An Effective Method for GPS GDOP Clustering

Using Ant Colony Optimization Algorithm

M. R. Mosavi
Department of Electrical Engineering
Iran University of Science and Technology, Narmak, Tehran 16846-13114, Iran

Abstract

Satellite geometry, which represents the geometry locations of the GPS satellites as seen by the receiver(s), plays a very important
role in the total positioning accuracy. The stronger the satellite geometry strength is, the higher the positioning accuracy obtained
An Effective Method for GPS GDOP Clustering
will be. As such, the overall positioning accuracy of GPS is measured by the combined effect of the unmodeled measurement
Colony
Optimization
Algorithm
errors and theUsing
effect ofAnt
the satellite
geometry.
This paper presents
satellites geometry clustering for good navigation satellites
ustering subset selection. This approach is based on clustering of the satellites Geometry Dilution of Precision (GDOP) factor using Ant
M. R. Mosavi
ithm
Colony Optimization
(ACO) algorithm which is used newly in solving data-clustering problems and developed from imitating
Associate Professor, Department of Electrical Engineering
the technique
of real ants
findingand
theTechnology,
shortest way
from their
nests
and food Iran
source. The proposed method use pheromone to
Iran University
of Science
Narmak,
Tehran
16846-13114,
evaluate individual colonys iterative
Without matrix inversion required, the ACO-based approach is capable of evaluating
Email:result.
[email protected]
ing
all subsets of satellites and hence reduces the computational burden. Simulation results show this method is more efficient to
6-13114, Iran
Abstract:upon
Satellite
whichinrepresents
the geometry
locations of the GPS satellites as
converge
the geometry,
optimal value
the GPS GDOP
clustering.
seen by the receiver(s), plays a very important role in the total positioning accuracy. The stronger

the satellite
geometry strength is, the higher the positioning accuracy obtained will be. As such,
Key
the GPS satellites
aswords: GPS GDOP, Clustering, and Ant colony optimization

the overall positioning accuracy of GPS is measured by the combined effect of the unmodeled
accuracy. The stronger
measurement errors and the effect of the satellite geometry. This paper presents satellites
ined will be. As such,
geometry clustering for good navigation satellites subset selection. This approach is based on
fect of the unmodeled
clustering of the satellites Geometry Dilution of Precision (GDOP) factor using Ant Colony
per presents satellites
Optimization (ACO) algorithm which is used newly in solving data-clustering problems and
approach is based on
developed from imitating the technique of real ants finding the shortest way from their nests and
tor using Ant Colony
food source. The proposed method use pheromone to evaluate individual colonys iterative
ustering problems
and
dimensional
position
1.
Introduction
result.
Without matrix inversion required, the ACO-based approach is capable
of evaluating
all and velocity information to users
y from their nests and
anywhere
in
the
world
subsets of satellites and hence reduces the computational burden. Simulation results show this [1].
ual colonys iterative
The
Global
Positioning
System
(GPS)
a global
method
is more
efficient to
converge
uponisthe
optimalnavigation
value in the GPS GDOP clustering.
pable of evaluating all
The Geometric Dilution of Precision (GDOP) is a
system
set
up
by
the
U.S.
Department
of
Defence.
It
consists
tion results show this
Keywords:
GPS
GDOP, Clustering,
andearth
Ant colony
optimization
geometrically determined factor that describes the effect of
of
24
at
least
satellites
orbiting
the
in
6
planes,
five
DOP clustering.

