Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Genetic Algorithms for Optimal Channel
Assignments in Mobile Communications
Lipo Wang*, Sa Li, Sokwei Cindy Lay, Wen Hsin Yu, and Chunru Wan
School of Electrical and Electronic Engineering
Nanyang Technological University
Block S2, Nanyang Avenue, Singapore 639798
* Corresponding author
Email: elpwang@ntu.edu.sg
Phone: +65 6790 6372
Fax: +65 6792 0415
ABSTRACT
The demand for mobile communication has been steadily increasing in recent years. With
the limited frequency spectrum, the problem of channel assignment becomes increasingly
important, i.e., how do we assign the calls to the available channels so that the
interference is minimized while the demand is met? This problem is known to belong to a
class of very difficult combinatorial optimization problems. In this paper, we apply the
formulation of Ngo and Li with genetic algorithms to ten benchmarking problems.
Interferencefree solutions cannot be found for some of these problems; however, the
approach is able to minimize the interference significantly. The results demonstrate the
2
effectiveness of genetic algorithms in searching for optimal solutions in this complex
optimization problem.
Keywords: genetic algorithms, channel assignment, mobile communications, wireless
network, optimization.
1 INTRODUCTION
As cellular phones become ubiquitous, there is a continuously growing demand for
mobile communication. The rate of increase in the popularity of mobile usage has far
outpaced the availability of the usable frequencies which are necessary for the
communication between mobile users and the base stations of cellular radio networks.
This restriction constitutes an important bottleneck for the capacity of mobile cellular
systems. Careful design of a network is necessary to ensure efficient use of the limited
frequency resources.
One of the most important issues on the design of a cellular radio network is to determine
a spectrumefficient and conflictfree allocation of channels among the cells while
satisfying both the traffic demand and the electromagnetic compatibility (EMC)
constraints. This is usually referred to as channel assignment or frequency assignment.
There are three types of constraints corresponding to 3 types of interference [1], namely:
1) Cochannel constraint (CCC)
3
• where the same channel cannot be assigned to certain pairs of radio cells
simultaneously
2) Adjacent channel constraint (ACC)
• where channels adjacent in the frequency spectrum cannot be assigned to
adjacent radio cells simultaneously
3) Cosite constraint (CSC)
• where channels assigned in the same radio cell must have a minimal
separation in frequency between each other.
One of the earlier aims of the channel assignment problem (CAP) is to assign the
required number of channels to each region in such a way that interference is precluded
and the frequency spectrum is used efficiently. This problem (called CAP1 in [2]) can be
shown to be equivalent to a graph coloring problem and is thus NPhard.
As demand for mobile communications grows further, interferencefree channel
assignments often do not exist for a given set of available frequencies. Minimizing
interference while satisfying demand within a given frequency spectrum is another type
of channel assignment problem (called CAP2 in [2]).
Over the recent years, several heuristic approaches have been used to solve various
channel assignment problems, including simulated annealing [3], neural networks
[4][5][2], and genetic algorithms [6][16]. In particular, [8],[13][16] used GA for CAP1.
[6], [7], and [11] formulated CAP2; however, they were interested only in interference
4
free situations. [12] gives a unique formulation of CAP2 in terms of GA; however, no
simulation results were presented. Ngo and Li [11] developed an effective GAbased
approach that obtains interferencefree channel assignment by minimizing interference in
a mobile network. They demonstrated that their approach efficiently converges to
conflictfree solutions in a number of benchmarking problems.
In this paper, we apply Ngo and Li's approach to several benchmarking channel
assignment problems where interferencefree solutions do not exist. The organization of
this paper is as follows. Section 2 states the channel assignment problem (CAP). Section
3 summarizes Ngo and Li's approach to solving CAP with genetic algorithms. Section 4
describes the tests carried out and results obtained, with many choices of parameters.
Finally, we conclude the paper in section 5.
2 CHANNEL ASSIGNMENT PROBLEM
The channel assignment problem arises in cellular telephone networks where discrete
frequency ranges within the available radio frequency spectrum, called channels, need to
be allocated to different geographical regions in order to minimize the total frequency
span, subject to demand and interferencefree constraints (CAP1), or to minimize the
overall interference, subject to demand constraints (CAP2). In this paper, we are
interested in CAP2, since it is more relevant in practice compared to CAP1.
5
There are essentially two kinds of channel allocation schemes  Fixed Channel Allocation
(FCA) and Dynamic Channel Allocation (DCA). In FCA the channels are permanently
allocated to each cell, while in DCA the channels are allocated dynamically upon request.
DCA is desirable, but under heavy traffic load conditions FCA outperforms most known
DCA schemes. Since heavy traffic conditions are expected in future generations of
cellular networks, efficient FCA schemes become more important [11]. The fixed
channel assignment problem, or in other words, assigning channels to regions in order to
minimize the interference generated has been shown to be a graph coloring problem and
is therefore NPhard.
A cellular network is assumed to consist of N arbitrary cells and the number of channels
available is given by M. The channel requirements (expected traffic) for cell j are given
by D
j
. Assume that the radio frequency (RF) propagation and the spatial density of the
expected traffic have already been calculated. The 3 types of constraints can be
determined. The electromagnetic compatibility (EMC) constraints, specified by the
minimum distance by which two channels must be separated in order that an acceptably
strong S/I ratio can be guaranteed within the regions to which the channels have been
assigned, can be represented by an N × N matrix called the compatibility matrix C.
In this matrix C:
• Each diagonal element C
ii
represents the cosite constraint (CSC), which is the
minimum separation distance between any two channels at cell i.
6
• Each nondiagonal element C
ij
represents the minimum separation distance in
frequency between any two frequencies assigned to cells i and j, respectively.
• Cosite constraint (CSC) is represented by C
ij
= 1.
• Adjacent channel constraint (ACC) is represented by C
ij
= 2.
• Cells that are free to use the same channels are represented by C
ij
= 0.
For example, suppose the number of cells in the network is N = 4, there are M = 11
channels available and the demand for the channels for each of these cells is given by D =
(1,1,1,3). Consider the compatibility matrix suggested by Sivarajan et al [1]:
C =





.

\

5 2 1 0
2 5 0 0
1 0 5 4
0 0 4 5
(1)
The diagonal terms C
ii
= 5 indicate that any two channels assigned to cell i must be at
least 5 frequencies apart in order that no cosite interference exists. Channels assigned to
cells 1 and 2 must be at least C
12
= 4 frequencies apart. Off diagonal terms of C
ij
= 1 and
C
ij
= 2 correspond to cochannel and adjacent channels constraints, respectively.
1 2 3 4 5 6 7 8 9 10 11
1
2
3
4
7
Figure 1. An interferencefree assignment for a 4cell and 11channel network
The solution space is represented by F as an N × M binary matrix, where N is the total
number of radio cells and M is the total number of available channels. Each element f
jk
in
the matrix is either one or zero such that
f
jk
=
¹
´
¦
0
1
if channel k is
¹
´
¦
)
`
¹
assigned not
assigned
to cell j (2)
This matrix F can be represented by the figure below.
Figure 2. Representation of F, an N × M binary matrix.
The cellular network is expected to meet the demand of the traffic and to avoid
interference. The first requirement imposes a demand constraint on F. Therefore, for cell
i , a total of d
i
channels are required . This implies that the total number of ones in row i
of F must be d
i
. If the assignment to cell i violates the demand constraint, then


.

