a novel routing algorithm for hierarchical wireless …...a novel routing algorithm for hierarchical...
TRANSCRIPT
A Novel Routing Algorithm forHierarchical Wireless Sensor Networks
Huang Lu
(Master’s Program in Computer Science)
Advised by Jie Li
Submitted to the Graduate School of
Systems and Information Engineering
in Partial Fulfillment of the Requirements
for the Degree of Master of Engineering
at the
University of Tsukuba
March 2009
平成 20 年度 システム情報工学研究科修士(工学)論文概要
A Novel Routing Algorithm for Hierarchical Wireless Sensor Networks
(階層型無線センサネットワークにおけるルーティングアルゴリズムに関する研究)
専攻名 コンピュータサイエンス 専攻
学籍番号 200720945 学生氏名 陸 璜
指導教員名 李 頡
無線センサネットワーク(Wireless Sensor Networks, WSNs)は無線のセンサノードで構成され
たネットワークである。そのネットワークは空間的に分布されたデバイスから構成され,各種の場
所で温度,音や動きなどの物理的な環境の状態をモニターする。典型的な無線センサネットワーク
構造では,センサノードは地域から情報を集めることと同時に,モニターされたデータを基地局に
送る。
無線センサネットワークでは,情報を収集し,低エネルギーの方法で情報のデータを基地局に送
ることと,ネットワークの生命時間を伸すことは重要な課題となっている。クラスタリングは低エ
ネルギー消耗を果たす方法の一つとなり,それはセンサノードをクラスターに分割することであ
り,各クラスターにはクラスタヘッドがある。いくつの関連文献が,無線センサネットワークにク
ラスタリングについて研究が行われていた。しかしながら,クラスタヘッドセンサノードから基地
局までのルーティングアルゴリズムについては,未だに研究が不十分である。
この論文では,無線センサネットワークのルーティング問題を研究する。特に,階層型無線セン
サネットワークのために新しいクラスタベースのルーティングアルゴリズムを提案する。このルー
ティングアルゴリズムでは,センサノードが基地局までの最小ホップ数によって各レベルの階層に
分けられる。クラスタヘッドセンサノードは,自主的に選択されて,マルチホップ通信を利用する
ことによりデータを基地局に送る。一方,非クラスタヘッドセンサノードはクラスタヘッドセンサ
ノードに直接通信する。本研究に対してシミュレーション実験を行った。シミュレーションの結果
は,提案されたルーティングアルゴリズムが大規模な無線センサネットワークに対して,ある既存
のルーティングアルゴリズムより良いネットワーク生命時間を得たことを示した。
Abstract
In wireless sensor networks (WSNs), gathering sensed information and transforming the
information data to the base station in an energy efficient manner and lengthening the
network life time are important issues. Clustering is an energy efficient way that divides
sensor nodes into many clusters, each of which has a cluster head. Some related work has
been conducted for clustering in WSNs. The routing from cluster head sensor nodes to
the base station, however, has not been well studied. In this thesis, we study the routing
problems in the WSNs. Specifically, we propose a novel cluster-based routing algorithm for
hierarchical WSNs. In our routing algorithm, we hierarchicalize sensor nodes into different
levels using hop number of transmission to the base station. Cluster head sensor nodes are
selected autonomously and transmit data to the base station using multi-hop transmission,
whereas non-cluster head sensor nodes transmit with cluster head sensor nodes directly. We
have conducted comprehensive simulation experiments. The simulation results show that,
for large scale WSNs, our proposed algorithm performs better than the typical existing
routing algorithms in terms of network lifetime.
Contents
1 Introduction 1
1.1 Introduction of Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . 1
1.2 Applications and Challenges of Wireless Sensor Networks . . . . . . . . . . 2
1.3 Clustering for Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . 3
1.4 Routing for Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . 4
1.5 Objective and Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Overview of Routing Algorithms in Wireless Sensor Networks 7
2.1 Categories in Routing Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Direct Routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Flat-based Routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Cluster-based Routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4.1 LEACH-like protocols for clustering . . . . . . . . . . . . . . . . . . 12
2.4.2 Multi-hop Data Forwarding in Cluster-based WSNs . . . . . . . . . 14
2.5 Time Synchronization Issue related to Routing . . . . . . . . . . . . . . . . 15
3 System Description 16
3.1 System Conception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 Clustering and Round Changing Mechanism . . . . . . . . . . . . . . . . . . 17
3.3 States of Sensor Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4 Proposed Routing Algorithm 19
4.1 Routing algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Alternative Sensor Node . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.4 Flow Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5 Performance Evaluation 27
5.1 Simulating A Sensor Node . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.2 Simulating A Wireless Sensor Network System . . . . . . . . . . . . . . . . 29
5.3 Performance Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.4 Simulation Details and Results . . . . . . . . . . . . . . . . . . . . . . . . . 33
i
5.4.1 The BS Stays in the Center of the Field . . . . . . . . . . . . . . . . 33
5.4.2 The BS Stays on the Margin of the Field . . . . . . . . . . . . . . . 36
5.4.3 Performance Result Analysis . . . . . . . . . . . . . . . . . . . . . . 36
6 Conclusion and Future Work 39
Acknowledgements 40
Bibliography 41
ii
List of Figures
1.1 Typical structure of a WSN . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Some current commercial sensor nodes . . . . . . . . . . . . . . . . . . . . . 2
1.3 Clustering sensor nodes for gathering data . . . . . . . . . . . . . . . . . . . 4
2.1 Classification of routing in WSNs . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 All nodes are ON and directly communicate with the base node . . . . . . . 10
2.3 An example of directed diffusion in WSNs . . . . . . . . . . . . . . . . . . . 11
2.4 The cluster-based hierarchy formed out of the network by using LEACH . . 13
3.1 System structure of a WSN . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 the operation time line of the system . . . . . . . . . . . . . . . . . . . . . . 17
4.1 Hierarchicalizing sensor nodes, clustering in each layer . . . . . . . . . . . . 20
4.2 Scheduling with TDMA in the system . . . . . . . . . . . . . . . . . . . . . 21
4.3 Routing for data transmission . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4 Alternative sensor node B for node A . . . . . . . . . . . . . . . . . . . . . 24
4.5 Algorithm flow chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.1 A simulated sensor node structure . . . . . . . . . . . . . . . . . . . . . . . 28
5.2 A sensor node structure under OMNeT++ . . . . . . . . . . . . . . . . . . 30
5.3 A WSN system structure with sensor nodes, a BS, and a manager . . . . . 30
5.4 The routing module in sensor node modules . . . . . . . . . . . . . . . . . . 31
5.5 A sensor network in simulating mode . . . . . . . . . . . . . . . . . . . . . . 32
5.6 Finding the optimum P with system energy dissipation . . . . . . . . . . . 34
5.7 Simulation topology with the BS in the center . . . . . . . . . . . . . . . . . 35
5.8 The number of alive nodes in the system with three routing algorithms, where
the BS is in the center of the field . . . . . . . . . . . . . . . . . . . . . . . 35
5.9 The comparison of system lifetime (the time of FND) in different routing
algorithms, where the BS is in the center of the field . . . . . . . . . . . . . 36
5.10 Simulation topology with the BS on the margin . . . . . . . . . . . . . . . . 37
5.11 The number of alive nodes in the system with three routing algorithms, where
the BS is on the margin of the field . . . . . . . . . . . . . . . . . . . . . . . 37
iii
5.12 The comparison of system lifetime (the time of FND) in different routing
algorithms, where the BS is on the margin of the field . . . . . . . . . . . . 38
iv
List of Tables
5.1 Simulation parameters in OMNet++ (the BS in center) . . . . . . . . . . . 34
5.2 Simulation parameters in OMNet++ (the BS on margin) . . . . . . . . . . 36
v
Chapter 1
Introduction
1.1 Introduction of Wireless Sensor Networks
A wireless sensor network (WSN) is a network of wireless sensor nodes, which consists
of spatially distributed devices using sensor nodes to monitor physical or environmental
conditions, such as temperature, sound, and motion, at different locations. Figure 1.1 shows
a typical sensor network structure, in which, sensor nodes are collecting the information
from the field, as well as transmitting sensed data back to the base station. Many sensor
systems are deployed in harsh, unattended and often adversarial physical environments such
as battle fields and desert [2].
