Embed Size (px)
Citation preview
Cooperative Routing in Underwater
Wireless Sensor Networks (UWSNs)
By: Hina Nasir
709-FBAS/MSCS/S13
Supervised by:
Dr. Nadeem Javaid
Assistant Professor, CAST,
COMSATS, Islamabad
Co- Supervised by:
Dr. Mohammad Sher
Dean, FBAS, IIU, Islamabad
Department of Computer Science
Faculty of Basic and Applied Sciences
International Islamic University Islamabad
2014
ii
Department of Computer Science
International Islamic University Islamabad
Date:
Final Approval
This is to certify that we have read the thesis submitted by Hina Nasir, 709-
FBAS/MSCS/S13. It is our judgment that this thesis is of sufficient standard to warrant
its acceptance by International Islamic University, Islamabad for the degree of MS
Computer Science.
Committee:
External Examiner:
Dr. Shahzad Saleem ___________________________
Assistant Professor
FAST National University of Computer and Emerging Sciences, Islamabad
Internal Examiner:
Ms. Ummarah Zahid ___________________________
Lecturer
IIU, Islamabad
Supervisor:
Dr. Nadeem Javaid ___________________________
Assistant Professor
CAST, COMSATS, Islamabad
Co-Supervisor:
Dr. Mohammad Sher ___________________________
Dean, FBAS
IIU, Islamabad
iii
This thesis is dedicated to my son.
iv
A dissertation Submitted To
Department of Computer Science,
Faculty of Basic and Applied Sciences,
International Islamic University, Islamabad
As a Partial Fulfillment of the Requirement for the Award of the
Degree of MS Computer Science.
v
Declaration
I hereby declare that this Thesis “Cooperative Routing in Underwater Wireless
Sensor Networks (UWSNs)” neither as a whole nor as a part has been copied out
from any source. It is further declared that I have done this research with the
accompanied report entirely on the basis of our personal efforts, under the proficient
guidance of my teachers especially my supervisors; Dr. Nadeem Javaid and Dr.
Mohammad Sher. If any part of the system is proved to be copied out from any source or
found to be reproduction of any project from any of the training institute or educational
institutions, I shall stand by the consequences.
___________________________
Hina Nasir
709-FBAS/MSCS/S13
vi
Acknowledgement
First of all I am obliged to Allah Almighty the Merciful, the Beneficent and the source of
all Knowledge, for granting me the courage and knowledge to complete this Project.
I am heartily grateful to my supervisor, Dr. Nadeem Javaid and Dr. Mohammad Sher,
whose encouragement, guidance and insightful criticism from the beginning to the final
level enabled me to have a deep understanding of the thesis.
I also offer my profound regard and blessing to everyone who supported me in any
respect, during and at the completion stage of this thesis work.
___________________________
Hina Nasir
709-FBAS/MSCS/S13
vii
Abstract
Mission critical applications impose the requirements of reliability and network
efficiency on Underwater Wireless Sensor Networks (UWSNs). Many cooperative
communication protocols are developed investigating physical and Media Access Control
(MAC) layer aspects to improve link efficiency, however, at network layer, it is still
largely un-explored. In this thesis, we propose a cooperative diversity routing protocol for
UWSNs to enhance network performance. Cooperation is employed at network layer in
existing non-cooperative routing protocol, Depth Based Routing (DBR), to increase its
reliability and throughput. Potential relays are selected on the basis of depth information.
Data from source node is cooperatively forwarded to the destination by relay nodes.
Simulation results show that Cooperative DBR (CoDBR) gives more throughput, more
packet acceptance ratio and less packet drop as compared to non-cooperative scheme.
Additionally, in this thesis, incremental relaying cooperative diversity with
retransmissions for UWSN is also studied and two routing protocols, based on
incremental relaying with cooperative retransmissions, are also proposed to enhance
reliability and throughput of the network. In the proposed model, feedback mechanism
indicates success or failure of data transmission. If direct transmission is successful, there
is no need of relaying by cooperative relay nodes. In case of failure, relays retransmit the
signal one by one till the desired signal quality is achieved at destination. Furthermore,
mathematical expression for the number of available relays along with closed-form
expression for outage probability is determined. Results show that the incremental
relaying with cooperative retransmissions can achieve less outage and more throughput
as compared to regular cooperative diversity networks.
TABLE OF CONTENTS
1 Introduction 1
1.1 Underwater Wireless Sensor Networks . . . . . . . . . . . . . . . . . 2
1.2 Error Control Methods . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Problem Statement and Proposed Solution . . . . . . . . . . . . . . 4
1.4 Report Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Literature Review 6
2.1 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Routing Protocols for Terrestrial WSNs . . . . . . . . . . . . 7
2.1.2 Routing Protocols for UWSNs . . . . . . . . . . . . . . . . . 7
2.1.3 ARQ Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.4 Cooperative Diversity Schemes . . . . . . . . . . . . . . . . 8
3 Analysis of Non-Cooperative Routing Protocols 10
3.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Proposed Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.1 Network Architecture . . . . . . . . . . . . . . . . . . . . . . 12
3.2.2 Protocol Details . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2.2.1 Optimal Forwarder Node Set Selection . . . . . . . 13
3.2.2.2 Forwarding Node Selection . . . . . . . . . . . . . 14
3.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . 14
4 CoDBR Protocol 21
4.1 Motivation of Co-DBR . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2 Proposed Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2.1 Channel Model . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.2.2 Proposed Scheme: CoDBR . . . . . . . . . . . . . . . . . . . 24
4.2.2.1 Path Setup Phase . . . . . . . . . . . . . . . . . . 25
4.2.2.2 Data Transmission Phase . . . . . . . . . . . . . . 25
viii
4.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . 27
5 ACE and E-ACE Protocol 33
5.1 Motivation of ACE and E-ACE . . . . . . . . . . . . . . . . . . . . 34
5.2 System Performance Analysis and Proposed Scheme . . . . . . . . . 34
5.2.1 System Model and Outage Performance Analysis . . . . . . 34
5.2.2 Determination of Number of Available Relays . . . . . . . . 36
5.2.3 Outage Probability . . . . . . . . . . . . . . . . . . . . . . . 37
5.2.4 Proposed Schemes . . . . . . . . . . . . . . . . . . . . . . . 40
5.2.4.1 Depth Exchange Phase . . . . . . . . . . . . . . . . 40
5.2.4.2 Path Establishment Phase . . . . . . . . . . . . . . 40
5.2.4.3 Data Transmission Phase . . . . . . . . . . . . . . 41
5.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . 43
5.3.1 Network Lifetime . . . . . . . . . . . . . . . . . . . . . . . . 45
5.3.2 Total Energy Consumption . . . . . . . . . . . . . . . . . . . 46
5.3.3 Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.3.4 Packet Drop . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3.5 Packet Acceptance Ratio . . . . . . . . . . . . . . . . . . . . 48
6 Conclusion and Future Work 50
6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
7 References 52
8 List of Publications 58
9 Appendices 60
.1 Area of Overlapping Region . . . . . . . . . . . . . . . . . . . . . . 61
.2 Derivation of fγ̄d . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
ix
LIST OF FIGURES
3.1 No. of nodes receiving data in DBR . . . . . . . . . . . . . . . . . . 11
3.2 Data transmission path in CDBR and CEEDBR . . . . . . . . . . . 13
3.3 DBR network lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 EEDBR network lifetime . . . . . . . . . . . . . . . . . . . . . . . . 15
3.5 DBR vs CDBR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.6 EEDBR vs CEEDBR . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.7 DBR packets dropped . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.8 EEDBR packets dropped . . . . . . . . . . . . . . . . . . . . . . . . 18
3.9 DBR end-to-end delay . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.10 EEDBR end-to-end delay . . . . . . . . . . . . . . . . . . . . . . . . 19
3.11 DBR energy consumption (J) . . . . . . . . . . . . . . . . . . . . . 19
3.12 EEDBR energy consumption (J) . . . . . . . . . . . . . . . . . . . . 20
4.1 Non-cooperative communication . . . . . . . . . . . . . . . . . . . . 22
4.2 Cooperative communication . . . . . . . . . . . . . . . . . . . . . . 22
4.3 CoDBR system model . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4 CoDBR multi-hop path from source to destination . . . . . . . . . . 25
4.5 Dead nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.6 Alive nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.7 Total energy consumption (J) . . . . . . . . . . . . . . . . . . . . . 29
4.8 Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.9 Packets dropped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.10 Packet acceptance ratio . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.11 Average end-to-end delay . . . . . . . . . . . . . . . . . . . . . . . . 32
5.1 System model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.2 Cooperative region and retransmission nodes . . . . . . . . . . . . . 37
5.3 Outage probability of incremental relaying with cooperative re-
transmissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
x
5.4 Packet is accepted by master node and ACK is sent to retransmis-
sion nodes indicating that no retransmission is required. . . . . . . 41
5.5 Packet is rejected by master node and asking for first retransmission
from R1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.6 Packet is again rejected and asking for second retransmission from
R2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.7 mth relay node performing data retransmission . . . . . . . . . . . . 42
5.8 ACE data transmission . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.9 E-ACE data transmission . . . . . . . . . . . . . . . . . . . . . . . 44
5.10 Network lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.11 Total energy consumption of the network (J) . . . . . . . . . . . . . 46
5.12 Throughput . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.13 Packets dropped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.14 Packet acceptance ratio . . . . . . . . . . . . . . . . . . . . . . . . . 49
.1.1 Area of overlapping region . . . . . . . . . . . . . . . . . . . . . . . 62
xi
LIST OF TABLES
4.1 CoDBR simulation parameters . . . . . . . . . . . . . . . . . . . . . 27
xii
Glossary
ACE Adaptive Cooperation in EEDBR. ix, 4, 32, 37, 38, 42, 44–46, 48
ACK Acknowledgement. 3, 40, 41
AF Amplify and Forward. 3, 22
ARQ Automatic Repeat reQuest. 2, 3, 8
AWGN Additive White Gaussian Noise. 22, 32
BER Bit Error Rate. 2, 3, 9, 21, 26, 29, 32, 40–42, 46, 48
BPSK Binary Pahse Shift Keying. 22, 32
CDBR Constraint Depth Based Routing. 11–15, 17, 19
CEEDBR Constraint EEDBR. 11, 12, 14, 15, 17, 19
CoDBR Cooperative Depth Based Routing. 4, 21–23, 26–29, 48
DBR Depth Based Routing. 4, 7, 11–13, 15, 17, 19, 21, 27, 29, 48
DEEC Distributed Energy Efficient Clustering. 7
DF Decode and Forward. 3, 48
DMC Distributed Multi-hop Cooperative communication. 7
E-ACE Enhanced Adaptive Cooperation in EEDBR. ix, 4, 32, 37, 38, 42–46, 48
EEDBR Energy Efficient Depth Based Routing. 7, 11–13, 15, 17, 38, 42, 43, 45,
46
LEACH Low Energy Adaptive Clustering Hierarchy. 7
MAC Media Access Control. 8
xiii
MRC Maximal Ratio Combining. 23, 26, 32, 33, 40, 41
NACK Negative Acknowledgement. 3, 40
PDF Probability Distribution Function. 36
REER Reliable Energy Efficient Routing. 7
SEP Stable Election Protocol. 7
SNR Signal-to-Noise Ratio. 8, 34–37, 40
TEEN TEEN. 7
UWSN Underwater Wireless Sensor Network. 2–4, 7, 8, 32, 48
WSN Wireless Sensor Network. 2
xiv
Chapter 1
Introduction
1
1.1 Underwater Wireless Sensor Networks
In recent years, deep sea exploration has caught a lot of attention because of its
usefulness regarding availability of resources, defense and transportation. Tra-
ditional methods of ocean exploration are time consuming, incur high costs and
human presence is also not possible due to harsh environment. For this reason,
idea of terrestrial WSN has been extended to underwater exploration and a new
class of UWSN has emerged [1].
