ←De pthT hreshold
5: dth1←F irst Dept hT hreshol d
6: dth2←Second Dept hT hreshold
7: dth3←T hird DepthT hreshol d
8: while dead 6=|N|do
9: if 0≤dead ≤Nth1then
10: dth =dth1
11: end if
12: if Nth1≤dead ≤Nth2then
13: dth =dth2
14: end if
15: if dead ≥Nth2then
16: dth =dth3
17: end if
18: end while
Figure 4.3: Eligible Neighbours with Range and Depth Threshold
neighbour nodes would always be more than depth threshold. Depth threshold is
employed so that next hop forwarder should be best and farthest node within the
transmission range as depicted in Figure 4.3. However when nodes start to die
out and network is changes from denser to sparser it is necessary to update depth
43
threshold. Depth threshold is updated depending on the death threshold values
given in 5.1. Algorithm 3 explains the updating of depth threshold.
4.4 The Channel Model
The mathematical model are used to calculate various parameters in underwater
networks. We have utilized two type of channel models for our scheme. In order to
find attenuation and transmission losses suffered by acoustic signal Thorp Attenu-
ation formula is utilized. Monterey-Miami Parabolic Equation is used to calculate
path loss. Both these formulas and other channel modeling characteristics are given
in details.
4.4.1 Thorp Attenuation Formula
Thorp attenuation considers propagation of sound waves as the molecular move-
ment of signals toward its contiguous particles. It mainly focuses on the band-
width efficiency as its attenuation loss are calculated on the basis of frequency and
depth. Underwater wireless channel is severely affected by issues like multi path
and fading. Signal to Noise Ratio (SNR) in UWSN is computed using passive sonar
equation [75] given below:
SNR =SL −T L −N L +DI (4.2)
where SL is source level of the target or noise generated by the target, TL is trans-
mission loss in aquatic environment, NL is noise loss and DI is directive index. DI
is the capability of receiving sensor to direct its antenna to avoid unwanted noise.
SNR value must be greater than detection threshold DT. All the above quantities
are measured in dBreµPa. Total attenuation of signal is sum of spreading loss and
signal absorption loss calculated by thorp model given in [76,77]. Transmission
44
Loss is given in following:
T L =10 log(d) + α×d×10−3(4.3)
Absorption loss α(f)is calculated by following equation.
αdB (f) =
0.11 f2
1+f2+44 f2
4100+f+
2.75 ×10−4f2+0.003 if f>0.4
0.002 +0.11(f
(1+f)) + 0.011 fif f<0.4
(4.4)
α(f)is calculated using absorption loss formula as following:
α(f) = 10
αdB (f)
10 (4.5)
Absorption coefficient is function of depth difference l and absorption loss given
in following equation.
Abc =l×αdB (4.6a)
Scattering loss is logarithmic function of depth difference l given in following equa-
tion.
Sc=k×10log(l)(4.6b)
where k is spreading coefficient with value 1.5 for practical spreading. Total atten-
uation A(l,f) is the sum of Absorption coefficient (Abc)and Scattering coefficient
(Sc)given by following formula.
A(l,f) = Abc +Sc(4.6c)
Ambient noise is calculated with adding the following four noises, turbulent noise
Nt(f), shipping noise Ns(f), wind noise Nw(f)and thermal noise Nth(f). These
45
noises are calculated using following formulas.
10log(Nt(f)) = 17 −30log(f)(4.7a)
10log(Ns(f)) = 40 +20(s−0.5) + 26 log(f)−60 log(f+0.03)(4.7b)
10log(Nw(f)) = 50 +7.5ω1/2+20 log(f)−40 log(f+0.04)(4.7c)
10log(Nth (f)) = −15 +20 log(f)(4.7d)
N(f) = Nt(f) + Ns(f) + Nw(f) + Nth(f)(4.8)
Passive sonar equation is used to find out Source Level SL to find intensity of trans-
mitted signal ITusing following formula:
IT=10SL/10 ×0.67 ×10−18 (4.9)
Transmission power of source is P
T(d)
P
T(d) = 2π×1m×H×IT(4.10)
Energy consumed in transmitting the k bits of packets is given as:
ET x(k,d) = 2P
T(d)×TT x (4.11)
where TT x is the transmission in seconds.
4.4.2 Monterey-Miami Parabolic Equation
Monterey-Miami Parabolic Equation is also an efficient and important formula to
calculate the path losses in underwater communication environment. It consist of
quite complex formulas mainly dependent on the depth, transmitted frequency and
distance between two nodes. The basic formula of MMPE consist of sum of differ-
46
ent chunks and is given in equation 4.12.
PL(t) = m(f,d,ds,dr) + ω(t) + n() (4.12a)
Where parameters are defined below:
PL(t): Propagation Loss between two nodes.
m(): Propagation loss with arbitrary and recurring factor; found from retrogression
of MMPE information.
f: Transmitted signal frequency in kHz.
ds: Depth of sender node in meters.
dr: Depth of receiver node in meters.
d: Distance between sender and receiver.
ω(t):Signal loss estimate due to wave passage
n(): Signal loss estimate due to arbitrary and random noise
MMPE model estimates the propagation loss with random and regular, periodic
components in the first slice of the equation. It performs nonlinear regression on
the data to obtain A(n) coefficients. Therefore, m(f,d,ds,dr)function computes
the propagation loss as follows:
m(f,d,ds,dr) = log
(d
0.9)A0dA9
sdA7((ds−dr)2)A10
(d×dr)10A5!+ f2A1
1+f2+41A1
4100 +f2+0.002+0.003!×d
914 +A16 ×dr+A8×dA6×dr+A8×d
(4.12b)
MMPE model approximates the losses due to wave mobility in the subsequent por-
tion of the equation. It counts the sinusoidal mobility of the water molecules around
the acoustic signal. The ω(t)computed in equation 4.12c function considers wave-
length, wave period and scale factor along with wave effects to approximate the
signal loss caused by the wave motion. It can be mathematically expressed as:
ω(t) = h(λω,t,dr,hω,Tω)E(t,Tω)(4.12c)
47
h(s): scale factor.
λω: Wavelength of ocean waves.
hω: Height of ocean wave in meters.
dr: Receiver’s depth in meters.
Tω: Wave period in seconds.
E(): Wave effects in nodes function.