Figure 1.1: Typical structure of a WSN
In a wireless sensor network, the individual nodes are capable of sensing their environ-
ments, processing the information locally, or sending it to one or more collection points
through a wireless link. Each node has a short-range transmission due to low RF (Radio
Frequency) transmit power [3]. Short-range transmission minimizes the possibility of the
transmitted signals being eavesdropped; also, it helps in prolonging the lifetime of the bat-
tery [8]. Figure 1.2 (redrawn from [1][2][4]) shows the examples of some of the commercial
1
sensor nodes.
Figure 1.2: Some current commercial sensor nodes
In some sensor system applications, the nodes are hard to reach and it is impossible
to replace their batteries. In other applications, the nodes must operate without battery
replacement for a long time. Such conditions make the system power consumption a very
crucial parameter. Thus, sensor nodes are severely constrained by the amount of battery
power, limiting the life time and quality of the network. So, it is crucial for WSNs to max-
imize nodes’ lifetimes, reduce bandwidth consumption by using local collaboration among
sensor nodes [1][5][7].
1.2 Applications and Challenges of Wireless Sensor Networks
Wireless sensor networks (WSNs) have significant impact upon the efficiency of military
and civil applications, which may be generally classified into two classes: data collection
and surveillance [2].
For data collection, the scientist, for example an environmental specialist, would want to
collect data from thousands of points spread throughout the area to recognize any changing
in the environment to keep the condition under control. In huge systems such as assembly
lines, shuttles, and airplanes, the sensor network could help the engineers to monitor the
system by periodically collecting data about the system’s modules.
Surveillance applications deploy sensor nodes by placing them at fixed locations through-
out an environment. All nodes continuously monitor the environment, and when there is
an abnormal, they send reports to the base station. Indoor Navigation is a project at MIT
[4], which deploys many sensor nodes that are placed at fixed locations inside a building
2
to survey all moving objects in the building, as well as help them to find out the right
destination.
When applying the applications which are addressed above, there are some challenges
to maintenance the sensor nodes in WSNs, as follows [1].
a. Small size of sensor nodes: A sensor node is very small in size, which leads to limit
battery capacity and lower hardware performance.
b. Overlapping sensing areas: Furthermore, the network is formed by randomly throwing
thousands or even millions of sensor nodes in an area. Surely, overlapping of sensing
areas (redundancy) among the nodes will happen.
c. Frequent change of network topology: Additionally, the network is usually installed in a
large area with many physical effects, such as earthquake, explosive, or etc. Therefore,
the topology of the network has to be changed frequently.
d. Wireless link: Sensor nodes typically communicate via wireless links. Due to the limited
bandwidth and interferences in the environment, problems such as dropped packets
and disconnected links may often arise.
To overcome the challenges listed above, it is desirable to build a sensor network that
employs multi-hop communication with, and is reconfigurable, and organizing (including
re-routing, redundancy reduction, and sensor deployment).
1.3 Clustering for Wireless Sensor Networks
Grouping sensor nodes into clusters has been widely pursued by the research community
in order to achieve the network scalability objective. Every cluster has a leader, often
referred to as the cluster-head (CH) as shown in Figure 1.3. The cluster membership may
be fixed or variable. Of the benefit, clustering may conserve communication bandwidth
since it limits the scope of inter-cluster interactions to CHs and avoids redundant exchange
of messages among sensor nodes. Moreover, a CH can aggregate the data collected by the
sensors in its cluster and thus decrease the number of relayed packets [5][9].
The LEACH (Low-Energy Adaptive Clustering Hierarchy) algorithm [6] is a well-known
clustering algorithm for WSNs of cluster-based architecture, by selecting the CHs in a round
changing mechanism. LEACH achieves improvement compared to direct transmissions, as
3
Figure 1.3: Clustering sensor nodes for gathering data
measured in terms of nodes’ lifetime. Following the idea of LEACH algorithm, a number
of algorithms have been presented such as PEGASIS (Power-Efficien Gathering in Sensor
Information Systems) [14], TEEN (a Routing Protocol for Enhanced Efficiency in Wireless
Sensor Networks) [15], APTEEN (A Hybrid Protocol for Efficient Routing and Compre-
hensive Information Retrieval in Wireless Sensor Networks) [16], PEACH (Power-efficient
and adaptive clustering hierarchy protocol for wireless sensor networks) [17] and an energy
efficient hierarchical clustering algorithm for wireless sensor networks [18], which use the
same concept with LEACH. In this thesis, for convenience, we call this kind of cluster-based
routing algorithms as LEACH-like protocols. However, the routing issue has not been well
studied in these LEACH-like algorithms. Because non-CH sensor nodes in a cluster use the
CH as the intermediary to the base station (BS), but for CH sensor nodes, there is nearly
no routing algorithm for them to transmit with the BS. Also in LEACH-like protocols, it is
assumed all sensor nodes could communicate with the BS using large transmission power. it
is unachievable in real world large range WSNs, because of not only the constrained range
of transmission in IEEE 802.15.4 for WSNs, but also the constraints of radio transmit
equipment [8][9].
1.4 Routing for Wireless Sensor Networks
Routing is the process of selecting paths in a network along which to send network
information. In WSNs, the routing algorithm specifies how sensor nodes communicate with
each other, which enables them to establish routes between sending sensor nodes to the
BS [9]. Routing in WSNs is challenging because of the inherent characteristics which are
4
different from other wireless networks. For example, due to the relatively large number of
sensor nodes, it is not possible to apply global IDs to all the sensor nodes, as the overhead of
maintenance is high. Besides, sensor nodes are constrained in terms of energy, processing,
and memory capacities [10]. We describe the details of routing algorithms and study the
taxonomy of routing algorithms in Chapter 2.
