KSCE Journal of Civil Engineering (2017) 21(2):535-545 Design Optimization and Applications in Civil Engineering

Copyright 2017 Korean Society of Civil Engineers

DOI 10.1007/s12205-017-1501-1 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Firefly Algorithm for Discrete Optimization Problems: A Survey

Surafel Luleseged Tilahun* and Jean Medard T. Ngnotchouye**
Received September 5, 2016/Revised October 21, 2016/Accepted November 2, 2016

Abstract

Firefly algorithm is a nature-inspired metaheuristic algorithm inspired by the flashing behavior of fireflies. It is originally proposed
for continuous problems. However, due to its effectiveness and success in solving continuous problems, different studies are
conducted in modifying the algorithm to suit discrete problems. Many engineering as well as optimization problems from other
disciplines involve discrete variables. Recent reviews on the application and modifications of firefly algorithm mainly focus on
continuous problems. This paper is devoted to the detailed review of the modifications done on firefly algorithm in order to solve
optimization problems with discrete variables. Hence, advances on the application of firefly algorithm for optimization problems
with binary, integer as well as mixed variables will be discussed. Possible future works will also be highlighted.
Keywords: firefly algorithm, optimization, nature-inspired algorithm, discrete optimization, discrete firefly algorithm

1. Introduction Firefly algorithm is one of these algorithm inspired by the

Optimization problems are problems of finding values for the
variables which will give an optimum functional value(s) of the
objective function(s). These kind of problems exists beyond our
daily activity. They are common problems in engineering, decision
science, agriculture, computer science, economics and many
other disciplines (Pike et al., 2014; Tilahun and Asfaw, 2012;
Hamad- neh et al., 2012; Ong abd Tilahun, 2011; Tilahun and
Ong, 2012b; Lucia and Xu, 1990; Joshi, 2013; Tilahun and Ong,
2012a; Parolo et al., 2009; Shimoyama et al., 2011). Based on
the decision variables an optimization problem can be classified
into three broad categories. These are problems with continuous
variable, non-continuous variables and mixed variables. Continuous
variables are when the variable can have any value in the given
interval and if that is not the case, i.e. if the variables can not take
some values in the given interval, then we have a problem with
discrete or non-continuous variables, this includes integer and
binary variables. If some of the decision variables can be assigned
with continuous and the rest with non-continuous values then it
is called a mixed problem. Most real optimization problems
involve non-continuous variables, like number of products and
human resource.
For an optimization problem there are different solution methods.
One class of the solution methods is metaheuristic algorithms.
These methods are non-deterministic solution methods which
search the solution space based on an educated guess and trial
and error approach incorporating a certain randomness terms.
flashing behavior of flashing bugs also called fireflies (Yang,
2008). It is an easy to implement algorithm which can easily be
parallelized and can also be used by non-expert practitioners. It is
also tested to be effected on problems from different problem
domain. Even though it has a number of strengths, it is prone to
parameter setting and also controlling the exploration and
exploitation of the search space is an issue, as in many metaheuristic
algorithms. To overcome these limitations different modifications
are proposed. In addition to modifying the algorithm, some
studies are suggest hybridizing firefly algorithm with other
metaheuristic or local search algorithms to produce a better
performing algorithm (for example Fister et al. (2013b); Tilahun
and Ong (2013); Rahmani and MirHassani (2014); Zhang et al.
(2016b); Huang et al. (2013)). These extensions of the standard
firefly algorithm are used to boost the performance of the
algorithm and also to make it suitable for different class of
problems including problems with non-continuous variables.
Different studies used firefly algorithm for optimization and
design engineering problems. There is a wide literature on this
issue, however some of them are mentioned here to motivate the
reader and also highlight the popularity of firefly algorithm in
engineering applications. Mohanty (2016) used the algorithm for
design optimization of a shell and tube heat exchanger from
economic point of view. In Sheikholeslami et al. (2015), firefly
algorithm has been hybridized with harmony search algorithm to
minimize the cost of reinforced concrete retaining walls. For the
standard pressure vessel design optimization, it has been found

*Postdoctoral researcher, School of Mathematics, Statistics and Computer Science, University of KwaZulu-Natal, 3209, Pietermaritzburg, South Africa
(Corresponding Author, E-mail: surafelaau@yahoo.com)
**Lecturer, School of Mathematics, Statistics and Computer Science, University of KwaZulu-Natal, 3209, Pietermaritzburg, South Africa (E-mail: ngno-
tchouye@ukzn.ac.za)

