You are on page 1of 3

International Journal of Advanced Research in Engineering Applications,

Volume 1, Issue 1, 6-8, 2014

Localization Algorithms under Correlated


Shadowing in Wireless Sensor Networks
Sameera V Mohd Sagheer
Department of Electronics and
Communication Engineering
KMCT College of Engineering for Women
Calicut, Kerala, India

Abstract- The existing models for radio channel shadowing was considered to be independent. However the
shadowing on real world links are dependent. In this paper we review different wireless sensor network localization
algorithms which considers the the shadowing between different links to be correlated. For this we have used the
NeSH model.The shadowing between the links is modelled as log normal shadowing path loss model. The
algorithms in the different works are analysed and their results are discussed.

Keywords Wireless networks, adhoc networks, networking, localization, correlated shadowing


1 INTRODUCTION

2 PROBLEM FORMULATION

Wireless sensor network has emerged as a major research


area.Determining the location of sthe sensor nodes in the
network has become an area of crucial research. Many
solutions have been put forth for this problem. An obvious
solution to this is to place GPS satellite receivers on the
sensors. However this method is expensive and hence
infeasible. A more realistic method is to assume that a small
have knowledge of their positions. These nodes are known
as anchor nodes. Then using the information that is
communicated between these known sensor nodes the
position of these unknown sensors are estimated. Currently
time of arrival (TOA) and received signal strength(RSS) are
used to determine the pairwise distances. Wireless sensor
localization problem deals with the determination of the
unknown position of sensor nodes using the position of
anchor nodes and the distance between the sensor nodes
(determined by TOA or RSS). The similar procedure has
been applied in molecular conformation and distance
geometry. Many localization algorithms have been
proposed. They vary in their approaches.

Consider a wireless sensor network in


with
nodes. Here the anchors whose positions are known
are m in number and is given by
for
. The sensor whose positions are known are in
number and is denoted by
for
. The
Euclidean distance between any two sensor nodes and
is represented by
. Let the distance between any sensor
and anchor node
which is unknown be represented by
. Let be the noisy estimate of
which is known
where
are the coordinates of sensor pairs. Let the set of
all anchor sensor pairs be denoted by
for which is
known. The sensor localization problem can be stated as
under:
Estimate the position of unknown sensor nodes
for
with overall localization error
minimized provided that we are given the position of the
anchor nodes
for
. The distance
estimate between the sensor nodes and between the
anchor and sensor for
.

www.ijarea.org

International Journal of Advanced Research in Engineering Applications,


Volume 1, Issue 1, 6-8, 2014

3 SHADOWING MODEL

The NeSH model formulates the shadowing on a


link
as a normalized integral of the
over the
link endpoints
.

3.1 Log normal shadowing


A number of algorithms have been proposed and many of
them used the log normal shadowing path loss model [6].
In this model, the distance estimates are given as the actual
distance corrupted by additive Gaussian noise with zero
mean and variance proportional to the distance. Consider a
multi link sensor network. The links present in the network
can be of three types: anchor to anchor, anchor to sensor or
sensor to sensor. We assume the link to be symmetric, i.e.
received power
from node 1 to 2 is same as the received
power
from node 2 to 1. Since all the links are
symmetric, we can assign a single received power (dBm)
for link . The log normal shadowing path loss model states
that the power at the receiver is inversely proportional to
the transmitter receiver separation distance raised to some
exponent (known as path loss exponent) and attenuated by
a log normally distributed random variable. That is, for link
, the received power in linear units Watts is given by

(equations 4)
where
is the transmit power,

Single link properties: The NeSH model agrees


with two important empirically observed link shadowing
properties.
Property 1: The variance of link shadowing is
approximately constant with path length.
Property 2: Shadowing losses are Gaussian.
Joint link property.
Given two links , belong to
are functions of
the same shadowing field
. Thus NeSH model
introduces correlation between them. The covariance
between
and
is given by

The received power between (1)two nodes


represented by
consists of losses due to both shadow
fading
and non shadow fading
. NeSH model
assumes that
is independent and is modeled as Gaussian
(2)
with zero mean and variance .

( )

is the path

loss at close in reference distance , is the unitless path


loss exponent, is the length of link and
is the
normally distributed random shadowing with zero mean
and as variance.

In short we can say that the total fading on a link


connecting two nodes is given as
where
is
shadow fading loss and
is non-shadow fading loss. It is
assumed that
and
are independent and Gaussian.
Hence
is also Gaussian with variance

3.2 NeSh Link Correlation Model

4 ALGORITHMS
It is assumed that the shadowing on the links are
correlated. Gudmundson [2] modeled shadow fading
correlation for a mobile link with common base station. But
this model does not provide correlation for links that [4]
does not have common end points. Hence this model
cannot be used in adhoc sensor networks. The NeSH model
considers shadowing losses to be a function of that
environment. Hence the losses in geographically proximate
links are correlated. The shadowing is caused by an
environment in an underlying spatial loss field
. The
NeSH model assumes that
is an isotropic wide sense
stationary Gaussian random field with zero mean and
exponentially decaying spatial correlation.

www.ijarea.org

Dhalaan [1] approaches this as a non convex optimization


problem. The localization problem can be written as convex
SDP.

where
be the vector of unknown
distance square
.
Vaghefi et al [10] defines a Cramer Rao Lower
Bound which is a lower bound on the performance of any
unbiased estimator. They formulated the maximum
likelihood estimator as a Semi Definite Program.
Yang et al [12] used correlation among
shadowing
(3)
losses to detect the location of the anchor nodes. An ML

International Journal of Advanced Research in Engineering Applications,


Volume 1, Issue 1, 6-8, 2014
Estimator and Kalman filter is combined to track mobile
location and shadowing losses. The ML location estimator
determines the target location. The Kalman filter
determines the shadowing losses.
In [7]the Stochastic Proximity Algorithm was used
for improved performance. The same correlated shadowing
model discussed earlier was used.

