A LSSVR Three-dimensional WSN Nodes Location
Algorithm Based on RSSI
Xuewen He
School of Mechanical and Electrical Engineering
Jiangxi University of Science and Technology
Ganzhou, 341000, China
e-mail: hxw993@vip.163.com
Yanmeng Wang
School of Mechanical and Electrical Engineering
Jiangxi University of Science and Technology
Ganzhou, 341000, China
e-mail:296134188@qq.com
Yong Xiao
School of Mechanical and Electrical Engineering
Jiangxi University of Science and Technology
Ganzhou, 341000, China
e-mail:xiaoyongcn@foxmail.com
Abstract—Nodes location is a key technology in the research of
Wireless Senor Network (WSN), in which the location method
that is based on RSSI ranging and Least-Square is the current hot
spot in research and application. Currently, with the tendency of
more and more applications of WSN in the three-dimensional
environment, the LSSVR three-dimensional WSN nodes location
algorithm based on RSSI is proposed, aiming at the problem of
the low accuracy of nodes location estimated by least square
method in applying RSSI which is easily influenced by many
factors and thus cause large ranging errors and difference. The
paper will also present the obvious advantages of the LSSVR
three-dimensional nodes location algorithm compared with Least-
square location algorithm after the simulation and regression
analysis of errors. The location errors acquired by the two kinds
of location algorithms under the circumstance of small ranging
error are similar. When the ranging error is over 16%, LSSVR
three-dimensional nodes location algorithm can greatly improve
positioning accuracy with an average reduction of 25% of
positioning error, sometimes even to an utmost of 50%, compared
with the traditional LS positioning algorithm.
Keywords: Wireless Senor Network; Nodes Location; Received
Signal Strength Indication (RSSI); Least-Square Support Vector
Regression (LSSVR); Regression Analysis
I. INTRODUCTION
In recent years, Wireless Sensor Network (WSN) is one
kind of network Wireless information perceptual technology
which develops more rapidly. In Wireless Sensor Network
applications, regarding many important application situations.
Such as target tracking, environment monitoring. The
positional information is very important to the sensor network.
The location of the event and the position of the gain
information node are important information include in the
wireless sensor node examination data. It is no significance
when there is no position detection information. So the wireless
sensor node must grasp their own position to be able to be clear
about the sources of it collect data. Obviously the wireless
sensor node location technology for data acquisition is the
premise of the location information. On the other hand, the
node localization information to entire network's routing
algorithm efficiency [1], the quality of network coverage [2]
and the network topology self-configuration [3] have the
influential role. Therefore, the localization problem of wireless
sensor network is one of the important supporting technology,
is also the present research hot spot and the difficulty [4]. To
realize accurate, highly efficient and reliable node localization
on target location, moving target tracking and improving the
efficiency of the routing has certain academic significance and
application value.
The current research and the application have had many
methods to realize the wireless sensor network node location.
However, these methods are mostly two-dimensional
environment for the application of two-dimensional
environment or solve the problem of nodes location. And there
are some deficiencies in some areas such as the network scale,
energy consumption, precision, environmental adaptability and
fault tolerance, and many other aspects. Therefore, the research
of new Wireless Sensor Network
Three-dimensional nodes location algorithm has a great
significance. Over all, Node localization method based on the
need for distance, the presence of anchor nodes, whether the
mobile node, centralized or distributed computing four
different aspects of node location to classify. Usually the node
localization method will carry on the division according to the
first kind of classified principle, divides into based on Range-
Based [5-7] and Range-Free [8-9] localization method. Based
on Range-Based localization method contains: Trilateral and
multilateral positioning method, accept signal Angle
positioning method. The former include: RSSI,
TOA/TDOA/RTOF, PDOA, NFER and so on. The latter refers
to the signal angle of arrival method (AOA). Based on Range-
Free localization method contains: Centroid positioning
method, DV-Hop algorithm, amorphous positioning algorithm,
APIT algorithm, etc. Generally, positioning methods based on
1889
978-1-4244-8165-1/11/$26.00 ©2011 IEEE