In this thesis, we allow multi-hop communication in routing, where the direct communi-
cation between CH sensor nodes and the BS is not possible due to the distance or obstacles
between them. Meanwhile, in WSNs, multi-hop data gathering usually causes a problem to
the sensor nodes which are close to the BS that, because of acting as intermediaries for data
transmission, their energy would exhaust faster. It is called self-induced black hole problem
[7], which would cause a transmission blank space around the BS, and if the blank space is
too wild for sensor nodes’ transmission, the WSN would lost its ability. As a method to this
issue, we proposed an approach using alternative sensor nodes (ASN) to reduce the effect of
the self-induced black hole problem in multi-hop data gathering transmission. Our approach
facing this issue is first to introduce the state of a sensor node. Then, we introduce our
method to change particular sensor nodes into ASNs, which also means changing the state
of particular sensor nodes. Although our approach could not solve this issue completely,
it could mitigate the effect caused by that, prolong the lifetime of sensor nodes, also could
prevent the sudden death of a sensor node.
1.5 Objective and Organization
The objective of this thesis is to study the routing issue for hierarchical WSNs. In
our proposal, we hierarchicalize sensor nodes into different levels using hop number of
transmission to the BS. CHs are selected autonomously and transmit with the base station
using multi-hop transmission, whereas non-cluster head sensor nodes transmit with CH
sensor nodes directly. The goal is to achieve the proposal, and find an energy efficient
routing algorithm for large scale WSNs.
The rest of this paper is organized as follows. Chapter 2 describes the details of net-
working in WSNs and some existing routing algorithms. In chapter 3, we describe the
mechanism of our system, the clustering algorithm from LEACH, and introduce the roles
of a sensor node for wireless sensor network systems. In chapter 4, we propose our power
efficient routing algorithm for hierarchical clustered WSNs. In chapter 5, we analyze our
5
algorithm with simulation. Chapter 6 offers conclusion and points of future work.
6
Chapter 2
Overview of Routing Algorithms in
Wireless Sensor Networks
When a WSN is activated, various tasks are performed to establish the necessary infras-
tructure to the network, such as distribution of sensor nodes and routing of data transmis-
sion, which will allow the sensor node to perform the applications normally. In particular,
each node must discover which other nodes it can directly communicate with, and its radio
power must ensure the connectivity.
In this chapter, we first provide an overview of routing algorithm for WSNs, then we focus
on the details of flat-based routing and cluster-based routing. In addition, we describe time
synchronization for ensuring the data transmission, and for checking the data information
(freshness).
2.1 Categories in Routing Algorithms
In general, routing in WSNs can be divided into flat-based routing, cluster-based routing,
and location-based routing depending on the network structure [10].
• In flat-based routing, all nodes are typically assigned equal roles or functionality to
perform the sensing and gathering tasks [12][13].
• In cluster-based routing, nodes will play different roles in the network; certain nodes
are used to process and send the information, while other nodes are used to perform
the sensing of the target [6][15].
• In location-base routing, sensor nodes’ positions are exploited to route data in the
network. The distance between neighboring nodes can be estimated on the basis of
7
incoming signal strengths [19][20].
A routing algorithm is considered adaptive if certain system parameters can be changed
to adapt to the current network conditions. Furthermore, these protocols can be classified
into multipath-based, query-based, negotiation-based, QoS-based, or coherent-based routing
techniques depending on the algorithm operation [9][10].
• In multipath-based routing, multiple paths are used rather than a single path in order
to enhance network performance, by maintaining multiple paths between the source
and destination [22][23].
• In query-based routing, the destination nodes propagate a query for data (sensing
task), and a node with this data sends the data which matches the query back to the
node that initiated the query [25].
• In negotiation-based routing, high-level data descriptors are used in order to eliminate
redundant data transmissions through negotiation [21][24].
• In QoS-based (quality of service) routing, the network has to balance between energy
consumption and data quality [26].
• In coherent-based routing, data processing is a major component in the operation of
wireless sensor networks [2][27].
The classification of routing protocols is shown in Figure 2.1.
Figure 2.1: Classification of routing in WSNs
Since this thesis is working on the cluster-based routing algorithm, we focus on the
categories in network structure. Before we describe the details of flat-based routing, we
8
first introduce the simplest way of data transmission, direct routing, which demands all
sensor nodes to have the ability to communicate with the BS directly. Afterward, we give
the overview of the flat-based routing algorithms and cluster-based routing algorithms,
which are related to our proposal.
2.2 Direct Routing
The simplest way for tracking an object is to turn on the sensor module of all nodes
in the network and have each node communicate directly with the base (shown in Figure
2.2), which is, the so-called Direct Communication method [9]. Direct Communication
is the simplest routing in flat-based routing algorithms, and this method renders optimal
accuracy in tracking objects, because of using one-hop communication with all sensor nodes.
However, direct communication is unrealistic in real-world applications because the base has
only a limited number of channels to communicate with the sensor nodes. In addition, the
communication distance of a sensor node is limited, so this method of direct communications
cannot apply to a large area.
2.3 Flat-based Routing
In flat networks, each node typically plays the same role and sensor nodes collaborate
together to perform the sensing task. Due to the large number of such nodes, it is not
feasible to assign a global identifier to each node. This consideration has led to data centric
routing, where the BS sends queries to certain regions and waits for data from the sensors
located in the selected regions. Since data is being requested through queries, attribute-
based naming is necessary to specify the properties of data. Early works on data centric
routing, e.g., directed diffusion [18] is shown to save energy through data negotiation and
elimination of redundant data.
C. Intanagonwiwat et. al. [9] proposed a popular data aggregation paradigm for WSNs,
called directed diffusion, and that becomes a popular and typical flat-based routing. Di-
rected diffusion is a data-centric and application-aware paradigm in the sense, that all
data generated by sensor nodes is named by attribute-value pairs. The main idea of the
data-centric paradigm is to combine the data coming from different sources and enroute
by eliminating redundancy, minimizing the number of transmissions; thus it saves network
9
Figure 2.2: All nodes are ON and directly communicate with the base node
10
energy and prolongs system lifetime. In directed diffusion, sensors measure events and cre-
ate gradients of information in their respective neighborhoods. The base station requests
data by broadcasting interests. Interest describes a task required to be done by the net-
work. Interest diffuses through the network hop-by-hop, and is broadcast by each node to
its neighbors. As the interest is propagated throughout the network, gradients are setup to
draw data satisfying the query that toward to the requesting node, for example, a BS may
query for data by disseminating interests and intermediate nodes propagate these interests.
Figure 2.3: An example of directed diffusion in WSNs
Each sensor that receives the interest setup a gradient toward the sensor nodes from
which it receives the interest. This process continues until gradients are setup from the
sources back to the BS. More generally, a gradient specifies an attribute value and a direc-
tion. Figure 2.3 shows an example of the working of directed diffusion ((a) sending interests,
(b) building gradients, and (c) data dissemination). When interests fit gradients, paths of
information flow are formed from multiple paths and then the best paths are determined.
In order to reduce communication costs, data is aggregated on the way. The goal is to find a
good aggregation tree which gets the data from source nodes to the BS. The BS periodically
refreshes and re-sends the interest when it starts to receive data from the source, which is
necessary because interests are not reliably transmitted throughout the network.
