Sie sind auf Seite 1von 11

Spectrum Allocation in Cognitive Radio Networks

Using Firefly Algorithm

Kiran Kumar Anumandla, Shravan Kudikala,


Bharadwaj Akella Venkata, and Samrat L. Sabat

School of Physics, University of Hyderabad, Hyderabad -500046, India


kirankumara@uohyd.ac.in, slssp@uohyd.ernet.in

Abstract. In cognitive radio network, Spectrum Allocation (SA) problem is a


NP-hard problem which needs to be solved in real time. In this work, a recent bio-
inspired heuristic Firefly algorithm (FA) is used for solving SA problem. Three
objective functions namely (a) Max-Sum-Reward (MSR), (b) Max-Min-Reward
(MMR) and (c) Max-Proportional-Fair (MPF) are optimized to maximize the
network capacity. The performance of FA is compared with Particle Swarm Op-
timization (PSO) and Artificial Bee Colony (ABC) algorithms in terms of qual-
ity of solution (network capacity) and timing complexity. The simulation result
reveals that the Firefly algorithm improved quality of solution and timing com-
plexity by 17% and 100% respectively compared to PSO, in contrast to 13% and
103% compared to ABC algorithm. FA proved to give maximum utilization of
network capacity by assigning conflict free channels to secondary users.

1 Introduction
Now a days, the numbers of wireless devices have increased which in turn resulted in in-
creased demand for radio spectrum. To eliminate the interference between the spectrum
users, current policies allocate fixed spectrum slice to each wireless application. Due to
the fixed licensing policy only 6% of spectrum is utilized temporally and spatially [15].
Studies have reported that the under-utilized spectrum can be utilized using cognitive
radio technology. In this technology, each cognitive radio user (Secondary user) can
adapt to various technologies and utilize the vacant spectrum with out any interference
to the licensed users (Primary users). Cognitive radio is built on a software-defined ra-
dio and work as an intelligent wireless communication system, that can sense, learn
and adapt to statistical variations in the input and utilize vacant spectrum efficiently [7].
By using four main functionalities such as spectrum sensing, spectrum management,
spectrum mobility and spectrum sharing, secondary users can opportunistically utilize
the spectrum holes. Secondary users must opportunistically utilize the vacant licensed
spectrum subject to the constraints imposed by licensed users. Due to spectrum hetero-
geneity, spectrum available to cognitive users vary with respect to location and time,
because of traffic variation of primary users. The interference between the neighboring
users can be avoided by proper coordination between them. This in turn will improve
the spectrum utilization. The main objective of the cognitive radio is to maximize the
spectrum utilization. This demands dynamic spectrum access, hence an efficient spec-
trum allocation algorithm need to be developed to support dynamic spectrum access
and to provide fairness across all the secondary users.

B.K. Panigrahi et al. (Eds.): SEMCCO 2013, Part I, LNCS 8297, pp. 366–376, 2013.
© Springer International Publishing Switzerland 2013
Spectrum Allocation in Cognitive Radio Networks Using Firefly Algorithm 367