535
 Surafel Luleseged Tilahun and Jean Medard T. Ngnotchouye
to be effective (Yang, 2010). In (Kazemzadeh-Parsi, 2014),
firefly algorithm was applied and compared with other algorithms to
result a promising result in engineering design optimization
problems and three sizing optimization of truss structures. The
discrete optimum design of new-generation beams problem was
solved using firefly algorithm (Erdal, 2016). There are many
similar applications and uses of firefly algorithm in different
aspects of engineering like in tower structures design (Talatahari
et al., 2014), system identification design problem (Upadhyay et
al., 2016), competitive location-design problem (Sadjadi et al.,
2016), power system design (Setiadi and Jones, 2016), antenna
design (Ram et al., 2014), engineering designing problems
(Francisco et al., 2015), groundwater remediation design
(Kazemzadeh-Parsi et al., 2015) and construction system
reliability analysis (Li et al., 2016b).
Recently, some survey and review papers have been publish,
mainly focusing on firefly algorithm for problems with continuous
search space (Fister et al., 2013a; Abdelaziz et al., 2015; Fister et
al., 2014; Yang and He, 2013; Johari et al., 2013; Fister et al.,
2015; Khan et al., 2016). However, in applications like in
engineering, the search space is not always well behaved and
continuous. This paper addresses this gap and discusses major
contribution and application of firefly algorithm for non-
continuous spaced problems. This may help practitioners to
identify the proper version of the algorithm suitable for their
problem.
Hence in this paper, the modifications proposed to make firefly
algorithm suitable for optimization problems with non-continuous
variables will be discussed. A discussion on the strength and
weakness of the modifications will be presented along with
possible future work. In the next section a general discussion on
optimization problems will be given followed by a discussion on
firefly algorithm. In Section 3, a discussion on modified versions
of firefly algorithms will be given followed by a summary and
discussion in Section 4. In Section 5 a conclusion will be
presented.
2. Preliminaries
2.1 Optimization Problems
A well-defined optimization problem needs to have a decision
variable x = (x1, x2, ..., xn), for which we search a value for; an
objective function, f(x), which is a function of the decision
variable and also used to measure the performance of the values
assigned to the decision variables; and a feasible region, S, from
which the decision variable can take values. A minimization
problem can then be given as in Eq. (1)
n uniform distribution between zero and one.
lim { f ( x )|x S R }
min (1)
x Therefore by putting Eqs. (4) and (5) together, the updating
The search space or the feasible region S can be continuous,
non-continuous or mixed i.e. continuous for some of the
variables and non-continuous for the rest, possibly binary or
integer.
A solution x* is said to be global (local) optimal solution for
536
the minimization problem given in Eq. (1) if and only if x S
and f( x* ) f ( x ) for all x S (for all x in the neighborhood of x*).
2.2 Firefly Algorithm
Nature has been a motivation to different metaheuristic
algorithms (Yang, 2008; Babaoglu et al., 2007; Miao, 2014;
Tilahun et al., 2012). The interaction between a predator and its
pray, the attraction between bodies and the swarm behavior of
fishes or birds can be mentioned as an example (Dorigo and
Stutzle, 2004; Kennedy and Eberhart, 1995; Tilahun and Ong,
2015; Yuce et al., 2013). Firefly algorithm is inspired by the
flashing behavior of flashing bugs also called fireflies. It is
proposed for optimization problems with continuous variables
(Yang, 2008). Randomly generated feasible solutions will be
considered as fireflies where their brightness is determined by
their performance on the objective function. The algorithm is
guided by three rules. The first rule is that fireflies are unisex,
which means any firefly can be attracted to any other firefly.
The second rule is the brightness of a firefly depends on its
performance in the objective function. The attraction of a
firefly depends on its brightness and decreases with distance.
This means, since we are considering a minimization problem,
a solution with smaller functional value is brighter. The light
intensity has an inverse square law as given in Eq. (2).
1
I ----2 (2)
r
where I is the intensity and r is the distance. The brightness
follows similar rule as the light intensity with respect to the
distance. Furthermore, suppose the light is passing through a
medium with a light absorption coefficient of . The brightness
of a firefly at a distance r can be summarized using the equation
given in Eq. (3).
r
2
= 0 e (3)
where is the brightness of the firefly at a distance r and 0 is the
brightness at the source, i.e. r = 0.
A solution xi will be attracted by a brighter firefly xj, this means
xi moves towards xj using Eq. (4).
r2
xi : = xi + 0 e ( xj xi ) (4)
In addition it will explore using a random movement given by
Eq. (5).
xi : = xi + ( rand( ) 0.5 ) (5)
for a step length for the random movement and rand( ) is a
vector of n vector whose entries are generated randomly from a
equation of a firefly xi is given as in Eq. (6).
r
2
xi : = x i + 0 e ( xj xi ) + ( rand ( ) 0.5 ) (6)
for a brighter firefly xj. If there is no brighter firefly than xi it will
only perform a random move as given in Eq. (5).
KSCE Journal of Civil Engineering
 Firefly Algorithm for Discrete Optimization Problems: A Survey

Table 1. The Standard Firefly Algorithm (based on (Yang, 2009))

Set algorithm parameters (, )
Set simulation set-up (Maximum number of iteration (MaxItr),
Number of initial solutions (N))
Randomly generate N feasible solutions (x1, x2, ..., xN)
Compute light intensity Ii for each xi based on the objective function f(x)
Rank or sort fireflies based on their light intensity i.e. Ii Ii 1, i
for iteration = 1 : MaxItr
for i=1:N-1
for j=i+1:N
if (Ii > I j)
move firefly i towards firefly j
attractiveness varies via exp(r2)
update solution and compute light intensity
end if
end for
end for
move xN randomly rank fireflies Fig. 1. S-shaped and V-shaped Transfer Functions (according to
end for Eq. (10) a = 2, 1, 1--- and 1--- for S1, S2, S3 and S4, respectively)
2 3
Report the best solution

1, rand < S ( xi ( k ) )
The algorithm is summarized in Table 1. xi ( k ) = (8)
0, otherwise
3. Modified Firefly Algorithms for Optimization In addition to using the sigmoid function, the tan hyperbolic
Problems with Discrete Variables function has also be used in some studies (Chandrasekaran and
Simon, 2012; Chandrasekaran et al., 2013; Crawford et al.,
Firefly algorithms has been modified and used for optimization 2014b). In (Crawford et al., 2014b), rather than using a random
problems with non-continuous variables. Based on the space number to determine the value of xi(k) as in Eq. (8), a given
where the updating is performed, these modifications can generally threshold, , is used, i.e. Eq. (9).
be categorized in to two categories. The first category is when
1, < tan xi ( k )
the updating is done in the continuous space and a discretization xi ( k ) = (9)
mechanism is used to change the values to discrete numbers 0, otherwise
whereas the second is when the updating is done on the discrete
Furthermore, in Liu et al. (2014) both the sigmoid as well as
space. In the second category there is no need to switch between
the tan hyperbolic functions are used in order to limit the value of
continuous and non-continuous spaces.
xi(k) in between 0 and 1. The two functions are shown in Fig. 1,
where the sigmoid function is an S-shaped function and the tan
3.1 Updating in the Continuous Space
hyperbolic function a V-shaped.
In this category the updating procedure of the standard firefly
In Crawford et al. (2014a), four variants of the sigmoid
algorithm will be used and the results will be converted to
functions and four variants of the tan hyperbolic functions are
discrete values.
given. The four S-shaped function is generalized and is given in
Perhaps the first modification of firefly algorithm for discrete
Eq. (10), for a = 2, 1, 1--- and 1--- .
problems, specifically for binary problems, is done for a job 2 3
scheduling problem in Sayadi et al. (2010). In this modification, 1
S ( xi ( k ) ) = --------------------
-
ax ( k )
(10)
after updating a solution xi using the updating equations of the 1+e i

standard firefly algorithm, each component k of the solution

The four V-shaped functions are also given in Eqs. ((11) -
vector xi will be converted to 0's and 1's based on the sigmoid
(14)).
function, given in Eq. (7), where the sigmoid function represents
the probability that xi(k) will be 1. 2
V1 ( xi ( k ) ) = erf ------ xi ( k ) (11)
1
S ( xi ( k ) ) = ------------------
-
x ( k )
(7)
1+e i

V2 ( xi ( k ) ) = tanh ( xi ( k ) ) (12)
A similar approach, using Eq. (8), is used in Palit et al. (2011);
xi ( k )
V3 ( xi ( k ) ) = ------------------------
Yang et al. (2013); Sayadi et al. (2013); Rajalakshmi et al. - (13)
2
(2015); Chhikara and Singh (2015) and Li et al. (2015). 1 + xi ( k )

Vol. 21, No. 2 / February 2017 537

 Surafel Luleseged Tilahun and Jean Medard T. Ngnotchouye

2 where 0 < <1.