geometry on the relationship between measurement error

control stations and user receiver equipment. These satellites
1. Introduction
and position determination error. It is used to provide an
transmit
signals
which
are
received
by
the
GPS
receiver.
By
The Global Positioning System (GPS) is a global navigation system set up by the U.S.
indication of the quality of the solution. Some of the GPS
measuring
the
time
of
arrival
of
the
signal,
the
users
distance
Department of Defence. It consists of 24 at least satellites orbiting the earth in 6 planes, five
receivers
may
be able to process all visible satellites due
(range)
each
thereceiver
satelliteequipment.
can be calculated.
Usingtransmit
the
control from
stations
andof
user
These satellites
signals
which
are not
received
set up by the U.S.
tousers
limited
number
of channels. Consequently, it is sometimes
by thetoGPS
receiver. By
measuring
thethe
time
of position
arrival of can
the signal,
the
distance
(range)
range
a
minimum
of
4
satellites,
user
be
earth in 6 planes, five
from each of
satellite
can be calculated.
Using the range
to aof
minimum
of 4 satellites,
the user
necessary
to select
the satellite subset that offers the optimal
calculated
in the
three
dimensions
using the following
system
als which are received
position
can
be
calculated
in
three
dimensions
using
the
following
system
of
non-linear
or acceptable solutions. The optimal satellite subset is
non-linear equations:
users distance (range)
equations:
sometimes obtained by minimizing the GDOP factor. The
of 4 satellites, the user
(1)
system of non-linear
most straightforward(1)approach for obtaining GDOP is to use
pi ( X X ) 2 (Y Y ) 2 ( Z Z ) 2 b ct
i
i
i
i
matrix inversion to all combinations and select the minimum
measurement
to i th
( X , Y , Z ) is user position (unknown),
Where p
pi isispseudorange
pseudorange
measurement
to satellite,
i-th satellite,
Where
one. However, the matrix inversion by computer presents a
i
(1)
Zi )user
is position
the i th satellite,
b )isisthe
clock bias
of receiver;burden
t j isto the navigation computer, especially
( X i , Yi , is
(X,Y,Z)
positionof(unknown),
( Xi ,Yi ,Z
position
of parameter
computational
i
er position (unknown),
the
i-th
satellite,
b
is
the
clock
bias
parameter
of
receiver;
number
of satellites is large [2].
able tothe
provide
accurate
clock bias of satellite, and c is speed of light. Using these equations it iswhen
of satellite,
and c isinformation
speed of to
light.
and velocity
usersUsing
anywhere in the world [1].
t j isis clock bias position
ter of receiver; . three-dimensional
The Geometric
Dilution
Precision
a geometrically
that describes
The factor
traditional
supervised clustering techniques assume
these
equations
it isofable
to (GDOP)
provide isaccurate
three-determined
ble to provide accurate

the effect of geometry on the relationship between measurement error and position determination
he world [1].
error.
It is used
provide
an indication of the quality of the solution. Some of the GPS receivers
2009 AARS,
Alltorights
reserved.
d factor that describes
may
not be able author:
to process
all visible satellites due to limited number of channels. Consequently,
*
Corresponding
[email protected]
position determination
it is sometimes necessary to select the satellite subset that offers the optimal or acceptable
e of the GPS receivers
solutions. The optimal satellite subset is sometimes obtained by minimizing the GDOP factor.
annels. Consequently,
optimal or acceptable
ing the GDOP factor.

An Effective Method for GPS GDOP Clustering Using Ant Colony Optimization Algorithm

prior knowledge to be available for all the thematic classes

that are present in the considered dataset. On the contrary,
unsupervised clustering does not require predefined
knowledge about the data. The main task of clustering with
unsupervised learning is to partition a given data set into
groups (clusters) such that the data points in a cluster are
more similar to each other than points in different clusters
[3]. We implement an unsupervised clustering method using
Ant Colony Optimization (ACO) algorithm for GPS GDOP.
ACO algorithm is the simulation of cooperation course of
real ant group. Each ant searches the solution independently
in the candidate solution space, meanwhile leaves some
information on the found solution. Ants leave more
information on the solution with better performance, which
has more possibility to be selected. All the solutions have the
same information on it in the elementary period of the
algorithm. With the progress of the computing, the better
solution gets more and more information. Therefore, the
algorithm reaches the optimization or approximately
optimization solution [4]. Compared to other optimization
algorithms such as Genetic Algorithm (GA), ACO has
searching capability of both local and global optimal through
adjusting parameters. It gives list of optimal points. ACO can
escape from local optimal and obtains global optimal. Whats
more, ACOs computing process is faster because it does not
have complex operators as aberrance and intercross [5].
The aim of this paper is solving GPS GDOP clustering
problem using a new approach named ACO. This paper is
organized as follows. In section II and section III, ACO and
its algorithm are reviewed, respectively. Section IV presents
ACO algorithm developed for GPS GDOP clustering. The
algorithm results using ACO are reported in section V.
Finally, conclusions are given in section VI.