The spectrum allocation model has been implemented by using game theory [16], auc-
tion and pricing mechanisms [8], graph coloring [29] and local bargaining algorithms[14].
This model can also be formulated as an optimization problem and it has been solved by
using different evolutionary algorithms like genetic algorithm (GA), quantum genetic
algorithm (QGA), particle swarm optimization (PSO) [28] and Artificial Bee Colony
(ABC) [22].
ABC algorithm is a recent swarm intelligent optimization algorithm inspired by the
foraging behavior of honey bees [9,10,11]. The colony consists of three groups of bee:
employed bees, onlookers and scouts. Each employed bee is associated with a food
source and onlooker bees contribute the information of the food sources found by em-
ployed bees to choose better one. If some food sources are not improved for more
cycles, the scouts are translated to a few employed bees which search for new food
sources. PSO is another general purpose swarm based optimization method [20,12,13],
inspired by foraging activity of birds. Each particle of PSO is influenced by the fittest
particle of the swarm as well as its history while traversing in the search space. Im-
proving the algorithm is as good as finding new algorithm, which outperforms present
techniques. Improving the algorithm may further improve the quality of solution at the
cost of increased complexity. Basically our motivation is real-time implementation of
SA problem in hardware (which is out of scope of the present paper), where compu-
tational complexity is a major constraint. So, we are limiting our studies to original
form of the algorithm. In this paper, we used Firefly algorithm to solve the spectrum
allocation problem and compared its performance with basic PSO and ABC in terms of
quality of solution and time complexity to obtain the results.
In recent years, Firefly algorithm has emerged as a heuristic algorithm to solve op-
timization problems [24]. The use of fireflies as an optimization tool was initially pro-
posed by Yang in 2008 [23] which imitate the social behavior of fireflies, according
to distinctive flashing and attraction properties of fireflies to protect themselves from
predators and absorb their prey. Improved variants of FA had been proposed in order to
compare the method with other meta-heuristic algorithms.
This paper aims to solve SA problem using FA and compare the performance with
PSO and ABC to maximize the network capacity. The rest of the paper is described
as follows. Section 2 provides the previous works related to the methods which solve
spectrum allocation problem and the applications of FA. The context of FA is explained
in Section 3. The SA model of cognitive radio network is described in Section 4. Cog-
nitive radio spectrum allocation based on firefly algorithm is presented in Section 5.
Section 6 provides experimental setup and simulation results followed by conclusions
in Section 7.

2 Related Work

In the literature, various evolutionary algorithms were used to solve spectrum alloca-
tion problem of cognitive radio networks. Zhijin Zhao et al., solved spectrum allocation
problem using GA, QGA and PSO algorithms. It has been shown that these evolution-
ary algorithms greatly outperform the commonly used color sensitive graph coloring
algorithm [28]. Fang Ye et al., proposed an improved genetic spectrum assignment
368 K.K. Anumandla et al.

model, in which genetic algorithm is divided into two sets of feasible spectrum as-
signment and randomly updates the spectrum assignment strategies [27]. The penalty
function is added to the utility function to achieve spectrum assignment strategy that
satisfies the interference constraints with better fitness. This method resulted better per-
formance than the conventional genetic and quantum genetic assignment model [27].
Xiaoya Cheng et al., proved that a biological inspired ABC optimization algorithm per-
forms better than GA for optimizing the spectrum allocation for fairness and efficiency
for cognitive users [22]. This paper defines a general framework for spectrum allo-
cation in cognitive radio system and optimize the allocation of spectrum for fairness
and efficiency. An Ant Colony System (ACS) technique based on the Graph Coloring
Problem (GCP) is proposed for spectrum allocation in CR network [4]. The perfor-
mance of ACS was compared with PSO for various number of secondary users, primary
users and available channels. ACS performed better than the other algorithms, but it re-
quires more execution time to converge the solution. Ahmad Ghasemi et al., proposed a
multi-objective spectrum allocation model is presented and a new spectrum allocation
method based on Differential Evolution (DE) algorithm with the multi-objective func-
tions as weighted summation of Max-Sum-Reward (MSR), Max-Min-Reward (MMR)
and Max-Proportional-Fair (MPF) [6].
In this work, a new population based meta-heuristic algorithm namely FA is used,
to solve the spectrum allocation problem. Nature inspired metaheuristic algorithms are
becoming popular to solve global optimization problems [26]. In literature, FA is be-
ing used for solving nonlinear design, structural optimization, wireless sensor network
localization and multi-modal optimization problems etc. It is reported that Firefly algo-
rithm performs better compared to other evolutionary algorithms in terms of quality of
solution and convergence rate.
FA is used to solve standard pressure vessel design problem and the solution provided
by FA is far better than other algorithms, and also proposed a few test functions to
validate the FA algorithm [25]. Firefly algorithm was used to solve Economic Emissions
Load Dispatch Problem [3]. The experimental result has clearly shown the efficiency
and success rates of the firefly algorithm compared to particle swarm optimization and
genetic algorithm for solving the particular optimization problem [1]. Amir Hossein
et al., applied firefly algorithm for mixed variable structural optimization [19]. The
FA code was tested to solve six structural optimization problems. The optimization
results indicated that FA is more efficient than other meta-heuristic algorithms such as
PSO, GA, SA and HS [5]. Saibal et al., performed a comparative study of FA with
PSO for noisy non-linear optimization problems [2]. It was proved that FA is more
powerful in terms of time taken to solve the optimization problem because of the effect
of attractiveness function which is unique to the firefly behavior [17]. The application
of FA algorithm for solving spectrum allocation problem is not yet explored.