\

−
¯
=
m
q
i jq
d f
1
0 (3)
The second requirement depends on the compatibility matrix C. It is made up of CSC,
CCC and ACC. In order to satisfy the CSC, if channel p is within distance C
ii
from an
f
jk
Channel Number
1 2 3 … m Row Sum
1 0 1 0 … 0 1 0 d
1
2 0 0 1 … 0 0 1 d
2
3 0 0 1 … 1 0 0 d
3
…
Cell
Number
n 1 0 … 1 0 1 d
n
8
already assigned channel q in cell i, then channel p must now be assigned to cell i. This
can be seen from the equation below.
¯
− +
≤ ≤
≠
− − =
) 1 (
1
) 1 (
ii
ii
c p
m q
p q
c p q
f
iq
> 0 (4)
To satisfy the requirements for CCC and ACC, if channel p in cell i is within distance C
ij
from an already assigned channel q in cell j, where C
ij
> 0 and i j, then channel p must
not be assigned to cell i. This is represented as shown.
¯
,
≠
=
n
c
i j
j
ij
0
1
¯
− +
≤ ≤
≠
− − =
) 1 (
1
) 1 (
ii
ii
c p
m q
p q
c p q
f
iq
> 0 (5)
Therefore, the cost function can be expressed as
C(F) =
¯¯
= =
n
i
m
p 1 1
\

¯
,
≠
=
n
c
i j
j
ij
0
1
¯
− +
≤ ≤
≠
− − =
) 1 (
1
) 1 (
ii
ii
c p
m q
p q
c p q
f
iq



.

f
ip
+
¯¯
= =
n
i
m
p 1 1
\

¯
− +
≤ ≤
≠
− − =
) 1 (
1
) 1 (
ii
ii
c p
m q
p q
c p q
f
iq



.

f
ip
+
¯
=
n
i 1


.

\

−
¯
=
m
q
i jq
d f
1
(6)
where and are weighting factors. Eq.6 comes from a combination of eqs.(3), (4), and
(5), since the 3 terms in eq.(6) become positive if any of the constraints represented by
eqs.(3), (4), and (5) are not satisfied.
9
3 SOLVING CHANNEL ASSIGNMENT PROBLEM WITH
GENETIC ALGORITHMS
We use a minimumseparation encoding scheme also by Ngo and Li [11]. In this scheme,
a pbit binary string represents an individual with q fixed elements and the minimum
separation between consecutive elements is represented by d
min
. The concept of this
scheme is to represent the solution in a way such that a one is followed by (d
min
– 1) zeros
encoded as a new “one”, denoted as . For example, an individual with p = 10 and q = 3
can be encoded as follows:
~ ~ ~
1 1 0 1
1000100100
Encoded Original
¬
The length of representation is substantially reduced.
Using the minimum separation scheme, the CSC requirement from the cost function
derived earlier can be eliminated and further reduces the search space. Hence, the cost
function of the channel assignment problem can be simplified to
C(F) =
¯¯
= =
n
i
m
p 1 1
\

¯
,
≠
=
n
c
i j
j
ij
0
1
¯
− +
≤ ≤
≠
− − =
) 1 (
1
) 1 (
ii
ii
c p
m q
p q
c p q
f
iq



.

f
ip (7)
The cost function can be further simplified by exploiting the symmetry of the
compatibility matrix C. Hence, the final cost function is represented by
10
C(F) =
¯ ¯ ¯¯
−
=
,
+ =
−
=
−
=
\

1
1
0
1
1
1
1
1
n
i
n
c
i j
c
p
p
q
ij
ij
f
jq
f
ip
+
¯ ¯
=
−
+ − =
m
c p
p
c p q
ij ij
1
1
f
jq
f
ip
+
¯
=
m
p 1
2
1
f
jq
f
ip


.

