Anchor node path planning for localization in wireless sensor networks Ketan Sabale 1 • S. Mini 1 Published online: 10 June 2017 Ó Springer Science+Business Media, LLC 2017 Abstract Localization is one of the most important chal- lenges of wireless sensor networks because the location information is typically used in other domains such as coverage, deployment, routing, and target tracking. There exist some localization algorithms that facilitate the sensor nodes to locate itself using the mobile anchor node posi- tion. Some crucial attempts have been made in the past for optimizing the mobile anchor node trajectory with good accuracy. This paper presents a novel path planning scheme, D-connect, which ensures the localization of all the sensor nodes with minimum trajectory length. The performance of the proposed scheme is evaluated through a series of simulations. Experimental results reveal that the shortest path for traversing the whole area can be traced with the minimum localization error using this method. It also shows that D-connect outperforms the existing meth- ods in terms of the anchor node trajectory length as well as the localization error. Keywords Sensor network Localization Anchor node Path planning mechanisms 1 Introduction Wireless sensor networks (WSNs) are composed of large collection of sensors that may be randomly deployed in a certain geographical region. The sensor nodes collect environmental data and forward that data to a remote device where the data is analyzed and processed. If the sensors cannot pass the information to the remote device directly, some intermediate nodes have to forward the data [8]. Sensor networks have a wide range of application areas such as home, environment monitoring, military surveil- lance, animal tracking, etc. WSNs can be used in the dis- aster relief services where human operations are difficult. Increased accuracy and minimizing time for location esti- mation are the important factors to be considered in emergency services. Sensor nodes have limited power resources, computational power, and memory availability [1]. Coverage, deployment, tracking and localization are some challenges in WSNs. Since the location information is used in other domains, it is necessary to determine the origin of the information, prior to any information pro- cessing. Localization can be defined as estimating the exact physical location of the sensor node in a certain geo- graphical area. Sensors can be located with the help of the Global Positioning System (GPS). Due to the high cost and poor performance of GPS indoors, it becomes inefficient to equip all sensor nodes with GPS [3]. There have been some localization algorithms that were proposed in the past, and are still being used to locate unknown sensors. The clas- sification of localization algorithms along several axes is presented in Fig. 1. In centralized algorithms, the sensor nodes send their data to the central processing unit where the data is ana- lyzed and processed to extract the positional information. The approach in which each sensor node can locate itself is & S. Mini [email protected]Ketan Sabale [email protected]1 Department of Computer Science and Engineering, National Institute of Technology Goa, Farmagudi, Goa, India 123 Wireless Netw (2019) 25:49–61 https://doi.org/10.1007/s11276-017-1538-6
13
Embed
Anchor node path planning for localization in wireless ...static.tongtianta.site/paper_pdf/8d0fe042-672f-11e... · methods in which a sensor node can locate itself by using the location
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Anchor node path planning for localization in wireless sensornetworks
Ketan Sabale1 • S. Mini1
Published online: 10 June 2017
� Springer Science+Business Media, LLC 2017
Abstract Localization is one of the most important chal-
lenges of wireless sensor networks because the location
information is typically used in other domains such as
coverage, deployment, routing, and target tracking. There
exist some localization algorithms that facilitate the sensor
nodes to locate itself using the mobile anchor node posi-
tion. Some crucial attempts have been made in the past for
optimizing the mobile anchor node trajectory with good
accuracy. This paper presents a novel path planning
scheme, D-connect, which ensures the localization of all
the sensor nodes with minimum trajectory length. The
performance of the proposed scheme is evaluated through a
series of simulations. Experimental results reveal that the
shortest path for traversing the whole area can be traced
with the minimum localization error using this method. It
also shows that D-connect outperforms the existing meth-
ods in terms of the anchor node trajectory length as well as
known as distributed localization. The main advantages of
the centralized approach is accuracy, precision and the
ability to process greater amounts of data. The disadvan-
tages of these algorithms are poor scalability and a single
point of failure. The distributed algorithms do not require a
central base station. In the distributed localization
approach, localization is done through node-to-node com-
munication. Localization which is carried out with the help
of signal properties is known as Range-based localization
[13]. The common techniques used in range-based local-
ization are Angle of Arrival (AOA), Time of Arrival
(TOA), Time Difference of Arrival (TDOA) and Received
Signal Strength Indicator (RSSI). These localization algo-
rithms are based on time and distance dependent mea-
surements. In the Angle of Arrival (AOA) method, the
location is estimated with the help of the angle at which a
signal arrives at a sensor. Distance information is obtained
in the Time of Arrival (TOA) and Time Difference of
Arrival (TDOA) methods by computing the transmission
time of the wireless signal. These approaches give better
location accuracy but use extra hardware. Received Signal
Strength Indicator (RSSI) is a measurement of the power
present in a received signal. There is no requirement of
extra hardware for estimating the distance using RSSI
technique. Estimated distance travelled by the signal up to
the receiver point is calculated with effective path loss.