V4 ( xi ( k ) ) = --- arc tan --- xi ( k ) (14)
2 Itr
= 0 (18)
The transformation functions are plotted and given in Fig. 1. After
Fig. 3 shows the adaptive step length proposed for different
the transformation of the values to the interval [0, 1], three methods
values of . Another modification on the random movement step
are used to convert the values to a binary number. These methods are
length, Eq. (19), is given in Bacanin and Tuba (2014).
1, rand < T ( x ( k ) ) x * ( k ), rand < T ( x ( k ) )
given by: x (k) = , x (k) = and
i i i
i i
1 -
0, otherwise 0, otherwise 4 -----------------
MaxItr
10
= --------- (19)
1
( x ( k ) ) , rand < T ( x ( k ) ) 9
x (k) =
i
i i
, where xi(k) is the kth component of
x ( k ), otherwise
i
Additionally, a scaling parameter, xmax xmin, multiplies the
xi from the previous iteration, x* the best solution so far from random movement (Bacanin et al., 2013; Baghlani et al., 2014).
memory and (xi(k))1 gives the complement of xi(k), that is if xi(k) In Liu et al. (2014), the randomness step length is modified
= 1 its complement will be 0 and if it is 0 its complement will using
be 1.
1
In addition to using sigmoid function for the conversion of the = 0 -------------------------------
MaxIter
- (20)
Itr ---------------------
variables to binary numbers, the updating formula was modified 2
1+e
in (Farhoodnea et al., 2014a, b). The distance is modified using
S ( r ) = tanh ( r ) for a parameter with value near 1 and based Furthermore, in Chhikara and Singh (2015), and is
on this the updating formula is done using modified based on the problem property.
In Costa et al. (2014), both and are made to change with
xi + ( xj xi ) + ( rand ( ) 0.5 ), rand < S ( r )
xi = (15) iteration using
x i, otherwise
Itr
= max -------------------- ( max min ) (21)
Hence, if firefly xi is closer to xj then it has less probability of MaxIter
moving. That may affect the quality of the solution as rather than
and
improving or exploring the solution will stay in its position. In
Costa et al. (2014), another version of S-shaped sigmoid Itr - ln ------------
-------------------- min
-
MaxIter max
function given in Eq. (16) is proposed. = max e (22)
S ( xi ( k ) ) = 0.5 ( 1 + erf ( xi ( k ) ) ) (16) In this modification decreases linearly and also decreases
but quicker than a linear function as shown in Fig. 2.
In addition to modification for binary problems, firefly algorithm
This implies that as the iteration increases the random movement
has also been used for optimization problems with integer valued
decreases and so does the attraction step length. In addition,
variables. In Bacanin et al. (2013) and Baghlani et al. (2014), the
Levy distribution is used to generate a random direction with a
same updating formulas as the standard firefly algorithm used
scaling parameter which is the difference between the maximum
and the results are rounded to the nearest integer value. A similar
and minimum values of the feasible region. However, it should
approach is used for the mixed integer problem in Gandomi et al.
be noted that Levy distribution will generate a random direction
(2011). A modification for binary mixed problem is presented in
with a step length and also is used to control the step length;
Bacanin and Tuba (2014). The modification is done using a new
hence adding additional scaling parameter may not be necessary.
updating formula given by
That is because it is possible to control the step length based on
1 and the random vector generated by Levy distribution. Additionally,
xi ( k ) = round -------------------------------------------------
- 0.06 (17)
1 + e x (k) + rand(x (k) x (k ))
i i j since is made to decrease through iteration, its effect may not
Another modification in this category is done in Marichelvam
et al. (2014) and de Oliveira and Schirru (2011). This modification
is based on a concept of random key (Bean, 1994). The method
uses a mapping of a random number space, [0, 1], to the integer
problem space. In other words, it encodes and decodes a solution
with real numbers and decimals between 0 and 1.
In addition to modifying the algorithm to suit the problem,
different additional modifications are proposed to increase the
effectiveness of the algorithm. In Gandomi et al. (2011); Bacanin
et al. (2013) and Baghlani et al. (2014), the random movement
step length was made adaptive as a function of the iteration Fig. 2. The Graph of Based on Eq. (22), with max = 2.5 and
number, Itr, using Eq. (18), for a new algorithm parameter , min = 0.1