(8)
In this genetic algorithm approach, the solution space represented by F, a N x M matrix is
treated as a chromosome in the population. This means that if a population is to contain n
chromosomes, there will be n F solution matrices in the population, each representing a
chromosome. These n F solution matrices are randomly generated and are all possible
solutions for the channel assignment problem. The number of chromosomes in a
population is stated by the population size, which is a parameter that should be
manipulated to obtain an optimized solution [17]. The setting of population size, for any
genetic algorithm, is generally quite ad hoc.
Each channel assignment in the network, represented by either a 1 or 0 in the F solution
matrix, represents the genes in each chromosome. These genes encode information about
which channels have been assigned and vice versa, forming the chromosomes, and thus
the F solution matrix. After randomly generating a population of chromosomes, the
fitness of each chromosome should be evaluated. Therefore, all F solution arrays in the
population are evaluated for their fitness values, by using the final cost function. The
lower the cost function value, the fitter the chromosome.
The next step in the genetic algorithm is to generate a new population, using genetic
algorithm operators, such as selection, crossover and mutation. The selection process
consists of selecting 2 parent chromosomes from a population according to their fitness,
11
i.e., individuals with better fitness have higher chances to be selected. Each F solution
array in the population stands a chance to be selected for crossover and mutation, as a
parent chromosome.
After selection and encoding, the selected parent chromosomes, or selected F solution
arrays (encoded), will undergo crossover, with a probability of crossover, and mutation
with a probability of mutation. Crossover probability and mutation probability are
parameters that should be manipulated to obtain an optimized solution. The settings of
these parameters, like the population size parameter, are on a trialanderror basis.
Therefore, after crossover and mutation, the new offspring of the parent chromosomes are
placed in the new population. For parents whereby no crossover or mutation is
performed, they will be placed in the new population too.
The selection, crossover and mutation processes will be repeated until the new
population, which has the same size as the old population, is formed. After this
procedure, all new rows in the F solution matrix, or chromosomes, will be used for a
further run of the entire genetic algorithm until an optimized solution is found.
4 SIMULATION RESULTS
The data sets used to test the performance of the above algorithm originated from various
papers. EX1 is the simple example shown in Fig.1 originally used in [1]. EX2 is a slightly
12
larger network similar to EX1 [2] (Table I). The second set of examples (HEX1HEX4)
considered here is based on the 21cell hexagonal mobile system studied in [18]. The
final set of test problems (Kunz1Kunz4) is generated by Kunz [4] from the
topographical data of an actual 24x21 km area around Helsinki, Finland. More details of
these data sets can be found in Table I, as well as the original papers [1], [2], [4], [18].
The costs of the solutions obtained using the above GAbased approach are also given in
Table I. The results for EX1, EX2, and Kunz4 show that interferencefree assignments
can be found, as evidenced by a zero objective value, whereas no interferencefree
assignments were found for other problems.
Problem Cell
No.
Channel
No.
Demand
Vector
Cost
Function
Ex1 4 11 D
1
0
Ex2 5 17 D
2
0
Hex1 21 37 D
3
39
Hex2 21 91 D
4
13.5
Hex3 21 21 D
5
46.5
Hex4 21 56 D
6
14.5
Kunz1 10 30 D
7
13.5
Kunz2 15 44 D
8
24
Kunz3 20 60 D
9
10
Kunz4 25 73 D
10
0
Table I. Problem specification and cost function values. The demand matrices D
1
,
D
2
,…, and D
10
are shown in the captions of Figures 615.
4.1 Convergence Behavior of Genetic Algorithm
13
0
50
100
150
200
250
300
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
6
0
0
7
0
0
8
0
0
9
0
0
1
0
0
0
1
1
0
0
CPU Time (sec)
C
o
s
t
F
u
n
c
t
i
o
n
Kunz 1
Kunz 2
Kunz 3
Kunz 4
Figure 3. A typical rate of convergence trajectory based on CPU time.
The CPU time taken for each problem is dependent on the size of the problem. A
problem with a larger network and more channels tends to take a longer CPU time (Fig.1)
or more generations (Figs. 3 and 4) for the minimum cost function to be found.
0
10
20
30
40
50
60
70
80
90
100
110
120
130
0
5
0
1
0
0
1
5
0
2
0
0
2
5
0
3
0
0
3
5
0
4
0
0
4
5
0
5
0
0
5
5
0
6
0
0
6
5
0
7
0
0
7
5
0
8
0
0
8
5
0
9
0
0
9
5
0
1
0
0
0
Generations
C
o
s
t
F
u
n
c
t
i
o
n
Kunz 2
Kunz 1
Figure 4. Convergence based on number of generations for Kunz1 and Kunz2 problems.
14
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
0
5
0
1
0
0
1
5
0
2
0
0
2
5
0
3
0
0
3
5
0
4
0
0
4
5
0
5
0
0
5
5
0
6
0
0
6
5
0
7
0
0
7
5
0
8
0
0
8
5
0
9
0
0
9
5
0
1
0
0
0
5
5
0
0
1
5
0
0
0
Generations
C
o
s
t
F
u
n
c
t
i
o
n
Kunz 3
Kunz 4
Figure 5. Convergence based on number of generations for Kunz3 and Kunz4 problems.
4.2 Solution Representation for Channel Assignment
We now present the optimal solutions of channel assignment obtained. Each dot in the
Figs. 615 represents a traffic demand for a particular cell and this demand would be
allocated to a channel in a manner such that the interferences are minimized. For
example, in problem EX1, there are three demands in cell four, as indicated in Fig.6.
These demands are then assigned to channel one, six, and eleven, respectively. We
present the detailed assignments for all the test problems, in case the reader wishes to
verify or compare with our results.
15
Channel Assignment for EX 1
0
1
2
3
4
5
6
7
8
9
10
11
12
0 1 2 3 4 5
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 6. Channel assignment for EX1. Demand matrix D
1
= {1,1,1,3}.
16
Figure 7. Channel assignment for EX2. Demand matrix D
2
= {2,2,2,4,3}.
Channel Assignment for EX 2
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
0 1 2 3 4 5
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
17
Channel Assignment for Hex 1
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
0 2 4 6 8 10 12 14 16 18 20 22
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 8. Channel assignment for Hex1. Demand matrix D
3
= {2,6,2,2,2,4,4,13,19,7,4,4,7,4,9,14,7,2,2,4,2}.
18
Channel Assignment for Hex 2
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
66
68
70
72
74
76
78
80
82
84
86
88
90
92
0 2 4 6 8 10 12 14 16 18 20 22
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 9. Channel assignment for Hex2. Demand matrix D
4
= {2,6,2,2,2,4,4,13,19,7,4,4,7,4,9,14,7,2,2,4,2}.
19
Channel Assignment for Hex 3
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
0 2 4 6 8 10 12 14 16 18 20 22
Cell number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 10. Channel Assignment for Hex3. Demand matrix D
5
= {1,1,1,2,3,6,7,6,10,10,11,5,7,6,4,4,7,5,5,5,6}.
20
Channel Assignment for Hex 4
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
0 2 4 6 8 10 12 14 16 18 20 22
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 11. Channel assignment for Hex4. Demand matrix D
6
= {1,1,1,2,3,6,7,6,10,10,11,5,7,6,4,4,7,5,5,5,6}.
21
Channel Assignment for Kunz 1
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
0 2 4 6 8 10
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