6 CONCLUSION
This paper reviewed localization algorithms that consider
correlated shadowing for localization in wireless sensor
networks. Most papers available in literature consider the
errors in ranges used for localization as independent. This
results in poor performance when the algorithms are
deployed in real conditions. The performance improves
when the correlation is considered.

5 SIMULATIONS, RESULT AND DISCUSSION


Dhalaan et al [1] placed the nodes in a 5
5 area. The
covariance matrix was formulated using the NeSH model.
Two nodes were assumed to be connected. The received
power between them was greater than the threshold power.
Two situations were consideredd. First in which the
covariance matrix was known, second in which the
covariance matrix was unknown. The root mean squared
distance was used as the measurement matrix.
Vaghefi et al [10] in his work fix the position of
anchor nodes and place the sensor nodes in an area of
. Uncorrelated and correlated shadowing were
considered. The cumulative distribution function of the
localization error was computed. The comparison of the
proposed algorithm was dodne with other existing
algorithms. The Cramer Rao Lower Bound was also
calculated. For uncorrelated case,
and
and
for correlated case,
and
. In both cases, the
proposed SDP was found to be close to that of the
Maximum Likelihood Estimator. It also performs better
than other SDP algorithms [5, 8, 3, 9, 11].
The average running time of estimators were also
compared in the same network having
and
. It is seen that among all the SDP algorithms compared
in the paper, the proposed SDP algorithm has highest
accuracy and lowest running time.
Sameera et al [7] solve the localization problem
using the Stochastic Proximity Embedding. Different
factors were considered, which included the standard
deviation of the error model, the number of anchor nodes
and various SPE parameters such as number of cycles,
steps, learning parameter etc. The proposed algorithm was
then compared with [1] and the result was found to be
superior. Here the performance metric used was
localization error.
Yang et al [12] considers a moving target and
tracking is done on it. Although the tracking algorithm
often does not give the exact path, it gives the correct
direction.

www.ijarea.org

REFERENCES
[1] Abdullah H Al-Dhalaan and Ioannis Lambadaris. Wireless
sensor network localization with spatially correlated shadowing. In
Communications (ICC), 2010 IEEE International Conference on,
pages 16. IEEE, 2010.
[2] Mikael Gudmundson. Correlation model for shadow fading in
mobile radio systems. Electronics letters, 27(23):21452146, 1991.
[3] Robin Wentao Ouyang, AK-S Wong, Chin-Tau Lea, and Victoria
Ying Zhang. Received signal strength-based wireless localization via
semidefinite programming. In Global Telecommunications
Conference, 2009. GLOBECOM 2009. IEEE, pages 16. IEEE, 2009.
[4] Neal Patwari and Piyush Agrawal. Nesh: A joint shadowing
model for links in a multi-hop network. In Acoustics, Speech and
Signal Processing, 2008. ICASSP 2008. IEEE International
Conference on, pages 28732876. IEEE, 2008.
[5] Neal Patwari, Alfred O Hero, Matt Perkins, Neiyer S Correal,
and Robert J Odea. Relative location estimation in wireless sensor
networks. Signal Processing, IEEE Transactions on, 51(8):2137
2148, 2003.
[6] Theodore S Rappaport et al. Wireless communications:
principles and practice, volume 2. prentice hall PTR New Jersey,
1996.
[7] V M Sameera, R Nandakumar, and Dr Suresh K. Localization
under correlated shadowing using modified stochastic proximity
embedding. In National Conference on Technological Trends, CET
Trivandrum.
[8] Hing Cheung So and Lanxin Lin. Linear least squares approach
for accurate received signal strength based source localization.
Signal Processing, IEEE Transactions on, 59(8):40354040, 2011.
[9] Reza Monir Vaghefi and R Michael Buehrer. Cooperative sensor
localization with nlos mitigation using semidefinite programming. In
Positioning Navigation and Communication (WPNC), 2012 9th
Workshop on, pages 1318. IEEE, 2012.
[10] Reza Monir Vaghefi and R Michael Buehrer. Received signal
strength-based sensor localization in spatially correlated shadowing.
In Acoustics, Speech and Signal Processing (ICASSP), 2013 IEEE
International Conference on, pages 40764080. IEEE, 2013.
[11] Gang Wang and Kehu Yang. A new approach to sensor node
localization using rss measurements in wireless sensor networks.
Wireless Communications, IEEE Transactions on, 10(5):13891395,
2011.
[12] Kai-Jie Yang and Yuh-Ren Tsai. Location tracking in mobile
networks under correlated shadowing effects. In Wireless
Communications and Networking Conference, 2009. WCNC 2009.
IEEE, pages 15. IEEE, 2009.

You might also like