538 KSCE Journal of Civil Engineering

 Firefly Algorithm for Discrete Optimization Problems: A Survey
Fig. 3. Different Modifications Proposed for ; is Based on Eq.
(18) for Different Values of , ' is Based on Eq. (20), '' is
Based on Eq. (21) and ''' is based on Eq. (19) (The initial
and Final Values are Fixed by, for All the Cases, 0 = max =
2.5 and min = 0.1
be seen due to this scaling parameter. The graph of different
modification of through iteration is demonstrated in Fig. 3.
3.2 Updating in the Discrete Space
In Durkota (2011) and Poursalehi et al. (2013), the standard
firefly algorithm is modified for loading pattern enhancement.
The generation of random solutions is done using random
permutation and the distance between fireflies are measured
using hamming distance. Hamming distance of two vectors xi
and xj is given by d = |H| where H is number of indices, k, for
which xi ( k ) xj ( k ) . The updating process is separated and made
sequentially; first the -step, a move due to the attraction and
then -step, a move due to the random movement. In the -step,
first same entries with same index for both fireflies, xi and xj will
be preserved and for the other components an entry from xj will
1
be copied if rand < , where = ---------------2 . If rand , then the
1 + rd
entry from xi will be used. After moving xi using the -step the
random movement or the -step will be used to update xi using xi
= round(xi + (rand( )-0.5)) with a swapping mechanism to
preserve feasibility. A similar approach, but different way of
computing is proposed in Ishikawa and Matsushita (2013). It is
computed based on the familiarity degree P, which is a random N
by N vector initially and updated by
1 filled using inversion mutation m times to generate m new
Pij = Pij + --------------------------------
ranki rankj
max { P } P
2 best N solutions will be selected to pass to the next iteration.
k ik ij
------------------------------------------
and max
k {P ik
}
=e
In addition to making the algorithm suitable for non-continuous
variables, in Ishikawa and Matsushita (2013), the randomness
parameter is made adaptive using Eq. (25), which is a function
of the iteration number and the dimension of the problem.
Itr
= n -------------------- n
MaxIter
As can be seen from Fig. (4), it decreases with iteration.
The modification proposed in Poursalehi et al. (2015) is
Vol. 21, No. 2 / February 2017
Fig. 4. The Step Length According to Eq. (25) for Dimensions of
n = 2, 3, 4, 5, 6, 7, 8, 9 and for 100 Iterations
similar with the modification proposed in Durkota (2011) and
Poursalehi et al. (2013). The only difference is in the updating
mechanism, which is done based on a given probability
0.5Itr
parameter , where = 0.5 + -----------------
MaxItr
- . It is the probability for a
firefly xi to follow another brighter firefly xj. However, if rand >
then firefly xi will move towards the brightest firefly of the
fireflies.
In Mamaghani and Hajizadeh (2014), after a randomly integer
coded initial solutions are generated, hamming distance is used
to compute the distance, r, between two solutions. Then a
random number R will be generated between 1 and r, and R
swaps will be done from the brighter firefly. A similar
modification is proposed in Osaba et al. (2016a). In Osaba et al.
(2016a), after computing the humming distance, r, of two
fireflies a random number R between 2 and rItr will be used
instead of r as in Mamaghani and Hajizadeh (2014). Only
improving solutions will be accepted.
For travel salesman problem, firefly algorithm has also been
modified (Jati and Suyanto, 2011). The distance between two
solution is computed using r = 10 A--n- , where A is number of
different arcs and n is the number of cities. A firefly i moves
towards brighter firefly j, m times where m is a new parameter to
determine the number of moves towards brighter fireflies.
Inversion mutation is used to preserve feasibility. That is an
initial chromosome is selected randomly and other entries will be
(23)
solutions. Once all the N solutions are moved m times then the
Another modification is proposed in Li et al. (2016a). The
updating for a solution xi towards a brighter firefly xj is given by
(24)
2
S ( k ), rand 0.5 < 0 e r
xi ( k ) = i (26)
xi ( k ), otherwise
xj ( k ), xj ( k ) xi ( k )
where Si ( k ) = , for each dimension k. In
0, otherwise
(25) addition, a firefly xi will be affected only by fireflies in its visual
range. That means for a firefly xi to move towards another firefly
xj, firefly xj should be brighter and also should be in xi's visual
range. The visual range, dv, is computed using Eq. (27).
539
 Surafel Luleseged Tilahun and Jean Medard T. Ngnotchouye
Fig. 5. The Visual Range of a Firefly with dvmax = 3 and dvmin = 0.2
for 100 Iterations
3 ( dvmax dvmin )Itr 2MaxItr
-------------------------------------------, Itr < --------------------
dv = 2 ( MaxItr 1 ) 3 (27)
dv , otherwise
max
As can be seen from Fig. 5, the visual range increases until it
reaches to the maximum value. This means that initially a
brighter firefly affects solution near its location where in latter
stages it can affect any firefly in the maximum allowed visual
range.
A modification based on the problem property for supplier
selection problem is also proposed in Kota (2012). In addition,
firefly algorithm has been modified and used for knapsack
problem, using priority based encoding (Baykasoglu and Ozsoydan,
2014). The discretization is done based on the problem property.
In addition, a firefly i will move towards a brighter firefly j if the
condition given in Eq. (28) is true.
mode (Itr 1, MaxItr )
-----------------------------------------------------
MaxItr
rand < ranki
The additional condition slowly vanishes as the iteration increases.
Furthermore, is modified for computational reason using
0
- , for a very small number ( is used to omit the
= ----------
+r
singularity case). A similar modification is used in Baykasoglu
and Ozsoydan (2015). In addition to the discretization done in
Baykasoglu and Ozsoydan (2014), the authors in (Baykasoglu
and Ozsoydan, 2015) proposed two additional moves after the
updates. The first one is a random flight by 10% of the top
fireflies with 0.45 probability. The move will be accepted only if
it is improving. The second is a local search of the brighter
firefly. After 10% of the iterations it will do a local search and
the update will be accepted if it is improving.
A recent modification in this category is presented in Osaba et
al. (2016b). The modification is done to make the algorithm
suitable for a vehicle routing problem modeling a newspaper
distribution system with recycling policy. Hamming distance is
used to compute the distance between fireflies. Based on the
distance between the solutions the number of entries to relocate
will be determined followed by a relocation step. Similarly,
based on the problem property a similar modification is proposed
in Singh et al. (2015). The modification is done for the disk
scheduling problem using edge based movement.
For a binary encoded bi-cluster problem, a binary firefly
algorithm is proposed in Thangavel and Rathipriya (2014). The
modification is proposed by using the crossover operator to
generate new fireflies based on the problem based formulation of
mining correlated bi-cluster problem. A binary firefly algorithm
is proposed in Zhang et al. (2016a). The modification is done by
changing the bit of the solution by the bits from the brighter
firefly based on a probability. The changing probability is
computed based on the attraction of the brighter firefly.
4. Summary and Discussion
4.1 Summary
Firefly algorithm has become popular in engineering applications
due to different reasons including its easiness for implementation
and suitability for parallelization. It has been modified to suit
problems arising from reality. Many applications involve discrete
variables and consequently different studies reported modified
versions of the algorithm for optimization problems with non-
continuous variable. These modifications can basically be
classified in two ways.
Based on the updating space, we can have two categories of
modifications, namely updating in continuous and discrete
space. Updating in continuous space refers to the updating of a
solution using the standard updating mechanism proposed for
continuous problems and the solution will later converted to
discrete values. Most applications in this category are done for
problems with binary variables. Once the update is done, the
resulting solution will be mapped to the interval [0, 1] and
different mechanisms based on probability will be used to
(28) change the value to either 0 or 1. These mechanisms includes
using sigmoid (Sayadi et al., 2010; Palit et al., 2011; Yang et al.,
2013; Sayadi et al., 2013; Rajalakshmi et al., 2015; Chhikara and
Singh, 2015; Li et al., 2015; Crawford et al., 2014a; Costa et al.,
2014; Liu et al., 2014; Far-hoodnea et al., 2014a,b) and tan
hyperbolic (Chandrasekaran and Simon, 2012; Chandrasekaran
et al., 2013; Crawford et al., 2014b,b; Liu et al., 2014) functions.
In the case of integer values rounding after using the standard
updating mechanism (Bacanin et al., 2013; Baghlani et al., 2014)
and the concept of random key (de Oliveira and Schirru, 2011;
Marichelvam et al., 2014) have been used. Rounding after
updating in the continuous space has also been used for mixed
problems (Gandomi et al., 2011; Bacanin and Tuba, 2014).
For the modifications where the updating is done in the
discrete space, no need of discretization is needed as the updated
solutions remain being a discrete value. In Durkota (2011);
Poursalehi et al. (2013); Ishikawa and Matsushita (2013) and
Poursalehi et al. (2015) hamming distance is used with a sequential
updating mechanism where the attraction term is applied first.
Hamming distance has also been used in Mamaghani and
Hajizadeh (2014) and Osaba et al. (2016a) with the updating
mechanism of copying entries from brighter fireflies. Similar
modification for a traveling salesman problem is proposed and
used in Jati and Suyanto (2011). By comparing the random step
540 KSCE Journal of Civil Engineering
 Firefly Algorithm for Discrete Optimization Problems: A Survey