= {10,11,9,5,9,4,5,7,4,8}.
Figure 12. Channel assignment for Kunz1. Demand matrix D
3
22
Channel Assignment for Kunz 2
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
0 2 4 6 8 10 12 14 16
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
= {10,11,9,5,9,4,5,7,4,8,8,9,10,7,7}.
Figure 13. Channel assignment for Kunz2. Demand matrix D
4
23
Channel Assignment for Kunz 3
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
0 2 4 6 8 10 12 14 16 18 20
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
= {10,11,9,5,9,4,5,7,4,8,8,9,10,7,7,6,4,5,5,7}
Figure 14. Channel assignment for Kunz3. Demand matrix D
5
24
Channel Assignment for Kunz 4
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
66
68
70
72
74
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Cell Number
C
h
a
n
n
e
l
N
u
m
b
e
r
Figure 15. Channel assignment for Kunz4. Demand matrix
D
6
={10,11,9,5,9,4,5,7,4,8,8,9,10,7,7,6,4,5,5,7,6,4,5,7,5}.
25
4.3 Crossover Parameter
This section presents the effect of different choices of crossover probabilities. Let us
consider population size = 20 and mutation probability = 0.004.
Table II. Effect of choices of probability of crossover on the interference results for the
Hex problem set.
Problem Probability of
Crossover
Cost Function
0.85 35
0.90 34
0.95 33
0.97 33.5
Hex 1
0.99 34
0.85 14
0.90 14
0.95 13.5
0.97 14.5
Hex 2
0.99 14.5
0.85 48.5
0.90 48
0.95 46.5
0.97 47
Hex 3
0.99 47.5
0.85 16
0.90 15
0.95 14.5
0.97 15
Hex 4
0.99 15.5
26
Table III. Effect of choices of probability of crossover on the interference results for the
Kunz problem set.
Different choices of probability of crossover gave quite different values of the cost
function. The results for the Hex and Kunz problem sets are presented in Tables II and
III. After many trials on varying the crossover probability, Tables II and III show that a
crossover probability of 0.95 tends to give the most satisfactory results (with minimum
interference value or cost function) for various problems.
Problem Probability of
Crossover
Cost Function
0.85 16
0.90 14.0
0.95 13.5
0.97 14.5
Kunz 1
0.99 15
0.85 28.5
0.90 28
0.95 26
0.97 28
Kunz 2
0.99 30
0.85 15
0.90 13.5
0.95 13
0.97 13.5
Kunz 3
0.99 16.5
0.85 7
0.90 6
0.95 3.5
0.97 5
Kunz 4
0.99 8
27
Detailed results of EX1 and EX2 are omitted, since zerointerference can be achieved
with a range of parameters tried. This may be because the problem sizes of EX1 and EX2
are small.
4.4 Mutation Parameter
This section presents the effect of different choices of mutation parameters. Let us
consider population size = 20 and probability of crossover = 0.95. The simulations were
performed by using two different methods of varying the mutation probability. In the first
method, an arbitrarily small probability of mutation, e.g., 0.003, is first selected to
generate the result. The value is then either increased or decreased to tune into a better
solution (with a lower interference value). In the second method, a large probability of
mutation, e.g., 0.0900, is initially chosen and gradually decreased until the minimum
interference value is found.
From the results obtained for Kunz problem set (Table IV), the second method proved to
be more efficient as the cost functions are smaller than the first method with fixed
mutation rates. A similar trend of results was found from the simulation of Hex problems.
The results also demonstrated that when the mutation probability is too small,
convergence to the minimum interference value is prevented. As the mutation probability
is increased, better cost function values can be obtained. However, increasing the
mutation probability beyond some critical value introduce too much randomness into the
population giving undesirable results with higher interference values (Table IV).
28
Table IV. Effects of different choices of probability of mutation on the interference
results.
Problem Probability of
Mutation
(fix rate)
Cost Function Probability of
Mutation
(decreasing
rate)
Cost Function
0.002 17 0.002 13.5
0.003 16.5 0.003 13.5
0.004 16 0.004 13
0.005 17 0.005 13.5
0.006 18 0.006 14
Kunz 1
0.007 19 0.007 14.5
0.002 28.5 0.002 28
0.003 28.5 0.003 24
0.004 28 0.004 26
0.005 29 0.005 26
0.006 29.5 0.006 27
Kunz 2
0.007 29 0.007 28
0.002 16 0.002 14.5
0.003 15.5 0.003 14
0.004 15 0.004 10
0.005 17 0.005 13
0.006 16 0.006 14
Kunz 3
0.007 17 0.007 14.5
0.002 8 0.002 6
0.003 6.5 0.003 6
0.004 6 0.004 0
0.005 5 0.005 3.5
0.006 7 0.006 4
Kunz 4
0.007 7.5 0.007 5
29
4.5 Population Size
After tuning into a satisfactory crossover and mutation probability, the effect of different
choices of population size is considered. The results tend to favor a population of 20 and
30. Greater populations lead to slower convergence and do not provide good solutions.
However, too small a population size would lead to unfavorable results as well. For
example, from Fig.16 below, a population size 20 proved to be the optimal value for of
the Kunz data set. The results are similar for the Hex data set; however, results are
independent of the population size for the EX data set.
0
5
10
15
20
25
30
35
40
Cost Function
4 10 20 30 50 70
Population
Kunz 1
Kunz 2
Kunz 3
Kunz 4
Figure 16. Effect of population size on the cost function.
30
4.6 Discussion
During the simulation, several parameters, i.e., crossover probability, mutation
probability and population size, need to be set. While experimenting on a particular
parameter, other parameters were kept constant to allow for comparisons.
The number of generations for different problems needs to be taken into account. For
example, for larger problems like Kunz 4, the number of generations needed to obtain a
satisfactory result is 100000 as compared to 50000 for Kunz 3.
For any genetic algorithms, the settings of these parameters are generally ad hoc. One
general rule was kept throughout the simulation as suggested in [17]. That is to use
relatively small population size, high crossover probability, and low mutation probability.
As can be seen from the results presented in this chapter, the algorithm achieved good
fitness values for all the benchmark problems.
5 Conclusions
In this paper, we applied Ngo and Li's GAbased approach to CAP2, i.e., channel
assignment problems in which the total interference is minimized while traffic demands
are satisfied within a given set of available channels. This approach permits the
satisfaction of traffic demand requirement and cosite constraint. It is achieved by the use
31
of a minimumseparation encoding scheme, which reduces the required number of bits
for representing the solution, and with unique genetic operators that kept the traffic
demand in the solution intact. This allowed the search space to be greatly reduced and
hence shorten the computation time. The simulations done on benchmark problems
showed that this approach could achieve desirable results.
Although we have tested a variety of choices of parameters, such as mutation rate, cross
probability, and population size, more such test with other choices of parameters should
be carried out. Implementations of GAs for DCA will also be studied in future work.
REFERENCES
[1] Sivarajan, K. N., McEliece, R. J.: Channel Assignment in Cellular Radio, Proceedings
IEEE Vehicular Technology Conference, 1989, 846850.
[2] Smith, K., Palaniswami, M.: Static and Dynamic Channel Assignment using Neural
Networks, IEEE Journal on selected areas in communications, vol. 15, no. 2, 1997,
238249.