2. Ant Colony Optimization

Ants can find the shortest path to food by laying a pheromone
(chemical) trail as they walk. Other ants follow the
pheromone trail to food. Ants that happen to pick the shorter
path will create a strong trail of pheromone faster than the
ones choosing a longer path. Since stronger pheromone
attracts ants better, more and more ants choose the shorter
path until eventually all ants have found the shortest path.
Consider the case of three possible paths to the food source
with one shorter than the others. Ants choose each path with
equal probability. Ants that went and returned on the shortest
path will cause it to have the most pheromone soonest.
Consequently new ants will select that path first and further
reinforce the pheromone level on that path. Eventually all
the ants will follow the shortest path to the food. The first
ACO algorithms were designed to solve the traveling
salesperson problem, because this problem closely resembles
finding the shortest path to a food source. Initial attempts at
an ACO algorithm were not very satisfying until the ACO

algorithm was coupled with a local optimizer. One problem

is premature convergence to a less than optimal solution
because too much virtual pheromone was laid quickly. To
avoid this stagnation, pheromone evaporation is implemented.
In other words, the pheromone associated with a solution
disappears after a period of time [6].

3. ACO Algorithm
The aim of GPS GDOP clustering is to obtain optimal
assignment of N satellites in one of the K clusters where N
is the number of satellites and K is the number of clusters.
Artificial ants used in algorithm are named as software ants
or agent and number of agents expressed with R. Ants start
with empty solution strings and in the first iteration the
elements of the pheromone matrix are initialized to the same
values. With the progress of iterations, the pheromone matrix
is updated depending upon the quality of solutions produced
[7].
Each agent used in algorithm is defined with solution strings
expressed with S of length N and at start of the algorithm
solution string of each element is empty. Each element of
string corresponds to one of the test samples and its value
includes which cluster it will be assigned.
To construct a solution, the agent uses the pheromone trail
information to allocate each element of string S to an
appropriate cluster label. At the start of the algorithm, the
pheromone matrix is initialized to some small value 0.
Hence, at first iteration each element of solution string S of
each agent is assigned randomly to one of the K clusters.
The trail value, ij at location (i,j) represents the pheromone
concentration of sample i associated to the cluster j. So, for
the problem of separating N samples into K clusters the size
pheromone matrix is NxK. Thus, each sample is associated
with K pheromone concentrations. The pheromone trail
matrix evolves as we iterate. At any iteration level, each
agent or software ants will develop solutions showing the
probability of each ant belonging to which cluster using this
pheromone matrix.
To generate a solution S , the agent selects cluster number for
each element of string S by the following way [7]:
Using probability q0 , cluster having the maximum
pheromone concentration is chosen ( q0 being a priori defined
number, 0<q0<1 , for example q0=0.98 ):
(i) First, to generate random numbers in the range [0,1] are r1
, r2 , ..., rN .
(ii) Second, if the random numbers are higher than the
threshold q0=0.98 :

following way [7]:

Using probability q0 , cluster having the maximum pheromone concentration is chosen ( q0
being a priori defined number, 0 q0 1 , for example
q0 0.of
98 Geoinformatics,
):
Asian Journal
Vol.10,No.4 (2010)
(i) First, to generate random numbers in the range [0,1] are r1, r2 ,, rN .
(ii) Second, if the random numbers are higher than the threshold q0 0.98 :

steps are repeated for certain number of iterations. The

iteration loop is carried out until no further improvement in
equation (2) is being made.

(1) If 0 ri i1 , then S (i) Cluster 1 .

(2) If i1 ri i1 i 2 , then S (i) Cluster 2 .