and attraction step lengths Li et al. (2016a) proposed a modified
firefly algorithm. Based on the properties of the problem,
different modifications are also proposed, for integer valued
problem (Kota, 2012; Osaba et al., 2016b; Singh et al., 2015) as
well as for binary problems (Baykasoglu and Ozsoydan, 2015,
2014). In Zhang et al. (2016a) bit level changing of entries from
the brighter firefly is also proposed for binary optimization
problems. A crossover operator is also used in binary coded
problem (Thangavel and Rathipriya, 2014).
The other possibly classification is based on the nature of the
problem variables. Hence, we can have three types under this
category, namely binary, integer and mixed, i.e. where mixed
refers to problems with variables whose components takes
discrete as well as continuous values. Firefly algorithm has been
modified to suit the need of solving binary problems (Sayadi et
al., 2010; Palit et al., 2011; Yang et al., 2013; Sayadi et al., 2013;
Rajalakshmi et al., 2015; Chhikara and Singh, 2015; Li et al.,
2015; Crawford et al., 2014a; Costa et al., 2014; Liu et al., 2014;
Farhoodnea et al., 2014a,b; Chandrasekaran and Simon, 2012;
Chan-drasekaran et al., 2013; Crawford et al., 2014b,b; Liu et
al., 2014; Zhang et al., 2016a; Thangavel and Rathipriya, 2014).
Most of the modifications applied to make firefly algorithm for
binary problems is by converting the real value to be in the
interval between zero and one, using sigmoid or tan hyperbolic
function, and different approaches can be used to convert it to
either 1 or 0. In addition some modifications are proposed based
on the formulated problem property.
Firefly algorithm has also been successful used for integer
valued problems (Ba-canin et al., 2013; Baghlani et al., 2014; de
Oliveira and Schirru, 2011; Marichelvam et al., 2014; Poursalehi
et al., 2013; Durkota, 2011; Ishikawa and Matsushita, 2013;
Poursalehi et al., 2015; Mamaghani and Hajizadeh, 2014; Osaba
et al., 2016a; Jati and Suyanto, 2011; Li et al., 2016a; Kota,
2012; Osaba et al., 2016b; Singh et al., 2015). For the algorithm
to suit integer valued optimization problems, updating mechanisms
like sequential updating, copying entries, priority based encoding
as well as problem specific updating mechanisms are used. For
the third type, which is mixed problem case, there are a number
of modifications (Gandomi et al., 2011; Bacanin and Tuba,
2014). However, it can be seen that a combination of one of the
modifications discussed above can be coined and used with the
standard firefly algorithm.
The modification based on these classifications is summarized
in Table 2.
Some of the modification is a generalized one where it can be
applied to other class of problems. However, in some case, the
modifications are problem dependent and needs additional
modifications to be applied to other problem domains, like Jati
and Suyanto (2011); Kota (2012); Baykasoglu and Ozsoydan
(2015, 2014) and Osaba et al. (2016b).

Table 2. Summary of Modified Versions of Firefly Algorithm foR Discrete Optimization Problems
Updating space
Continuous Discrete
Sayadi et al. (2010) Baykasoglu and Ozsoydan (2015)
Problem based
Palit et al. (2011) Baykasoglu and Ozsoydan (2014)
Yang et al. (2013) Entry replacing Zhang et al. (2016a)
Sayadi et al. (2013)
Crawford et al. (2014a)
Sigmoid Costa et al. (2014)
Liu et al. (2014)
Farhoodnea et al. (2014a,b)
Binary Rajalakshmi et al. (2015)
Chhikara and Singh (2015)
Crossover Thangavel and Rathipriya (2014)
Li et al. (2015)
Chandrasekaran and Simon (2012)
Chandrasekaran et al. (2013)
Tan hyperbolic Crawford et al. (2014b)
Crawford et al. (2014b)
Liu et al. (2014)
Bacanin et al. (2013) Poursalehi et al. (2013)
Rounding
Baghlani et al. (2014) Durkota (2011)
Sequential
Ishikawa and Matsushita (2013)
Poursalehi et al. (2015)
Mamaghani and Hajizadeh (2014)
Copying
Integer Osaba et al. (2016a)
de Oliveira and Schirru (2011)
Random key Systematic Li et al. (2016a)
Marichelvam et al. (2014)
Jati and Suyanto (2011)
Kota (2012)
Problem based
Singh et al. (2015)
Osaba et al. (2016b)
Gandomi et al. (2011)
Rounding
Mixed Bacanin and Tuba (2014)
Problem based Bacanin and Tuba (2014)