[3] Mathar, R., Mattfeldt, J.: Channel assignment in cellular radio networks, IEEE Trans.
Veh. Technol., Vol.42, Feb, 1993, 1421.
[4] Kunz, D.: Channel assignment for cellular radio using neural networks, IEEE
Trans.Veh. Technol., vol. 40, Feb 1991, 188193.
32
[5] Funabiki, N., Takefuji, Y.: A neural network parallel algorithm for channel
assignment problems in cellular radio networks, IEEE Trans. Veh. Technol., Vol.41,
No.4, 1992, 430437.
[6] Lai W.K., Coghill G.G.: Channel Assignment for a Homogeneous Celular Network
with Genetic Algorithms, Conf. Intelligent Information Systems, 1994, 263267.
[7] Lai W.K., Coghill G.G.: Channel assignment through evolutionary optimization,
IEEE Trans. Veh. Technnol., vol. 45, no. 1, 1996, 9196.
[8] JaimesRomero, F.J., MunozRodriguez, D., Tekinay, S.: Channel assignment in
cellular systems using genetic algorithms, IEEE 46th Vehicular Technology
Conference, vol. 2, 1996, 741 745.
[9] Kim, J.S., Park, S.H., Dowd, P.W., Nasrabadi, N.M.: Channel assignment in cellular
radio using genetic algorithm, Wireless Persona Commun, vol.3, no.3, Aug.1996,
273286.
[10] Cuppini, M.: A genetic algorithm for channel assignment problems, Eur. Trans.
Telecomm. Related Technol., vol.45, no.1, 1996, 9196.
[11] Ngo, C. Y., Li, V.O.K.: Fixed channel assignment in cellular radio networks using a
modified genetic algorithm, IEEE Trans. Veh. Technol., Vol.47, No.1, 1998, 163
172.
[12] Smith, K. A.: A genetic algorithm for the channel assignment problem, IEEE Global
Technology Conference, vol. 4, 1998, 20132017.
[13] Beckmann, D., Killat, U.: A new strategy for the application of genetic algorithms to
the channel assignment problem, IEEE Trans. Veh.Technol., Vol. 48, No. 4, July
1999, 12611269.
33
[14] Thavarajah, A., Lam, W.H.: Heuristic approach for optimal channel assignment in
cellular mobile systems, IEE Proceedings Communications, vol. 146 3, June, 1999,
196 –200.
[15] Chakraborty, G., Chakraborty, B.: A genetic algorithm approach to solve channel
assignment problem in cellular radio networks, Proc. 1999 IEEE MidnightSun
Workshop on Soft Computing Methods in Industrial Applications, 1999, 34–39.
[16] Williams, M.: Making the best use of the airways: an important requirement for
military communications, Electronics & Communication Engineering Journal,
vol.12, no.2, April, 2000, 7583.
[17] Goldberg D. E.: Genetic algorithms in search, optimization, and machine learning.
Reading, Mass.: AddisonWesley, 1989.
[18] A. Gamst, “Some lower bounds for a class of frequency assignment problems",
IEEE Trans. Veh. Tech., vol. VT35, pp. 814, Feb., 1986.
effectiveness of genetic algorithms in searching for optimal solutions in this complex optimization problem.
Keywords: genetic algorithms, channel assignment, mobile communications, wireless network, optimization.
1
INTRODUCTION
As cellular phones become ubiquitous, there is a continuously growing demand for mobile communication. The rate of increase in the popularity of mobile usage has far outpaced the availability of the usable frequencies which are necessary for the communication between mobile users and the base stations of cellular radio networks. This restriction constitutes an important bottleneck for the capacity of mobile cellular systems. Careful design of a network is necessary to ensure efficient use of the limited frequency resources.
One of the most important issues on the design of a cellular radio network is to determine a spectrumefficient and conflictfree allocation of channels among the cells while satisfying both the traffic demand and the electromagnetic compatibility (EMC) constraints. This is usually referred to as channel assignment or frequency assignment.
There are three types of constraints corresponding to 3 types of interference [1], namely: 1) Cochannel constraint (CCC)
2
•
where the same channel cannot be assigned to certain pairs of radio cells simultaneously
2) Adjacent channel constraint (ACC) • where channels adjacent in the frequency spectrum cannot be assigned to adjacent radio cells simultaneously 3) Cosite constraint (CSC) • where channels assigned in the same radio cell must have a minimal separation in frequency between each other.
One of the earlier aims of the channel assignment problem (CAP) is to assign the required number of channels to each region in such a way that interference is precluded and the frequency spectrum is used efficiently. This problem (called CAP1 in [2]) can be shown to be equivalent to a graph coloring problem and is thus NPhard.
As demand for mobile communications grows further, interferencefree channel assignments often do not exist for a given set of available frequencies. Minimizing interference while satisfying demand within a given frequency spectrum is another type of channel assignment problem (called CAP2 in [2]).
Over the recent years, several heuristic approaches have been used to solve various channel assignment problems, including simulated annealing [3], neural networks [4][5][2], and genetic algorithms [6][16]. In particular, [8],[13][16] used GA for CAP1. [6], [7], and [11] formulated CAP2; however, they were interested only in interference
3
free situations. 2 CHANNEL ASSIGNMENT PROBLEM The channel assignment problem arises in cellular telephone networks where discrete frequency ranges within the available radio frequency spectrum. The organization of this paper is as follows. In this paper. or to minimize the overall interference. since it is more relevant in practice compared to CAP1. we apply Ngo and Li's approach to several benchmarking channel assignment problems where interferencefree solutions do not exist. subject to demand and interferencefree constraints (CAP1). we are interested in CAP2. called channels. Section 2 states the channel assignment problem (CAP). Finally. [12] gives a unique formulation of CAP2 in terms of GA. we conclude the paper in section 5. Ngo and Li [11] developed an effective GAbased approach that obtains interferencefree channel assignment by minimizing interference in a mobile network. Section 4 describes the tests carried out and results obtained. with many choices of parameters. need to be allocated to different geographical regions in order to minimize the total frequency span. 4 . subject to demand constraints (CAP2). Section 3 summarizes Ngo and Li's approach to solving CAP with genetic algorithms. They demonstrated that their approach efficiently converges to conflictfree solutions in a number of benchmarking problems. however. In this paper. no simulation results were presented.
The fixed channel assignment problem. while in DCA the channels are allocated dynamically upon request. In FCA the channels are permanently allocated to each cell. can be represented by an N × N matrix called the compatibility matrix C. A cellular network is assumed to consist of N arbitrary cells and the number of channels available is given by M. assigning channels to regions in order to minimize the interference generated has been shown to be a graph coloring problem and is therefore NPhard. The channel requirements (expected traffic) for cell j are given by Dj. Since heavy traffic conditions are expected in future generations of cellular networks. DCA is desirable. but under heavy traffic load conditions FCA outperforms most known DCA schemes. specified by the minimum distance by which two channels must be separated in order that an acceptably strong S/I ratio can be guaranteed within the regions to which the channels have been assigned. efficient FCA schemes become more important [11]. In this matrix C: • Each diagonal element Cii represents the cosite constraint (CSC). which is the minimum separation distance between any two channels at cell i. Assume that the radio frequency (RF) propagation and the spatial density of the expected traffic have already been calculated.There are essentially two kinds of channel allocation schemes . The electromagnetic compatibility (EMC) constraints. 5 . The 3 types of constraints can be determined.Fixed Channel Allocation (FCA) and Dynamic Channel Allocation (DCA). or in other words.