K 1
(K) If ij ri 1 , then S (i) Cluster K .
j 1
(iii) Third, if the random numbers are less than q0 0.98 , choose the

Most of existing ACO algorithms uses some local search

procedures to develop the generated solutions discovered by
the software ants. Local search helps to generate better
maximum number of
solutions, if the heuristic information can not be discovered
(iii)
if the random numbers are less than q0=0.98 ,
eachThird,
test sample.
easily. Local search is applied on a few percent R generated
Each
agentthe
selects
a clusternumber
number with
a probability
value for each element of S string to form
choose
maximum
of each
test sample.
solutions.
in terms of
its own solution string S . The quality of constructed solution string S is measured

theEach
valueagent
of objective
a givenwith
data-clustering
problem.
selects function
a clusterfornumber
a probability
valueThis objective function is
local
is implemented on top L (L<R )
defined
as
the
sum
of
squared
Euclidian
distances
between
each object Aand
the search
center procedure
of
for each element of S string to form its own solution string
belonging cluster.
solutions with highest fitness values (lowest values of
S. The quality of constructed solution string S is measured in
We altered the cluster number of each
partitionedfunction).
into K
A given dataset of N objects {x1, x2 ,, x N } in R n dimensional space to beobjective

terms of the value of objective function for a given data-

sample in the solution string with certain threshold

clusters.
The mathematical
formulation
of function
the data-clustering
problem
clustering
problem. This
objective
is defined
as thecan be described as [7]:
sum of squared Euclidian distances between each object and
the center of belonging cluster.

A given dataset of N objects {x1 , x2 , ..., xN } in Rn dimensional

space to be partitioned into K clusters. The mathematical
formulation of the data-clustering problem can be described
as [7]:

2
K N n
Min F (W , M ) = wij xiv m jv
j =1i =1 v=1

(2)

Where xiv is a value of v-th attribute of i-th sample, M is a

cluster center matrix of size K x n, mjv is an average of the
v-th attribute values of all samples in the cluster j, W is a
weight matrix of size N x K, wij is an associated weight of
object xi with cluster j which can be assigned as:
1 ; if object i is contained in cluster j
wij =
0 ; otherwise

(3)

Where:
K
wij =1 ; i = 1, , N and j = 1, , K
j =1

(4)

With local search probability threshold pls in [0,1], a neighbor

of Sk , k=1,2, ..., L is generated as:
(i) k=1 .
(ii) Let St be a temporary solution and assign St(i) = Sk (i)
(for i=1,...,N).
(iii) For each element i of St , draw a random number r in
[0,1]. If r pls, an integer j in the rang [1,K], such that
Sk (i) j is randomly selected and let St(i) = j .
(iv) Calculate cluster centers and weights associated with
solution string St and find its objective function value as
Ft . If Ft is less than Fk , then Sk= St and Fk= Ft .
(v) k=k+1 . If k L go to step (ii), else stop.

The center of each cluster, mj can be obtained as:

N
wij xiv
m jv = i =1
; j = 1,, K and v = 1,, n
N
w

i =1 ij

probability, pls a priori defined number in the range [0,1].

First, to generate random numbers in the range [0,1] are r1 ,
r2 , ..., rN . Second, if the random numbers are less than the
threshold probability pls , these elements have to be assigned
to different number. In the local search procedure, the
objective function values of top L agents are computed
again. These solutions can be accepted only if there is an
improvement on the fitness, namely, if the newly computed
objective function value is lower than the first computed
value, newly generated solution replaces the old one. The
local search algorithm can be written as follows [7]:

(5)

Then, the elements of the population, namely agents are

sorted increasingly by the objective function values. Because,
the lower objective function value, the higher fitness to the
real solution, namely, lower objective function values are
more approximated to real solution values. After generating
the solutions of R agents, a local search is performed to
further improve fitness of these solutions. The pheromone
matrix is then updated depending on the quality of solutions
produced by the agents. Then, the agents build improved
solutions depending on the pheromone matrix and the above