Since we have undertaken the continuous node movement in aqueous environment,
the effect on wave height and wavelength depicted in h(s) is due to movement of
receiver node. Scale factor is estimated using equation in []. Movement of signal is
also scaled by distance given in h(s).
h(Tω,λω,t,hω,dr) = (hω(1−(2dr
λω)))
0.5!×
sin(2π(mod(Tω))
Tω
)
The mathematical expression e() in the last slice of MMPE basic equation is for
random background noise. Random noise function considers random Gaussian dis-
tribution. It depends on the ratio of distance between communicating nodes and
source transmitter range.
n() = 20(d
RT×NR
)(4.12d)
n(): random noise function.
d: Euclidean distance between sending node S and receiving node R in meters.
TR: transmission range of source in meters.
NR: random number from a gaussian distribution with 0 mean and 1 variance.
4.5 Improvements in Terms of Forwarding Function
Forwarding function is the criteria for a node to decide whether to forward the delay
sensitive packets or not. Forwarding decision done by each node irrespective of its
48
own type. This means a normal node would not broadcast its own packet to sink
but if a delay sensitive packet is received it would also be eligible for forwarding
and forward the packet if it comes out to be the best forwarding node. Likewise
iAMCTD forwarding function in ESDR is defined differently for different depth
nodes. SQI based forwarding function is redefined to ensure the comparable values
of forwarding function.
SQI0=log10 (LSNR)×Er×RT
li
(4.13)
SQI based forwarding function range is always in interval [0 to 491.91]. Upper limit
is far less than defined in previous schemes and is more comparable to the range of
ECF and DDF based forwarding function. ECF based forwarding function FFis
directly proportional to Erand inversely proportional to depth difference previous
and current node depths Dp−Dcis defined below:
ECF 0= [ pv×RT
Dp−Dc
]Er(4.14)
where RTis transmission range Eris remaining energy of a certain node. Forward-
ing function is directly dependent on remaining energy weighted by a factor given
in square brackets whose range is described as follows: Neighbour selection pro-
cess and depth threshold ensure that Dth <(Dp−Dc)<RT. Initially Dth is set
to 60mand is subsequently changed to 40mand 20mafter certain percentage of
deaths, and RTis constant=100. So for above mentioned values factor in square
brackets in equation 4.14 has range, from 1 to 1.66 for dth =60, from 1 to 2.5 for
dth =40 and from 1 to 5 for dth =60. This eventually gives value of forwarding
function always between 0 to 5 ×Eowhich is comparable with the range of SQI’.
DDF is also modified taking newly defined and linearized version of SQI.
DDF0=SQI ×li
Di
(4.15)
49
SQI based forwarding function is employed when depth of nodes is less than D0
1=
250m, ECF based forwarding decision is used if node is in between D0
1=250mand
D0
2=750m. Nodes with depth more than D0
2forward the packet using DDF based
forwarding function.
4.6 Improvements in Terms of Holding Time
Holding time is given to nodes with lesser values of respective forwarding function.
Holding time calculation is defined by the formula given in equation.
H0
T= (1−FF)HT(4.16a)
HTis the maximum value of holding time. It is defined in following equation as the
ratio of consumed energy per unit power.
HT=Eo−Er
P
T
(4.16b)
So if a certain node has less value of remaining energy it would have higher value
of holding time and refrain to forward the packet.
4.7 Mobility of Sink
In conventional terrestrial wireless sensor networks mobile sinks are deployed to
collect data from relatively farther nodes [78,79]. This data collection reduces the
forwarding load on intermediated nodes employed in multi hop communications It
also provides variety of possibilities for routing in a network. The role of these data
collecting devices is of prime importance otherwise all the data is to be forwarded
in multi-hop fashion. Hence courier nodes and AUV are employed for data gath-
ering applications so they are very crucial for underwater networks [80]. Courier
50
nodes can move vertically up and down and there is no facility of horizontal move-
ment while AUV can be moved to any desired place with proper tour planning [81].
Anyhow there is no energy constraint on both of these devices. Mobility of courier
nodes has not been exploited to its full potential in previous work. In previous
scheme mobility of these nodes starts after certain percentage of death. Attaining
the global knowledge of deaths and updating that to CNs itself has communication
overhead. Additionally employing a resource in deep water and waiting for some
certain level of calamity is not a wise option. Mobility of these nodes is needed to
be exploited as per network requirements. So it is necessary to plan a better mo-
bility pattern of CN to get full use of it. We have reconsidered the deployment and
mobility of CNs. Moreover event segregation dictates and selects limited number of
nodes to forward to CNs. Hence mobility of courier node is influenced by density
of normal nodes and their rate of packet generation. In our proposed environment
horizontal movement is not required while vertical movement of each courier node
is explained in two different ways.
4.7.1 Synchronized and Uniform Mobility of AUV
In this type of mobility four courier nodes deployed on four different depths move in
elliptical path. The motion is uniform in the sense that speed of each node remains
same in its sojourn tour on elliptical paths. By the term synchronized mean that
phase relation between any two courier remains same throughout the tour as shown
in Figure 4.4a. Elliptical path of courier node iin yz plane is defined in following
equation.
(z−ki)2
a2+(y−hi)2
b2=1 (4.17a)
Elliptical path is planned in such a way that two ellipses have some overlapping re-
gion. This allows better coverage to the whole region and also facilitates to respec-
tive lower courier node to send its data to upper node. If the length of overlapping
51
Depth-(m)
0 100 200 300 400 500 600 700 800 900 1000
-50
0
50 AUV1
AUV2
AUV3
AUV4
AUV Sojourn
AUV
(a) Uniform mobility of AUV
Depth-(m)
0 100 200 300 400 500 600 700 800 900 1000
-50
0
50 AUV1
AUV2
AUV3
AUV4
AUV Sojourn
AUV
Very Critical Node
Critical Node
Normal Node
(b) Uniform mobility of AUV with node deployed nodes
Figure 4.4: Synchronized mobility of AUV
region on zaxis is xvertex ais related with xby following equation:
a=3
8x+125 (4.17b)
This relation reveals that for x=0 a becomes equal to 125mmaking length of
major axis of the each ellipse to 2a=250. x is subsequently changed to 20mand
40mmaking a=133mand 140m. The resulting ellipses are shown in Figure 4.4.