UWSN consists of large number of sensor nodes under the water with sink(s)
located at surface. Sensors are deployed randomly over an area to perform collab-
orative monitoring tasks. These networks offer variety of applications like assisted
navigation, environmental monitoring, resource investigation, tactical surveillance
and disaster prevention [2].
The study of terrestrial WSN has been going on since many years. However, the
ideas developed for terrestrial WSNs cannot be directly applied to UWSNs. The
main reason is the environment in which sensors are required to operate in. Ter-
restrial WSN use radio waves for communication. However, radio waves are not
very suitable choice underwater because of high attenuation, absorption and scat-
tering. These impairments restricts distant transmissions when there is large area
to monitor in sea. Alternatively, acoustic waves becomes preferable choice due to
their favorable propagation characteristics underwater [3]. However, UWSNs have
some design challenges because of which the design of routing protocols requires
different approach than traditional ways applied to terrestrial WSN. Firstly, acous-
tic waves suffer long propagation delay as compared to radio waves. Furthermore,
energy constraint sensor nodes, dynamic network topology, huge monitoring area
and low bandwidth make design of a routing protocol more challenging task [1,2,4].
Acoustic signal is also affected by severe path loss, multi-path fading, reflection
and refraction from surface and sea bed, and aquatic noises [1, 3, 5, 7, 8]. These
channel impairments introduce high BER in acoustic transmission and lowers the
quality of signal.
1.2 Error Control Methods
In literature [5,9,10], it has been discussed that Automatic Repeat reQuest (ARQ)
and cooperative diversity schemes are efficient way to improve signal quality and
2
reduce BER in noisy channels. ARQ is an error control method to achieve reli-
able data transmission over unreliable link. It uses acknowledgement signal from
destination which indicates success or failure of data delivery. In case of failure,
data is retransmitted. Basic idea of ARQ is to retransmit data in case of failure.
Short feedback message from destination, in the form of ACK/NACK is utilized
to indicate success or failure of transmission [5].
The other error control method discussed in literature is cooperative diversity. Co-
operation is defined as a group of entities working together to achieve a common
goal by sharing each other’s resources. Diversity is achieved by multiple transmit
and receive antennas, however, this solution is costly for underwater sensor nodes.
Alternate solution is that each node has single antenna and use nearby node’s an-
tenna in transmitting data. Basic idea behind cooperative diversity is to transmit
same data over multiple paths. A system, with a source-destination pair and a
relay, creates a simple cooperative network. When a source node broadcasts its
data, it is received by destination node and also overheard by relay nodes in its
locality. The relay nodes forward the overheard data to the destination as a replica
of original data [3]. Therefore, at the destination, instead of single faded copy of
data, there are multiple independently faded copies of original data. In this way,
source node uses relay’s antenna to forward its data to destination. This is also
termed as cooperative diversity. At destination, independently received copies are
combined using diversity combining techniques to get the best out of them.
Cooperative diversity relaying techniques are divided into two categories: fixed
relaying and incremental relaying [9]. In fixed relaying, two commonly used tech-
niques are Amplify and Forward (AF) and Decode and Forward (DF). In AF, relay
node only amplifies the received signal and forwards it to destination, whereas,
in DF, data at relay is decoded, corrected, recoded and forwarded to destina-
tion [11]. In incremental relaying, source broadcasts data to destination and relay.
Feedback is generated from the destination about the success or failure of data.
Relay retransmits data to destination in case of negative acknowledgement using
AF or DF relaying technique, otherwise, source continues with the next packet.
Incremental relaying is an ARQ scheme with cooperation on demand. It is also
termed as Hybrid ARQ (H-ARQ) scheme [9].
3
1.3 Problem Statement and Proposed Solution
Significant research [3,10,12–15] has been conducted at physical and MAC layer in
cooperative transmission for UWSN. Efficient cooperative routing protocol is also
required at network layer. To the best of our knowledge cooperation at network
layer is still largely unexplored to have reliable data delivery and less packet drop
due to high Bit Error Rate (BER) in acoustic channel.
In this thesis, we propose a novel protocol, called Cooperative Depth Base Routing
(CoDBR) for UWSNs. CoDBR employs cooperation in Depth Based Routing
(DBR) [4] protocol, in order to increase reliability and throughput efficiency of
the network. Keeping the idea of DBR in mind, CoDBR performs data routing on
the bases of depth of sensor node. Relays are selected on the basis on minimum
depth neighbour.
Secondly, we propose cooperative retransmission protocols: Adaptive Cooperation
in EEDBR (ACE) and Enhanced-ACE (E-ACE) for UWSNs. In ACE protocol,
retransmission mechanism is incorporated in a cooperative manner to enhance re-
liability of an existing routing protocol called Energy Efficient Depth Based Rout-
ing (EEDBR) [1]. It is an adaptive cooperative retransmission protocol based on
incremental relaying, in which nodes cooperate when retransmission is required.
Retransmission from relay nodes is performed only when destination receives er-
roneous copy. Relay nodes are at less distance from destination, therefore, they
consume less energy in communication. Further, load balancing is achieved by
allowing nodes other than source node to retransmit data. In our work, the idea
of cooperative retransmission is taken from reference [16]. ACE allows only two
retransmissions in case of erroneous reception of data at destination. Enhanced
version of ACE, E-ACE is also proposed. In E-ACE, instead of only two retrans-
missions, m number of retransmissions from relay nodes are allowed. Increased
number of retransmissions helps to achieve more reliability and throughput at the
cost of increased energy consumption. Furthermore, our work presents the outage
performance analysis of incremental relaying with cooperative retransmissions in
UWSNs. Outage is defined as non-availability of signal at destination due to er-
rors introduced in signal on its way from source to destination. Source broadcasts
data to destination and relays. If destination receives erroneous signal, then re-
lay is responsible to retransmits the signal and both direct and relayed signal are
combined at destination using a diversity combining technique. If signal quality is
still not sufficient, second relay is held responsible for retransmission of data. This
4
process continues till the destination has received the signal with acceptable qual-
ity or all available relays are expired. We also calculated the number of available
relays and closed-form expression for outage probability is also derived.
1.4 Report Outline
Rest of the thesis is organized as follows. Chapter 2 describes the related work
done in the domain of cooperative routing. chapter 3 presents the analysis of
non cooperative routing protocols. Chapter 4 consists of proposed cooperative
diversity scheme. Chapter 5 presents proposed incremental relaying cooperative
diversity protocols along with outage performance analysis. Chapter 6 concludes
the the thesis.
5
Chapter 2
Literature Review
6
2.1 Related Work
This chapter provides an overview of related work in the field of terrestrial WSNs
and UWSNs.
2.1.1 Routing Protocols for Terrestrial WSNs
Many terrestrial routing protocols such as LEACH,TEEN, SEP and DEEC were
investigated in [17–20]. These routing protocols present an efficient solution to the
problem of routing in the case of terrestrial WSNs. For example, in the case of
LEACH, cluster-heads are formed and updated in each round. Cluster-heads are
rotated in each round based upon a threshold probability. This leads to an even
distribution of energy consumption for all the sensor nodes. LEACH is an example
of proactive protocols. In TEEN, the idea has been further developed from LEACH
to accommodate reactive networks. SEP and DEEC are heterogeneous aware
protocols in terms of residual energy of the sensor nodes. The main problem in
direct implementation of these protocols to UWSN is that these protocols have
been designed for static network topologies. In the case of UWSN, main feature
is that the network topology is dynamic in nature. Our protocol can be easily
adjusted to accommodate the dynamic nature of underwater sensor nodes.
2.1.2 Routing Protocols for UWSNs
Localization free non cooperative UWSN routing protocols are proposed in DBR
[4], EEDBR [1], iAMCTD [6] and AMCTD [7]. DBR uses only local depth infor-
mation of sensor nodes and forwards data towards sink located at surface using
greedy approach. It is a receiver based approach in which the nodes having smaller
depth participate in forwarding the data packet. Here, redundant transmissions
consume a lot of energy. DBR is improved in EEDBR, where local depth infor-
mation along with residual energy of sensor nodes is used to select the optimal
forwarder to achieve load balancing. Redundant transmissions are controlled by
introducing holding time for forwarding nodes based on residual energy and depth
information. AMCTD achieves network efficiency by adaptive depth threshold to
cope with sparse condition of network. Optimal weight functions for load balanc-
ing and on spot data collection using courier nodes to increase throughput are also
incorporated in this protocol.
7
REER [21] and DMC [22] are cooperative routing protocols for terrestrial networks
which improve reliability and throughput efficiency of the network.
2.1.3 ARQ Schemes
Different ARQ schemes are used to achieve reliability in data transmission. Few
of them are discussed in this subsection.
Cooperative ARQ (C-ARQ) [16] scheme is proposed for addressing cooperation
at MAC layer. In this scheme, cooperative nodes are used for retransmission
of data packet to enhance reliability and throughput efficiency. C-ARQ is fur-
ther improved in [8]. In this paper, authors proposed a retransmission proto-
col called Cooperative-Hybrid Automatic Repeat reQuest (C-HARQ) using Rate-
Compatible Punctured Convolution (RCPC) codes to maximize throughput and
energy efficiency of the network. Valera et al. in [23] presented opportunistic
multi-hop ARQ scheme. This scheme shows significant improvement in terms of
throughput efficiency at the cost of increased end to end delay. Cooperation-
based ARQ strategies are extensively studied in [24–26] for terrestrial networks
and proves to be very efficient in combating channel fading effects.
2.1.4 Cooperative Diversity Schemes
Taking advantage of broadcast nature of wireless transmission, cooperative com-
munication is proposed as a powerful technique to reduce fading in harsh underwa-
ter environment. Few cooperative schemes are presented in this subsection. In [3],
Suhail et al. present a contemporary overview of underwater acoustic communica-
tion (UWAC) and investigate physical layer aspects on cooperative transmission
techniques. It demonstrates the superiority of cooperative UWAC systems over
their point-to-point counterparts.
In [27], authors develop a relay selection criterion called Cooperative Best Re-
lay Assessment (COBRA) for UWSN. A best relay selection algorithm based on
COBRA criterion is also proposed. This algorithm only requires the statistical
information of the channel instead of the instantaneous channel state. COBRA
improves network performance in terms of throughput and delivery ratio with long
propagation delays.
Tan et al. in [28] presented a distributed cooperative scheme which includes net-
8
working protocols and cooperative transmissions to improve average energy con-
sumption, packet delivery ratio, and end-to-end delay. Relays are selected on the
basis of SNR and distance from sink.