At the end of any iteration level each agent generates the

solution using the information derived from updated
pheromone matrix. After the local search procedure, the
pheromone trail matrix is updated. Such a pheromone
updating process reflects the usefulness of dynamic
information provided by software ants. The pheromone
matrix used in ACO algorithm is a kind of adaptive memory
that contains information provided by the previously found
superior solutions and is updated at the end of the iteration.
The pheromone updating process used in this algorithm
includes best L solutions discovered by R agents at iteration
level t. This L agent mimics the real ants pheromone
deposition by assigning the values of solutions. The trail
information is updated using the following rule as [7]:

An Effective Method for GPS GDOP Clustering Using Ant Colony Optimization Algorithm

The flowchart of ACO algorithm developed for solving dataclustering problem and explained in detail above is shown in
Local Best
Global Best
(t + 1) = (1 ) (t ) + 0.3 ij
+ 0.6 ij
; i(6)
= 1, 2, , N and j = 1, 2, , K
ij
ij
Figure 1.
Global Best
+ 0.6 ij
; i = 1, 2, , N and j = 1, 2, , K

Where is a persistence or trail and lies between [0,1] and

(1-) is the evaporation rate. Higher value of suggests that
the information gathered in the past iterations is forgotten
faster.
An optimal solution is that solution which minimizes the
objective function value. If the value of best solution in
memory is updated with the best solution value of the current
iteration if it has a lower objective function value than that of
the best solution in memory, otherwise the best solution in
memory kept. This process explains that an iteration of the
algorithm is finished. Algorithm iterates these steps
repeatedly until a certain number of iterations and solution
having lowest function value represents the optimal
partitioning of objects of a given dataset into several groups.
Summary of the ACO algorithm for clustering problem is as
follows [8]:
(1) Generation of new R solutions by software ants using the
modified pheromone trail information available from
previous iteration.
(2) Performing local search operation on the newly generated
solutions.
(3) Updating pheromone trail matrix.

Figure1. The flow chart of ACO algorithm developed for

solving data-clustering problem

4. GPS GDOP Clustering using ACO

Algorithm
Good satellite geometry is obtained when the satellites are
spread out in the sky. In general, the more spread out the
satellites are the sky, the better the satellite geometry will be,
and vice versa. The satellite geometry effect can be measured
by a single dimensionless number called the Dilution of
Precision (DOP). The lower the value of the DOP is, the
better the geometric strength will be, and vice versa. The
DOP number is computed based on the relative receiversatellite geometry at any instance (i.e., it requires the
availability of both the receiver and the satellite coordinates).
Approximate values for the coordinates are sufficient though,
which means that the DOP value can be determined without
making any measurements. As a result of the relative motion
of the satellites and the receiver(s), the value of the DOP will
change over time. The changes in the DOP value, however,
will generally be slow except in the following two cases: (1)
a satellite is rising or falling as seen by the users receiver,
and (2) there is an obstruction between the receiver and the
satellite (e.g., when passing under a bridge).
In practice, various DOP forms are used, depending on the
users need. For example, for the general GPS positioning
purposes, a user may be interested in examining the effect of
the satellite geometry on the quality of the resulting threedimensional position (latitude, longitude, and height). This
could be done by examining the value of the Position DOP
(PDOP). In other words, PDOP represents the contribution
of the satellite geometry to the three-dimensional positioning
precision. PDOP can be broken into two components:
Horizontal DOP (HDOP) and Vertical DOP (VDOP). The
former represents the satellite geometry effect on the
horizontal component of the positioning accuracy, while the
latter represents the satellite geometry effect on the vertical
component of the positioning accuracy. Because a GPS user
can track only those satellites above the horizon, VDOP will
always be longer then HDOP. As a result, the GPS height
solution is expected to be less precise than the horizontal
solution. The VDOP value could be improved by
supplementing GPS with other sensors, for example, the
pseudolites. Other commonly used DOP forms include the
Time DOP (TDOP) and the Geometric DOP (GDOP). GDOP
represents the combined effect of the PDOP and the TDOP.
The geometry matrix G is given by [9]:
(Cos ( E1) * Sin( Az1)) (Cos ( E1) * Cos ( Az1)) Sin( E1)
(Cos ( E 2) * Sin( Az 2)) (Cos ( E 2) * Cos ( Az 2)) Sin( E 2)
G=
(Cos ( E 3) * Sin( Az 3)) (Cos ( E 3) * Cos ( Az 3)) Sin( E 3)