11
2.4 Cluster-based Routing
Cluster-based routing, originally proposed in wireline networks, are well-known tech-
niques with special advantages related to scalability and efficient communication. As such,
the concept of cluster-based routing is also utilized to perform energy-efficient routing in
WSNs. In this architecture, higher energy nodes can be used to process and send the in-
formation while low energy nodes can be used to perform the sensing in the proximity of
the target. This means that creation of clusters and assigning special tasks to cluster heads
(CH) can greatly contribute to overall system scalability, lifetime, and energy efficiency.
Cluster-based routing is an efficient way to lower energy consumption within a cluster and
by performing data aggregation and fusion in order to decrease the number of transmitted
messages to the BS. Hierarchical routing is mainly two-layer routing where one layer is
used to select CHs and the other layer is used for routing. However, most transmission
algorithms provided in cluster-based routing are little about routing, rather, focusing on
the method of clustering the sensor nodes, from which leads us to propose the novel routing
algorithm for cluster-based WSNs in this thesis [9][10].
2.4.1 LEACH-like protocols for clustering
The classic cluster-based algorithm called LEACH (Low-Energy Adaptive Clustering
Hierarchy), originally developed by W. Heinzelman, A. Chandrakasan and H. Balakrishnan
from MIT [6], builds a cluster-based hierarchy for the network. LEACH consists of 2
phases: set-up phase and steady-phase. In the set-up phase, sensors may elect randomly
among themselves a local CH with a certain probability. By doing so, the network may
balance energy dissipation across the whole network. After the CHs are selected, the heads
advertise to all sensor nodes in the network that they are the new CHs. Once the nodes
receive the advertisements, they decide which head they belong to. In the steady-phase,
sensors sense and transmit data to the BS through their CHs. After a certain period spent
in the steady-state, the network goes into the set-up phase again and enters another round
of selecting CHs. Figure 2.4 shows that the cluster-based hierarchy is formed out of a WSN
using LEACH. In the figure, the blue are CHs and the red circle is the base station.
As we described in section 3 of chapter 1, a number of routing protocols have been
proposed with the same concept from LEACH (e.g.,[14][15][16][17][18]), and we call them
12
Figure 2.4: The cluster-based hierarchy formed out of the network by using LEACH
13
LEACH-like protocols, in which sensor nodes are clustered into groups. LEACH-like pro-
tocol is more realistic than the direct diffusion methods because of its using multi-hops to
communicate. However, most LEACH-like protocols own the same limitations with LEACH
as follow:
a. LEACH assumes that all nodes have enough power to directly communicate with the
BS. Such an assumption is not true when the sensor network is applied to a large
area, which has also been addressed in chapter 1.
b. CH nodes communicating directly with the BS, possibly causes channel overload at
the BS.
c. Because cluster heads are randomly self-elected, in some areas within the network
there may not exist any cluster head.
2.4.2 Multi-hop Data Forwarding in Cluster-based WSNs
In cluster-based wireless sensor networks, Multi-hop data transmission is used for cluster
head nodes to transmit data to the BS, where direct communication is not possible due to
the distance between them. The solutions to this issue could be grouped into the following
two models [7].
• Multi-hop Planar Model:
In this model, a cluster head sensor node transmits data to the BS by forwarding its
data to one of its neighbors, which is closer to the BS. Following that, in turn the data
is sent to a neighbor sensor node which is yet closer to the BS. Thus the information
data travels from the cluster head node to neighbor sensor nodes until it arrives at
the BS. There are some protocols employing this model, such as [14], [15] and [28].
• Cluster-based Hierarchical Model:
In the hierarchical model, the network is broken into clustered layers. Data travel
from a lower cluster heads to a higher one. With this approach, the transmission hops
from one layer to another by hoping from sensor nodes in different layers, and covers
large distance. There are some protocols employing this model, e.g., [29] and [30].
In this thesis, we are facing and trying to solve some issues using our proposed routing
algorithm addressed above, which employs the multi-hop planar model for cluster-based
14
WSNs. Meanwhile, in the concept of our algorithm, only a part of sensor nodes could
communicate with the BS directly in their transmission range, and others need routing to
find the BS.
2.5 Time Synchronization Issue related to Routing
Since the sensor nodes in a WSN operate independently, their clocks may not be, or
stay, synchronized with one another [1]. This can cause difficulties when trying to integrate
and interpret data information sensed at different nodes. For example, if a moving car is
detected at two different times along a road, before we can even tell in what direction the
car is moving, we have to be able to meaningfully compare the detection times. And clearly,
we must be able to transform the two time readings into a common frame of reference before
we can estimate the speed of the car.
Configuring a beam-forming array or setting a TDMA (Time Division Multiple Access)
radio schedule are just two more examples of situations, in which collaborative sensing
requires the nodes agreeing on a common time frame. In our proposed algorithm, aiming to
avoid collisions and to get time synchronization, we use TDMA to synchronize the clocks
in all sensor nodes in the system.
TDMA is a channel access method for shared medium (usually radio) networks. It allows
several users to share the same frequency channel by dividing the signal into different time
slots. The users transmit in rapid succession, one after the other, each using his own time
slot. This allows multiple stations to share the same transmission medium (e.g., radio
frequency channel) while using only a part of its channel capacity [5].
15
Chapter 3
System Description
This section describes our system conception, round changing mechanism and clustering
algorithm for choosing the CHs in each round. After that, a short introduction is given to
four kinds of roles that a sensor node could act in WSNs.
3.1 System Conception
A large scale WSN may consist of hundreds or thousands of sensor nodes, and be dis-
tributed in a large area. Since the sensor nodes which are far away from the BS could not
communicate with the BS directly in the transmission range, all sensor nodes could reach
the BS use multi-hop communication, as shown in Figure 3.1.
Figure 3.1: System structure of a WSN
We consider the WSN system in which the data being sensed by sensor nodes is trans-
16
mitted to a base station, and there is only one base station with sufficient energy near part
of the sensor nodes and immobile. Because sensor nodes first transmit sensed data to the
CHs, so the CH sensor nodes perform data fusion. In addition, sensor nodes are desired to
be designed as cheap and energy-efficient as possible, we assume that all sensor nodes use
symmetric radio channel in the network, they are distributed randomly and densely, and
their energy is constrained.
3.2 Clustering and Round Changing Mechanism
A WSN has restricted energy resource, and energy consumption patterns in WSNs are
data processing and transmission. Data transmission cost is generally more expensive than
data processing. Thus the method that intermediate node (e.g., a CH) integrates data and
sends it to the BS is preferred than the method that each sensor node directly sends data to
the BS [9]. Sensor node is standing by for energy saving except sensing and sending data.
Figure 3.2: the operation time line of the system
The operation of the mechanism is divided by rounds as shown in figure 3.2, and each
round includes set-up phase for constructing clusters and steady-state for transmitting data
from sensor nodes to the BS via CHs. In each round, this algorithm elects CHs repeatedly,
and in each frame, a sensor node transmit the data to its CH.