Cooperative communication concept is applied at physical layer by Gao et al.
in [12]. In this scheme relay partner nodes are selected on the bases of minimum
propagation delay and SNR among the relay nodes.This scheme achieves low BER
in noisy underwater channel. A relay-aided protocol, Asynchronous Amplify and
Forward (AsAP), is proposed in [13]. It achieves reliable data communications and
solve time synchronization difficulties of UWSN. All relays amplify with fixed am-
plification factor and forward the received signal to the destination asynchronously
without any time coordination with other relays. It is further modified in [14],
with adaptive amplification based on instantaneous source- relay channel state
information.
In [10], asynchronous cooperative transmission technique using Underwater Am-
plify and Forward (UAF) and Underwater Decode and Forward (UDF) are used
to improve network performance. Coordinated Transmission-MAC protocol(CT-
MAC) addressing low bandwidth, low energy and long propagation delay challenge
at MAC layer is proposed in [15].
9
Chapter 3
Analysis of Non-Cooperative Routing Protocols
10
3.1 Motivation
In this section we thoroughly analyze the deficiencies of DBR and EEDBR which
leads to the development of CDBR and CEEDBR protocols.
In DBR, all neighbor nodes, that are above the defined threshold receive data
and only one node forwards the data. In case of large no. of neighboring nodes,
there will be more energy consumption as all of them are receiving the data from
the source node. Another problem with DBR is that, there is unnecessary data
forwarding. DBR is a receiver based approach. When source node broadcast data,
all nodes above the depth threshold becomes optimal forwarder and the node with
lowest depth wins the competition to become next forwarder and intimates rest of
the forwarders to stop further transmission. However, because of long propagation
delays, nodes may not receive the intimation on time and forward the data as well
when their holding time expires.
Fig. 3.1 shows that the source node forwards data to all its neighbors. In DBR 8
neighbors receive data hence their receiving energy is consumed and only neighbor
2 will be the forwarder because it has minimum depth among all. Rest of the
nodes will discard the data. This is not an optimal solution since lot of energy is
wasted in this way. A better solution can be obtained by restricting the number of
receiving nodes and ultimately selecting one to forward the data to the next hop
neighbor. This approach has been followed in the case of CDBR and CEEDBR.
Receiving nodes are restricted to n nodes. This set of receiving nodes is called
8
1
7
9
5
4 3
2
6
A
Figure 3.1: No. of nodes receiving data in DBR
optimal forwarder node set. This not only increases network life time, but also
restricts unnecessary data forwarding by selecting only one node as a forwarder.
One forwarder is selected based on lowest depth among the forwarder node set
that forwards data to the destination. This greatly enhances the network lifetime
and makes the network more suitable for applications where network lifetime is of
11
critical importance.
3.2 Proposed Scheme
In this section, we present the working of our protocol in detail.
3.2.1 Network Architecture
This protocol uses the same network architecture as that of DBR and EEDBR.
The sensor nodes are deployed under the water randomly. It is assumed that the
nodes do not change their depth and horizontal mobility of nodes is also ignored.
A number of sinks are deployed on the water surface and the sensor nodes are
responsible for delivering the sensed data to the sinks. The sinks are equipped
with Radio Frequency (RF) and acoustic modems. The sensor nodes under the
water are equipped with acoustic modems. The nodes communicate with each
other and the sinks using the acoustic modems. The sinks communicate with each
other and the on-shore data center using the RF Modems. Data reaching any
of the sinks is considered as data delivered. This is because the velocity of RF
signals is very high as compared to acoustic signals. So data reaching any of the
sinks can be efficiently communicated to other sinks without much delay. This is
obvious from the fact that sound propagates (at a speed of 1.5 x 103 m/s in water)
five times slower than radio (at a speed of 3 x 108 m/s in air). Furthermore, it is
also assumed that the sensor nodes are equipped with depth sensors which can be
used to know the depth information. Just like DBR, the proposed protocol only
needs to know the depth information of itself and neighboring nodes [4].
3.2.2 Protocol Details
This section elaborates the complete working of CDBR and CEEDBR. It is a lo-
calization free protocol and nodes are equipped with depth sensors only therefore
it is important to exchange depth information among the local neighbors. For this
purpose, all the nodes exchange their depth information among neighboring nodes
at the start of the network. Once all nodes know the depth information of their
neighboring nodes, a path is established from source to destination to transmit
data as shown in fig. 3.2.
12
SINK
Figure 3.2: Data transmission path in CDBR and CEEDBR
Both protocols consists of following phases
1. Optimal forwarder node set selection
2. Forwarding node selection
3.2.2.1 Optimal Forwarder Node Set Selection
In this phase, source node identifies its neighbours. The nodes having depth lower
than the depth of source node are identified as the neighbors. The number of
neighboring nodes is further constrained by applying a global parameter of depth
threshold (dth). This allows only those nodes to receive the data which are at
a depth difference more than dth. Among the identified neighbors, source node
identifies a set of nodes known as a optimal forwarder node set. These are called
optimal forwarders because they are considered best candidates to receive data
from source node and forward it to the destination. In the neighbor identification
phase, it is important to know whether the source is within the range of any sink
or not. In case a sink is in its close vicinity, the data is delivered directly to the
sink. If there is no sink in the range of source node, then it is forwarded to its
next hop forwarder node set. Finally, one node out of this node set is selected to
broadcast data to next hop forwarder node set. The selection criteria for forwarder
node set is based on key idea of DBR and EEDBR. For CDBR it is nodes with
minimum depth among the neighbouring nodes and for CEEDBR it is nodes with
13
maximum weight based upon depth and residual energy. The number of optimal
forwarder node set can be adjusted depending on the application.
The number of nodes in forwarder node set is n. Changing this parameter has lots
of implications on the way the protocol performs. As this parameter is changed
it effects the consumption of energy in the network. Value of n can be adjusted
to select 5, 3 or 2 forwarding nodes. Lower value of n leads to reduced energy
consumption in the network as compared to higher values. If all the in-range
neighbors were to receive data as in the case of DBR and EEDBR then energy
consumption is more and leads to smaller network lifetime. Constraining the
number of forwarding nodes will lead to an increase in network lifetime which is
the goal of our proposed scheme.
3.2.2.2 Forwarding Node Selection
In this scheme, the source first identifies a set of nodes in its transmission range
known as optimal forwarder node set. All the nodes in this set receive the sensed
information broadcasted by source node. In CDBR among the forwarder node
set, a node with minimum depth is selected for data forwarding. In the case of
CEEDBR, among the forwarder node set, a weight is assigned to the nodes based
upon depth and residual energy. The node will have maximum weight if it has
minimum depth and highest residual energy among the neighbor nodes. The node
with maximum weight is the candidate for data forwarding. It is also important
to check whether the node is alive or not. In this way data is forwarded from one
group of forwarder node set to the next group until it reaches the sink.
3.3 Simulation Results and Discussions
The simulations are performed in MATLAB with initial energy of 20J per node,
total number of nodes is 200 and maximum number of rounds is 4000. There are
total of 4 sinks at the surface. CDBR5 and CEEDBR5 are the plots for number of
receiving nodes as 5. CDBR2 and CEEDBR2 are the plots for number of receiving
nodes as 2.
In the above fig 3.3 and fig. 3.4 the plot of network lifetime for CDBR and
CEEDBR are compared to DBR and EEDBR respectively. The maximum lifetime
14
0 500 1000 1500 2000 2500 30000
50
100
150
200
Rounds
De
ad
no
de
s
CDBR5CDBR2DBR
Figure 3.3: DBR network lifetime
0 500 1000 1500 2000 2500 30000
50
100
150
200
Rounds
De
ad
no
de
s
CEEDBR5CEEDBR2EEDBR
Figure 3.4: EEDBR network lifetime
15
for DBR is around 2225 rounds compared to CDBR which is round 3000 rounds.
In case of EEDBR and CEEDBR all the nodes die at around 3000 rounds but the
rate at which the nodes die in the case of EEDBR is high. In case of CDBR there is
a considerable improvement in the network lifetime as number of forwarding nodes
is decreased. Although there is no improvement in the network lifetime in the case
of CEEDBR, but the nodes die at a much reduced rate so considerable portion
of network is alive for majority of the time. The increase in network lifetime in
CDBR and decrease in rate of CDBR and CEEDBR is because less number of
nodes is involved in data forwarding so total energy consumption is less.
It can be concluded from the above graphs that network lifetime can be consid-
erably improved by limiting the number of forwarding nodes. The bar plots were
obtained for CDBR and CEEDBR with forwarding nodes equal to 5, 3 and 2.
The simulation was run for a total of 8000 rounds. The fig. 3.5 and fig. 3.6
plots show that number of dead nodes in 8000 rounds is decreasing as the number
of forwarding nodes are reduced from 5 to 2. This shows that the network life
time is increasing as the number of forwarding nodes are reduced. This is because
as the number of forwarding nodes is decreased, there are fewer nodes which are
responsible for forwarding the data. This leads to reduced energy drainage of
the forwarding nodes. It can be concluded from the graphs that by limiting the
number of forwarding nodes, the overall network life time can be improved.
DBR hmax=5 hmax=3 hmax=20
100
200
300
400
De
ad
no
de
s
Number of dead nodes in 8000 rounds
Figure 3.5: DBR vs CDBR
In fig. 3.7 and fig. 3.8, as the number of forwarding nodes is decreased the packet
drop increases as there are less number of forwarding nodes in any given time
within a round. This leads to less number of packets reaching the sink. Network
16
EEDBR hmax=5 hmax=3 hmax=20
100
200
300
400
De
ad
no
de
s
Number of dead nodes in 8000 rounds
Figure 3.6: EEDBR vs CEEDBR
quality can be improved by allowing more number of nodes to forward the data.
Thus, network lifetime is improving at the cost of network quality which cannot
be improved by limiting the number of forwarding nodes. It can also be inferred
that an optimum value of number of forwarding nodes can be adjusted for an
acceptable level of packet drop at the cost of network lifetime.
0 500 1000 1500 2000 2500 30000
1
2
3
4
5x 10
5
Rounds
Pa
cke
ts d
rop
pe
d
CDBR5CDBR2DBR
Figure 3.7: DBR packets dropped
From the fig. 3.9, fig. 3.10 plot of end-to-end delay it can be observed that from
about 500 rounds till 2000 rounds the delay for DBR is less than both the plots of
CDBR. Similar situation can be observed for the plots if EEDBR versus CEEDBR.
The delay for both the cases of CDBR and CEEDBR is almost the same. The
delay for DBR and EEDBR drops to 0 at about 2000 rounds when almost all the
nodes die at that point. The delay for the two cases of CDBR and CEEDBR drops
to 0 at 3000 rounds when all the nodes die. From 500 to 3000 rounds the delay for
17
0 500 1000 1500 2000 2500 30000
1
2
3
4
5
6x 10
5
Rounds
Pa
cke
ts d
rop
pe
d
CEEDBR5CEEDBR2EEDBR
Figure 3.8: EEDBR packets dropped
CDBR and CEEDBR is more than DBR and EEDBR respectively because more
number of nodes is alive. Since more nodes are present to forward the data so
naturally delay time is increased.