(Cos ( E 4) * Sin( Az 4)) (Cos ( E 4) * Cos ( Az 4)) Sin( E 4)

1
1
1

(7)

Asian Journal of Geoinformatics, Vol.10,No.4 (2010)

Figure 2. The geometry of the visible GPS constellation

Where E and Az are elevation and azimuth of satellite,

respectively. The GDOP factor is obtained as:
GDOP = trace(GT G ) 1 =

trace[adj (GT G )]
det(GT G )

(8)

The GDOP factor determines the magnification factor of the

measurement noise that is translated into the derived solution.
The concept of GDOP is a powerful tool for GPS. All
receivers use some algorithm based on GDOP to select the
best set of satellites to track among the group of up to 11
satellites in view. Positioning accuracy can then be estimated
as the ranging accuracy multiplied by a dilution factor. This
dilution factor (DOP) depends solely on geometry. Typically,
variations in geometry are far greater than variations in
ranging accuracy for the nominal satellite constellation. The
GDOP concept also quantizes the effect when the nominal
satellites are not in view. The GDOP calculation for those
satellites still being tracked will give the multiplier on
ranging accuracies to yield positioning accuracies [9]. Figure
2 shows Good DOP and Poor DOP.
Since GT G is a 4 x 4 matrix, it has four eigenvalues, i
(i=1,2,3,4). It is known that the four eigenvalues of (GT G)-1
will be i-1 . Based on the fact that the trace of a matrix is
equal to the sum of its eigenvalues, equation (8) can be
represented as [2]:

GDOP = 1 + 1 + 1 + 1
1
2
3
4

(9)

The mapping is performed by defining the four variables:

f ( ) = + + + = trace (GT G )
1
1
2
3
4

(10)

f ( ) = 2 + 2 + 2 + 2 = trace [(GT G ) 2 ]
2
3
4
2
1

(11)

f ( ) = 3 + 3 + 3 + 3 = trace [(GT G )3 ]
2
3
4
3
1

(12)

f ( ) = = det (GT G )
4
1 2 3 4

(13)

The GDOP can be viewed as a functional R4R1 mapping

from f directly to GDOP, i.e., GDOP = fn ( f ) :
T
T
Inputs: ( x1, x2 , x3 , x4 ) = ( f1, f 2 , f3 , f 4 )

Output: y=GDOP
The mapping from f to GDOP is highly non-linear and
cannot be determined analytically, but can be precisely
clustered using the ACO algorithm. Figure 3 shows the block
diagram of GPS GDOP clustering using ACO algorithm.

Figure 3. The block diagram of GPS GDOP clustering using ACO algorithm

An Effective Method for GPS GDOP Clustering Using Ant Colony Optimization Algorithm

5. Simulations and Results

A DOP value of less than 2 is considered excellent-about as
good as it gets, but it doesnt happen often, usually requiring
a clear view of the sky all the way to the horizon. DOP values
of 2 to 3 are considered very good. DOP values of 4 or below
are frequently specified when equipment accuracy
capabilities are given. DOP values of 4 to 5 are considered
fairly good and would normally be acceptable for all but the
highest levels of survey precision requirements. A DOP
value of 6 would be acceptable only in low precision
conditions, such as in coarse positioning and navigation.
Position data generally should not be recorded when the
DOP value exceeds 6. The selection of the threshold dose not
closely relate to the clustering algorithm parameters and
performance.
The GDOP classifier can be employed for selecting one of
the acceptable subsets (e.g., one with GDOP factor
sufficiently small) of satellites for navigation use. In the
present study, the ACO classifier has three types of output
based on the GDOP factors (the small, medium, and large
GDOP). Similarly, clustering into more groups is feasible
based on the same philosophy.
With the aim of generating the optimal solutions of the
presented ACO algorithm developed for solving GPS GDOP
clustering problem, an application program was written with
Microsoft Visual C#.Net. Number of clusters, agents, local
search agents, iterations and initial pheromone values,
evaporation rate of pheromone and some values needed for
the algorithm are listed in Table 1.
Table 1. The parameters of ACO
Table 1: The parameters of ACO
Parameters
Value
3
K
10
R
2
L
1000
T
0
0.01
0