Adjacent channel constraint (ACC) is represented by Cij = 2.1. 1 1 2 3 4 2 3 4 5 6 7 8 9 10 11 6 . Consider the compatibility matrix suggested by Sivarajan et al [1]: 5 4 0 0 C= 4 5 0 1 0 0 5 2 0 1 2 5 (1) The diagonal terms Cii = 5 indicate that any two channels assigned to cell i must be at least 5 frequencies apart in order that no cosite interference exists. Cells that are free to use the same channels are represented by Cij = 0. Channels assigned to cells 1 and 2 must be at least C12 = 4 frequencies apart.3). suppose the number of cells in the network is N = 4. For example. respectively. • • • Cosite constraint (CSC) is represented by Cij = 1.• Each nondiagonal element Cij represents the minimum separation distance in frequency between any two frequencies assigned to cells i and j.1. there are M = 11 channels available and the demand for the channels for each of these cells is given by D = (1. respectively. Off diagonal terms of Cij = 1 and Cij = 2 correspond to cochannel and adjacent channels constraints.
an N × M binary matrix. Representation of F. CCC and ACC. If the assignment to cell i violates the demand constraint. In order to satisfy the CSC. then m q =1 f jq − d i 0 (3) The second requirement depends on the compatibility matrix C. An interferencefree assignment for a 4cell and 11channel network The solution space is represented by F as an N × M binary matrix. It is made up of CSC. a total of di channels are required . if channel p is within distance Cii from an 7 . fjk 1 1 Cell Number 2 3 … 0 0 0 1 Channel Number 2 1 0 0 0 3 0 1 1 … … … … … 0 0 1 1 1 0 0 0 m 0 1 0 1 Row Sum d1 d2 d3 dn n Figure 2. Therefore. This implies that the total number of ones in row i of F must be di .Figure 1. for cell i . The first requirement imposes a demand constraint on F. The cellular network is expected to meet the demand of the traffic and to avoid interference. Each element fjk in the matrix is either one or zero such that fjk = assigned 1 if channel k is not assigned 0 to cell j (2) This matrix F can be represented by the figure below. where N is the total number of radio cells and M is the total number of available channels.
Eq.already assigned channel q in cell i. where Cij > 0 and i not be assigned to cell i. p + ( cii −1) q = p − ( cii −1) q≠ p 1≤ q ≤ m fiq > 0 (4) To satisfy the requirements for CCC and ACC.(3). if channel p in cell i is within distance Cij from an already assigned channel q in cell j. (4).(6) become positive if any of the constraints represented by eqs. since the 3 terms in eq. and (5) are not satisfied. then channel p must now be assigned to cell i.6 comes from a combination of eqs. and (5). (4). n j =1 j ≠i cij 0 p + ( cii −1) q = p − ( cii −1) q≠ p 1≤ q ≤ m j. This can be seen from the equation below. This is represented as shown. the cost function can be expressed as n m n j =1 j ≠i cij 0 p + ( cii −1) q = p − ( cii −1) q≠ p 1≤ q ≤ m n m p + ( cii −1) q = p − ( cii −1) q≠ p 1≤ q ≤ m C(F) = fiq fip + fiq fip i =1 p =1 i =1 p =1 + n i =1 m q =1 f jq − d i (6) where and are weighting factors.(3). 8 . then channel p must fiq > 0 (5) Therefore.
an individual with p = 10 and q = 3 can be encoded as follows: Original 1000100100 Encoded 1 0 1 1 ~ ~ ~ The length of representation is substantially reduced. The concept of this scheme is to represent the solution in a way such that a one is followed by (dmin – 1) zeros encoded as a new “one”. Using the minimum separation scheme.3 SOLVING CHANNEL ASSIGNMENT PROBLEM WITH GENETIC ALGORITHMS We use a minimumseparation encoding scheme also by Ngo and Li [11]. Hence. denoted as . the cost function of the channel assignment problem can be simplified to n m n j =1 j ≠i cij 0 p + ( cii −1) q = p − ( cii −1) q≠ p 1≤ q ≤ m C(F) = fiq fip (7) i =1 p =1 The cost function can be further simplified by exploiting the symmetry of the compatibility matrix C. the final cost function is represented by 9 . a pbit binary string represents an individual with q fixed elements and the minimum separation between consecutive elements is represented by dmin. In this scheme. For example. Hence. the CSC requirement from the cost function derived earlier can be eliminated and further reduces the search space.
which is a parameter that should be manipulated to obtain an optimized solution [17]. each representing a chromosome. crossover and mutation. The next step in the genetic algorithm is to generate a new population. These genes encode information about which channels have been assigned and vice versa. there will be n F solution matrices in the population. a N x M matrix is treated as a chromosome in the population. is generally quite ad hoc.C(F) = n −1 n cij −1 p −1 i =1 j = i +1 p =1 q =1 cij 0 fjq fip + m p −1 fjq fip + p = cij q = p − cij +1 1 2 m p =1 fjq fip (8) In this genetic algorithm approach. Each channel assignment in the network. for any genetic algorithm. Therefore. forming the chromosomes. The setting of population size. the fitter the chromosome. and thus the F solution matrix. using genetic algorithm operators. The lower the cost function value. 10 . After randomly generating a population of chromosomes. the fitness of each chromosome should be evaluated. The number of chromosomes in a population is stated by the population size. represents the genes in each chromosome. by using the final cost function. the solution space represented by F. such as selection. all F solution arrays in the population are evaluated for their fitness values. This means that if a population is to contain n chromosomes. represented by either a 1 or 0 in the F solution matrix. These n F solution matrices are randomly generated and are all possible solutions for the channel assignment problem. The selection process consists of selecting 2 parent chromosomes from a population according to their fitness.
or chromosomes. EX1 is the simple example shown in Fig. will be used for a further run of the entire genetic algorithm until an optimized solution is found. Therefore. as a parent chromosome. The selection. the new offspring of the parent chromosomes are placed in the new population. like the population size parameter. The settings of these parameters. all new rows in the F solution matrix. For parents whereby no crossover or mutation is performed.i.. Each F solution array in the population stands a chance to be selected for crossover and mutation. which has the same size as the old population. will undergo crossover. is formed. After selection and encoding. individuals with better fitness have higher chances to be selected. with a probability of crossover. are on a trialanderror basis. after crossover and mutation.1 originally used in [1]. EX2 is a slightly 11 . the selected parent chromosomes. and mutation with a probability of mutation. they will be placed in the new population too. 4 SIMULATION RESULTS The data sets used to test the performance of the above algorithm originated from various papers. or selected F solution arrays (encoded). Crossover probability and mutation probability are parameters that should be manipulated to obtain an optimized solution. After this procedure.e. crossover and mutation processes will be repeated until the new population.