Vol. 21, No. 2 / February 2017 541

 Surafel Luleseged Tilahun and Jean Medard T. Ngnotchouye
4.2 Discussion
The two popular approach of converting a real value to a
binary is based on either sigmoid or tan hyperbolic function.
Both functions have same use of changing a real value to be in
the interval [0, 1]. The difference between them is that sigmoid
function is an S shaped and tan hyperbolic is a V shaped
transformation functions. Hence, for two values x1 and x2 if x1 <
x2, the x2 has higher probability of becoming one in sigmoid
function. However, in the case of tan hyperbolic, for the two
numbers, x1 and x2, x2 will have higher probability of becoming 1
if |x1| < |x2|. That means the sigmoid function tends to map values
to one and zero based on a particular value x'' (also based on
S(x'')), as shown in Fig. 6(a), whereas in the case of the tan
hyperbolic, it tends to map the value to zero and one based on the
value of V'' as demonstrated in Fig. 6(b).
Exploring the conditions on when to choose the V-shaped
mapping over the S-shaped or the other way around needs
further study. That can possibly be done by analyzing the search
space and values of x and its mappings. This may help practitioners
to choose one of these mapping while updating on continuous
space while using of firefly algorithm for a binary or mixed
problem.
Using the standard firefly algorithm and converting the results
may be easier compared to going through conversion and
working in the discrete space. However, it also has its own
drawback. The solution found for the continuous version of the
problem can be different from the discrete version of the
problem. For example; consider the function given in Fig. 7, the
solution for the continuous case is around x = 5.5 and the nearest
integer is x = 5 and x = 6. However, for the integer valued
problem the solution is at x = 1.
Updating in the discrete space usually is done based on the
coding used, which is based on the properties of the problems.
The distance between two fireflies are measured based on the
difference in the entries or based on the difference sequence of
Fig. 6. (a) Vertical, (b) Horizontal Cuts for Probabilities of Mapping
Values to 1 for Sigmoid (S-shaped) and Tan Hyperbolic (V-
shaped) Mappings
542
Fig. 7. Discrete Versus Continuous Space Search where the Black
Circles Representing the Integer Functional Value and the
Red Graph is the Corresponding Continuous Function
the entries. In the updating stage different swapping approaches
are used. A research on which approach is better has not been
studied and can be listed as one of possible future works.
Modifying an algorithm to suit a given problem has also been
done and it is often effective for that particular class of problem.
Another possible future work is generalizing these approaches so
that a given modification in one problem domain can be used in
another.
Apart from making the algorithm suitable for problems with
non-continuous variables the algorithm parameters are modified.
Making the randomness step length decreasing with iteration is
a good idea in order to archive quality solutions. In addition it has
also been modified based on the problem dimension, however,
appropriate scaling parameter needs to accompany the modification.
Perhaps it is also a promising issue to explore the adaptive step
length which varies based on the performance and also which
can possibly be increasing and also decreasing based on its
current status, as one of the possible future works.
Another possible future work is to merge the search mechanism
both on the discrete as well as the continuous space. Perhaps the
advantage of the search on continuous space and discrete space
merged together can produce an efficient approach.
5. Conclusions
Firefly algorithm is originally proposed for continuous problems.
However, due to the existence of discrete optimization problem
in a wide range of applications, it has been modified and used for
discrete problems in different studies. The modification basically
can be categorized into two, based on the updating space used.
The first category is where the updating mechanism is done on
the continuous space and the result is converted to the discrete
values. Popular ways in this category is using sigmoid and tan
hyperbolic function to convert the real values to binary values
based on probabilistic conditions. Rounding the resulting real
solution to the nearest integer values is also another commonly
used approach for integer as well as mixed problems in this
category. The second category is when the updating is done in
the discrete space. Systematic updating mechanisms, like swapping,
sequential updating and copying entries, are used mainly based
on the behavior of the problem variable. Apart from these,
KSCE Journal of Civil Engineering
 Firefly Algorithm for Discrete Optimization Problems: A Survey
parameter modifications have also been proposed to boost the
performance of the algorithm. Adaptive step length is one of the
main centers of interest in this regard. A number of possible
research directions have also been discussed.
Acknowledgements
We would like to thank Dr. Nawaf Hamadneh, for the useful
discussions and support in preparing this paper.
References
Abdelaziz, A., Mekhamer, S., Badr, M., and Algabalawy, M. (2015).
The firefly metaheuristic algorithms: Developments and applications.
International Electrical Engineering Journal (IEEJ), Vol. 6, No. 7,
pp. 1945-1952.
Babaoglu, O., Binci, T., Jelasity, M., and Montresor, A. (2007). Firefly-
inspired heartbeat synchronization in overlay networks. First
International Conference on Self-Adaptive and Self-Organizing Systems
(SASO 2007), pp. 77-86. IEEE.
Bacanin, N. and Tuba, M. (2014). Firefly algorithm for cardinality
constrained mean-variance portfolio optimization problem with
entropy diversity constraint. The Scientific World Journal, Vol.
2014, 16 pages, DOI: 10.1155/2014/721521.
Bacanin, N., Brajevic, I., and Tuba, M. (2013). Firefly algorithm applied to
integer programming problems. Recent Adv Math, pp. 143-148.
Baghlani, A., Makiabadi, M. H., and Sarcheshmehpour, M. (2014).
Discrete optimum design of truss structures by an improved firefly
algorithm. Advances in Structural Engineering, Vol. 17, No. 10, pp.
1517-1530, DOI: 10.1260/1369-4332.17.10.1517.
Baykasoglu, A. and Ozsoydan, F. B. (2014). An improved firefly
algorithm for solving dynamic multidimensional knapsack problems.
Expert Systems with Applications, Vol. 41, No. 8, pp. 3712-3725,
DOI: 10.1016/j.eswa.2013.11.040.
Baykasoglu, A. and Ozsoydan, F. B. (2015). Adaptive firefly algorithm
with chaos for mechanical design optimization problems. Applied
Soft Computing, Vol. 36, No. C, pp. 152-164, DOI: 10.1016/
j.asoc.2015. 06.056.
Bean, J. C. (1994). Genetic algorithms and random keys for sequencing
and optimization. ORSA Journal on Computing, Vol. 6, No. 2,
pp. 154-160.
Chandrasekaran, K. and Simon, S. P. (2012). Network and reliability