0 500 1000 1500 2000 2500 30000
5
10
15
20
25
30
35
40
Rounds
Ave
rag
e e
nd
−to
−e
nd
de
lay (
se
c)
CDBR5CDBR2DBR
Figure 3.9: DBR end-to-end delay
From both the plots fig. 3.11, fig. 3.12 it can be assessed that in the initial rounds,
less than 1000, the energy consumption of CDBR is less than DBR and similarly
is the case for CEEDBRs energy which is less than EEDBR. This is because,
in the initial rounds greater number of nodes is alive and there are fewer data
forwarding nodes in the case of CDBR and CEEDBR. As the number of rounds
progress nodes are dying at a greater frequency in the case of DBR and EEDBR.
As a result after 1000 rounds energy consumption of CDBR and CEEDBR exceeds
DBR and EEDBR till about 3000 rounds when all the nodes die. In all the above
cases the total initial energy of nodes is a constant. In case of DBR and EEDBR
18
0 500 1000 1500 2000 2500 30000
5
10
15
20
25
30
35
40
Rounds
Ave
rag
e e
nd
−to
−e
nd
de
lay (
se
c)
CEEDBR5CEEDBR2EEDBR
Figure 3.10: EEDBR end-to-end delay
the nodes consume all their energy and die out in about 2000 rounds. While in
the case of CDBR and CEEDBR, nodes consume all their energy and die out in
around 3000 rounds. This is because the nodes are dying out at a greater rate in
the case of DBR and EEDBR. The plots show an increase in lifetime for CDBR
and CEEDBR.
0 500 1000 1500 2000 2500 30000
2
4
6
8
10
Rounds
En
erg
y c
on
su
mp
tio
n (
J)
CDBR5CDBR2DBR
Figure 3.11: DBR energy consumption (J)
19
0 500 1000 1500 2000 2500 30000
1
2
3
4
5
6
7
8
Rounds
En
erg
y c
on
su
mp
tio
n (
J)
CEEDBR5CEEDBR2EEDBR
Figure 3.12: EEDBR energy consumption (J)
20
Chapter 4
CoDBR Protocol
21
4.1 Motivation of Co-DBR
This section highlights the deficiencies in DBR protocol that becomes our motiva-
tion to develop CoDBR. DBR is a non-cooperative routing protocol. It follows a
receiver based approach in which source node broadcasts its data to all its neigh-
bors and one forwarder is selected based upon minimum depth. Hence, data is
routed from source to destination over a single noisy link in a multi-hop fashion.
Due to noise and multi-path fading in underwater environment, signal suffers high
BER.
Figure 4.1: Non-cooperative communication
Figure 4.2: Cooperative communication
Consider a scenario in which there is a single link and a school of fish as an
obstacle between a source and a destination node as shown in fig. 4.1. Presence
of such obstacle may create two problems: link breakage or high BER. Both
contribute to low reliability of network, thus making it unsuitable for mission
critical applications where data loss is unaffordable.
However, in the second scenario, fig. 4.2, it is observed that by sending same data
over multiple paths, we achieve path diversity. Multiple independent faded copies
of data are received, hence there is less data loss. In case there is a link breakage
or high BER, the availability of other paths increase the chance of data reception
at destination. Further, if no link failure occurs, best may still be extracted out
of multiple faded copies at the destination.
22
CoDBR aims to solve these problems via cooperative diversity. Like in DBR,
CoDBR also selects forwarder node along with two relays based on minimum
depth that cooperatively forward data to the destination. It increases the rate of
successful data delivery to the destination because in case of link failure, at least
one link is capable of delivering the data successfully to the destination. Secondly,
even if there is no link failure, however it still suffers high Bit Error Rate then
diversity can help to mitigate fading.
4.2 Proposed Scheme
This section gives details about the proposed scheme. Channel model describes the
network topology. Next sub section describes the working of proposed scheme. The
proposed scheme consists of two phases namely path setup and data transmission
phase. The detail of both these phases is elaborated in this section.
4.2.1 Channel Model
In CoDBR, each source node has two relays and a destination node as shown in
fig. 4.3.
R2
S D
R1
Figure 4.3: CoDBR system model
Source node broadcast its data to the two relay nodes and a destination node. Two
relays R1 and R2 forward data to the destination using AF technique. Three re-
ceived copies are combined at the destination using diversity combining technique.
Protocol assumes that relay nodes are in perfect synchronization with each other.
Binary Phase Shift Keying (BPSK) modulation scheme is used for modulating the
23
transmitted signal. Channel suffers Rayleigh fading with Additive White Gaus-
sian Noise (AWGN) noise. Equations (1-5) [29] describe the relationship between
transmitted and received signal at relays and destination nodes.
Ysd = Xsgsd + nsd (4.1)
Ysr1 = Xsgsr1 + nsr1 (4.2)
Ysr2 = Xsgsr2 + nsr2 (4.3)
The received signals at the destination sent by relays are
Yr1d = βYsr1gr1d + nr1d (4.4)
Yr2d = βYsr2gr2d + nr2d (4.5)
where Xs is the original signal. Ysd, Ysr1, Ysr2 are received signal by destination,
R1 and R2 respectively. Yr1d and Yr2d are received signal at destination sent by
relays. nsd, nsr1, nsr2 are channel noise from source to destination, source to R1
and source to R2 respectively. nr1d, nr2d represents channel noise from relays to
destination link. gsrd, gsr1, gsr2 are channel gain from source to destination, source
to R1 and source to R2 respectively. gr1d, gr2d represents channel gain from relays
to destination link. β is the amplification factor. The three independently faded
copies of same data are combined at destination using Maximal Ratio Combining
(MRC) technique.
4.2.2 Proposed Scheme: CoDBR
CoDBR is a localization free protocol and only depth information of sensor node is
used in routing the data. In CoDBR, all nodes exchange their depth information
among their neighboring nodes at the start. Source node identifies its neighbors
and registers them in its neighbor list. The node with lowest depth is selected by
the source node from its neighbor list for next hop destination. Source further se-
lects two relays from the neighbor list with second and third lowest depth. CoDBR
makes selection of relays and next hop destination on the basis of minimum depth
neighbor. In the data forwarding phase, source node broadcasts data to the next
hop destination, R1 and R2. The two relays then retransmit the received signal to
the next hop destination using amplify and forward cooperative scheme. Hence,
at the destination there are three independently faded copies of same data which
24
are combined using maximal ratio combining technique. CoDBR scheme consists
of two phases
1. Path setup phase
2. Data transmission phase
4.2.2.1 Path Setup Phase
In this phase, a multi-hop path is established from each source node to sink node
as shown in fig. 4.4. Source node first checks if it is in the vicinity of sink node
and selects sink as its next hop. Source further selects two relays based on lowest
depth to cooperatively forward data to sink. If it is not in vicinity of sink, then
source node selects its next hop destination based on the lowest depth neighbor
node among the neighboring nodes in its transmission range. Since network is
sparse and nodes are randomly deployed, therefore source node looks for nearby
relays. In case of more than two neighbors, relays are selected on the basis of
lowest depth. Algorithm 1 gives details of path setup phase.
4.2.2.2 Data Transmission Phase
In this phase, data is transmitted from source to sink through the path that is
established in path setup phase. Source node broadcasts data to relays and next
hop. Relays retransmit the same data using Amplify and Forward scheme. AF
is used because path loss, fading and noise weaken the signal which needs to be
Surface Buoy
Source / Destination
Relay
Node
De
pth
LinkTransmission Range
Figure 4.4: CoDBR multi-hop path from source to destination
25
Algorithm 1 Path Setup Algorithm
S = Total no. of Nodesfor i = 1 to S do
SINKREACHED=false
while not(SINKREACHED) doif Ri >0 and NextHop = SINK then
Find neighbors N for iSort N in ascending order Depth wiseif N >=2 then
Make 1st neighbor as relay1Make 2nd neighbor as relay 2SINKREACHED=true
else if N <=1 then
Make 1st neighbor as relay1SINKREACHED=true
else
break
end if
else if Ri >0 and not (NextHop = SINK) then
Find neighbors N for iSort N in ascending order Depth wiseif N >=3 then
Make 1st neighbor as NextHopMake 2nd neighbor as relay 1Make 3rd neighbor as relay 2
else if N <=2 then
Make 1st neighbor as NextHopMake 2nd neighbor as relay 1
else if N <=1 then
Make 1st neighbor as NextHopelse
break
end if
else if Ri <0 then
break
end if
end while
end for
26
amplified. When relays transmit data to destination, they do not aggregate their
own sensed data with the data of the source node. They only forward the amplified
version of data sent by the source node. Their own data is transmitted at their
own turn.
At the destination, three data copies, i.e. data from source to destination, source
to R1 to destination and from source to R2 to destination are combined using
MRC technique. Destination calculates the BER of received data and checks
against the threshold T. T is the maximum allowable error rate in data. If BER is
less than or equal to T, packet is accepted, otherwise, packet is dropped. CoDBR
has multi-hop path so this process is repeated at each next hop destination till
the sink is reached. Protocol assumes that each node is sending single packet per
round and in case of packet drop there is no retransmission.
4.3 Simulation Results and Discussions
Simulations are performed in MATLAB. Total number of sensor nodes are 225
which are randomly deployed underwater at an area of 500m x 500m. Four sinks
are located at an equal horizontal distance of 100m on the surface. Each node
has fixed transmission range of 100 m. Data packet size is 1000 bits and control
packet is of 48 bits. Initial energy of each node is 70 Joules. It is assumed that
each alive node will send 1 packet per round. Table 1 summarizes the simulation
parameters. BER threshold T is 0.50.
Table 4.1: CoDBR simulation parameters
Parameter ValueNetwork Size 500 m x500 mTotal Nodes 225Initial Energy 70 JPacket size 1000 bitsFrequency 30Hz
T 0.5No. of Sinks 4
Transmission Range 100m
Following evaluation metrics are considered to evaluate CoDBR.
Network Lifetime: It is the time from the start of the network till the death of
last node.
Total Energy Consumption: It is the total energy consumed by all the alive
27
nodes in one round. It includes transmission, reception and sensing energy.
Throughput: It is the total no of packets successfully received by the sink per
round.
Packet Drop: It is defined as the total no. of packets sent by the nodes but not
received by the sink per round.
Packet Acceptance Ratio: It is defined as the ratio of total no. of packets
received by the sink to the total no. of packets sent to the sink per round.
Average End-to-End Delay: It represents the average time taken by packet to
travel from source to sink. It is measured in Seconds.
1. Network Lifetime
Fig. 4.5 and 4.6 shows the life time of CoDBR and DBR. CoDBR dies out
earlier than DBR, because in DBR only source node is transmitting data to
its next hop neighbor. However, CoDBR is using source node along with
two relay nodes to transmit data to the next hop. So, CoDBR is consuming
three times more transmission energy than DBR. This shows the trade-off
between energy conservation and reliability. Fig. 4.7 shows the total energy
consumption by both protocols. CoDBR is consuming approximately three
times more energy than DBR in stable region. As, the nodes start to die
after 300 rounds, the total energy consumption tends to decrease. At the
end of 1300 rounds total energy consumption of CoDBR is less than DBR.
This is because CoDBR and DBR has 105 and 203 alive nodes at that time,
respectively. Second reason is that in CoDBR because nodes transmit most
of the time without relays when few nodes are left alive and there is no
unnecessary data forwarding, so energy consumption is less than DBR. Near
the end of simulation, throughput of both the schemes is almost similar, so
total energy consumption is also alike.