3 Firefly Algorithm
This section briefs about the nature of firefly algorithm. Fireflies produce short and
rhythmic flashes. These flashes are to attract female partner (communication) and to
Spectrum Allocation in Cognitive Radio Networks Using Firefly Algorithm 369

attract potential prey. In addition, flashing may also serve as a protective warning mech-
anism. Light intensity from a particular distance r from a light source obeys inverse
square law, as distance increases light intensity decreases. Furthermore, the air absorbs
light which becomes weaker and weaker as the distance increases. These two combined
factors make most fireflies visible only to a limited distance. The FA has mainly two
important issues, change in light intensity, formulation of attractiveness. The attractive-
ness of a firefly is calculated by its brightness, which in turn is corresponding fitness
of objective function. As the light intensity and the attractiveness decreases as the dis-
tance from the source increases. So the light intensity and attractiveness are considered
as monotonically decreasing functions. The light source obeys inverse square law and
light intensity is a function of distance expressed as [24]

I(r) = Io e−γ r
2
(1)

where I is the light intensity, Io is the original light intensity and γ is light absorption
coefficient. As a firefly’s attractiveness is proportional to the light intensity seen by
adjacent fireflies, so it is defined as.

β (r) = βo e−γ r
2
(2)

where βo is attractiveness at r = 0
If firefly j attracts the firefly i, then it moves towards firefly j and the state of firefly
i can be described as

xi = xi + βoe−γ ri j (x j − xi ) + αεi
2
(3)

where xi and x j are the locations of firefly i and firefly j. α is randomization parameter
and εi is a vector of random numbers with uniform distribution. The pseudo code for
FA is given in Algorithm 1.

4 Spectrum Allocation Model of Cognitive Radio Network

In this section we describe about the SA model of cognitive radio network architecture.
Assume a network of N secondary users (1 to N), M spectrum channels and K primary
users (1 to K). The general spectrum allocation model consists of channel availability
matrix L = {ln,m |ln,m ∈ {0, 1}}N×M , where ln,m = 1 if and only if channel m is available
to user n, and ln,m = 0 otherwise. The channel reward matrix B = {bn,m }N×M where
bn,m represents the reward that can be obtained by user n using the channel m. The
interference constraint matrix C = {cn,p,m |cn,p,m ∈ {0, 1}}N×N×M represents the inter-
ference constraint among secondary users, where cn,p,m = 1 if users n and p would
interfere with each other if they use channel m simultaneously and cn,p,m = 0 otherwise
and cn,p,m = 1 − ln,m if n = p [18].
In real-time applications, the spectrum environment changes slowly while users per-
form network-wide spectrum allocation operation quickly. Here we assume that the
location, available spectrum, etc. are static during the spectrum allocation, thus L, B
370 K.K. Anumandla et al.