Range-free techniques do not need extra hardware but
localization depends on the connectivity of the network.
Localization based on range-free techniques in which the
anchor node moves along a hexagonal pattern is discussed
in [18]. Cost-effective ways of localization with less
accuracy are provided by range-free techniques. The
methods in which a sensor node can locate itself by using
the location information of some specific nodes is known
as Anchor-based approach. The position of Anchor nodes
is predefined or can be located with the help of GPS. An
anchor node is also referred to as a Beacon or a Reference
node. Anchor-free localization does not depend on the
anchor nodes. In this approach, each node computes the
relative coordinates by measuring the distance to its
neighboring nodes by using either range-free or range-
based techniques.
Usually WSN is used in remote geographical areas
where human operations are impossible. It is infeasible to
deploy beacons at known positions. So beacon nodes must
be equipped with GPS receivers. Cost effective WSN is
dependent on the minimum number of anchor nodes used.
So the motivation behind designing the D-connect trajec-
tory is to locate all the sensor nodes with the help of a
single beacon node. The beacon node broadcasts its loca-
tion information while travelling in the region of interest.
Beacons do not broadcast constantly. Advertising Interval
describes the time between each broadcast. Stability of the
signal depends on the Advertising Interval. The signal is
more stable for shorter intervals. It is very beneficial to use
such anchor nodes for localization. The fundamental issue
is to find an optimum path for a mobile beacon trajectory in
the region of interest. Before defining any path planning
mechanism, certain properties of optimum path planning
mechanisms should be investigated. Due to the poorly
designed trajectory, some sensor nodes may not be
localized.
The existing anchor node trajectories are different from
one another in the pattern they follow. The Mobile Anchor
Centroid Localization [8] traverses the region along a spiral
path. Due to the spiral nature of the path taken by the
anchor node, sensor nodes present in the corner of the
network do not get sufficient number of beacon positions,
which leads to an increase in the localization error. [9]
presents the Hilbert curve approach that solves the local-
ization and coverage problems. In this approach the
unknown sensor estimates its position by using h keys.
Scan and Double Scan [11] minimizes the anchor node
trajectory length but increases the localization time as the
Fig. 1 Classification of
localization algorithms
50 Wireless Netw (2019) 25:49–61
123
sensor has to wait for non collinear positions for location
estimation. The Z-curve [10] follows the path in a Z pattern
while LMAT [12] follows an equilateral triangle pattern.
All these trajectories differ in terms of construction and
work, but all try to achieve the same goal. To the best of
our knowledge, no literature provides a sufficient and
optimal trajectory to solve the problems of localization and
coverage.
In this paper, we propose a path planning scheme,
D-connect, for anchor-based localization using a range-
based technique. It guarantees the localization of all
unknown sensor nodes from a certain geographical region
with minimized localization error. It uses two signals with
two different transmitted signal powers which are required
to increase the accuracy while taking care of the
collinearity issue. For maximum accuracy, the anchor node
has to travel near the boundary of the region. In D-connect,
the increased power of the signal resolves this problem. For
locating any unknown sensor accurately, at least three non-
collinear beacon signals are required. If the sensor node
receives more than three beacon node positions, at least
one non-collinear position is required to eliminate any
collinear beacon.
The rest of the paper is organized as follows: Sect. 2
summarizes the related work on different existing local-
ization algorithms and path planning mechanisms with
more clarity. Section 3 defines the problem and describes
the D-connect method. The experimental results are
reported and discussed in Sect. 4. Sect. 5 concludes the
paper.
2 Related work
There have been several research efforts on tackling
problems related to localization in WSN. The various
hierarchical architectures of WSN are presented in [2].
Most existing localization schemes for WSNs are mainly
classified into two groups, computation based and range
based. A detailed classification is provided in [3]. Dis-
tributed computation based methods are discussed in [4, 5]
and [6]. The Monte-Carlo Localization algorithm is pre-
sented in [4]. The Monte-Carlo algorithm estimates the
position of an unknown sensor by considering the near and
the farther anchor node constraints. At first, the sensor node
constructs a possible location set which denotes the pos-
sible location of the sensor. In the filtering phase the
locations which are not satisfying the anchor node con-
straints are removed and the average of the remaining
location set gives the final estimated location of the
unknown node. The efforts for increasing the efficiency of
the Monte-Carlo algorithm are done in [5]. Drawing sam-
ples is a time consuming process, so Monte-Carlo
Localization Boxed algorithm constrains the area from
which the sensor draws samples. This method is known as
Monte-Carlo Localization Boxed (MCB) algorithm. The
increased accuracy and reduced localization time can be
obtained by using relay nodes. Self Localization
Scheme using relay nodes and anchor nodes is presented in
[6]. Relay nodes are also sensor nodes which get their
positional information from the anchor node. Sensor nodes
calculate their position by the received information about
the relay node position. The various conditions for relay
node selection are discussed in [6].