2. Throughput
Fig. 4.8 shows that CoDBR outperforms DBR in the stable period with
83% more throughput. When all nodes are alive throughput of CoDBR is
220 and DBR is 120. At round 644, number of packets received at sink in
CoDBR is close to DBR. At this time, alive nodes of CoDBR are 179 and
DBR has 222 alive nodes. This shows that packet drop rate of DBR is more
than CoDBR. After 1400 rounds, throughput of CoDBR is constant to five
because nodes do not find any neighbours to transport data to sink. Only
nodes close to sink gets their data successfully received at sink and high
28
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
150
200
250
Rounds
De
ad
no
de
s
CoDBRDBR
Figure 4.5: Dead nodes
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
150
200
250
Rounds
Alive
no
de
s
CoDBRDBR
Figure 4.6: Alive nodes
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
2
4
6
8
10
12
14
Rounds
En
erg
y c
on
su
mp
tio
n (
J)
CoDBRDBR
Figure 4.7: Total energy consumption (J)
29
packet drop occurs for rest of the alive nodes data. Fig 4.9 confirms the
observation that in CoDBR out of 95 packets sent to sink ,90 packets are
dropped and only 5 packets reach to sink. At 2500 round both lines are close
because of few nodes left alive and most of the packets are dropped due to
non availability of neighbours to carry data to sink.
Fig. 4.9 is actually the difference of total no. packet sent to sink per round
to the total no. packets received by the sink per round. As each node sends 1
packet per round so this graph is a difference of alive nodes and throughput.
Packet drop in CoDBR is very less because, CoDBR drops packet only when
none out of 3 links are available or Bit error rate of the combined signal e.g.
source to destination, source to relay1 to destination and source to relay2
to destination is greater than 50%. This result supports the reliability of
CoDBR. High packet drop of DBR is due to single poor link having BER
greater than 50% most of the time.
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
150
200
250
Rounds
Pa
cke
ts r
ece
ive
d a
t sin
k
CoDBRDBR
Figure 4.8: Throughput
3. Packet Acceptance Ratio
Fig. 4.10 is about packet acceptance ratio. When all nodes are alive, CoDBR
has double packet acceptance ratio than DBR. As nodes start to die, accep-
tance ratio starts decreasing because less packets are sent to sink and more
packets are dropped. At the end PAR of CoDBR is higher than DBR because
more packets are delivered to sink as compared to DBR.
4. Delay
Fig. 4.11 shows that average end-to-end delay of DBR is less than CoDBR
30
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
50
100
150
200
Rounds
Pa
cke
ts d
rop
pe
d
CoDBRDBR
Figure 4.9: Packets dropped
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
0.2
0.4
0.6
0.8
1
Rounds
Pa
cke
t a
cce
pta
nce
ra
tio
CoDBRDBR
Figure 4.10: Packet acceptance ratio
because it is sending data to next hop without waiting to receive data from
relay nodes in the next time slot. Delay increases because, this value is the
delay of all the packets sent to sink per round. When throughput starts
decreasing, delay increases because, packet is dropped somewhere in the
middle of the path from source to sink and its time is added in total delay.
31
0 500 1000 1500 2000 2500 3000 3500 4000 4500 50000
2
4
6
8
10
Rounds
Ave
rag
e e
nd
−to
−e
nd
de
lay (
se
c)
CoDBRDBR
Figure 4.11: Average end-to-end delay
32
Chapter 5
ACE and E-ACE Protocol
33
5.1 Motivation of ACE and E-ACE
In literature little attention has been paid to performance analysis of incremental
relaying for UWSN. In [30], Ikki et al. presented end to end performance anal-
ysis of incremental relaying cooperative diversity over rayleigh fading channels.
Closed form expression have been determined for BER, outage probability and
average achievable rate. Ikki et al. in [31] also presented performance analysis
of incremental relaying in which the best relay among multiple available relays
retransmits the source signal. Closed-form expressions for the BER, the outage
probability and average channel capacity are determined. Duy et al. [32] derived
closed-form expressions of outage probability and average channel capacity. They
exploited multiple relays and select one of the relays to retransmit signal.
The above mentioned research , focus on single retransmission in case of failure
of direct transmission. Due to severe multi-path fading, path-loss and noise, link
quality is very poor in water. In some conditions data may not reach the destina-
tion. In case, it reaches destination, quality is so poor that makes it useless and
protocol suffers from high packet drop and degraded throughput performance.
Therefore, single retransmission may not be sufficient to achieve desired signal
quality at receiver and require more number of retransmissions. This is the major
motivation behind our research.
5.2 System Performance Analysis and Proposed Scheme
This section presents outage performance analysis of incremental relaying coop-
erative diversity with cooperative retransmissions. Details of proposed protocols
are also given.
5.2.1 System Model and Outage Performance Analysis
Fig. 5.1 shows the proposed system model for a single source destination pair.
This model consists of a Source (S), Destination (D) and Relays (R), where
R = {R1, R2, ...Rm} and 1 ≤ m ≤ 250. Each node is equipped with a sin-
gle omni directional antenna. Each link is assumed to have rayleigh fading and
AWGN. BPSK modulation scheme is used in the proposed model. Relays use AF
cooperative diversity relaying technique. MRC is used at destination as a diversity
combining technique.
34
D
R
D
R3D
D
�2D
S�
�R�
��m
���
Source/ DestinationRelays
Figure 5.1: System model
The proposed system is supposed to follow incremental relaying cooperative di-
versity. Communication process takes place in two phases. In the first phase, S
broadcasts its signal to D and R. If signal received at destination is of sufficient
quality, relays are not supposed to retransmit data to destination. On the other
hand, if destination receives low quality signal, then relays retransmit the signal
one by one till the desired signal quality is achieved or all the relays are expired.
Quality of a signal is measured in terms of SNR threshold, γ0. The value of γ0
depends on the sensing environment. Too low value of γ0 causes direct transmission
for most of the time and too high value results in many retransmissions. The
received signals at relays and destination are mathematically represented as:
ySD(t) = hSD(√P )x(t) + n0(t) , (5.1)
ySRi(t) = hSRi
(√P )x(t) + nj(t) , (5.2)
yRiD(t) = hRiD(√P )xs(t) + ni(t) , (5.3)
xs(t) = GySRi(t) . (5.4)
Where i=1→ m and G is the amplification factor.
Using MRC at destination, the signal at D is given as:
yD(t) = ySD + yRiD . (5.5)
Here hSD, hSRiand hRiD are channel coefficients and n0, ni and nj represents
channel noises. x(t) is transmitted signal and P is its power. m is maximum
35
number of relays present in the common region of source and destination’s trans-
mission range which is also called cooperative region. These relays are responsible
for retransmission in case of erroneous reception at D.
5.2.2 Determination of Number of Available Relays
Relays help in retransmission of data when direct transmission has SNR less than
γ0. Relays are present in region common to S and D’s transmission range called
cooperative region. Maximum number of retransmissions depends on the number
of relays present in the cooperative region. Therefore, for m number of relays,
maximum allowable retransmissions are also m.
In this thesis, we assume that nodes are distributed uniformly over an area, A, with
density, ρ. Highlighted region in fig. 5.2 is the overlapping region in which nodes
can directly communicate to source and destination. In order to find number of
nodes in cooperative region, it is required to find the area, A, of that region. Let
the distance between source and destination is dSD. The node density is given as:
ρ =No. ofNodes
A. (5.6)
From Appendix .1, area of cooperative region with same transmission range, R, is
calculated as:
A = 2R2 cos−1(dSD2R
)− dSD2
√
4R2 − d2SD . (5.7)
Total number of nodes present in cooperative region is ρ× A. Since source and
destination nodes are also included in that area, therefore, total number of re-
transmission nodes are given as:
m = ρ× A− 2 , (5.8)
m =
[
ρ×[
R2 cos−1(dSD2R
)− dSD2
√
4R2 − d2SD
]]
− 2 . (5.9)
36
Cooperative Region Retransmission nodes
Source/ destination
dSD DS
R
Figure 5.2: Cooperative region and retransmission nodes
5.2.3 Outage Probability
Outage is defined as non-availability of signal at D. In the proposed model, outage
occurs when direct transmission along with all the retransmissions fail to achieve
the desired SNR threshold at the destination.
The expression for outage probability (Pout) can be written as:
Pout = Pr(γSD ≤ γ0)Pr(γSR1D + γSD ≤ γ0|γSD ≤ γ0)
Pr(γSR1D + γSR2D + γSD ≤ γ0|γSR1D
+γSD ≤ γ0)Pr(γSR1D + γSR2D + γSR3D
+γSD ≤ γ0|γSR1D + γSR2D + γSD ≤ γ0) · · ·
Pr(m∑
i=1
γSRiD + γSD ≤ γ0|m−1∑
i=1
γSRiD
+γSD ≤ γ0) , (5.10)
The first term, Pr(γSD ≤ γ0), in equation 5.10 represents the failure probability
of direct link (S → D)which requires the first relay to retransmit the signal.
Therefore, first relay is needed to retransmit the signal. Second term in equation
5.10, Pr(γSR1D + γSD ≤ γ0|γSD ≤ γ0), represents the probability that combined
signal (S → R1 → D and S → D) at the destination is below γ0 when direct
transmission has already suffered from outage. Similarly, third term, Pr(γSR1D +
γSR2D+γSD ≤ γ0|γSR1D+γSD ≤ γ0), shows that second retransmission is also failed
to achieve SNR above γ0, provided that the first retransmission is also in outage. In
37
third term, second retransmission is combined with the first retransmission along
with directly transmitted signal by using MRC. These retransmissions continue till
SNR above γ0 is achieved at destination or all available relays are utilized. When
mth retransmission fails to achieve SNR greater than γ0, outage is considered to
be occurred.
By using law of conditional probability, equation 5.10 can be reduced to:
Pout = Pr(γSD ≤ γ0)×Pr(γSR1D + γSD ≤ γ0)
Pr(γSD ≤ γ0)× Pr(γSR2D + γSR1D + γSD ≤ γ0)
Pr(γSR1D + γSD ≤ γ0)×
Pr(γSR3D + γSR2D + γSR1D + γSD ≤ γ0)
Pr(γSR2D + γSR1D + γSD ≤ γ0)× · · · ×
Pr(γSRmD + γSRm−1D + · · ·+ γSD ≤ γ0)
Pr(γSRm−1D + γSRm−2D + · · ·+ γSD ≤ γ0)
= Pr(γSRmD + γSRm−1D + · · ·+ γSD ≤ γ0)
= Pr(
m∑
i=1
γSRiD + γSD ≤ γ0) . (5.11)
Now we find a closed form expression for outage probability. In order to calculate
a closed form expression, m is limited to 3 for simplicity. Hence, Pout can be
expressed as:
Pout = Pr(γSR1D + γSR2D + γSR3D + γSD ≤ γ0) . (5.12)
To calculate Pout, it is required to know the output SNR at the destination. Since
MRC is used at destination, the SNR at destination is the sum of direct signal, γSD
and relayed signals, γSRiD. Where γSRiD is the equivalent SNR of S → Ri → D
[30]. The equivalent SNR can be written as [34]:
γSRiD =γSRi
γRiD
γSRi+ γRiD + 1
, (5.13)
A tight upper bound for γSRiD is given by [33]:
γ̄Ri= γ̄SRiD = min(γ̄SRi
, γ̄RiD) , (5.14)
where γ̄Riis minimum of γ̄SRi
and γ̄RiD. It is assumed that γ̄ follows exponential
distribution. PDF is also exponentially distributed with mean γ̄. For sum of expo-
nentially distributed independent random variables, their PDF is the convolution
of these variables.