In this type of mobility vertex length a=140mand co-vertices length b=50mare
used. In order to calculate time period of a node it is necessary to have knowledge
of perimeter (circumference) of an ellipse. There is no straightforward formula to
calculate this and following infinite series is utilized.
p=π(a+b)
∞
∑
n=10.5
n2
hn(4.17c)
52
The parameter h is defined as follows:
h=(a−b)2
(a+b)2(4.17d)
We have employed the following formula by great Indian mathematician Srinavasan
Ramanejan [82] to calculate the perimeter of the each elliptical.
p≈π[3(a+b)−p(3a+b)(a+3b)] (4.18)
Time period of each nodes is dependent on uniform tangential speed of courier
node during its sojourn tour. In synchronized and uniform mobility pattern speed
of a node i is vimaking time period:
Ti=p
vi
(4.19)
Mobility of CN is envisioned in such a way than CN1is in phase with CN3and CN2
is in phase with CN4. Initially CN1and CN3are at bottom end of their respective
elliptical paths and CN2and CN4are at the top end of their corresponding elliptical
paths. Every node moves in clockwise direction and reaches to next destination
after time Ti/4. In this way a node in its one revolution has four points to stop each
after phase angle of θ/2. Their motion is synchronized in such a way that when
ihiki
1 0 140
2 0 380
3 0 620
4 0 860
Table 4.1: Centers of elliptical paths
CN1is about the top. Each node moves upward to their predefined destination,
transfer data to next CN, and move back to initial stage. This motion pattern is
synchronized in such a way that when a certain courier node is about to reach its
destination next forwarder has just started its tour, first node find next CN to take
its data to next CN. 53
4.7.2 Adaptive Mobility of Courier Nodes
In this mobility pattern node may accelerate or decelerate its motion on the basis
of node density and data generated requirements. In this motion node moves up
and down on a vertical line and its speed is influenced by node density and rate of
traffic generated by normal nodes. Mobility of CN1and CN3are in same direction
while CN2and CN4are in same direction. But both pairs of courier node move
opposite to each other to ensure that these nodes must overlap and communicate
at the overlapping area. Motion of a courier node cis governed by the following
Depth-(m)
0 100 200 300 400 500 600 700 800 900 1000
-50
0
50
CN1CN2CN3CN4
AUV or CN Sojourn
AUV or CN
(a) Adaptive mobility of Courier Nodes
Depth-(m)
0 100 200 300 400 500 600 700 800 900 1000
-50
0
50
CN1CN2CN3CN4
AUV Sojourn
AUV
Very Critical Node
Critical Node
Normal Node
(b) Adaptive mobility of Courier Nodes with deployed nodes
Figure 4.5: Adaptive mobility of AUV/CN
equation. Let a courier node cis at position yc; its new position ytafter time tcan
be determined by following equation.
dy2
c
dt2+1
t
dyc
dt +yc=1
t2×yt(4.20)
In this equation left term (second degree differential term) corresponds the acceler-
ation of courier node c. If this value is set to zero motion becomes to unform. For
54
adaptive mobility of the nodes this factor is dependent node density in the upcom-
ing coverage region of courier node c. Upcoming coverage regions is double of the
transmission range of courier node. This process is explained as: Initially turns its
transmission range to double of the present values and broadcast its arrival to its
future region. Receiving nodes acknowledge with their depth information, rate of
data generation and elapsed time since their previous transfer of data to any courier
node. Courier node calculate the mean of the depth information of only those nodes
which are currently not in its range. This provides that spatial information to esti-
mate its next possible location in its sojourn tour. It also utilizes elapsed time data
to calculate the allowed time to reach the estimated location. Using the depth differ-
ence between current depth and estimated depth; and remaining time node courier
node decides new velocity to reach at next location.
yt=1
Nd
Nd
∑
1
Di(4.21)
where Ndrepresent total number of data generating nodes farther than current trans-
mission range of courier node Node utilizes elapsed time and average traffic gen-
eration of all nodes Ndto determine remaining time trto reach new location. This
remaining time tralong with new location distance ytare utilized to estimate what
the next courier node velocity vcwould be optimum.
vc=yt
tr
(4.22)
Variations of vcduring its sojourn tour make sure the adaptive and accelerated mo-
bility of courier nodes for reaching at optimum new location in time.
55
CHAPTER 5
Performance Evaluation
In this chapter we present the simulation setup and performance comparison of the
ESDR with contemporary schemes. This chapter is summarized in the following
two points.
•In the first section simulation setup with necessary parameter is explained.
•The second section of this chapter is dedicated for performance evaluation.
Initially evaluation metrics are defined and then each one is compared with
other protocols.
5.1 Simulation Setup
We have considered the deep underwater environment of volume 100×100×1000m3
in which 224 nodes are randomly deployed. Analytical calculations are programmed
in MATLAB using channel model equations. Generally in UWSN nodes can be di-
vided into two major groups on the basis of traffic generation. In some applications
only a fraction of all deployed nodes are event sensing nodes which can generate
data packets. Rest of the nodes are used to forward or broadcast those packets. In
other group all the deployed nodes can sense and communicate at a time. Number
of nodes in former group can be fix and preselected based on their depth, or can
be varying and adaptively selected based on some threshold value of the sensed
attribute. We have chosen the second group in our simulations in which all the de-
ployed 224 nodes are data generating nodes and also act as forwarder to the received
data packets. In simulation scenario each node is allowed to generate exactly one
packet in one round.
Simulation setup for random event generation of ESDR can be with many different
56
possibilities. What would be the type of a certain node and how many number of
nodes would normal, very critical and critical. This number can be fixed or varying
(as in present case we have chosen fixed number of node in each type). For example
in this scenario nodes are selected randomly as critical, very critical and normal.
This election is repeated after 100 rounds however it is ensured that total number
of nodes in each type remains constant. So every time in a network ratio of very
critical, critical and normal nodes is constant and it remains uniform for the whole
lifetime of network. One can argue about varying number of delay sensitive nodes
because who knows how much events can be delay sensitive. But here in order
to simplify the analysis and simulation setup, so far constant ratio of delay critical
nodes is used. We have compared the performance of ESDR with DBR, EEDBR
and iAMCTD. In order to ensure fair comparison all the comparing schemes are
implemented in same sensing volume. Necessary parameters with their values are
given in Table 3.1.
5.2 Performance Evaluation Metrics
In order to compare the performance of underlaying schemes following evaluation
metrics are considered.
Stability period: Time from the start of network operation till the death of first
nodes is termed as stability period, whereas time from the death of first node till the
death of last node is termed as instability period.
Halflife: Halflife of network is defined as the time till the death of half nodes.
Lifetime: Lifetime of network is the time till the death of all nodes.
Average end to end Delay Average end to end delay is mean of accumulated sum
of propagation and transmission delays of all nodes.
Throughput: Throughput is considered in terms of following two aspects:
Packet sent to and received at sink: Number of packets sent to sink are depen-
57
dent on type of application. In event driven or threshold sensitive applications some
selected nodes are packet generator nodes, while in other type all the nodes have
equal probability to generate the packet after some regular interval and sent to sink.