Algorithm 1. Pseudo-code for Firefly Algorithm (FA) [24]


Step 1: Initialize the control parameter values of the FA : light absorption coefficient γ ,
attractiveness β , randomization parameter α , maximum number of iterations tmax , the number of
fireflies NP, domain space D

Step 2: Define objective function f (− →x ), −


→x = (x1 , x2 , x3 , ..., xd ), Generate the initial location of
fireflies xi (i = 1, 2, ...NP) and set the iteration number t = 0
Step 3:
while t ≤ tmax do
for i=1 to NP //do for each individual sequentially do
for j=1 to NP //do for each individual sequentially do
compute light intensity Ii at xi is determined by f (xi )
if Ii ≤ I j , then
Move firefly i towards j as described in Equation 3
endif
Attractiveness varies with distance r via e−γ r
Evaluate new solutions and update light intensity
Check the updated solutions are within limits
end for
end for
Step 3.1:
Rank the fireflies and find the current best;
Increase the Iteration Count
t=t+1
end while

and C are constant in an allocation period. The conflict free spectrum assignment ma-
trix A = {an,m |an,m ∈ {0, 1}}N×M , where an,m = 1 if channel m is allocated to secondary
user n, and an,m = 0 otherwise. A must satisfy the interference constraints defined by C:

an,m .a p,m = 0, i f cn,p,m = 1, ∀1 ≤ n, p ≤ N, 1 ≤ m ≤ M (4)


For the given L and C, spectrum allocation is to maximize network utilization U(R) and
the optimal conflict free channel assignment matrix A∗ :

A∗ = argmax U(R) (5)


A∈(L,C)

Here, we consider three fitness functions as in [28].


1. Max-Sum-Reward (MSR): It maximizes the total spectrum utilization in the system
regardless of fairness. This optimization problem is expressed as:
N M
MSR : U(R) = ∑ ∑ an,m .bn,m (6)
n=1 m=1

2. Max-Min-Reward (MMR): It maximizes the spectrum utilization of the user with


the least allotted spectrum. This optimization problem is expressed as:
Spectrum Allocation in Cognitive Radio Networks Using Firefly Algorithm 371

M
MMR : U(R) = min ∑ an,m .bn,m
1≤n≤N m=1
(7)

3. Max-Proportional-Fair (MPF): It maximizes the fairness for single-hop flows and


the corresponding fairness-driven utility optimization problem expressed as:
N M
MPF : U(R) = ( ∏ ( ∑ an,m .bn,m ))1/N (8)
n=1 m=1

5 Spectrum Allocation Based on Firefly Algorithm

In the proposed FA-based spectrum allocation, each population specifies the conflict
free channel assignment matrix. Here we propose to encode only those elements such
that ln,m = 1 and it refers to the dimension of the population. The value of every element
in the population is randomly generated that satisfies interference constraints C. The
value of L, B and C are initialized as [18]. The proposed FA-based spectrum allocation
algorithm proceeds as follows:

1. Given L = {ln,m |ln,m ∈ {0, 1}}N×M , B = {bn,m }N×M and


C = {cn,p,m |cn,p,m ∈ {0, 1}}N×N×M , set the dimension of the population as D =
∑Nn=1 ∑M m=1 ln,m , and set L1 = (n, m)|ln,m = 1 such that the elements in L1 are ar-
ranged in ascending order with n and m.
2. Initialize the control parameters of the algorithm α , β , γ and tmax .
3. Generate the initial location of fireflies randomly Xi = [x1,i , ...., x3,i , ..., xD,i ] where
xd,i ∈ 0, 1, and i ∈ (1 . . . NP)
4. Map the population xd,i to an,m , where (n, m) is the dth element of L1 for all d ∈
1 . . . D and i ∈ (1 . . . NP). The complete A matrix should satisfy the constraint matrix
C, if any violations are there then one of the user will get the channel m depends on
their reward value and the corresponding element of the matrix A is set to 1 or 0.
5. Compute the fitness of the each individual of the current population and ranking
fireflies according to their intensity (fitness values).
6. Find the current best solution and move all fireflies to the better locations as defined
in Algorithm 1.
7. If it reaches the predefined maximum generation then derive the assignment matrix
as mentioned in the step 4 and stop the process else go to step 4 and continue.