Channel No. as well as the original papers [1]. and D10 are shown in the captions of Figures 615. and Kunz4 show that interferencefree assignments can be found.5 24 10 0 Problem specification and cost function values. Finland.larger network similar to EX1 [2] (Table I). Problem Cell No. [4]. 4.5 14. 4 5 21 21 21 21 10 15 20 25 11 17 37 91 21 56 30 44 60 73 D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 0 0 39 13. The costs of the solutions obtained using the above GAbased approach are also given in Table I. The results for EX1. [2].1 Convergence Behavior of Genetic Algorithm 12 . D2 . [18]. EX2.5 13.….5 46. The final set of test problems (Kunz1Kunz4) is generated by Kunz [4] from the topographical data of an actual 24x21 km area around Helsinki. Demand Cost Vector Function Ex1 Ex2 Hex1 Hex2 Hex3 Hex4 Kunz1 Kunz2 Kunz3 Kunz4 Table I. The demand matrices D1. The second set of examples (HEX1HEX4) considered here is based on the 21cell hexagonal mobile system studied in [18]. whereas no interferencefree assignments were found for other problems. as evidenced by a zero objective value. More details of these data sets can be found in Table I.
130 120 110 100 90 80 70 60 50 40 30 20 10 0 50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Cost Function Kunz 2 Kunz 1 Generations Figure 4. A problem with a larger network and more channels tends to take a longer CPU time (Fig. A typical rate of convergence trajectory based on CPU time. The CPU time taken for each problem is dependent on the size of the problem.300 Cost Function 250 200 150 100 50 0 10 0 20 0 30 0 40 0 50 0 60 0 70 0 80 0 90 0 10 00 11 00 Kunz 1 Kunz 2 Kunz 3 Kunz 4 CPU Time (sec) Figure 3. 10 00 0 13 .1) or more generations (Figs. 3 and 4) for the minimum cost function to be found. Convergence based on number of generations for Kunz1 and Kunz2 problems.
there are three demands in cell four. 615 represents a traffic demand for a particular cell and this demand would be allocated to a channel in a manner such that the interferences are minimized. 4.0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 5500 Generations Figure 5. in case the reader wishes to verify or compare with our results. six.6. We present the detailed assignments for all the test problems. respectively. For example. 14 15000 140 135 130 125 120 115 110 105 100 95 90 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 Cost Function Kunz 3 Kunz 4 . as indicated in Fig. Each dot in the Figs. and eleven. in problem EX1.2 Solution Representation for Channel Assignment We now present the optimal solutions of channel assignment obtained. These demands are then assigned to channel one. Convergence based on number of generations for Kunz3 and Kunz4 problems.
1.1. Demand matrix D1= {1.Channel Assignment for EX 1 12 11 10 9 Channel Number 8 7 6 5 4 3 2 1 0 0 1 2 3 4 5 Cell Number Figure 6.3}. Channel assignment for EX1. 15 .
16 .2. Channel assignment for EX2.3}.4.2.Channel Assignment for EX 2 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 0 1 2 3 4 5 Cell Number Channel Number Figure 7. Demand matrix D2= {2.
9.2.6.Channel Assignment for Hex 1 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 22 Cell Number Channel Number Figure 8.19.13.2}.4.2.14.7.4.4. Channel assignment for Hex1.4.2.7. 17 .7.2.4. Demand matrix D3 = {2.2.4.
4.2.4.2.Channel Assignment for Hex 2 92 90 88 86 84 82 80 78 76 74 72 70 68 66 64 62 60 58 56 54 52 50 48 46 44 42 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Channel Number 0 2 4 6 8 10 12 14 16 18 20 22 Cell Number Figure 9.2.6.2}.7.7.9. 18 .4.2. Channel assignment for Hex2.13.7.19. Demand matrix D4 = {2.4.4.14.4.2.
2.1.1.7.Channel Assignment for Hex 3 22 21 20 19 18 17 16 15 Channel Number 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 0 2 4 6 8 10 12 14 16 18 20 22 Cell number Figure 10.5.7.6}. Demand matrix D5 = {1.6.11. 19 .7.4.3.5.10.4.10. Channel Assignment for Hex3.6.5.6.5.
Channel assignment for Hex4.5.7.4.Channel Assignment for Hex 4 58 56 54 52 50 48 46 44 42 40 38 Channel Number 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 18 20 22 Cell Number Figure 11.7.7.5.5.4.3.1. 20 .10.6.6.11.6}. Demand matrix D6 = {1.2.1.10.6.5.
7.4.Channel Assignment for Kunz 1 30 28 26 24 22 Channel Number 20 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 Cell Number Figure 12. Channel assignment for Kunz1.9.8}.11. Demand matrix D3 = {10. 21 .5.4.9.5.
4. 22 .4.10.7}.7.Channel Assignment for Kunz 2 44 42 40 38 36 34 32 30 Channel Number 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 0 2 4 6 8 10 12 14 16 Cell Number Figure 13.8.9.9. Demand matrix D4 = {10.7.5.8.11.5. Channel assignment for Kunz2.9.
5.4. Demand matrix D5 = {10.Channel Assignment for Kunz 3 60 58 56 54 52 50 48 46 44 42 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Channel Number 0 2 4 6 8 10 12 14 16 18 20 Cell Number Figure 14.5.7.4.5.7} 23 .7.8.5.9.11.7.9.8.6.4.10. Channel assignment for Kunz3.9.
Demand matrix D6={10.5.9.9.8.7.10. Channel assignment for Kunz4.5.6.11.4.5.4.8.7.5}.9.4. 24 .5.6.7.Channel Assignment for Kunz 4 74 72 70 68 66 64 62 60 58 56 54 52 50 48 46 44 42 40 38 36 34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0 Channel Number 0 2 4 6 8 10 12 14 16 18 20 22 24 26 Cell Number Figure 15.7.4.7.5.
5 16 15 14.99 0.99 0.5 15 15.85 0.95 0.5 34 14 14 13.95 0.95 0.95 0.97 0. Effect of choices of probability of crossover on the interference results for the Hex problem set.5 48 46.97 0.5 Hex 2 0.85 0.5 14. 25 .97 0.5 47 47.99 Cost Function Hex 1 35 34 33 33. Let us consider population size = 20 and mutation probability = 0.85 0.004.90 0.4.97 0. Problem Probability of Crossover 0.90 0.3 Crossover Parameter This section presents the effect of different choices of crossover probabilities.90 0.5 14.5 48.90 0.99 Hex 3 Hex 4 Table II.85 0.
5 5 8 Kunz 2 0.99 0.95 0.95 0. Different choices of probability of crossover gave quite different values of the cost function.5 15 28.85 0.85 0.99 Cost Function 16 14.97 0.0 13.5 14.99 0.90 0.97 0.90 0.97 0.90 0.95 tends to give the most satisfactory results (with minimum interference value or cost function) for various problems.85 0.99 Kunz 3 Kunz 4 Table III.95 0.85 0.5 7 6 3. Tables II and III show that a crossover probability of 0.95 0.90 0.5 28 26 28 30 15 13.97 0. After many trials on varying the crossover probability. The results for the Hex and Kunz problem sets are presented in Tables II and III. Effect of choices of probability of crossover on the interference results for the Kunz problem set. 26 .5 13 13.Problem Kunz 1 Probability of Crossover 0.5 16.