Number of packets successively received at sink is less than number of packets sent
to sink.
PDR: Packet Delivery Ratio: PDR is the ratio of number of packets received at
sink to number of packets sent to sink. In ideal case where no packet is dropped
PDR is equal 1, otherwise its value is less than 1 and greater than 0.
Transmission LossTransmission loss is the average decrease in sound energy/intensity
level as it propagates through water.
Each metric is explained in details in the following subsections.
Table 5.1: Parameters values used in simulations
Parameter Name Value
Initial Energy (Eo) 70J
Volume of the Network 100 ×100 ×1000×
Data Packet 64B
Control Packet 8B
Number of nodes |N|224
Number of courier nodes |C|4
Maximum Depth (Dmax) 1000m
First Depth Margin (D01) 250m
Second Depth Margin (D02) 750m
First Death Margin (Nth1) 75
Second Death Margin (Nth2) 150
Depth threshold 1 (dth1) 60m
Depth threshold 2 (dth2)40m
Depth threshold 3 (dth3) 20m
Transmission Range (RT) 100m
Priority value for very critical nodes(pvc ) 2
Priority value for critical nodes(pcr ) 1.5
Priority value for normal nodes (pvc ) 1
pvc 0.111
pcr 0.222
pno 0.666
Speed of acoustic signal 1500m/s
5.2.1 Network Lifetime
Network lifetime is very crucial evaluation metric for the progress of any UWSN
scheme. It is always required to prolong the lifetime with some efficient routing
decisions considering the affecting parameters. Network lifetime of the underly-
58
ing schemes is given in Figure 5.1. Stability period, halflife and lifetime of all
comparing schemes are also given in Figure 5.2. Four bars in three groups are for
stability period, halflife and lifetime of their respective scheme. Stability period of
ESDR is greater than iAMCTD after that iAMCTD shows somewhat stable behav-
ior and there occurs rapid deaths in ESDR. This is due the difference of courier node
mobility strategy between iAMCTD and ESDR. In iAMCTD courier nodes mobil-
ity is associated with the certain percentage of deaths. Hence courier nodes come
into action as nodes have depleted a big portion of their energy and about to die
out. Courier nodes data collection reduces large distance transmissions and bring
somewhat stable behaviour in network. Anyhow this rescue mechanism could not
prolong lifetime to very much extent. Halflife of both schemes is almost similar but
after that iAMCTD nodes die out very rapidly whereas ESDR prolongs its lifetime
to appreciable extent. Motion of courier nodes in ESDR is influenced by relative
distribution and data generation of normal nodes. This motion is cyclic in nature in
which, in each new cycle courier node utilizes record of its previous sojourn stay
and estimates the better new position. This gradual improving mobility mechanism
results into the deterrent behaviour of network. This type of mobility strategy may
not stop the deaths but the lifetime of more important nodes (with higher rate of
data generation) is prolonged while the death of lesser important nodes is tolerated
for the overall benefit of network. iAMCTD nodes die out very early due to lack of
network healing mechanism. Over all comparison proves that ESDR outperforms
iAMCTD, EEDBR and DBR by considerable margin. Since all the comparing
schemes have their own lifetime which is dependent on their network characteris-
tics. Hence lifetime cannot serve as a fair independent quantity for comparison of
remaining evaluation metrics. So in order to have a common independent quantity
of same length we have taken certain but equal number of ascending samples of
each respective metric, from the timeline of all the comparing scheme. Then these
taken samples of that metric are plotted and compared. So samples in timeline of
59
Timeline ×104
0 0.5 1 1.5 2 2.5 3
0
50
100
150
200
250
Nodes
DBR
EEDBR
iAMCTD
ESDR
Figure 5.1: Lifetime
Stability Halflife Lifetime
Rounds
×104
0
0.5
1
1.5
2
2.5
3
1227
4099
13059
1646
5215
17412
1798
10972
12984
2999
10552
29941
DBR
EEDBR
iAMCTD
ESDR
Figure 5.2: Stability Period, half-life and life-time
following figures actually refers to the equally spaced and same number of samples
of that evaluation metric taken in each scheme own lifetime. Stability period of
60
ESDR is 17.68% more than iAMCTD while halflife is 3% greater, and lifetime is
29.36% is greater than iAMCTD. Over all comparison proves that ESDR outper-
forms iAMCTD, EEDBR and DBR by considerable margin.
5.2.2 Average end to end delay
Average end to end delay is calculated using following procedure in each schemes.
Two type of delay constitutes in operation of model.
Transmission delay: Transmission delay caused by a node nto a packet of length
lis given by following formula.
Dn
t=l
r(5.1)
Where r is data rate with value 250000 bps.
Propagation delay: Major part of delay is constituted by propagation delay given
by following equation. Propagation delay of an acoustic signal moving with veloc-
ity vbetween a sender i and receiver j separated by distance si,jis given by:
Di,j
p=si,j
v(5.2)
Speed of acoustic signal is given in Table. Sum of both type of delay constitute in
one hop delay than sum of all hops makes total delay in a packet transmission from
a certain node it is also termed as end to end delay.
Di,j
n=Di,j
t+Di
p(5.3)
Delay for by all nodes in a round are measured and stored in a vector and then mean
of this gives average end to end delay. In this way average end to end delay in round
is calculated and recorded for the whole network lifetime.
Dn
e2e=
hmax
∑
h=1
Dh(5.4)
61
De2e=∑nmax
n=1Dn
e2e
nmax
(5.5)
Timeline (Samples)
2 4 6 8 10 12 14 16 18 20
Delay (Seconds)
0
0.1
0.2
0.3
0.4
0.5
0.6
DBR
EEDBR
iAMCTD
ESDR
Figure 5.3: Average end to end delay: Comparison
From this vector of whole lifetime twenty samples are taken in ascending order
and plotted is shown in Figure 5.3. It is clear from Figure that ESDR perform
better than its counterparts in terms of average end to end delay. Packet segregation
causes significant decrease in delay. ESDR put comparatively lesser forwarding
overhead on nodes and hence these nodes causes less retransmission and delay. It is
also important to note that delay in DBR is lesser than EEDBR as EEDBR selects
forwarders on the base of energy so a node with lesser depth but more energy can
become forwarder, and eventually involving more number of nodes in forwarding
process, which eventually accumulates into larger transmission delays.