6 Experimental Setup and Results

To evaluate the FA algorithm for solving spectrum allocation problem, we setup the
objective functions by assuming that the network is noiseless and static environment.
The entire experiment is carried out in MATLAB software. In this setup, we consider
that the network has N secondary users, K primary users and M channels in network.
Each primary user selects the channel from the available list and having the protection
range of dP which is constant. Each secondary user can adjusts its communication range
372 K.K. Anumandla et al.

with in the bounds of dmin and dmax to avoid the interference between the secondary and
primary users.
By setting the values of N=20, M=20, K=20, dP =2 , dmin =1 and dmax =5, the channel
availability, reward, constraint matrices are derived as in pseudo-code of Appendix 1
[18]. For comparison purpose the same problem was solved by using PSO and ABC al-
gorithms. The parameters of PSO algorithm are defined as follows: Number of particles
NP=20, maximum iterations as 500, weighting coefficients c1 , c2 are set to 0.9 and iner-
tial weight ω =0.3. The algorithmic parameters of ABC are set as: colony size NP=20,
number of cycles=500 and limit=100. For FA number of fireflies NP=20, maximum
iterations as 500, α =0.25, β =0.2 and γ =1 [23].

Table 1. Performance analysis of FA, PSO and ABC

FA PSO ABC
Fitness Function
Time Reward std % Time Reward std % Time Reward std %
MSR 2.520e+4 2587 2.34 5.098e+4 2488 2.35 5.109e+4 2195 1.78
MMR 2.532e+4 60.8 15.18 5.111e+4 54.4 30.82 5.245e+4 46.5 8.19
MPF 2.525e+4 118.8 3.26 5.103e+4 101.3 3.89 5.134e+4 105 4.15

In this experimental setup, both algorithms were run for 20 independent runs. Indi-
vidually three fitness functions MSR, MMR and MPF as in Equation (6), (7) and (8) are
optimized using PSO, ABC and Firefly algorithms. The results are tabulated in Table.1.
This table describes about the mean timing complexity of the algorithm, mean reward
value (quality of service) and standard deviation (in percentage) of 20 runs. The compu-
tational timing complexities of FA, PSO and ABC are evaluated as [21]. The following
test code ( Algorithm 2 ) was used for evaluating timing complexity.

Algorithm 2. Test Code


for i=1 to 1000
x= double(5.55);
x=x+x; x=x./2; x=x*x; x=sqrt(x); x=log(x);
x=exp(x); y=x/x;
end for

The timing complexity (execution time) T of each algorithm is calculated as:


T2 − T1
T= (9)
T0
where T2 is the total execution time of the optimization problem, T1 is the time re-
quired for objective function alone and T0 is the execution time of test code. T1 is ob-
tained by evaluating each objective function for 3000 iterations and T2 is the time for
total execution time including objective function over 3000 function evaluations.
Spectrum Allocation in Cognitive Radio Networks Using Firefly Algorithm 373

The convergence graphs under a fixed network topology is shown in Fig.1, Fig.2 and
Fig.3. These figures corresponding to MSR, MMR and MPF objective functions respec-
tively, optimized by FA, PSO and ABC algorithms. The FA algorithm achieved a max-
imum of ~17% and ~13% performance improvement in quality of service (MPF) and
~100%, ~103% improvement in timing complexity with respect to PSO and ABC algo-
rithms. Fig.4 shows the convergence graphs for N=10, M=10 and K=10, corresponding
to three fitness functions (MSR value is divided by number of secondary users).

2600

2500

2400 FA
Reward (MSR)

PSO
2300
ABC

2200

2100

2000

1900
0 50 100 150 200 250 300 350 400 450 500
No. of Iterations

Fig. 1. Convergence Graph (Maximum Sum Reward)

70
65
60
55
Reward (MMR)

50
45
40
FA
35 PSO
30 ABC

25
20
0 50 100 150 200 250 300 350 400 450 500
No. of Iterations

Fig. 2. Convergence Graph (Max Min Reward)

Fig.5 shows the convergence graphs for N=5, M=5 and K=5, corresponding to three
fitness functions. In this figure, PSO performs better compared to FA and ABC, be-
cause of lower dimension of the problem. From all the convergence graphs, it can be
concluded that FA has attained higher reward value. This confirms the superiority of FA
algorithm having high convergence speed, quality of solution and low timing complex-
ity with respect to PSO and ABC algorithms.
374 K.K. Anumandla et al.