38
Since γ̄d = γ̄SD + γ̄R1+ γ̄R2
+ γ̄R3. PDF fγ̄d, of γ̄d is given as:
fγ̄d =γ̄2SD [exp (−t/γ̄SD)− exp (−t/γ̄R3
)]
(γ̄SD − γ̄R1)(γ̄SD − γ̄R2
)(γ̄SD − γ̄R3)−
γ̄SDγ̄R2[exp (−t/γ̄R2
)− exp (−t/γ̄R3)]
(γ̄SD − γ̄R1)(γ̄SD − γ̄R2
)(γ̄R2− γ̄R3
)+
γ̄R1[υ + τ ]
(γ̄SD − γ̄R1)(γ̄R1
− γ̄R2). (5.15)
Where,
υ =γ̄R2
[exp (−t/γ̄R2)− exp (−t/γ̄R3
)]
γ̄R2− γ̄R3
, (5.16)
and
τ =γ̄R1
[exp (−t/γ̄R1)− exp (−t/γ̄R3
)]
γ̄R1− γ̄R3
. (5.17)
The detailed derivation of fγ̄d is given in Appendix .2.
By integrating equation .2.11 and doing some necessary simplification, we get
closed form expression for Pout as:
Pout = 1 +γ̄2SD (γ̄R3
exp(−γ0/γ̄R3)− γ̄SDexp(−γ0/γ̄SD))
(γ̄SD − γ̄R2)(γ̄SD − γ̄R3
)(γ̄SD − γ̄R1)
+
γ̄R2(γ̄R2
exp(−γ0/γ̄R2)− γ̄R3
exp(−γ0/γ̄R3))
(γ̄SD − γ̄R2)(γ̄R2
− γ̄R3)(γ̄SD − γ̄R1
)+
γ̄R1γ̄R2
(γ̄R3exp(−γ0/ ¯γR3
)− γ̄R2exp(−γ0/ ¯γR2
))
(γ̄R2− γ̄R3
)(γ̄R1− γ̄R2
)(γ̄SD − γ̄R1)
+
γ̄2R1
(γ̄R1exp(−γ0/ ¯γR1
)− γ̄R3exp(−γ0/ ¯γR3
))
( ¯γR1− ¯γR3
)( ¯γR1− ¯γR2
)( ¯γSD − ¯γR1)
. (5.18)
Fig. 5.3 shows the outage probability vs SNR of incremental relaying with
5 10 15 20 25 30 35 4010
−15
10−10
10−5
100
SNR (dB)
Out
age
prob
abili
ty
γ0=5
Single RetransmissionThree Retransmissions
Figure 5.3: Outage probability of incremental relaying with cooperative retransmissions
cooperative retransmissions. This simulation is conducted for the case when
39
γSD 6= γR16= γR2
6= γR3and γ0 is set to 5 dB. Fig. 5.3 clearly shows that more
number of retransmissions reduces outage probability. In case of error after first
retransmission, second retransmission may help the system to get out of outage
and same is the case with third retransmission.
5.2.4 Proposed Schemes
This section gives detail about proposed protocols; ACE and E-ACE. ACE allows
retransmissions from only two relays, whereas, E-ACE allows m relays to perform
retransmission. The proposed schemes work in time slot, such that one time slot
is the duration in which each node takes its turn for data transmission. Every slot
consists of three phases given as:
1. Depth exchange phase
2. Path establishment phase
3. Data transmission phase
5.2.4.1 Depth Exchange Phase
ACE and E-ACE are localization free routing protocols. Nodes are equipped with
inexpensive depth sensors. Each node broadcasts its depth information to all nodes
in its transmission range via small hello packet. Neighbour nodes are identified
on the basis of depth information. This process is repeated for all the nodes and
information regarding local neighbours is stored in each node’s database.
5.2.4.2 Path Establishment Phase
Once a node knows its neighbours, a multi-hop path is established from source
to sink. Path establishment phase has two main objectives: (i) identification of
the next destination and (ii) identification of relays that act as cooperative nodes
for the retransmission of data. Source node identifies its neighbours with the help
of depth information of other nodes. Nodes that have depth lower than that of
source node are included in the Forwarding Neighbour (FN) list and are called
forwarding neighbours. FNs are potential candidates for the next hop destination.
Neighbours with depth greater than source node are neglected. EEDBR algorithm
is followed to select the master node from FN list for the next hop destination. In
40
this process, the node having lesser depth and highest residual energy is selected
as master node.
Relay nodes are identified after the selection of master node. Relay nodes are
identified among the nodes that lie in the cooperative region as shown in fig. 5.2.
Nodes that are deployed in cooperative region are known as cooperative nodes. In
ACE, only two out of m nodes that are present in the cooperative region, act as
cooperative nodes. Whereas, in E-ACE, all the nodes present in the cooperative
region are candidates for retransmission of data. The nodes in cooperative region
may vary for each S-D pair. More m results in increased retransmissions thereby,
reducing outage probability. This is also true for vice versa. Selection process of
master node and retransmission nodes continues till sink is approached.
5.2.4.3 Data Transmission Phase
DATA
ACK
ACK
DATA
DATA
DATA
DATA ACK
ACK
Source node
Figure 5.4: Packet is accepted by master node and ACK is sent to retransmissionnodes indicating that no retransmission is required.
NACK 1Source node
DATA
R1
R2
R3
Rm
Figure 5.5: Packet is rejected by master node and asking for first retransmission fromR1.
In this phase, data is transmitted from source to sink through the path which
is established in path establishment phase. Source node broadcasts its data to
master and cooperative nodes. Data on its way from source to destination suffers
fading due to multi-path propagation and noise present in the water. These factors
41
NACK 2
Source node
DATA
R1
R2
R3
Rm
Figure 5.6: Packet is again rejected and asking for second retransmission from R2.
NACK m
Source node DATA
R1
R2
R3
Rm
Figure 5.7: mth relay node performing data retransmission
introduce high BER in the signal. In both protocols, data received at the master
node in direct transmission is compared with the data sent by the source node
and BER is calculated. If BER is less than or equal to maximum allowable BER,
E, data packet is accepted. On the acceptance of packet, master node sends ACK
signal to retransmission nodes as presented in fig. 5.4. Upon reception of ACK
signal, retransmission nodes discard the data.
However, if BER is greater than E, a negative acknowledgement signal, NACK,
is sent by master node to retransmission nodes. Fig. 5.5 shows that, for the first
retransmission, NACK1 is sent to one of the retransmission nodes. In response to
this NACK1, data is amplified and forwarded to master node by R1. This amplified
data may also suffer fading and noise. Therefore, at master node, direct signal
from source to master node and relayed signal from source to relay to master node
are combined using MRC to achieve acceptable SNR. At the master node, BER
is calculated and compared with the predefined threshold, E. If it is less than or
equal to E, data packet is accepted and ACK signal is sent to retransmission nodes.
Retransmission nodes discard the data on reception of ACK signal. However, if
BER is greater than E, NACK2 is sent to another retransmission node, which,
amplifies and forwards the data to master node as shown in fig. 5.6. Selection of
different retransmission nodes allows balanced energy consumption. If same node
does multiple retransmissions, it depletes its energy quickly and creates an energy
42
Direct
transmission
If BER<= E
First
retransmission
No
Accept packet
If BER<=E
Yes
Second
retransmission
No
If BER<=E
YesYes
Reject packet
No
Start
End
Figure 5.8: ACE data transmission
hole. At master node, three signals i.e., direct signal, and two relayed signals
are combined using MRC and the process of calculating BER is repeated. This
time, if BER is less than or equal to E, data packet is accepted and retransmission
nodes discard the data on reception of ACK from master node. If BER is not
acceptable, third retransmission node repeats the same process in E-ACE and the
process continues till all the retransmission nodes are expired or acceptable BER
is achieved. However, in case of ACE, after the failure of two retransmissions, data
packet is dropped. Fig. 5.8 and fig. 5.9 presents data transmission of a single
packet for ACE and E-ACE respectively.
5.3 Simulation Results and Discussions
In this section, simulation results of ACE and E-ACE are presented. Performance
of the proposed protocols is measured in terms of network lifetime, throughput,
total number of packets dropped, packet acceptance ratio and total energy con-
43
Direct
transmission
If BER <= E
Reject packet
If
retransmission nodes
are available
Yes
Retransmission
Yes
If BER<=E
NoNo
Accept packet
Figure 5.9: E-ACE data transmission
sumption of the network. Each plot is taken against time in seconds. A time slot
is the maximum time taken by data packet to reach from source to sink. Following
performance metrics are considered:
1. Network Lifetime: It is the time duration from the start of the network
till the death of the last node. It is measured in terms of number of dead
nodes/time slot.
2. Throughput: It is defined as the total number of packets successfully received
at the sink. It is measured in packets/time slot.
3. Packet Drop: It is the number of packets that are not successfully received
at the sink. Packet drop is measured as packets/time slot.
4. Packet Acceptance Ratio: It is the ratio of packets received at sink to the
total number of packets sent towards sink.
5. Total Energy Consumption: It is the total energy consumed by all the nodes
during transmission, reception, idle time. It is measured in Joules.
44
0 100 200 300 400 500 600 7000
50
100
150
200
250
Time (sec)
De
ad
no
de
s
E−ACEACEEEDBR
Figure 5.10: Network lifetime
In the simulations, total 250 nodes are randomly deployed at an area of 500 m
x 500 m. The network is homogenous and each node has initial energy of 30
Joules with a fixed transmission range of 100 m. 4 Sinks, which are not energy
constraint, are deployed on the surface of water at an equal distance of 100 m.
Underwater nodes are equipped with acoustic modems and sinks are equipped
with both acoustic and radio modems to communicate with underwater nodes
and surface sinks respectively. The size of data packet is 1000 bits and that of
control packet is 48 bits. Maximum allowable BER is set to 0.35. m is set to 3.
5.3.1 Network Lifetime
Fig. 5.10 shows network lifetime of proposed protocols which are compared with
EEDBR. Number of dead nodes are plotted against time. Network lifetime of
ACE and E-ACE is less than EEDBR because they consume more energy in
retransmissions in case of erroneous reception of data at destination. Difference
in lifetime of EEDBR and ACE is greater, whereas, the difference in lifetime of
ACE and E-ACE is very small. This is due to the fact that ACE allows only
two relays and hence, maximum two retransmissions can be done, whereas, E-
ACE allows m relays to retransmit data packet. However, most of the time two
retransmissions are enough to achieve acceptable BER at destination and in few
cases, there is a requirement for third or more retransmissions. This also confirms
that our assumption of keeping m=3 for simulation purpose is correct.