62
Lifetime (Samples)
2 4 6 8 10 12 14 16 18 20
Delay (Seconds)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Very critical
Critical
Ordinary
Figure 5.4: Delay comparison in ESDR
5.2.3 Throughput
Throughput of the entire network is measured in terms of total number of packets
received at BS. The comparison for total number of packets successively received at
BS is shown in Figure 5.5. ESDR outperforms its counterparts in terms of through-
put by considerable margin. This is because packet loss in ESDR is minimum due
to packet segregation strategy. Two third of generated packets are sent to courier
nodes available in underwater environment. There is very minute packet loss in this
transmission since it requires maximum one hop communication. Only one third of
total generated packets are broadcasted to surface sink. In previous scheme holding
time value was depending on forwarding function which itself was having inconsis-
tent values. In ESDR Improved forwarding function results into optimum values of
holding time which eventually ensures decreased collision and lesser packet loss.
So overall comparison, depicted in Figure 5.5 results into increased throughput of
ESDR. Another parameter to get the insight in terms of throughput is Packet De-
63
Timeline (Samples)
2 4 6 8 10 12 14 16 18 20
Packets
×106
0
0.5
1
1.5
2
2.5
3
DBR
EEDBR
iAMCTD
ESDR
Figure 5.5: Throughput
livery Ratio (PDR) shown in Figure 5.6. PDR throughout the lifetime is calculated
down-sampled and plotted for all comparing schemes. Improved forwarding func-
tion and efficient adaptive mobility of courier node cause the increase PDR value
for proposed scheme. Moreover since approximately 66% of the generated packets
are transmitted to courier nodes which may cause delay but ensure reliable delivery
of packets destinations resulting with improvement in PDR and throughput as in
Figure 5.6.
5.2.4 Transmission Loss
Transmission loss is dependent on; number of transmissions that a packet undergo
in multi-hop communication, attenuation loss of the signal, and bandwidth effi-
ciency. It is calculated using thorp attenuation formula given in Equation 4.3.
Major contribution to transmission loss is the spreading of the acoustic wave as
it propagates away from the source for longer distances in underwater and cause
64
Timeline (Samples)
2 4 6 8 10 12 14 16 18 20
Ratio
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
DBR
EEDBR
iAMCTD
ESDR
Figure 5.6: Packet Delivery Ratio
greater transmission losses. Transmission loss for all comparing schemes is calcu-
lated and samples from whole lifetime taken in ascending order are plotted in Figure
5.7. ESDR performs best among all comparing scheme in terms of transmission
loss as well. This is due the fact that a major part of data generated is transferred
to patrolling sinks and is not suffered to long distances spreading losses. More-
over premature death of intermediate nodes in previous schemes cause increase in
transmission loss for their respective protocols.
5.2.5 Path Loss
Path Loss in underwater is dependent on the distance or depth difference between
two communicating nodes. Path Loss is affected due to wave movement and is
calculated using MMPE model explained in previous chapter. Nodes in ESDR un-
dergo comparatively lesser distance communication due to proper availability of
courier nodes and efficient forwarding plan. Hence in ESDR path loss values are
65
Timeline (Samples)
2 4 6 8 10 12 14 16 18 20
Transmission Loss (dB)
0
10
20
30
40
50
60
70
80
DBR
EEDBR
iAMCTD
ESDR
Figure 5.7: Transmission Loss
comparatively lesser than its counter parts for the whole time line of each respec-
tive scheme as shown in Figure 5.8. On the other hand in DBR and EEDBR
broadcasting load on intermediate nodes cause their premature death which result
larger distance communication and eventually increase in path loss values. Path
Loss is iAMCTD is more than ESDR because mobility of sinks in iAMCTD starts
after certain percentage of deaths; before their motion node have to suffer longer
distance communications. Adaptive mobility of sink ESDR ensure comparatively
lesser distance communications making lesser values of path loss.
66
Timeline
2 4 6 8 10 12 14 16 18 20
Path Loss (dB)
10
20
30
40
50
60
70
80
90
100
DBR
EEDBR
iAMCTD
Proposed
Figure 5.8: Path Loss
67
CHAPTER 6
Conclusion and Future Work
In this work, we have proposed a novel and energy efficient approach to accommo-
date delay sensitivity requirements with increased network lifetime for Under Water
Sensor Network. We have formulated a strategy in which three types of events with
different probability of occurrence are generated and their efficient forwarding is
planned. Forwarding functions are analyzed and redefined and holding time calcu-
lation process is also improved. However probability of occurrence for each type
of event remains unchanged throughout the lifetime. In future we intend to imple-
ment event segregation approach with varying provability of occurrence of normal,
critical and very critical events. In this work generated events type (and hence node
type) is randomly selected but in future type of event selection is intended to be
related with it depth, remaining energy or any other relevant parameter.
68
REFERENCES
[1] Einstein qoute about curiosiy, 2016. URL http://www.brainyquote.
com/quotes/quotes/a/alberteins145949.html?src=t_
curiosity. Accessed: 2016-04-18.
[2] G.A. Boxshall, J Mees, and et al. World register of marine species, 2016. URL
http://www.marinespecies.org. Accessed: 2016-04-18.
[3] A study about the divesity of ocean, 2016. URL http:
//school.discoveryeducation.com/schooladventures/
planetocean/ocean.html. Accessed: 2016-04-18.
[4] Ian F Akyildiz, Dario Pompili, and Tommaso Melodia. Challenges for effi-
cient communication in underwater acoustic sensor networks. ACM Sigbed
Review, 1(2):3–8, 2004.
[5] Dario Pompili, Tommaso Melodia, and Ian F Akyildiz. Three-dimensional
and two-dimensional deployment analysis for underwater acoustic sensor net-
works. Ad Hoc Networks, 7(4):778–790, 2009.
[6] Ian F Akyildiz and Mehmet Can Vuran. Wireless underwater sensor networks.
Wireless Sensor Networks, pages 399–442, 2010.
[7] Ian F Akyildiz. Wireless sensor networks in challenged environments such as
underwater and underground. In Proceedings of the 17th ACM international
conference on Modeling, analysis and simulation of wireless and mobile sys-
tems, pages 1–2. ACM, 2014.
[8] JV Anand and S Titus. Analysis of depth based routing protocols in under wa-
ter sensor networks. International Review on Computers and Software (IRE-
COS), 8(10):2389–2402, 2013.
[9] H. L. Ko, M. S. Kim, S. G. Lee, Dae-Young Cho, J. W. Park, and Y. K.
Lim. The performance analysis of diversity combined with beamforming
technologies for underwater channel environments using experimental data.
In OCEANS’11 MTS/IEEE KONA, pages 1–5, Sept 2011.