120

115

110

Reward (MPF)
105

100

95 FA
PSO
90
ABC
85

80
0 50 100 150 200 250 300 350 400 450 500
No. of Iterations

Fig. 3. Convergence Graph (Max Proportional Fair Reward)

90
MSR
80

70
MPF
60
Reward

50

40 MMR
FA
PSO
30
ABC

20
0 50 100 150 200 250 300 350 400 450 500
No. of Iterations

Fig. 4. Convergence Graph

60
55
50 MSR

45
40 MPF
Reward

35
30
25
FA
20 MMR PSO
15 ABC

10
0 10 20 30 40 50 60 70 80 90 100
No. of Iterations

Fig. 5. Convergence Graph


Spectrum Allocation in Cognitive Radio Networks Using Firefly Algorithm 375

7 Conclusion
Spectrum allocation (SA) for cognitive radio network is computationally complex, real
time NP-hard optimization problem. In this work, the SA problem is solved by using
three evolutionary methods: particle swarm optimization (PSO), Artificial Bee Colony
(ABC) and firefly algorithm (FA). Recent literature reported that FA’s dominance over
PSO and ABC on test functions due to its unique characteristic of attraction towards
global best population member. In PSO, particles are oriented towards global best par-
ticle irrespective of distance between them. But FA introduces a distance paradigm
that has implicit local as well as global search motivation, thereby maintaining diver-
gence. In this paper, we focused on performance evaluation of three methods to the SA
problem in-terms of critical characteristics of optimization algorithms such as accuracy,
convergence speed and repeatability. In all aspects, result shows that FA’s performance
is superior to that of PSO and ABC, proving maximum utilization of network capacity
by optimizing MSR, MMR and MPF utilization objective functions and provide conflict
free channel assignment to secondary users.

Acknowledgment. The authors are thankful to the University Grants Commission


(UGC), Government of India for providing necessary support to carry out this research
work.

References
1. Apostolopoulos, T., Vlachos, A.: Application of the firefly algorithm for solving the eco-
nomic emissions load dispatch problem. International Journal of Combinatorics 2011 (2010)
2. Chai-Ead, N., Aungkulanon, P., Luangpaiboon, P.: Bees and Firefly algorithms for noisy
non-linear optimization problems. In: Proceedings of the International Multi Conference of
Engineering and Computer Scientists, vol. 2 (2011)
3. Dhillon, J.S., Parti, S.C., Kothari, D.P.: Stochastic economic emission load dispatch. Electric
Power Systems Research 26(3), 179–186 (1993)
4. Koroupi, F., Talebi, S., Salehinejad, H.: Cognitive radio networks spectrum allocation: An
ACS perspective. Scientia Iranica 19(3), 767–773 (2012)
5. Gandomi, A.H., Yang, X.-S., Alavi, A.H.: Mixed variable structural optimization using Fire-
fly Algorithm. Computers & Structures 89(23), 2325–2336 (2011)
6. Ghasemi, A., Qassemi, F.: Multi-Objective Spectrum Allocation Based on Differential Evo-
lution Algorithm between Cognitive Radio Users. SWISS Journal of Electrical and Computer
Applications 1 (2012)
7. Haykin, S.: Cognitive Radio: Brain-Empowered Wireless Communications. IEEE Journal on
Selected Areas in Communications 23(2), 201–220 (2005)
8. Huang, J., Berry, R.A., Honig, M.L.: Auction-Based Spectrum Sharing. Mobile Networks
and Applications 11(3), 405–418 (2006)
9. Karaboga, D.: An idea based on honey bee swarm for numerical optimization. Techn. Rep.
TR06, Erciyes Univ. Press, Erciyes (2005)
10. Karaboga, D., Basturk, B.: A powerful and efficient algorithm for numerical function opti-
mization: artificial bee colony (ABC) algorithm. Journal of Global Optimization 39(3), 459–
471 (2007)
11. Karaboga, D., Basturk, B.: On the performance of artificial bee colony (ABC) algorithm.
Applied Soft Computing 8(1), 687–697 (2008)
376 K.K. Anumandla et al.