constrained unit commitment problem using binary real coded
firefly algorithm. Electrical Power and Energy Systems, Vol. 43,
No. 1, pp. 921-932, DOI: 10.1016/j.ijepes.2012.06.004.
Chandrasekaran, K., Simon, S. P., and Padhy, N. P. (2013). Binary real
coded firefly algorithm for solving unit commitment problem.
Information Sciences, Vol. 249, pp. 67-84, DOI: 10.1016/j.ins.
2013.06.022.
Chhikara, R. R. and Singh, L. (2015). An improved discrete firefly and
t-test based algorithm for blind image steganalysis. 2015 6th
International Conference on Intelligent Systems, Modelling and
Simulation, pp. 58-63. IEEE.
Costa, M. F. P., Rocha, A. M. A. C., Francisco, R. B., and Fernandes, E.
M. G. P. (2014). Heuristic-based firefly algorithm for bound
constrained nonlinear binary optimization. Advances in Operations
Research, Vol. 2014 , 12 pages.
Crawford, B., Soto, R., Olivares-Suarez, M., and Paredes, F. (2014b).
A binary firefly algorithm for the set covering problem. R. Silhavy
Vol. 21, No. 2 / February 2017
et al., editor. Modern Trends and Techniques in Computer Science,
Advances in Intelligent Systems and Computing 285. Switzerland:
Springer, pp. 65-73.
Crawford, B., Soto, R., Olivares-Suarez, M., Palma, W., Paredes, F.,
Olguin, E., and Norero, E. (2014a). A binary coded firefly algorithm
that solves the set covering problem. Romanian Journal of
Information Science and Technology, Vol. 17, No. 3, pp. 252-264.
de Oliveira, I. M. S. and Schirru, R. (2011). A modified firefly algorithm
applied to the nuclear reload problem of a pressurized water
reactor, International Nuclear Atlantic Conference - INAC 2011;
October 24-28, 2011; Belo Horizonte, MG, Brazil.
Dorigo, M. and Stutzle, T. (2004). Ant colony optimization, Scituate,
MA: Bradford Company.
Durkota, K. (2011). Implementation of discrete firefly algorithm for
the qap problem within the seage framework. Bachelor thesis,
Czech Technical University, Prague, pp. 52.
Erdal, F. (2016). A firefly algorithm for optimum design of new-
generation beams. Engineering Optimization, pp. 1-17.
Farhoodnea, M., Mohamed, A., Shareef, H., and Zayandehroodi, H.
(2014a). Optimum placement of active power conditioner in
distribution systems using improved discrete firefly algorithm for
power quality enhancement. Applied Soft Computing, Vol. 23, pp.
249-258.
Farhoodnea, M., Mohamed, A., Shareef, H., and Zayandehroodi, H.
(2014b). Optimum placement of active power conditioners by a
dynamic discrete firefly algorithm to mitigate the negative power
quality effects of renewable energy-based generators. Electrical
Power and Energy Systems, Vol. 61, pp. 305-317, DOI: 10.1016/
j.ijepes.2014.03.062.
Fister, I., Perc, M., and Kamal, S. M. (2015). A review of chaos-based
firefly algorithms: perspectives and research challenges. Applied
Mathematics and Computation, Vol. 252, pp. 155-165, DOI:
10.1016/j.amc.2014.12.006.
Fister, I., Yang, X.-S., and Brest, J. (2013a). A comprehensive review
of firefly algorithms. Swarm and Evolutionary Computation, Vol.
13, pp. 34-46, DOI: 10.1016/j.swevo.2013.06.001.
Fister, I., Yang, X.-S., Brest, J., and Fister Jr, I. (2013b). Memetic self-
adaptive firefly algorithm. Swarm Intelligence and Bio-inspired
Computation: Theory and Applications, pp. 73-102.
Fister, I., Yang, X.-S., Fister, D., and Fister Jr, I. (2014). Firefly
algorithm: a brief review of the expanding literature. Cuckoo
Search and Firefly Algorithm, pp. 347-360. Springer, DOI: 10.1007/
978-3-319-02141-6-17.
Francisco, R. B., Costa, M. F. P., and Rocha, A. M. A. (2015). A firefly
dynamic penalty approach for solving engineering design problems.
AIP Conference Proceedings (pp. 1400101), AIP Conference
Proceedings Vol. 1648.
Gandomi, A. H., Yang, X.-S., and Alavi, A. H. (2011). Mixed variable
structural optimization using firefly algorithm. Computers and
Structures, Vol. 89, Nos. 23-24, pp. 2325-2336, DOI: 10.1016/
j.compstruc. 2011.08.002.
Hamadneh, N., Tilahun, S. L., Sathasivam, S., and Ong, H. C. (2012).
Learning logic programming in radial basis function network via
genetic algorithm. Journal of Applied Sciences, Vol. 12, No. 9, pp.
840-847, DOI: 10.3923/jas.2012.840.847.
Huang, S.-J., Liu, X.-Z., Su, W.-F., and Yang, S.-H. (2013). Application of
hybrid firefly algorithm for sheath loss reduction of underground
transmission systems. IEEE Transactions on Power Delivery, Vol.
28, No. 4, pp. 2085-2092.
Ishikawa, M. and Matsushita, H. (2013). Discrete firefly algorithm
543
 Surafel Luleseged Tilahun and Jean Medard T. Ngnotchouye
using familiarity degree. In: 2013 Shikoku-Section Joint Convention
Record of the Institute of Electrical and Related Engineers,
Tokushima, pp. 1-1.
Jati, G. K. and Suyanto (2011). Evolutionary discrete firefly algorithm
for travelling salesman problem. A. Bouchachia, editor. ICAIS
2011, LNAI 6943. Springer-Verlag, pp. 393-403, DOI: 10.1007/978-
3-642-23857-4-38.
Johari, N. F., Zain, A. M., Noorfa, M. H., and Udin, A. (2013). Firefly
algorithm for optimization problem. Applied Mechanics and
Materials, pp. 512-517. Trans Tech Publ Vol. 421, DOI: 10.4028/
www.scientific.net/AMM.421.512.
Joshi, R. V. (2013). Optimization techniques for transportation problems
of three variables. IOSR Journal of Mathematics, Vol. 9, No. 1, pp.
46-50.
Kazemzadeh-Parsi, M. (2014). A modified firefly algorithm for engineering
design optimization problems. Iranian Journal of Science and
Technology, Vol. 38, No. M2, pp. 403-421.
Kazemzadeh-Parsi, M. J., Daneshmand, F., Ahmadfard, M. A., Adamowski,
J., and Martel, R. (2015). Optimal groundwater remediation design of
pump and treat systems via a simulationoptimization approach and
firefly algorithm. Engineering Optimization, Vol. 47, No. 1, pp. 1-
17.
Kennedy, J. and Eberhart, R. (1995). Particle swarm optimization.
Neural Networks, 1995. Proceedings, IEEE International Conference
on Vol. 4, pp. 1942-1948, DOI: 10.1109/ICNN.1995.488968.
Khan, W. A., Hamadneh, N. N., Tilahun, S. L., and Ngnotchouye, J. M.
T. (2016). A review and comparative study of firefly algorithm and
its modified versions. O. Baskan (Ed.), Optimization Algorithms-
Methods and Applications, pp. 281-313, InTech, DOI: 10.5772/
62472.
Kota, L. (2012). Optimization of the supplier selection problem using
discrete firefly algorithm. Advanced Logistic Systems, Vol. 6, No. 1,
pp. 117-126.
Li, M., Zhang, Y., B. Zeng, H. Z., and Liu, J. (2016a). The modified
firefly algorithm considering fireflies visual range and its application in
assembly sequences planning. Int J Adv Manuf Technol, Vol. 82,
No. 5, pp. 1381-1403, DOI: 10.1007/s00170-015-7457-8.
Li, X.-K., Gu, C.-H., Yang, Z.-P., and Chang, Y.-H. (2015). Virtual ma-
chine placement strategy based on discrete firefly algorithm in cloud
environments. Wavelet Active Media Technology and Information
Processing (ICCWAMTIP), 2015 12th International Computer
Conference on, pp. 61-66. IEEE.
Li, Y., Yu, Y., and Zhao, J. (2016b). Construction system reliability
analysis based on improved firefly algorithm. The Open Civil