[10] Jim Partan, Jim Kurose, and Brian Neil Levine. A survey of practical issues
in underwater networks. ACM SIGMOBILE Mobile Computing and Commu-
nications Review, 11(4):23–33, 2007.
[11] Zaihan Jiang. Underwater acoustic networks–issues and solutions. Interna-
tional journal of intelligent control and systems, 13(3):152–161, 2008.
[12] RB Manjula and Sunilkumar S Manvi. Issues in underwater acoustic sensor
networks. International Journal of Computer and Electrical Engineering, 3
(1):101, 2011.
69
[13] Dr. S. Murugappan U.Devee Prasan. Underwater sensor networks :archi-
tecture, research challenges and potential applications. International Jour-
nal of Engineering Research and Applications(IJERA), "2":"251–256", April
"2012". ISSN "2248-9622".
[14] John Heidemann, Milica Stojanovic, and Michele Zorzi. Underwater sensor
networks: applications, advances and challenges. Philosophical Transactions
of the Royal Society of London A: Mathematical, Physical and Engineering
Sciences, 370(1958):158–175, 2012.
[15] Guangjie Han, Chenyu Zhang, Lei Shu, Ning Sun, and Qingwu Li. A survey
on deployment algorithms in underwater acoustic sensor networks. Interna-
tional Journal of Distributed Sensor Networks, 2013, 2013.
[16] Peng Jiang, Xingmin Wang, and Lurong Jiang. Node deployment algorithm
based on connected tree for underwater sensor networks. Sensors, 15(7):
16763–16785, 2015.
[17] Tariq Ali, Low Tang Jung, and Ibrahima Faye. Classification of routing algo-
rithms in volatile environment of underwater wireless sensor networks. Inter-
national Journal of Communication Networks and Information Security, 6(2):
129, 2014.
[18] Emad Felemban, Faisal Karim Shaikh, Umair Mujtaba Qureshi, Adil A
Sheikh, and Saad Bin Qaisar. Underwater sensor network applications: A
comprehensive survey. International Journal of Distributed Sensor Networks,
2015:1–14, 2015.
[19] Mohsin Murad, Adil A Sheikh, Muhammad Asif Manzoor, Emad Felemban,
and Saad Qaisar. A survey on current underwater acoustic sensor network
applications. International Journal of Computer Theory and Engineering, 7
(1):51, 2015.
[20] M Kiranmayi and A Kathirvel. Underwater wireless sensor networks: appli-
cations, challenges and design issues of the network layer-a review”. Inter-
national Journal of Emerging Trends in Engineering Research, 3(1):05–11,
2015.
[21] Arijit Khan and Lawrence Jenkins. Undersea wireless sensor network for
ocean pollution prevention. In Communication Systems Software and Middle-
ware and Workshops, 2008. COMSWARE 2008. 3rd International Conference
on, pages 2–8. IEEE, 2008.
[22] G Acar and AE Adams. Acmenet: an underwater acoustic sensor net-
work protocol for real-time environmental monitoring in coastal areas. IEE
Proceedings-Radar, Sonar and Navigation, 153(4):365–380, 2006.
[23] Dario Pompili, Tommaso Melodia, Ian F Akyildiz, et al. A resilient routing
algorithm for long-term applications in underwater sensor networks. In Proc.
of Mediterranean Ad Hoc Networking Workshop (Med-Hoc-Net), 2006.
70
[24] Dan Chen, Zhixin Liu, Lizhe Wang, Minggang Dou, Jingying Chen, and Hui
Li. Natural disaster monitoring with wireless sensor networks: a case study of
data-intensive applications upon low-cost scalable systems. Mobile Networks
and Applications, 18(5):651–663, 2013.
[25] Azzedine Boukerche, Begumhan Turgut, Nevin Aydin, Mohammad Z Ahmad,
Ladislau Bölöni, and Damla Turgut. Routing protocols in ad hoc networks: A
survey. Computer Networks, 55(13):3032–3080, 2011.
[26] Reetika Khetarpal and Bhushan Dua. Routing protocols in mobile ad hoc
networks: A survey. IJCER, 2(3):54–59, 2013.
[27] Alex Hinds, Michael Ngulube, Shaoying Zhu, and Hussain Al-Aqrabi. A
review of routing protocols for mobile ad-hoc networks (manet). International
Journal of Information and Education Technology, 3(1):1, 2013.
[28] VG Muralishankar and EGDP Raj. Routing protocols for manet: A literature
survey. International Journal of Computer Science and Mobile Applications,
2(3):18–24, 2014.
[29] Geetha Jayakumar and G Gopinath. Ad hoc mobile wireless networks routing
protocols–a review. Journal of Computer science, 3(8):574–582, 2007.
[30] Ramandeep Kaur and Chandan Sharma. Review paper on performance anal-
ysis of aodv, dsdv, olsr on the basis of packet delivery. IOSR Journal of Com-
puter Engineering (IOSR-JCE), 11(1):51–55, 2013.
[31] Bahuguna Renu, Tayal Pranavi, et al. Routing protocols in mobile ad-hoc
network: a review. In Quality, Reliability, Security and Robustness in Hetero-
geneous Networks, pages 52–60. Springer, 2013.
[32] Philippe Jacquet, Paul Mühlethaler, Thomas Clausen, Anis Laouiti, Amir
Qayyum, and Laurent Viennot. Optimized link state routing protocol for ad
hoc networks. In Multi Topic Conference, 2001. IEEE INMIC 2001. Tech-
nology for the 21st Century. Proceedings. IEEE International, pages 62–68.
IEEE, 2001.
[33] Charles E Perkins and Pravin Bhagwat. Highly dynamic destination-
sequenced distance-vector routing (dsdv) for mobile computers. In ACM SIG-
COMM computer communication review, volume 24, pages 234–244. ACM,
1994.
[34] Mehran Abolhasan, Tadeusz Wysocki, and Eryk Dutkiewicz. A review of
routing protocols for mobile ad hoc networks. Ad hoc networks, 2(1):1–22,
2014.
[35] David B Das. Ad-hoc on demand routing protocol for mobile ad hoc networks.
draft-ietf-manet-dsr-09. txt, 2003.
71
[36] Hiroaki Higaki and Shingo Umeshima. Multiple-route ad hoc on-demand
distance vector (mraodv) routing protocol. In Parallel and Distributed Pro-
cessing Symposium, 2004. Proceedings. 18th International, page 237. IEEE,
2004.
[37] David B Johnson. The dynamic source routing protocol for mobile ad hoc
networks. draft-ietf-manet-dsr-09. txt, 2003.