95.g. As the mutation probability is increased. an arbitrarily small probability of mutation.g. is initially chosen and gradually decreased until the minimum interference value is found.003. better cost function values can be obtained. 0. A similar trend of results was found from the simulation of Hex problems. 0. e. The value is then either increased or decreased to tune into a better solution (with a lower interference value). 4. In the first method.0900.Detailed results of EX1 and EX2 are omitted.4 Mutation Parameter This section presents the effect of different choices of mutation parameters. a large probability of mutation. The results also demonstrated that when the mutation probability is too small. Let us consider population size = 20 and probability of crossover = 0. e. From the results obtained for Kunz problem set (Table IV).. the second method proved to be more efficient as the cost functions are smaller than the first method with fixed mutation rates. since zerointerference can be achieved with a range of parameters tried. 27 . However. increasing the mutation probability beyond some critical value introduce too much randomness into the population giving undesirable results with higher interference values (Table IV). This may be because the problem sizes of EX1 and EX2 are small. convergence to the minimum interference value is prevented.. is first selected to generate the result. The simulations were performed by using two different methods of varying the mutation probability. In the second method.
004 0.002 0.006 0.007 0.5 14 14.002 0.003 0.5 Probability of Mutation (decreasing rate) 0.5 15 17 16 17 8 6.5 28 29 29.5 28.004 0.002 0.005 0.003 0.007 17 16.006 0.007 0.006 0.003 0.5 13 13.002 0.004 0.002 0. Effects of different choices of probability of mutation on the interference results.5 6 5 7 7.003 0.Problem Probability of Mutation (fix rate) Cost Function Kunz 1 0.003 0.005 0.004 0.005 0.5 16 17 18 19 28.007 Kunz 3 Kunz 4 Table IV.003 0.002 0.007 0.5 14 10 13 14 14.007 0.004 0.002 0.5 29 16 15.005 0.005 0.002 0.006 0.005 0.006 0.004 0.003 0.5 6 6 0 3. 28 .004 0.007 Cost Function 13.005 0.006 0.007 0.003 0.006 0.5 13.5 28 24 26 26 27 28 14.5 4 5 Kunz 2 0.004 0.005 0.006 0.
from Fig. however. Greater populations lead to slower convergence and do not provide good solutions. results are independent of the population size for the EX data set. 29 . The results tend to favor a population of 20 and 30. 40 35 30 25 Cost Function 20 15 10 5 0 Kunz 1 Kunz 2 Kunz 3 Kunz 4 4 10 20 30 50 70 Population Figure 16.5 Population Size After tuning into a satisfactory crossover and mutation probability. For example.16 below. However. too small a population size would lead to unfavorable results as well. The results are similar for the Hex data set. the effect of different choices of population size is considered.4. a population size 20 proved to be the optimal value for of the Kunz data set. Effect of population size on the cost function.
For any genetic algorithms.. 5 Conclusions In this paper. For example.6 Discussion During the simulation. the algorithm achieved good fitness values for all the benchmark problems. the settings of these parameters are generally ad hoc. As can be seen from the results presented in this chapter. and low mutation probability. channel assignment problems in which the total interference is minimized while traffic demands are satisfied within a given set of available channels. crossover probability. mutation probability and population size. we applied Ngo and Li's GAbased approach to CAP2. several parameters. need to be set. the number of generations needed to obtain a satisfactory result is 100000 as compared to 50000 for Kunz 3. This approach permits the satisfaction of traffic demand requirement and cosite constraint. It is achieved by the use 30 . One general rule was kept throughout the simulation as suggested in [17]. That is to use relatively small population size. for larger problems like Kunz 4. other parameters were kept constant to allow for comparisons.e.4. The number of generations for different problems needs to be taken into account. While experimenting on a particular parameter. i. i.. high crossover probability.e.
The simulations done on benchmark problems showed that this approach could achieve desirable results.. 1421..42. Palaniswami. 1989.. crossprobability. N. 15. and with unique genetic operators that kept the traffic demand in the solution intact. Implementations of GAs for DCA will also be studied in future work. Mattfeldt.: Channel assignment for cellular radio using neural networks. Technol. 238249. REFERENCES [1] Sivarajan. R.of a minimumseparation encoding scheme.: Channel assignment in cellular radio networks. This allowed the search space to be greatly reduced and hence shorten the computation time. [2] Smith. J. such as mutation rate. Feb 1991. 31 . Feb. 188193.: Channel Assignment in Cellular Radio. which reduces the required number of bits for representing the solution. K. Proceedings IEEE Vehicular Technology Conference. R.Veh. more such test with other choices of parameters should be carried out. 1997. IEEE Journal on selected areas in communications. Although we have tested a variety of choices of parameters. Veh. IEEE Trans. [4] Kunz. 40. J. vol. and population size. 1993.. D. IEEE Trans. no.: Static and Dynamic Channel Assignment using Neural Networks. vol. 2. McEliece. Vol. K. M. 846850. Technol.. [3] Mathar.
July 1999. IEEE Trans. Vol. 20132017. [6] Lai W.O. No. IEEE Trans. Vol. 430437. [10] Cuppini. 48. F. 263267. IEEE Global Technology Conference... Y.: A genetic algorithm for channel assignment problems..G.: Fixed channel assignment in cellular radio networks using a modified genetic algorithm. Coghill G. 273286. 4. Vol. Eur. [9] Kim.M. Aug. Takefuji.G. no.. Tekinay. Wireless Persona Commun. [7] Lai W. M.. D.45. vol. Conf. [8] JaimesRomero. N. Telecomm. Veh. Nasrabadi.47.. no. vol. J. [13] Beckmann.J. Technol. 1992.[5] Funabiki. vol. Dowd.. 1994. 1996.. 9196.: A genetic algorithm for the channel assignment problem.. No. Technnol. 32 . N. Coghill G. 9196. no. 163172. vol.K. Y. No. [11] Ngo. 2. Technol. D. 12611269. 45. IEEE 46th Vehicular Technology Conference. V.: A new strategy for the application of genetic algorithms to the channel assignment problem. 1998.1996. IEEE Trans.1. 4..4.S.: Channel assignment in cellular radio using genetic algorithm. Trans.1.K. Killat.: Channel Assignment for a Homogeneous Celular Network with Genetic Algorithms. U. vol.W. Park. Li. Veh. Intelligent Information Systems. MunozRodriguez..41. Veh... S.3. 1. 741 745. K.. 1996.: Channel assignment in cellular systems using genetic algorithms.: A neural network parallel algorithm for channel assignment problems in cellular radio networks. Related Technol.: Channel assignment through evolutionary optimization.K. 1998. [12] Smith. 1996.. C. IEEE Trans. A. P.Technol.3. Veh. S.H.
and machine learning. 33 .H. [16] Williams. 1999. [17] Goldberg D.. Reading. W.2. G. Mass. 7583. 1989. no. pp. B. Proc.. [15] Chakraborty. [18] A.. E. Gamst. optimization.. VT35. Veh. M. Lam. IEEE Trans. 2000. IEE Proceedings Communications. A. vol. June.12. April.: Genetic algorithms in search. Electronics & Communication Engineering Journal. “Some lower bounds for a class of frequency assignment problems".: AddisonWesley.[14] Thavarajah. 814. 1986. Feb. 146 3.: Making the best use of the airways: an important requirement for military communications. 34–39. Tech. vol. 1999. vol. 196 –200.: A genetic algorithm approach to solve channel assignment problem in cellular radio networks. Chakraborty. 1999 IEEE MidnightSun Workshop on Soft Computing Methods in Industrial Applications.: Heuristic approach for optimal channel assignment in cellular mobile systems.
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.