We use dynamic clustering method LEACH to all sensor nodes to increase energy effi-
ciency by data integration [6]. For fair energy consumption, each node is randomly elected
as a CH in rounds. With CHs, Other non-CH sensor nodes join clusters using one-hop
transmission. In this method, CHs are randomly selected in rounds. In order to select
CHs in a round, each sensor node determines a random number and compare that with a
threshold. If the number is less than the threshold, the node becomes a CH for the current
round.
17
3.3 States of Sensor Nodes
The sensor nodes in a WSN are classified into four main states: sensing only, relaying
only, sensing-relaying, and inactivity [11]. In the sensing state, the node sensing circuitry
is on and it sends data toward the gateway sensor node or the BS. In the relaying state, the
node relays the data from other active nodes. When a node is both sensing the target and
relaying messages from other nodes, it is considered in the sensing-relaying state. Otherwise,
the node is considered inactive and can turn off its sensing and communication circuitry.
In hierarchical cluster-based architecture, the CH sensor node in a cluster could be at
the states of relaying only and sensing- relaying. Whereas the non-CH sensor nodes can be
at the any state of the four main states. The determination for the nodes’ state is done
at each node itself and the gateway sensor node based on the current organization, node
battery level and routing algorithm.
18
Chapter 4
Proposed Routing Algorithm
This section describes the operation details of the proposed routing algorithm and other
features. First, we describe the operation of our routing algorithm, which consists of five
steps in a round. Afterward, we introduce the details of proposed ASN (Alternative Sensor
Node) function, which is used in routing for saving the low energy sensor nodes. Last, we
collect the notations and draw the algorithm flow chart.
4.1 Routing algorithm
The operation of our proposed routing algorithm works in rounds as the round changing
mechanism described in chapter 3. In each round, the BS uses broadcasting to count hop
numbers and hierarchicalizes sensor nodes. Then we cluster sensor nodes into clusters, and
select transmission routes for data transmission to the BS. Overall, the algorithm is divided
into following five steps for conducting the data transmission in one round.
Step 1: Counting Hop Number
In this step, broadcasting from the BS is used to count the hop number for each sensor
node to the BS. Broadcasting is used, because the same broadcasting message from the BS
could reach a sensor node through many routes. After received broadcasting messages, each
sensor node uses a routing table to cache all the routes and their hop numbers from the BS
to itself in its memory, and then finds out the smallest hop number.
Step 2: Hierarchicalizing Sensor Nodes
Since we want to hierarchicalize sensor nodes into different layers to set the level of sensor
nodes, we hierarchicalize sensor nodes depending on the smallest hop number from the BS
19
to each sensor node. Set sensor nodes into one layer which communicate with the BS via
the same hope number. In figure 4.1, we show a image that how the sensor nodes would be
organized into hierarchical architecture.
Figure 4.1: Hierarchicalizing sensor nodes, clustering in each layer
Step 3: Clustering in the System
• Use the concept of LEACH Algorithm [6] to decide cluster head (CH) nodes in the
system, shown in Figure 4.1. As we addressed in section II, since each sensor node has
selected a value within 0 and 1 at the beginning of the round, it compares its selected
value with a threshold T (n), and decide whether to become the CH or not. If a sensor
node becomes a CH node independently, then it announces that to the sensor node
near itself by 1-hop transmission. The threshold T (n) is set as follows:
T (n) =P
1 − P ×(
rmod⌊
1
P
⌋) · · · ∀n ∈ G (4.1)
T (n) = 0 · · · ∀n /∈ G (4.2)
where T (n) is the threshold for node n to decide whether to turn into a CH with
the value from 0 to 1. P is a priori determined value that represents the desired
20
percentage of CH nodes during a round (e.g., P = 10%), r is the round number of the
current round, we have 20 frames per round, and G is the set of nodes that have not
been CH nodes in the last ⌊ 1
P⌋ rounds.
• For recognize each CH node in their layers, we could mark them as:
Level1CH1, Level1CH2, . . .
Level2CH1, Level2CH2, . . .
• The other non-CH nodes join a cluster using 1 hop communication, depending on the
strongest transmission signal.
• Because there could be sensor nodes that could not find a CH node in their transmis-
sion range, so we called them non-clustered sensor nodes.
Step 4: Transmission and Scheduling in Clusters
A CH node schedules all sensor nodes in its cluster with TDMA (Time-Division Multiple
Access) medium access control to avoid collision and to ensure the transmission in a cluster,
as shown in figure 4.2. Then, non-CH nodes could transmit the sensed data to the CH node.
Figure 4.2: Scheduling with TDMA in the system
21
Step 5: Selecting Transmission Routes
We establish transmission routes for CH nodes and non-clustered sensor node’s data
transmission, which concerns:
• First, minimize the hop number.
In step 2, the minimized hop number to the BS for each sensor node has been
counted by broadcasting. So, we select the transmission routes with smallest hop
number Hroute.
Hroute = Min {Hi} (4.3)
where Hi is the hop number of route i from from a sensor node to the BS.
• Second, maximize the energy in selected routes.
The method is that, calculate the lowest energy node in all routes from the selected
routes above with the hop number Hroute. Then, select the route with the highest
value of the lowest energy node. The route with the largest Eroute would be the route
for transmission.
Eroute = Max{
Min{
E′
ij
}}
(4.4)
where E′
ij is the energy of the lowest energy node j in route i.
• Third, establish routes for transmission with the method addressed above, for all CH
nodes and non-clustered sensor nodes, as the result in Figure 4.3.
• Last, discover the alternative sensor node (ASN) while transmitting data.
In the next section, we give a particular account of the ASN function. Here, we try
to find if there exists an ASN n (Nalt) for the sensor node m near the route. If so, the
sensor node m stops routing function and change its state into other states for saving
energy.
Nalt =
{
exist | (Smn <R
2)⋂
(En > Em)
}
(4.5)
where S is the distance between two sensor nodes, E is the energy of a sensor node,
and R is the transmission range for a sensor node.
• By then, the routing algorithm in one round is accomplished with the five steps
described above.
22
Figure 4.3: Routing for data transmission
4.2 Alternative Sensor Node
In this thesis, we use routing algorithm for selecting the transmission routes. There could
be sensor nodes with high energy near the transmission routes. To reduce the possibility of
the occurrence to the self-induced black hole problem, we introduce the mechanism of the
alternative sensor node (ASN) for a sensor node in its transmission range as follows.
• Suppose that node C transmits with node D via node A, and node B is in the
transmission range of A, C, and D.
• Determine the a distance threshold S = R2
for alternative sensing range, where R is
the transmission range for a sensor node.
Here, we talk about the reason why we decide S = R2
as the distance threshold
for ASN operation. Because the threshold should neither be too small nor too large.
Suppose that, if the threshold is determined larger, or nearly the whole transmission
range, the routing would lost the ability of communicating with the sensor nodes on the
opposite side of the original rout, which is selected for ASN operation. Meanwhile,
if the distance threshold is determined smaller, or smaller than one quarter of the
transmission range, the ASN operation would lost many opportunities of seeking the
23
ASN sensor node, since the distance for the ASN operation is small. Thus, we decide
the distance threshold for the ASN operation as R2
of the sensor nodes’ transmission
range.