The efficiency of anchor based localization algorithms is
dependent on the trajectory of the anchor node. For
defining any new beacon node trajectory certain conditions
must be satisfied by the trajectory. First of all, trajectory
should pass closely to the unknown sensor for best position
estimation. Also each sensor node should have at least
three non-collinear anchor node positions to locate itself.
[7] illustrates the conditions that are to be satisfied by the
anchor node trajectory. The various schemes to reduce the
trajectory length of the anchor node are discussed in [8–11]
and [12]. [8] presents the trajectory in a spiral form. The
position estimation of sensor nodes is done with the help of
the range-free localization technique called Centroid
algorithm. The position of each sensor is calculated by
taking the average of the total received beacon messages in
the time interval t. The length of the Spiral trajectory is
more than all other trajectories. The localization algorithm
that uses the Spiral trajectory is known as the Mobile
Anchor Centroid Localization (MACL). The trajectory
based on the Hilbert space filling curve is presented in [9].
The Hilbert space filling curve is a one-dimensional curve,
which visits every point exactly once without crossing
itself within a two or three-dimensional space. The Hilbert
curve is generated recursively. A superior path planning
mechanism called Z-curve is explained in [10]. Z-curve
handles the collinearity issue occurring in the anchor node
trajectory by using the determinant of the matrix that
contains consecutive beacon positions received by the
sensor node. The received beacon positions are said to be
non-collinear if the determinant of the matrix is non-zero.
The Z-curve trajectory is tested for an obstacle presence
scenario. Scan and Double Scan methods are explained in
[11]. The Scan method has the disadvantage of collinearity.
In the Scan method, the mobile beacon node travels along
one dimension and when it reaches the end of the network
it travels along the second dimension where the length of
the path along the second dimension is equal to the reso-
lution. The procedure is repeated till the entire network is
traversed. The Double Scan method traverses the network
along both directions. The collinearity problem of the Scan
method is resolved by the Double Scan strategy up to some
extent. But the length of the Double Scan trajectory is
Wireless Netw (2019) 25:49–61 51
123
double, compared to the Scan mehod for the same reso-
lution. The Hilbert curve method overcomes the disad-
vantages of the Scan and Double Scan methods. Since the
Hilbert curve trajectory takes more turns, it gives better
position estimation compared to the Scan and Double Scan
trajectories. As Hilbert curve connects the centers of two
successive cells in the network, it will never move along
the border of the deployment area. This is a drawback of
the Hilbert curve method. Localization with a Mobile
Anchor node based on Trilateration (LMAT) is presented
in [12]. In LMAT trajectory, an anchor node moves along
the boundaries of the region based on an equilateral tri-
angle pattern.
After designing the optimum trajectory for an anchor
node, the next main task is to estimate the physical position
of the unknown sensor nodes using anchor node positions.
For estimating the position of an unknown node various
range-free and range-based methods can be used. The
range-free and range-based techniques are discussed in
[13]. The cost effective method in range-based localization
algorithm called RSSI is presented in detail in [14] and
[15]. Distance estimation using RSSI is dependent on path
loss. Various propagation models for mobile communica-
tion are discussed in [16]. Path loss and fading are the main
characteristics of the radio channel. The RSSI calculations
are basically influenced by path loss and fading. Free space
model, Two ray ground model and Log-normal shadowing
model are the RSSI propagation models used in wireless
sensor networks. When the transmitter and receiver have a
clear unobstructed line of sight between them, the free
space propagation model is used [16]. The Two ray ground
model is considered only when there exists a single direct
path between the transmitter and the receiver for the
propagation of the radio signal. The directed path and a
ground reflected propagation between the transmitter and
the receiver is considered in two ray propagation model.
The Log-normal shadowing model is the most suitable ra-
dio propagation model as it provides a number of param-
eters for configuration for different environments (indoor
and outdoor). This study mainly focuses on the develop-
ment of optimal anchor node trajectory for localization of
unknown sensors using the Log normal shadowing model.
3 Proposed work
3.1 Problem statement
Given a geographic region R, and a single anchor node A to
locate m sensor nodes S ¼ fS1; S2; S3; :::; Smg, the objectiveis to identify the minimum length trajectory for anchor
node A, such that all unknown sensor nodes are located
with minimum localization error.
If the mobile anchor node A, is at position ðx1; y1Þ andthe sensor node Si, ð1� i�mÞ is at location ðx2; y2Þ then A
can locate sensor Si iff sensor node Si lies within the
communication range r of A. That is,ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðx1 � x2Þ2 þ ðy1 � y2Þ2q
� r ð1Þ
where r is the communication range of anchor node A.
Let (X, Y) and ðxi; yiÞ represent the estimated and orig-
inal coordinates of sensor Si, respectively. Then the