[38] Guangjie Han, Jinfang Jiang, Lei Shu, Yongjun Xu, and Feng Wang. Local-
ization algorithms of underwater wireless sensor networks: A survey. Sensors,
12(2):2026–2061, 2012.
[39] Ying Guo and Yutao Liu. Localization for anchor-free underwater sensor net-
works. Computers & Electrical Engineering, 39(6):1812–1821, 2013.
[40] Guangjie Han, Aihua Qian, Chenyu Zhang, Yan Wang, and Joel JPC Ro-
drigues. Localization algorithms in large-scale underwater acoustic sensor
networks: a quantitative comparison. International Journal of Distributed
Sensor Networks, 2014, 2014.
[41] Guangjie Han, Chenyu Zhang, Lei Shu, and J.J.P.C. Rodrigues. Impacts of de-
ployment strategies on localization performance in underwater acoustic sensor
networks. Industrial Electronics, IEEE Transactions on, 62(3):1725–1733,
March 2015. ISSN 0278-0046. doi: 10.1109/TIE.2014.2362731.
[42] Sudip Misra, Tamoghna Ojha, and Ayan Mondal. Game-theoretic topology
controlfor opportunistic localizationin sparse underwater sensor networks.
Mobile Computing, IEEE Transactions on, 14(5):990–1003, 2015.
[43] Edward Carlson, Pierre-Philippe Beaujean, Edgar An, et al. Location-aware
routing protocol for underwater acoustic networks. In OCEANS 2006, pages
1–6. IEEE, 2006.
[44] KY Foo, PR Atkins, T Collins, C Morley, and J Davies. A routing and
channel-access approach for an ad hoc underwater acoustic network. In
OCEANS’04. MTTS/IEEE TECHNO-OCEAN’04, volume 2, pages 789–795.
IEEE, 2004.
[45] Peng Xie, Jun-Hong Cui, and Li Lao. Vbf: vector-based forwarding protocol
for underwater sensor networks. In Networking 2006. Networking technolo-
gies, services, and protocols; performance of computer and communication
networks; mobile and wireless communications systems, pages 1216–1221.
Springer, 2006.
[46] Nicoletta Nicolaou, ALEX SEE, Peng Xie, Jun-Hong Cui, and Dario Mag-
giorini. Improving the robustness of location-based routing for underwater
sensor networks. In OCEANS 2007-Europe, pages 1–6. IEEE, 2007.
72
[47] Peng Xie, Zhong Zhou, Nicolas Nicolaou, Andrew See, Jun-Hong Cui, and
Zhijie Shi. Efficient vector-based forwarding for underwater sensor net-
works. EURASIP J. Wirel. Commun. Netw., 2010:4:1–4:13, April 2010. ISSN
1687-1472. doi: 10.1155/2010/195910. URL http://dx.doi.org/10.
1155/2010/195910.
[48] Xiong Xiao, Xiao Peng Ji, Guang Yang, and Yan Ping Cong. Le-vbf:
Lifetime-extended vector-based forwarding routing. In Computer Science
Service System (CSSS), 2012 International Conference on, pages 1201–1203,
Aug 2012. doi: 10.1109/CSSS.2012.304.
[49] Peng Xie, Zhong Zhou, Zheng Peng, Jun-Hong Cui, and Zhijie Shi. Void
avoidance in three-dimensional mobile underwater sensor networks. In Wire-
less Algorithms, Systems, and Applications, pages 305–314. Springer, 2009.
[50] Josep Miquel Jornet, Milica Stojanovic, and Michele Zorzi. Focused beam
routing protocol for underwater acoustic networks. In Proceedings of the third
ACM international workshop on Underwater Networks, pages 75–82. ACM,
2008.
[51] Nitthita Chirdchoo, Wee-Seng Soh, and Kee Chaing Chua. Sector-based
routing with destination location prediction for underwater mobile networks.
In Advanced Information Networking and Applications Workshops, 2009.
WAINA’09. International Conference on, pages 1148–1153. IEEE, 2009.
[52] Rodolfo WL Coutinho, Luiz FM Vieira, Antonio Loureiro, et al. Dcr: Depth-
controlled routing protocol for underwater sensor networks. In Computers and
Communications (ISCC), 2013 IEEE Symposium on, pages 000453–000458.
IEEE, 2013.
[53] Dongseung Shin, Daeyoup Hwang, and Dongkyun Kim. Dfr: an efficient
directional flooding-based routing protocol in underwater sensor networks.
Wireless Communications and Mobile Computing, 12(17):1517–1527, 2012.
[54] Yen-Da Chen, Chan-Ying Lien, Ching-Hung Wang, and Kuei-Ping Shih.
Darp: A depth adaptive routing protocol for large-scale underwater acoustic
sensor networks. In OCEANS, 2012-Yeosu, pages 1–6. IEEE, 2012.
[55] Zonglin Li, Nianmin Yao, and Qin Gao. Relative distance based forward-
ing protocol for underwater wireless networks. International Journal of Dis-
tributed Sensor Networks, 2014, 2014.
[56] Zhong Zhou, Zheng Peng, Jun-Hong Cui, and Zhijie Shi. Efficient multipath
communication for time-critical applications in underwater acoustic sensor
networks. Networking, IEEE/ACM Transactions on, 19(1):28–41, 2011.
[57] Hai Yan, Zhijie Jerry Shi, and Jun-Hong Cui. Dbr: depth-based routing for
underwater sensor networks. In NETWORKING 2008 Ad Hoc and Sensor Net-
works, Wireless Networks, Next Generation Internet, pages 72–86. Springer,
2008. 73
[58] Liu Guangzhong and Li Zhibin. Depth-based multi-hop routing protocol
for underwater sensor network. In Industrial Mechatronics and Automation
(ICIMA), 2010 2nd International Conference on, volume 2, pages 268–270.
IEEE, 2010.
[59] Haitao Yu, Nianmin Yao, Tong Wang, Guangshun Li, Zhenguo Gao, and
Guozhen Tan. Wdfad-dbr: Weighting depth and forwarding area division dbr
routing protocol for uasns. Ad Hoc Networks, 2015.
[60] Uichin Lee, Paul Wang, Youngtae Noh, Luiz FM Vieira, Mario Gerla, and Jun-
Hong Cui. Pressure routing for underwater sensor networks. In INFOCOM,
2010 Proceedings IEEE, pages 1–9. IEEE, 2010.
[61] Youngtae Noh, Uichin Lee, Paul Wang, Brian Sung Chul Choi, and Mario
Gerla. Vapr: void-aware pressure routing for underwater sensor networks.