• Assume that node A finds out node B in its transmission range, and the distance
between two sensor nodes A and B is SAB, shown in figure 4.4.
Figure 4.4: Alternative sensor node B for node A
• If SAB < S, and EB > EA, the node B becomes the ASN of node A. Thus the
transmission route from C to D is changed from C → A → D into C → B → D.
In this work, the ASN function is done using broadcasting by each sensor node to its
neighbor nodes. From figure 4.4 we can see that, after node A broadcasts message to all
the neighbor nodes in its transmission range, the node B which fits the requirements of
ASN function, returns a message to node A, C and D, thus the new rout is formed. In
this method, a sensor node changes the current state into others depending on the ASN
operation addressed above.
4.3 Notation
We show all notations used in our proposal below for reviewing and understanding the
formulas above.
T (n) - the threshold for selecting CH nodes in round n.
R - the transmission range of a sensor node.
P - A priori determined number that determines the
average percentage of CHs during a round.
24
r - the round number of the current round.
G - the set of nodes that have not been CHs in the last⌊
1
P
⌋
rounds.
Sij - the distance between sensor node i and j.
Eij - the energy of sensor node j on rout i.
Hi - the hop number from node i to the BS.
N - a capital character which points to a sensor node.
4.4 Flow Chart
For convenience, we also show the flow chart of our proposed algorithm in round changing
system in Figure 4.5.
The flow chart consists of there parts in one round time, which includes routing for the
set up of communication, clustering for gathering sensor nodes, and routing for transmitting
data. The first routing part is for the establishment of all sensor nodes to communicate with
the BS, and the counting of the hop numbers. The clustering part is to cluster all sensor
nodes in to clusters in different layers. The latter routing part is to ensure the information
data is transmitted from sensor nodes to the BS.
25
Figure 4.5: Algorithm flow chart
26
Chapter 5
Performance Evaluation
In this section, we present some results obtained by simulation. We use the network sim-
ulator OMNeT++ [31] with C++ programming to simulate our routing algorithm, which
is an open-architecture environment with GUI support. Its primary application area is the
simulation of communication networks, IT systems, querying networks, hardware architec-
tures, or even business processes. The main reason to select OMNeT++ for the simulation
is that it enables the placement of modules separately at any place.
The simulation modules we have built to test the performance of the routing algorithm,
generally consist of three sub-modules: a sensor node module, a sensor network system
module, and a routing protocol module. The source code of the simulations for this work
could be downloaded from my homepage, which is linked under our OSDP laboratory
homepage (http://www.osdp.cs.tsukuba.ac.jp).
5.1 Simulating A Sensor Node
According to the framework downloaded from homepage of OMNeT++ [31], and a
generic sensor node designed by the Sensor Network Research Group at Louisiana State
University [32], we have built a simulated sensor node as illustrated in Figure 5.1.
• The Layer 0 module represents the physical layer of a sensor node. It is responsible
for making connections between the node and its neighbors, and forwarding messages
from according to the routing protocol module.
• The MAC module represents pre-processing packet layers. It consists of gates (in/out)
and queues (incoming queue and outgoing queue). When the queue is full, it deletes
some of the oldest messages to make room in the queue for new messages. It helps
27
Figure 5.1: A simulated sensor node structure
to evaluate performance of the node. In the current simulation, we do not consider
those features, so this module is disabled to speed up the simulation.
• The Application module represents the application layer. Note that, each time after
sending a message, the module automatically sends a DECREASE ENERGY message
to the energy module (through the coordinator) to let the module decrease the energy
by a number of energy units.
• The Coordinator module is an interface to connect all modules together. It categorizes
an incoming message in order to deliver it to the right module. For example, when
receiving a DECREASE ENERGY message, it will forward the message to the energy
module.
• The Sensor module represents the sensor board in a sensor node. If the parameter
of SENSOR SWITCH is ON (=1), the module consumes energy, so, after an interval
(timer), the module sends a DECREASE ENERGY message to the energy module
(through the Coordinator). When the timer ticks, the waiting timer decreases. If the
waiting timer is zero, the module will turn off (SENSOR SWITCH parameter is set
to 0).
• The Radio module represents the radio board in a sensor node. If RADIO SWITCH
parameter is ON (=1), the module consumes energy, so, after an interval (timer), the
module sends a DECREASE ENERGY message to the energy module (through the
Coordinator).
28
• The Energy module represents the battery in a sensor node. If the module receives a
DECREASE ENERGY message, it decreases the energy level by a number of energy
units.
• Parameters of the sensor node:
1. CNNCTVTY: Maximum connections a node has.
2. OCCUPATION: Task of the node
3. PX: Position by X.
4. PY: Position by Y .
5. ID: ID of node.
6. FATHER ID: ID of node for forwarding messages.
7. SENSING RADIUS: Radius of the effective zone within which the node can
sense.
8. COMMUNICATION RADIUS: Radius of the effective zone within which the
node can communicate.
9. ENERGY: Energy level.
10. SENSOR SWITCH: Turn ON/OFF the sensor module.
11. RECEIVER SWITCH: Turn ON/OFF the radio receiver module.
12. POWER SWITCH: Turn ON/OFF the node.
Figure 5.2 is an illustration of the sensor node structure built using the OMNeT++ tool
set.
5.2 Simulating A Wireless Sensor Network System
A WSN system includes an array of sensor nodes. To simulate such a WSN in the
simulation environment, we need a module called manager [32] to help simulate tasks such
as making connections among the nodes, making connections between the nodes and the
object, and saving the simulation results. Also, it manages the broadcast from the base to
all nodes (creates connections from the base to all nodes). Finally, the manager module
controls the power switch (POWER SWITCH parameter) of all nodes in the network.
29
Figure 5.2: A sensor node structure under OMNeT++
Figure 5.3 shows a sensor network structure with an intruder under OMNeT++. Since
the sensor nodes in the system could use multi-hop transmission to communicate with the
BS, each sensor module uses a routing protocol module (a sub-mudule) to link with its
neighbors, as shown in figure 5.4.
Figure 5.3: A WSN system structure with sensor nodes, a BS, and a manager
To construct the sensor network, the manager module starts by first reading data in the
network.opp file, which stores network configuration information, including sensor node and
30
Figure 5.4: The routing module in sensor node modules
object positions, tasks, and routing tables. Then, it makes connections among nodes and
the BS using our proposed routing algorithm. Figure 5.5 shows a WSN system in simulating
mode, where the BS is in the center of the field, sensor nodes transmit the data via the CH
nodes (color in blue).
The following are the set of parameters for setting up the sensor network:
1. ZONE X: Area by x.
2. ZONE Y: Area by y.
3. FILE PATHS: Paths to input/output files for the simulation.
4. NNODES: Number of sensor node in the network.
5. NODE SENSING RADIUS: Sensing radius of each node.
6. NODE COMMUNICATION RADIUS: Communication radius of each node.
5.3 Performance Metrics
Overall, there are two types of metrics that we use to measure the performance as follow:
system lifetime (the time of FND (first node dies) and the number of alive nodes.