45
0 50 100 150 200 250 300 350 400 4500
2
4
6
8
10
12
Time (sec)
En
erg
y c
on
su
mp
tio
n (
J)
E−ACEACEEEDBR
Figure 5.11: Total energy consumption of the network (J)
5.3.2 Total Energy Consumption
Fig. 5.11 shows total energy consumption of proposed protocols in comparison
to EEDBR. This energy includes all the energy required for communication in
network. Energy consumption of E-ACE and ACE is greater than EEDBR due
to more allowed retransmissions. In the beginning, E-ACE shows less energy
consumption as compared to ACE. This is because, in initial time slots, E-ACE
has less throughput as compared to ACE as shown in fig. 5.12. Less through-
put contributes to low energy consumptions because less packets are transmitted.
However, E-ACE and ACE have almost same energy consumption after few sec-
onds.
In later time slots, energy consumption of three protocols is almost same because,
ACE and E-ACE have very few alive nodes to send data to sink. EEDBR has 200
nodes alive at 300 seconds, whereas, ACE and E-ACE have not more than 100
alive nodes.
Fig. 5.10 and fig. 5.11 show the drawback of E-ACE. These figures also show the
trade-off between energy consumption and reliability. More number of retrans-
mission causes extra energy consumption, thereby, reducing network lifetime.
46
0 50 100 150 200 250 300 350 400 4500
50
100
150
200
250
Time (sec)
Th
rou
gh
pu
t (P
acke
ts)
E−ACEACEEEDBR
Figure 5.12: Throughput
5.3.3 Throughput
Fig. 5.12 shows the performance of ACE and E-ACE in terms of throughput. E-
ACE has 91.2% more throughput as compared to EEDBR and 2.4% less through-
put as compared to ACE. Difference in throughput of E-ACE and ACE is low
because third retransmission is rarely invoked. Most of the time, two retrans-
missions are enough to get acceptable BER. In initial seconds, E-ACE show less
throughput as compared to ACE because, network conditions e.g., noise, fading
etc may not be suitable enough to produce acceptable BER even after three re-
transmissions. However, in later seconds, out performance of E-ACE is clearly
visible.
It is observed that throughput can be increased by using more cooperative nodes.
However, with more retransmissions, more energy is consumed. After 200 seconds,
ACE and E-ACE has less throughput as compared to EEDBR because they have
less number of alive nodes available to send data to sink and EEDBR has more
alive nodes. More alive nodes can send more data to sink n increase throughput.
Since the load on nodes is equally distributed therefore, in sparse network, nodes
may not find relays or next destination to send data to sink and die out because
of idle sensing energy consumption. This is the main reason of low throughput
and low packet acceptance ratio as shown in later time slots of fig. 5.14.
47
0 50 100 150 200 250 300 350 400 4500
20
40
60
80
100
120
140
Time (sec)
Pa
cke
ts d
rop
pe
d
E−ACEACEEEDBR
Figure 5.13: Packets dropped
5.3.4 Packet Drop
Fig. 5.13 compares E-ACE with ACE with EEDBR in terms of packet drop. It
is actually the difference between total number of packets sent to the sink to the
total number of packets successfully received at the sink. In the simulation, packet
is considered dropped when BER at destination is greater than the threshold after
double or triple retransmissions. Packet is also dropped when there is no neighbor
to provide cooperative retransmission. As each node sends single packet per time
slot and there is no data aggregation by relay nodes, therefore, maximum number
of packets sent are equal to total number of nodes.
E-ACE shows very less packet drop. Packet drop in case of ACE is also very
negligible and close to E-ACE. Whereas, EEDBR has packet drop close to 50.8%.
Fig. 5.13 stats show performance improvement of E-ACE compared to EEDBR.
Packet drop is increased in the end of simulation, because of fewer number of alive
nodes for cooperation in EEDBR.
5.3.5 Packet Acceptance Ratio
Fig. 5.14 demonstrates another parameter i.e., packet acceptance ratio. E-ACE
outperforms EEDBR with packet acceptance ratio close to 96.4% . Whereas,
EEDBR has lowest packet acceptance ratio of approximately 50.6%. This means
that less than half of the packets are successfully delivered to sink and rest of
the packets are dropped. ACE shows acceptance ratio of 98% in initial seconds
and then after 180 seconds, there is a drop off in it. This drop is because of two
48
0 50 100 150 200 250 300 350 400 4500
0.2
0.4
0.6
0.8
1
Time (sec)
Pa
cke
t a
cce
pta
nce
ra
tio
E−ACEACEEEDBR
Figure 5.14: Packet acceptance ratio
reasons: (1) Less relays are available to provide cooperative retransmissions and
(2) Since fading is considered, therefore, network condition is not suitable to get
BER more than threshold even after two retransmissions.
More acceptance ratio means more reliable network. In later time slots, EEDBR
has more packet acceptance ratio because it has more alive nodes to forward
data to sink. Retransmission mechanism clearly enhances network performance in
terms of reliability.
49
Chapter 6
Conclusion and Future Work
50
6.1 Conclusion
In this thesis, a novel cooperative routing protocol, CoDBR has been presented to
increase reliability and throughput efficiency of the network. Simulation results
show that, CoDBR proved to be beneficial for mission critical applications. It
offers 90% less packet drop and 83% more throughput compared to DBR in noisy
underwater environment. However, it consumes more energy and more end-to-end
delay.
Also, incremental relaying with cooperative retransmission protocols; ACE and
E-ACE, have been proposed for UWSN along with outage performance analysis.
Closed-form expression for outage probability has also been determined and ex-
pression for number of available relays has also been derived. Results show that
incremental relaying with three retransmissions show very less outage probability
as compared to regular incremental relaying cooperative diversity network having
only one retransmission mechanism. The proposed model and protocols have been
validated via simulations which show better performance than compared proto-
cols in terms of reliability and throughput efficiency. However, this reliability is
achieved at the cost of decreased network lifetime due to increased energy con-
sumption.
6.2 Future Work
In future, better criteria for the selection of relays and destination may be proposed
to achieve better load balancing of network. The protocols are restricted to send
single packet and data aggregation is not performed. Throughput can be improved
by allowing relays to aggregate their own data while transmitting source node data
to destination. Future works also include derivation of expression for outage, BER
and capacity of incremental relaying with m retransmissions with DF relaying.
51
Chapter 7
References
52
References
[1] A. Wahid, S. Lee, H. J. Jeong and D. Kim, “An energy efficient localization-
free routing protocol for underwater wireless sensor networks”, International
Journal of Distributed Sensor Networks, PP. 223-234, 2011.
[2] I. F. Akyildiz, P. Dario, and M. Tommaso, “Underwater acoustic sensor net-
works: research challenges ”, Ad hoc networks, Vol. 3, No. 3, PP. 257-279,
2005.
[3] S. Al-Dharrab, M. Uysal and T. M. Duman, “Cooperative underwater acous-
tic communications [Accepted From Open Call]”, IEEE Communications Mag-
azine, Vol. 51, No. 7, PP. 146,153, July 2013.
[4] H. Yan, Z. J. Shi and J. H. Cui, “DBR: depth-based routing for underwa-
ter sensor networks”, NETWORKING Ad Hoc and Sensor Networks, Wireless
Networks, Next Generation Internet, PP. 72-86, 2008.
[5] A. Babiker, and Z. Nordin , “Energy Efficient Communication for Underwater
Wireless Sensors Networks”, Energy Efficiency in Communications and Net-
53
works, PP. 979-953, 2012.
[6] N. Javaid, M. R. Jafri, Z. A. Khan, U. Qasim, T. A. Alghamdi, and M.
Ali, “iAMCTD: improved Adaptive Mobility of Courier Nodes in Threshold-
Optimized DBR Protocol for Underwater Wireless Sensor Networks”.
[7] M. R. Jafri , S. Ahmed, N. Javaid, Z. Ahmad, and R. J. Qureshi, “AM-
CTD: Adaptive Mobility of Courier nodes in Threshold-optimized DBR Pro-
tocol for Underwater Wireless Sensor Networks”, IEEE Eighth International
Conference on Broadband and Wireless Computing, Communication and Ap-
plications (BWCCA), 2013.
[8] A. Ghosh, J. W. Lee and H. S. Cho, “Throughput and Energy Efficiency of a
Cooperative Hybrid ARQ Protocol for Underwater Acoustic Sensor Networks”,
Sensors, Vol. 13, No. 11, PP. 15385-15408, 2013.
[9] M. M. Fareed, and A. Mohamed-Slim, “Efficient Incremental Relaying”. IEEE
International Symposium on Information Theory Proceedings (ISIT), PP. 1964-
1668, 2013.
[10] P. Wang, W. Feng, L. Zhang and V. O. Li. “Asynchronous cooperative trans-
mission in underwater acoustic networks”, IEEE Symposium on Underwater
Technology (UT) and Workshop on Scientific Use of Submarine Cables and
Related Technologies (SSC), PP. 1-8, 2011.
[11] M. J. Jannati, D. A. Abdollah, and T. V. Vahid, “Implementation of co-
operative virtual MISO communication in underwater acoustic wireless sensor
networks”, International Journal of Computer Science Issues (IJCSI), Vol. 8,
No. 4, 2011.
[12] C. Gao, Z. Liu, B. Cao and L. Mu, “Relay selection scheme based on propa-
gation delay for cooperative underwater acoustic network”, IEEE International
Conference on Wireless Communications and Signal Processing (WCSP), PP.
1-6, 2013.
[13] R. Cao, L. Yang and F. Qu, “On the capacity and system design of relay-aided
underwater acoustic communications”, IEEE conference on Wireless Commu-
nications and Networking Conference (WCNC), PP. 1-6, 2010.
[14] X. Cheng, R. Cao, F. Qu and L. Yang, “Relay-aided cooperative Underwater
Acoustic Communications: Selective relaying”, IEEE OCEANS, PP. 1-7, 2012.
54
[15] Y. Luo, L. Pu, Z. Peng, Z. Zhou and J. H. Cui,“CT-MAC: a MAC protocol for
underwater MIMO based network uplink communications”, Proceedings of the
Seventh ACM International Conference on Underwater Networks and Systems,
PP. 23, 2012.
[16] J. W. Lee, J. Y. Cheon, and H. S. Cho, “A cooperative ARQ scheme in
underwater acoustic sensor networks”, IEEE OCEANS, PP.1-5, 2010.
[17] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan, “Energy-efficient
communication protocol for wireless microsensor networks”, Proceedings of the
33rd IEEE Annual Hawaii International Conference on System Sciences, PP.
10-pp, 2000.
[18] A. Manjeshwar, and DP Agrawal, “TEEN: A Routing Protocol for Enhanced
Efficiency in Wireless Sensor Networks”, Parallel and Distributed Processing
Symposium, Vol. 3, 2001.
[19] G. Smaragdakis, I. Matta, and A. Bestavros, “SEP: A stable election pro-
tocol for clustered heterogeneous wireless sensor networks”, Boston University
Computer Science Department, 2004.
[20] L. Qing, Q. Zhu, and M. Wang, “Design of a distributed energy-efficient clus-
tering algorithm for heterogeneous wireless sensor networks”. Computer com-
munications, Vol. 29, No. 12, PP. 2230-2237, 2006.
[21] M. Chen, T. Kwon, S. Mao and Y. Yuan, “Reliable and energy-efficient rout-
ing protocol in dense wireless sensor networks”,International Journal of Sensor
Networks, Vol. 4, No. 1, PP. 104-117, 2008.