Mobile Computing, IEEE Transactions on, 12(5):895–908, 2013.
[62] Tiansi Hu and Yunsi Fei. Qelar: a machine-learning-based adaptive rout-
ing protocol for energy-efficient and lifetime-extended underwater sensor net-
works. Mobile Computing, IEEE Transactions on, 9(6):796–809, 2010.
[63] S. Gopi, G. Kannan, D. Chander, U.B. Desai, and S.N. Merchant. Pulrp:
Path unaware layered routing protocol for underwater sensor networks. In
Communications, 2008. ICC ’08. IEEE International Conference on, pages
3141–3145, May 2008. doi: 10.1109/ICC.2008.591.
[64] Sarah Gopi, Kannan Govindan, Deepthi Chander, Uday B Desai, and SN Mer-
chant. E-pulrp: Energy optimized path unaware layered routing protocol for
underwater sensor networks. Wireless Communications, IEEE Transactions
on, 9(11):3391–3401, 2010.
[65] Abdul Wahid and Dongkyun Kim. An energy efficient localization-free rout-
ing protocol for underwater wireless sensor networks. International journal
of distributed sensor networks, 2012, 2012.
[66] Muhammad Ayaz and Azween Abdullah. Hop-by-hop dynamic addressing
based (h2-dab) routing protocol for underwater wireless sensor networks. In
Information and Multimedia Technology, 2009. ICIMT’09. International Con-
ference on, pages 436–441. IEEE, 2009.
[67] Muhammad Ayaz, Azween Abdullah, Ibrahima Faye, and Yasir Batira. An
efficient dynamic addressing based routing protocol for underwater wireless
sensor networks. Computer Communications, 35(4):475–486, 2012.
[68] Abdul Wahid and Dongkyun Kim. Connectivity-based routing protocol for
underwater wireless sensor networks. In ICT Convergence (ICTC), 2012 In-
ternational Conference on, pages 589–590. IEEE, 2012.
[69] Stefano Basagni, Chiara Petrioli, Roberto Petroccia, and Daniele Spaccini.
Channel-aware routing for underwater wireless networks. In OCEANS, 2012-
Yeosu, pages 1–9. IEEE, 2012. 74
[70] Stefano Basagni, Chiara Petrioli, Roberto Petroccia, and Daniele Spac-
cini. Carp: A channel-aware routing protocol for underwater acous-
tic wireless networks. Ad Hoc Networks, 34:92 – 104, 2015. ISSN
1570-8705. doi: http://dx.doi.org/10.1016/j.adhoc.2014.07.014. URL
http://www.sciencedirect.com/science/article/pii/
S1570870514001450. {ADVANCES} {IN} {UNDERWATER} {COM-
MUNICATIONS} {AND} {NETWORKS}.
[71] ZhangBing Zhou, Beibei Yao, Riliang Xing, Lei Shu, and Shengrong Bu. E-
carp: an energy efficient routing protocol for uwsns in the internet of under-
water things. 2015.
[72] Angelo Capossele, Gianluca De Cicco, and Chiara Petrioli. R-carp: A rep-
utation based channel aware routing protocol for underwater acoustic sensor
networks. 2015.
[73] Mohsin Raza Jafri, Shehab Ahmed, Nadeem Javaid, Zainal Ahmad, and
RJ Qureshi. Amctd: Adaptive mobility of courier nodes in threshold-
optimized dbr protocol for underwater wireless sensor networks. In Broad-
band and Wireless Computing, Communication and Applications (BWCCA),
2013 Eighth International Conference on, pages 93–99. IEEE, 2013.
[74] Nadeem Javaid, Mohsin Raza Jafri, ZA Khan, Umar Qasim, Turki Ali Al-
ghamdi, and Muhammad Ali. Iamctd: Improved adaptive mobility of courier
nodes in threshold-optimized dbr protocol for underwater wireless sensor net-
works. International Journal of Distributed Sensor Networks, 2014, 2014.
[75] Jing Han, ChunXia Meng, and Feng Cao. Simulated research on passive sonar
range using different hydrographic conditions. In MATEC Web of Confer-
ences, volume 35, page 04003. EDP Sciences, 2015.
[76] Milica Stojanovic. On the relationship between capacity and distance in an un-
derwater acoustic communication channel. ACM SIGMOBILE Mobile Com-
puting and Communications Review, 11(4):34–43, 2007.
[77] Albert F Harris III and Michele Zorzi. Modeling the underwater acoustic
channel in ns2. In Proceedings of the 2nd international conference on Per-
formance evaluation methodologies and tools, page 18. ICST (Institute for
Computer Sciences, Social-Informatics and Telecommunications Engineer-
ing), 2007.
[78] Shiqing Shen, Andong Zhan, Panlong Yang, and Guihai Chen. Exploiting sink
mobility to maximize lifetime in 3d underwater sensor networks. In Commu-
nications (ICC), 2010 IEEE International Conference on, pages 1–5. IEEE,
2010.
[79] Jalaja Janardanan Kartha and Lillykutty Jacob. Delay and lifetime perfor-
mance of underwater wireless sensor networks with mobile element based
data collection. International Journal of Distributed Sensor Networks, 2015:
28, 2015. 75
[80] Jin Wang, XQ Yang, ZQ Zhang, Bin Li, and Jeong-Uk Kim. A survey about
routing protocols with mobile sink for wireless sensor network. International
Journal of Future Generation Communication and Networking, 7(5):221–228,
2014.
[81] MJ Jalaja and Lillykutty Jacob. Adaptive data collection in sparse underwater
sensor networks using mobile elements. In Ad-hoc Networks and Wireless,
pages 57–65. Springer, 2014.
[82] Mark B Villarino. Ramanujan’s perimeter of an ellipse. arXiv preprint
math/0506384, 2008.
76
Abbreviations
UWSNs: Underwater Wireless Sensor Networks
DBR: Depth Based Routing
EEDBR:Eergy Efficient Depth Based Routing
IAMCTD:Improved Adaptive Mobility of Courier nodes in Threshold-optimized
DBR
ESDR: Event Segregation based Delay sensitive Routing
MANETs: Mobile Ad-hoc Networks
LSNR: Localized free Signal to Noise Ratio
LASR: Localized Aware Source Routing
SQI: Signal Quality Index
DDF: Depth Dependent Function
ECF: Energy Cost Function
AUV: Autonomous Underwater Vehicle
CN: Courier Node
MMPE: Monterey-Miami Parabolic Equation
77