12. Kennedy, J., Eberhart, R.: Particle swarm optimization. In: Proceedings of IEEE International
Conference on Neural Networks, vol. 4, pp. 1942–1948 (1995)
13. Kennedy, J.: The particle swarm: social adaptation of knowledge. In: Proceedings of IEEE
International Conference on Evolutionary Computation, pp. 303–308. IEEE (1997)
14. Cao, L., Zheng, H.: Distributed spectrum allocation via local bargaining. In: Proceedings of
Second Annual IEEE Communications Society Conference on Sensor and Ad Hoc Commu-
nications and Networks, 2005 (SECON), pp. 475–486 (2005)
15. McHenry, M.: Spectrum white space measurements. In: New America Foundation Broad-
band Forum., vol. 1 (2003)
16. Nie, N., Comaniciu, C.: Adaptive channel allocation spectrum etiquette for cognitive radio
networks. In: Proceedings of First IEEE International Symposium on New Frontiers in Dy-
namic Spectrum Access Networks (DySPAN), pp. 269–278 (2005)
17. Pal, S.K., Rai, C.S., Singh, A.P.: Comparative Study of Firefly Algorithm and Particle Swarm
Optimization for Noisy Non-Linear Optimization Problems. International Journal of Intelli-
gent Systems and Applications (IJISA) 4(10), 50 (2012)
18. Peng, C., Zheng, H., Zhao, B.Y.: Utilization and Fairness in Spectrum Assignment for Op-
portunistic Spectrum Access. Mobile Networks and Applications 11(4), 555–576 (2006)
19. Rao, S.S., Rao, S.S.: Engineering optimization: theory and practice. John Wiley & Sons
(2009)
20. Shi, Y., Eberhart, R.: A modified particle swarm optimizer. In: Proceedings of IEEE Interna-
tional Conference on Evolutionary Computation, pp. 69–73. IEEE (1998)
21. Suganthan, P.N., Hansen, N., Liang, J.J., Deb, K., Chen, Y.P., Auger, A., Tiwari, S.: Prob-
lem definitions and evaluation criteria for the CEC 2005 special session on real-parameter
optimization. KanGAL Report 2005005 (2005)
22. Cheng, X., Jiang, M.: Cognitive Radio Spectrum Assignment Based on Artificial Bee Colony
Algorithm. In: Proceeding of IEEE 13th International Conference on Communication Tech-
nology (ICCT), pp. 161–164 (2011)
23. Yang, X.-S.: Nature-inspired metaheuristic algorithms. Luniver Press (2008)
24. Yang, X.-S.: Firefly Algorithms for Multimodal Optimization. In: Watanabe, O., Zeugmann,
T. (eds.) SAGA 2009. LNCS, vol. 5792, pp. 169–178. Springer, Heidelberg (2009)
25. Yang, X.-S.: Firefly algorithm, stochastic test functions and design optimisation. Interna-
tional Journal of Bio-Inspired Computation 2(2), 78–84 (2010)
26. Yang, X.-S.: Review of meta-heuristics and generalised evolutionary walk algorithm. Inter-
national Journal of Bio-Inspired Computation 3(2), 77–84 (2011)
27. Ye, F., Yang, R., Li, Y.: Genetic spectrum assignment model with constraints in cognitive
radio networks. International Journal of Computer Network and Information Security (IJC-
NIS) 3(4), 39 (2011)
28. Zhao, Z., Peng, Z., Zheng, S., Shang, J.: Cognitive radio Spectrum Allocation using Evo-
lutionary Algorithms. IEEE Transactions on Wireless Communications 8(9), 4421–4425
(2009)
29. Zheng, H., Peng, C.: Collaboration and fairness in opportunistic spectrum access. In: Pro-
ceedings of IEEE International Conference on Communications (ICC), vol. 5, pp. 3132–3136
(2005)

Das könnte Ihnen auch gefallen