• System lifetime (the time of FND) - The definition of system lifetime can be
used to determine how alive a system is. It is difficult to have a precise criterion to
31
Figure 5.5: A sensor network in simulating mode
define the WSN lifetime, because users can define system lifetime differently based on
different applications. There exist many lifetime definitions, such as the time when
the first node dies (FND), the time when half of the nodes die, the time some fraction
of the nodes die, or the time that the network breaks in two or more segments [2]. We
use the most general definition of the lifetime in this thesis, the time FND. Therefore,
in order to maximize the system lifetime, we simply maximize the time of FND in a
WSN system.
• The number of alive nodes - To keep the ability of sensing and gathering data in
a WSN, it is essential to find how many sensor nodes are still functional in the entire
system. The ability of a WSN depends on the set of alive nodes (nodes that have not
failed) which can effectively communicate the information to the BS. Therefore, we
evaluate the functionality of the WSN system depending on counting the number of
alive nodes in the system.
Considering the energy consumption with and without clustering, routing or round
changing mechanism in multi-hop transmission WSNs, we evaluated our proposed rout-
32
ing algorithm by comparing it with other two data transmission algorithms, simple directed
diffusion and non-clustering routing algorithm. Simple directed diffusion is an improvement
of the classical flooding mechanism [9], which is a mechanism including neither routing nor
clustering algorithm. In this mechanism, each node runs a start-up routine, then starts the
transmission. If the radio channel is busy, the node backs off; otherwise, the node transmits
its data to the next-hop node. The sensor nodes which act as intermediaries relay received
data only from the higher layers to the lower layers or the BS, in which way data are trans-
mitted and directed to the BS. Non-clustering routing algorithm is an arrangement of our
proposal, in which each node runs the same algorithm as in chapter 4 without clustering
process, which means each node runs a start-up routine, then only uses the round changing
mechanism and routing algorithm for data transmission.
5.4 Simulation Details and Results
In our simulation experiments, we use a network where 200 nodes were randomly dis-
tributed in a 400m × 400m area, with a fixed BS located in the center of the sensing
field. The transmission range of a sensor node is 100m, and all sensor nodes periodically
sense events and transmit the data packet to the BS. The initial energy of each sensor
node is 0.5J , and the BS has unlimited energy. each data message is 512 bytes long, and
the packet header for each type of packet is 25 bytes long. The communication energy
parameter-electronics energy is set as: Eelec = 50nJ/bit, and the energy for data aggrega-
tion is set as: Eaggr = 5nJ/bit. We show a screen shot of our simulations in figure 5.5. For
convenience, we show a part of parameters in table 5.1.
Here, we look forward to find a proper desired percentage of CH nodes P during one
round. Thus, in the first simulation, P is set as 5%, 10%, 15%, which aims at 5 percent,
10 percent, and 15 percent nodes for CHs per round. The optimum P is expectated to
obtain the lowest system energy dissipation. The sumulation result shows that, P = 10%
is assumed to be the optimum percentage of CH nodes, as shown in figure 5.6.
Then, two case of WSN’ structures are considered for simulation: the BS stays in the
center of the simulation field, as well as the BS stays on the margin of the simulation field.
5.4.1 The BS Stays in the Center of the Field
The simulation parameters are show in table 5.1.
33
Figure 5.6: Finding the optimum P with system energy dissipation
Table 5.1: Simulation parameters in OMNet++ (the BS in center)
Map size 400m × 400m
BS position (200m, 200m)
Number of nodes 200
Transmit range 100m
Node distribution Random
P 10%
Message packet size 512byte
Node initial energy 0.5J
Energy for data aggregation 5nJ/bit
Communication energy parameter 50nJ/bit
Test items System lifetime, time of FND
34
Figure 5.7: Simulation topology with the BS in the center
Figure 5.8: The number of alive nodes in the system with three routing algorithms, wherethe BS is in the center of the field
35
Figure 5.9: The comparison of system lifetime (the time of FND) in different routing algo-rithms, where the BS is in the center of the field
5.4.2 The BS Stays on the Margin of the Field
The simulation parameters are show in table 5.2.
Table 5.2: Simulation parameters in OMNet++ (the BS on margin)
Map size 400m × 400m
BS position (0m, 200m)
Number of nodes 200
Transmit range 100m
Node distribution Random
P 10%
Message packet size 512byte
Node initial energy 0.5J
Energy for data aggregation 5nJ/bit
Communication energy parameter 50nJ/bit
Test items System lifetime, time of FND
5.4.3 Performance Result Analysis
Figure 5.8 and figure 5.11 show the comparison of system lifetime using our proposed
routing algorithm versus other two algorithms which are introduced above. Except for the
latter parts of the results, the simulation results demonstrate that the system lifetime of
our proposed routing algorithm is generally longer than that of non-clustering routing and
simple directed diffusion. However, because of the self-induced black hole problem, from
36
Figure 5.10: Simulation topology with the BS on the margin
Figure 5.11: The number of alive nodes in the system with three routing algorithms, wherethe BS is on the margin of the field
37
Figure 5.12: The comparison of system lifetime (the time of FND) in different routingalgorithms, where the BS is on the margin of the field
the figure we can see that, the result becomes into nearly straight line near the end of the
simulations. The reason is that sensor nodes in certain part of the system (far away from
the BS) could not transmit the data packets to the BS, since the intermediaries died already
for multi-hop transmission to the BS.
A confidence interval gives an estimated range of values which is likely to include an
unknown population parameter, the estimated range being calculated from a given set of
sample data [33]. We applied confidence Intervals to our simulation results, and the certain
percentage (confidence level) is set to 95%. Figure 5.9 and figure 5.12 present the time when
the first node dies (FND) in different data transmission algorithms, which also indicates the
balance of energy consumption in the system. The results demonstrate that our proposed
routing algorithm has a better energy consumption balance in the system than the other
two, especially, times better than simple directed diffusion, because the FND appears the
latest in our proposed routing algorithm.
38
Chapter 6
Conclusion and Future Work
In this thesis, we have introduced a novel power efficient routing algorithm for hierar-
chically clustered WSNs. The simulation results show that, compared with simple directed
diffusion and non-clustering routing algorithm, our proposed routing algorithm prolongs
the system lifetime for large scale multi-hop transmission WSNs.
Our future plan includes the improvements to our simulation experiments with more vari-
ations for better comparison results, such as changing the position of the BS, and changing
the probability of becoming CHs from all sensor nodes. Then, we will consider more about
the issue to the the self-induced black hole problem in the multi-hop transmission.
39
Acknowledgements
First of all, I would like to give my thankful words to Professor Jie Li of the Graduate
School of Systems and Information Engineering at the University of Tsukuba, my supervisor,
for all his guidance and advice throughout my research and life at the University of Tsukuba.
Furthermore, I would like to thank Associate Professor Shigetomo Kimura and Associate
Professor Masahiro Mambo for reviewing my thesis.
Then, I would like to thank the most members in my laboratory, the Laboratory of
Operating System and Distributed Processing (OSDP). Especially, to Wei Jing and Ruidong
Li, who helped a lot when I first came to this lab, to Yu Gu and Khoriba Ghada, who helped
a lot on my coding in C++ program, simulations and comments in English writing. More
over, I would like to thank all my friends in the University of Tsukuba for their friendship
and encouragement.