[22] M. Chen, M. Qiu, L. Liao, J. Park and J. Ma, “Distributed multi-hop co-
operative communication in dense wireless sensor networks”, The Journal of
Supercomputing, Vol. 56, No. 3, PP. 353-369, 2011.
[23] A. Valera , P. W. Lee, H. Tan, H. Liang, and W. K. Seah, “Implementation
and evaluation of multihop ARQ for reliable communications in underwater
acoustic networks”, IEEE OCEANS, PP. 1-6, 2009.
[24] H. Chen, C. Yueming, Y. Weiwei, Z. Dongmei, and H. Yingbo, “Throughput
and energy efficiency of a novel cooperative ARQ strategy for wireless sensor
networks”, Computer Communications, Vol. 35, No. 9, PP. 1064-1073, 2012.
55
[25] M. Tacca, M. Paolo, and F. Andrea, “Cooperative and reliable ARQ protocols
for energy harvesting wireless sensor nodes”, IEEE Transactions on Wireless
Communications, Vol. 6, No. 7, PP. 2519-2529, 2007.
[26] J. J. Alcaraz, and G. H. Joan, “Performance of single-relay cooperative ARQ
retransmission strategies”, IEEE Communications Letters, Vol. 13, No. 2, PP.
121-123, 2009.
[27] Y. Luo, L. Pu, Z. Peng, Z. Zhou, J. H. Cui and Z. Zhang, “Effective Relay
Selection for Underwater Cooperative Acoustic Networks”, IEEE 10th Interna-
tional Conference on Mobile Ad-Hoc and Sensor Systems (MASS), PP. 104-112,
2013.
[28] D. D. Tan, T. T. Le and D. S. Kim, “Distributed cooperative transmission
for underwater acoustic sensor networks”, IEEE Wireless Communications and
Networking Conference Workshops (WCNCW), PP. 205-210, 2013.
[29] K. El-Darymli, “Amplify-And-Forward cooperative relaying for a linear wire-
less sensor network”, IEEE International Conference on Systems Man and Cy-
bernetics (SMC), PP. 106-112, 2010.
[30] S. S. Ikki, and M. H. Ahmed, “Performance analysis of incremental-relaying
cooperative-diversity networks over Rayleigh fading channels”. Communica-
tions, IET, Vol. 5, No. 3, PP. 337-349, 2011.
[31] S. S. Ikki, and M. H. Ahmed, “Performance analysis of cooperative diversity
with incremental-best-relay technique over Rayleigh fading channels“, IEEE
Transactions on Communications, Vol. 59, No. 8, PP. 2152-2161, 2011.
[32] T. T. Duy, H. Y. Kong, “Performance analysis of hybrid decode-amplify-
forward incremental relaying cooperative diversity protocol using SNR-based
relay selection”, Journal of Communications and Networks, Vol. 14, No. 6, PP.
703-709, 2012.
[33] S. Ikki and M. H.Ahmed, “Performance analysis of cooperative diversity wire-
less networks over Nakagami-m fading channel”, IEEE Communications Letters,
Vol. 11, No. 4, PP. 334-336, 2007.
[34] M. O. Hasna, and M. S. Alouini, “End-to-end performance of transmis-
sion systems with relays over Rayleigh-fading channels”, IEEE Transactions
on Wireless Communications, Vol. 2, No. 6, PP. 1126-1131, 2003.
56
[35] X. Huang,H. Zhai, and Y. Fang, “Robust cooperative routing protocol in mo-
bile wireless sensor networks”, IEEE Transactions on Wireless Communications,
Vol. 7, No. 12, PP. 5278-5285, 2008.
57
Chapter 8
List of Publications
58
List of Publications
[1] H. Nasir, N. Javaid H. Ashraf, S. Manzoor, Z. A. Khan, U. Qasim and M.
Sher, “CoDBR: Cooperative Depth Based Routing for Underwater Wireless Sen-
sor Networks”, 9th IEEE International Conference on Broadband and Wireless
Computing, Communication and Applications (BWCCA’14), 2014.
[2] H. Nasir, N. Javaid, M. Murtaza, S. Manzoor, Z. A. Khan, U. Qasim, and M.
Sher, “ACE: Adaptive Cooperation in EEDBR for Underwater Wireless Sen-
sor Networks”, 9th IEEE International Conference on Broadband and Wireless
Computing, Communication and Applications (BWCCA’14), 2014.
[3] S. Mahmood, H. Nasir, S. Tariq, H. Ashraf, M. Pervaiz, Z. A. Khan and
N. Javaid, “Forwarding Nodes Constraint based DBR (CDBR) and EEDBR
(CEEDBR) in Underwater WSNs”, The 9th International Conference on Future
Networks and Communications (FNC’14), Vol. 24, PP. 228-235, 2014.
59
Chapter 9
Appendices
60
.1 Area of Overlapping Region
Fig. .1.1 shows the overlapping area of same transmission ranges of two sensors.
Transmission range is denoted by R. Therefore, the area of overlapping region is
found subtracting area of triangle SEG from area of sector [35].
Area of sector is given by:
AS =θ
2R2 , (.1.1)
where θ is the angle of sector in radians. In order to find θ, we know that triangle
SEG is an isosceles triangle with height dSD/2. Using half angle identity:
cos(θ
2) =
dSD/2
R=
dSD2R
. (.1.2)
solving for θ, we get:
θ = 2 arccos(dSD2R
) . (.1.3)
Putting equation .1.3 in equation .1.1, we get:
AS =2 arccos(dSD
2R)
2R2 = R2 arccos(
dSD2R
) . (.1.4)
Now area of triangle SEG is given as:
AT =1
2base ∗ height . (.1.5)
Using Pythagorean theorem to find OE of the right angle triangle SOE.
OE =
√
R2 − d2SD4
. (.1.6)
Therefore, base of the triangle SEG is 2 ∗OE.
base =√
4R2 − dSD2 , (.1.7)
Area of the triangle SEG is given as:
AT =1
2
√
4R2 − d2SD ∗ dSD2
=dSD4
√
4R2 − dSD2 . (.1.8)
61
Area of the overlapping region is 2 ∗ (AS − AT ).
A = 2 ∗[
R2 arccos(dSD2R
)− dSD4
√
4R2 − dSD2
]
,
A = 2R2 arccos(dSD2R
)− dSD2
√
4R2 − d2SD . (.1.9)
θ/�
θ
/�
d
R
R
�
E
G
OD
Figure .1.1: Area of overlapping region
.2 Derivation of fγ̄d
γ follows exponential distribution with mean γ̄. By definition of PDF:
fγ̄SD=
1
γ̄SDexp(−γSD/γ̄SD) , (.2.1)
fγ̄R1=
1
γ̄R1
exp(−γR1/γ̄R1
) , (.2.2)
fγ̄R2=
1
γ̄R2
exp(−γR2/γ̄R2
) , (.2.3)
fγ̄R3=
1
γ̄R3
exp(−γR3/γ̄R3
) . (.2.4)
62
fγ̄d is calculated by the convolution of above equations. By parts convolution,
fγ̄1(x) is derived as [30]:
fγ̄1(x) =1
γ̄SDexp(−x/γ̄SD) ∗
1
γ̄R1
exp(−x/γ̄R1)
=
∫ t
0
1
γ̄SDexp(−x/γ̄SD)×
1
γ̄R1
exp(−(t− x)/γ̄R1)dx
=1
γ̄SDγ̄R1
∫ t
0
exp(−γ̄R1
x− γ̄SDt + γ̄SDx
γ̄SDγ̄R1
)dx
=1
γ̄SDγ̄R1
exp(−γ̄R1
x−γ̄SDt+γ̄SDx
γ̄SD γ̄R1
)
(γ̄SD − γ̄R1)/γ̄SDγ̄R1
t
0
=1
γ̄SD − γ̄R1
exp(−t/γ̄SD)− exp(−t/γ̄R1) . (.2.5)
Similarly, convolution of fγ̄1 and fγ̄R2is calculated as:
fγ̄2(x) =1
γ̄SD − γ̄R1
exp(−x/γ̄SD)− exp(−x/γ̄R1) ∗ 1
γ̄R2
exp(−x/γ̄R2)
=
∫ t
0
1
γ̄SD − γ̄R1
exp(−x/γ̄SD)− exp(−x/γ̄R1)× 1
γ̄R2
exp(−(t− x)/γ̄R2)dx
=1
γ̄R2(γ̄SD − γ̄R1
)
∫ t
0
[[
exp(−γ̄R2
x− γ̄SDt+ γ̄SDx
γ̄R2γ̄SD
)
]
−[
exp(−γ̄R2
x− γ̄R1t+ γ̄R1
x
γ̄R1γ̄R2
)
]]
dx (.2.6)
=
[
γ̄SD
γ̄SD−γ̄R2
a− γ̄R1
γ̄R1−γ̄R2
b]
(γ̄SD − γ̄R1)
. (.2.7)
where,
a = exp(−t/γ̄SD)− exp(−t/γ̄R2) (.2.8)
b = exp(−t/γ̄R1)− exp(−t/γ̄R2
) . (.2.9)
63
Third convolution of fγ̄2 with fγ̄R3is calculated as:
fγ̄3(x) =
[
γ̄SD
γ̄SD−γ̄R2
a− γ̄R1
γ̄R1−γ̄R2
b]
(γ̄SD − γ̄R1)
∗ 1
γ̄R3
exp(−x/γ̄R3)
=
∫ t
0
[
γ̄SD
γ̄SD−γ̄R2
a− γ̄R1
γ̄R1−γ̄R2
b]
(γ̄SD − γ̄R1)
× 1
γ̄R3
exp(−(t − x)/γ̄R3)
dx
=γ̄SD
γ̄R3(γ̄SD − γ̄R2
)
∫ t
0
[
exp(−γ̄R3
x−γ̄SDt+γ̄SDx
γ̄SDγ̄R3
)− exp(−γ̄R3
x−γ̄R2t+γ̄R2
x
γ̄R2γ̄R3
)]
dx
γ̄SD − γ̄R1
− γ̄R1
γ̄R3(γ̄R1
− γ̄R2)
∫ t
0
[
exp(−γ̄R3
x−γ̄R1t+γ̄R1
x
γ̄R1γ̄R3
)− exp(−γ̄R3
x−γ̄R2t+γ̄R2
x
γ̄R2γ̄R3
)]
dx
γ̄SD − γ̄R1
,
(.2.10)
Integrating and doing some necessary manipulations, PDF of fγ̄d is given by:
fγ̄d =γ̄2SD [exp (−t/γ̄SD)− exp (−t/γ̄R3
)]
(γ̄SD − γ̄R1)(γ̄SD − γ̄R2
)(γ̄SD − γ̄R3)−
γ̄SDγ̄R2[exp (−t/γ̄R2
)− exp (−t/γ̄R3)]
(γ̄SD − γ̄R1)(γ̄SD − γ̄R2
)(γ̄R2− γ̄R3
)+
γ̄R1[υ + τ ]
(γ̄SD − γ̄R1)(γ̄R1
− γ̄R2). (.2.11)
Where:
υ =γ̄R2
[exp (−t/γ̄R2)− exp (−t/γ̄R3
)]
γ̄R2− γ̄R3
, (.2.12)
and
τ =γ̄R1
[exp (−t/γ̄R1)− exp (−t/γ̄R3
)]
γ̄R1− γ̄R3
. (.2.13)
64