0.01
0.98
q
0
p
ls

0.01

The choice of the algorithm parameters is very important. In

this paper, the proposed algorithm parameters selection was
based on the experimental results. It is seen that if a
sufficiently large number of iterations is applied, very good
clustering can be achieved. The simulations results show that
correct clustering percentage of proposed method is more
than %96.
The reference [2] classified GPS GDOP using four type
neural networks. The proposed method in our paper using

ACO has better accuracy than the proposed models in

reference [2], since error of classification using ACO model
is lower than other models.

6. Conclusions
In order to improve the performance of a GPS receiver, it is
important to minimize GDOP. It is also an important
consideration to design a GPS receiver so that it can always
receive signals from four or more GPS satellites. Recent
developments in the heuristic optimization techniques are
concentrated on natural and biological systems. The ACO is
one of the most popular optimization techniques in the area
of swarm intelligence. The real power of ants resides in their
colony brain and pheromone-driven communication with in
the colony. In this paper, a method based on ACO algorithm
was proposed for GPS GDOP clustering. Without matrix
inversion required, the ACO-based approach was capable of
evaluating all subsets of satellites and hence reduced the
computational burden. Simulation results on the developed
program demonstrated that the proposed method was
preferable and effectiveness, so that correct clustering
percentage of proposed method is more than %96.

Acknowledgment
This research was supported by Iran University of Science
and Technology grants.

References
[1] K. D. McDonald, The Modernization of GPS: Plans,
New Capabilities, and the Future Relationship to
Galileo, Journal of Global Positioning System, Vol.1,
No.1, pp.1-17, 2002.
[2] D. J. Jwo and C. C. Lai, Neural Network-based GPS
GDOP Approximation and Classification, Journal of
GPS Solutions, Vol.11, No.1, pp.51-60, 2007.
[3] R. Xu and D. Wunsch, Survey of Clustering Algorithms,
IEEE Transactions on Neural Networks, Vol.16, No.3,
pp.645-678, 2005.
[4] X. Zhang, H. Peng, and Q. Zheng, A Novel Ant Colony
Optimization Algorithm for Clustering, IEEE
Conference on Signal Processing (ICSP 2006), 2006.
[5] W. Huang, J. Gou, and H. Wu, An ACO-based Approach
to Improve C-means Clustering Algorithm, IEEE
Conference on Computational Intelligence for Modelling
Control and Automation, 2006.
[6] M. Dorigo and L. M. Gambardella, Ant Colony System:
A Cooperative Learning Approach to the Traveling
Salesman Problem, IEEE Transactions on Evolutionary

Asian Journal of Geoinformatics, Vol.10,No.4 (2010)

Computation, Vol.1, No.1, pp.53-66, 1997.

[7] N. Kekec, N. Yumusak, and N. Celebi, Data Mining and
Clustering with Ant Colony Optimization, 5th
Symposium on Intelligent Manufacturing Systems,
pp.1178-1190, 2006.
[8] M. Dorigo and C. Blum, Ant Colony Optimization
Theory: A Survey, Journal of Theoretical Computer
Sciences, Vol.344, pp.243-278, 2005.

[9] B. W. Parkinson and J. J. Spilker, Global Positioning

System: Theory and Applications, The American
Institute of Aeronautics and Astronautics, 1996.
[10] R. Yarlagadda, I. Ali, N. Al-Dhahir, and J. Hershey,
GPS GDOP Metric, IEE Proc.-Radar, Sonar Navig.,
Vol.147, No.5, pp.259-264, 2000.