Finally, most of all, I would like to thank my parents, Huilin Lu and Maolin Huang, for
all their support throughout my schooling career and their financial backup for my living
in Japan.
40
Bibliography
[1] F. Zhao, and L. Guibas. ”Wireless Sensor Networks: An Information Processing Ap-
proach”, The Morgan Kaufmann Publishers, 2004.
[2] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci. ”Wireless sensor net-
works: a survey”, Computer Networks 38, 2002.
[3] Definition of frequency bands. Retrieved from http://www.vlf.it/frequency/bands.html.
[4] Cricket Project Group (2005). ”The Cricket Indoor Location System” website, MIT.
Retrieved from http://nms.lcs.mit.edu/projects/cricket/.
[5] A. A. Abbasi, and M. Younis. ”A survey on clustering algorithms for wireless sensor
networks”, Computer Communications, Volume 30, Issues 14-15, 2007.
[6] W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan. Energy efficient communi-
cation protocol for wireless microsensor networks”, HICSS’00, 2000.
[7] J. Ibriq, and I. Mahgoub. ”Cluster-Based Routing in Wireless Sensor Networks: Issues
and Challenges”, SPECTS’04, pp.795-796, 2004.
[8] ZigBee Alliance. Retrieved from http://www.zigbee.org/en/index.asp.
[9] K. Akkaya, and M. Younis. ”A survey on routing protocols for wireless sensor networks”,
Ad Hoc Networks 3, 2005.
[10] J.N. Al-Karaki, and A.E. Kamal. ”Routing techniques in wireless sensor networks: a
survey”, Wireless Communications, IEEE Volume 11, Issue 6, 2004.
[11] M. Younis, M. Youssef and K. Arisha. ”Energy-aware routing in cluster-based sensor
networks”, MASCOTS, 2002.
41
[12] D. Braginsky and D. Estrin. ”Rumor Routing Algorithm for Sensor Networks”, Proc.
1st Wksp. Sensor Network and Apps., Atlanta, GA, 2002.
[13] C. Schurgers and M.B. Srivastava. ”Energy Efficient Routing in Wireless Sensor Net-
works”, MILCOM Proc. Commun. for Network-Centric Ops.: Creating the Info. Force,
McLean, VA, 2001.
[14] S. Lindsey and C. Raghavendra. ”PEGASIS: Power-Efficien Gathering in Sensor In-
formation Systems”, IEEE Aerospace Conf. Proc., vol. 3, 9-16, 2002.
[15] A. Manjeshwar and D. P. Agarwal. ”TEEN: a Routing Protocol for Enhanced Efficiency
in Wireless Sensor Networks”, 1st Int’l. Wksp. on Parallel and Distrib. Comp. Issues in
Wireless Networks and Mobile Comp., 2001.
[16] A. Manjeshwar and D. P. Agarwal. ”APTEEN: A Hybrid Protocol for Efficient Routing
and Comprehensive Information Retrieval in Wireless Sensor Networks”, Proc. Int’l.
Parallel and Distrib. Proc. Symp., 2002.
[17] S. Yia, J. Heoa, Y. Choa and J. Hong. ”PEACH: Power-efficient and adaptive clustering
hierarchy protocol for wireless sensor networks”, Computer Communications Volume 30,
Issues 14-15, 2007.
[18] S. Bandyopadhyay and E. J. Coyle. ”An energy efficient hierarchical clustering al-
gorithm for wireless sensor networks”, IEEE INFOCOM 2003. Twenty-Second Annual
Joint Conference of the IEEE Computer and Communications Societies, 2003.
[19] F. Ye, A. Chen, S. Lu and L. Zhang. ”A Scalable Solution to Minimum Cost Forwarding
in Large Sensor Networks”, Proc. 10th Int’l. Conf. Comp. Commun. and Networks, 2001.
[20] W. Heinzelman, J. Kulik and H. Balakrishnan ”Adaptive Protocols for Information
Dissemination in Wireless Sensor Networks”, Proc. 5th ACM/IEEE Mobicom, Seattle
WA, 1999.
[21] IEEE 802.15. Retrieved from http://www.ieee802.org/15/.
[22] Q. Fang, F. Zhao and L. Guibas. ”Lightweight Sensing and Communication Protocols
for Target Enumeration and Aggregation”, Proc. 4th ACM MOBIHOC, 2003.
42
[23] J. N. Al-Karaki and A. E. Kamal. ”On the Correlate Data Gathering Problem in
Wireless Sensor Networks”, Proc. 9th IEEE Symp. Comp. and Commun., Alexandria,
Egypt, 2004.
[24] J. Kulik, W. R. Heinzelman and H. Balakrishnan. ”Negotiation-Based Protocols for
Disseminating Information in Wireless Sensor Networks”, Wireless Networks vol. 8,
2002.
[25] J. N. Al-Karaki, R. Ul-Mustafa, A. E. Kamal. ”Data Aggregation in Wireless Sen-
sor Networks - Exact and Approximate Algorithms”, Proc. IEEE Wksp. High Perf.
Switching and Routing 2004, Phoenix, AZ, 2004.
[26] S. Tilak, N. Abu-Ghazaleh and W. Heinzelman ”A Taxonomy of Wireless Micro-sensor
Network Models”, ACM SIGMOBILE Mobile Comp. Commun. Rev., vol. 6, no. 2, 2002.
[27] Y. Yu, D. Estrin and R. Govindan. ”Geographical an Energy-Aware Routing: A
Recursive Data Dissemination Protocol for Wireless Sensor Networks”, UCLA Comp.
Sci. Dept. tech. rep., UCLA-CSD TR-010023, 2001.
[28] C. Cano, B. Bellalta, P. Villalonga and J. Perello. ”Multihop Cluster-based Archi-
tecture for Sparse Wireless Sensor Networks”, Wireless Conference, EW 2008. 14th
European, 2008.
[29] Y. Jia, L. Zhao and B. Ma. ”A Hierarchical Clustering-based Routing Protocol for
Wireless Sensor Networks Supporting Multiple Data Aggregation Qualities”, Interna-
tional Journal of Sensor Networks, Volume 4, Numbers 1-2, pp. 79-91(13), 2008.
[30] H. CHENG, G. YANG and S. HU. ”A Novel Hierarchical Routing Protocol for Wireless
Sensor Networks”, The Journal of China Universities of Posts and Telecommunications
Volume 15, Issue 3, Pages 75-81, 2008.
[31] OMNeT++. Retrieved from http://www.omnetpp.org/.
[32] Sensor Network Research Group at Louisiana State University. ”Sim-
ulating Wireless Sensor Networks with OMNeT++”. Retrieved from
http://bit.csc.lsu.edu/sensor web/final papers/SensorSimulator-IEEE-Computers.pdf.
43
[33] H. Kameda, K. Kino and J. Li. System Performance Evaluation, Theory and Applica-
tions, Kyoritu Press (in Japanese), 253 pages, 1998.
44