Engineering Journal, Vol. 10, pp. 189-199.
Liu, J. J., Hou, L., and Wang, X. Y. (2014). A discrete firefly algorithm
for the scaffolding modular construction in mega projects, The 31st
International Symposium on Automation and Robotics in
Construction and Mining (ISARC 2014).
Lucia, A. and Xu, J. (1990). Chemical process optimization using newton-
like methods. Computers chem. Engng., Vol. 14, No. 2, pp. 119-
138.
Mamaghani, A. S. and Hajizadeh, M. (2014). Software modularization
using the modified firefly algorithm. 8th Malaysian Software
Engineering Conference (MySEC), IEEE, pp. 321-324, DOI:
10.1109/MySec.2014.6986037.
Marichelvam, M. K., Prabaharan, T., and Yang, X.-S. (2014). A discrete
firefly algorithm for the multi-objective hybrid flowshop scheduling
problems. IEEE Transactions on Evolutionary Computation, Vol. 18,
No. 2, pp. 301-305, DOI: 10.1109/TEVC.2013.2240304.
544
Miao, Y. (2014). Resource scheduling simulation design of firefly algorithm
based on chaos optimization in cloud computing. International
Journal of Grid Distribution Computing, Vol. 7, No. 6, pp. 221-228,
DOI: 10.14257/ijgdc.2014.7.6.18.
Mohanty, D. K. (2016). Application of firefly algorithm for design
optimization of a shell and tube heat exchanger from economic point
of view. International Journal of Thermal Sciences, Vol. 102, pp.
228-238.
Ong, H. C. and Tilahun, S. L. (2011). Integration fuzzy preference in
genetic algorithm to solve multiobjective optimization problems.
Far East Math. Sci., Vol. 55, No. 2, pp. 165-179.
Osaba, E., Carballedo, R., Yang, X.-S., and Diaz, F. (2016a). An
evolutionary discrete firefly algorithm with novel operators for
solving the vehicle routing problem with time windows. In Nature-
Inspired Computation in Engineering, pp. 21-41. Springer.
Osaba, E., Yang, X.-S., Diaz, F., Onieva, E., Masegosa, A. D., and
Perallos, A. (2016b). A discrete firefly algorithm to solve a rich
vehicle routing problem modelling a newspaper distribution system
with recycling policy. Soft Computing, pp. 1-14.
Palit, S., Sinha, S. N., Molla, M. A., Khanra, A., and Kule, M. (2011).
A cryptanalytic attack on the knapsack cryptosystem using binary
firefly algorithm. International Conference on Computer &
Communication Technology (IC- CCT), IEEE, pp. 428-432.
Parolo, G., Ferrarini, A., and Rossi, G. (2009). Optimization of tourism
impacts within protected areas by means of genetic algorithms.
Ecological Modelling, Vol. 220, No. 8, pp. 1138-1147, DOI:
10.1016/j.ecolmodel.2009.01.012.
Pike, J., Bogich, T., Elwood, S., Finnoff, D. C., and Daszak, P. (2014).
Economic optimization of a global strategy to address the
pandemic threat. Proceeding of the National Academy of Science,
Vol. 111, No. 52, pp. 18519-18523, DOI: 10.1073/pnas.1412661112.
Poursalehi, N., Zolfaghari, A., and Minuchehr, A. (2013). Multi-
objective loading pattern enhancement of PWR based on the
discrete firefly algorithm. Annals of Nuclear Energy, Vol. 57, pp.
151-163, DOI: 10.1016/j.anucene.2013.01.043.
Poursalehi, N., Zolfaghari, A., and Minuchehr, A. (2015). A novel
optimization method, effective discrete firefly algorithm, for fuel
reload design of nuclear reactors. Annals of Nuclear Energy, Vol.
81, pp. 263-275, DOI: 10.1016/j.anucene.2015.02.047.
Rahmani, A. and MirHassani, S. (2014). A hybrid firefly-genetic
algorithm for the capacitated facility location problem. Information
Sciences, Vol. 283, pp. 70-78.
Rajalakshmi, N., Subramanian, D. P., and Thamizhavel, K. (2015).
Performance enhancement of radial distributed system with
distributed generators by reconfiguration using binary firefly algorithm.
J. Inst. Eng. India Ser. B., Vol. 96, No. 1, pp. 91-99, DOI: 10.1007/
s40031-014-0126-8.
Ram, G., Mandal, D., Kar, R., and Ghoshal, S. P. (2014). Design of
non-uniform circular antenna arrays using firefly algorithm for side
lobe level reduction. International Journal of Electrical, Electronic
Science and Engineering, Vol. 8, No. 1, pp. 40-45.
Sadjadi, S. J., Ashtiani, M. G., Ramezanian, R., and Makui, A. (2016).
A firefly algorithm for solving competitive location-design problem: A
case study. Journal of Industrial Engineering International, pp. 1-11.
Sayadi, M. K., Hafezalkotob, A., and Naini, S. G. J. (2013). Firefly-
inspired algorithm for discrete optimization problems: An application to
manufacturing cell formation. Journal of Manufacturing Systems,
Vol. 32, No. 1, pp. 78-84, DOI: 10.1016/j.jmsy.2012.06.004.
Sayadi, M. K., Ramezanian, R., and Ghaffari-Nasab, N. (2010). A discrete
firefly meta-heuristic with local search for makespan minimization
KSCE Journal of Civil Engineering
 Firefly Algorithm for Discrete Optimization Problems: A Survey
in permutation flow shop scheduling problems. International
Journal of Industrial Engineering Computations, Vol. 1, No. 1, pp.
1-10, DOI: 10.5267/j.ijiec.2010.01.001.
Setiadi, H. and Jones, K. O. (2016). Power system design using firefly
algorithm for dynamic stability enhancement. Indonesian Journal
of Electrical Engineering and Computer Science, Vol. 1, No. 3, pp.
446-455.
Sheikholeslami, R., Khalili, B. G., Sadollah, A., and Kim, J. H. (2015).
Optimization of reinforced concrete retaining walls via hybrid
firefly algorithm with upper bound strategy. KSCE Journal of Civil
Engineering, pp. 1-11.
Shimoyama, K., Seo, K., Nishiwaki, T., Jeong, S., and Obayashi, S.
(2011). Design optimization of a sport shoe sole structure by
evolutionary computation and finite element method analysis.
Proceedings of the Institution of Mechanical Engineers. Part P:
Journal of Sports Engineering and Technology, Vol. 225, No. 4, pp.
179-188.
Singh, A., Thapar, S., Bhatia, A., Singh, S., and Goyal, R. (2015). Disk
scheduling using a customized discrete firefly algorithm. Cogent
Engineering, Vol. 2, No. 1, pp. 1-12.
Talatahari, S., Gandomi, A. H., and Yun, G. J. (2014). Optimum design
of tower structures using firefly algorithm. The Structural Design
of Tall and Special Buildings, Vol. 23, No. 5, pp. 350-361.
Thangavel, K. and Rathipriya, R. (2014). Mining correlated bicluster
from web usage data using discrete firefly algorithm based biclustering
approach. International Journal of Mathematical, Computational,
Physical and Quantum Engineering, Vol. 8, pp. 706-711.
Tilahun, S. L. and Asfaw, A. (2012). Modeling the expansion of
Prosopis juliflora and determining its optimum utilization rate to
control its invasion in afar regional state of Ethiopia. International
Journal of Applied Mathematical Research, Vol. 1, No. 4, pp. 726-
743, DOI: 10.14419/ijamr.v1i4.200.
Tilahun, S. L. and Ong, H. C. (2012a). Bus timetabling as a fuzzy
multiobjective optimization problem using preference based genetic
algorithm. Promet Traffic & Transportation, Vol. 24, No. 3, pp.
183-191, DOI: 10.7307/ptt.v24i3.311.
Tilahun, S. L. and Ong, H. C. (2012b). Fuzzy preference of multiple
decision- makers in solving multi-objective optimisation problems
using genetic algorithm. Maejo International Journal of Science
and Technology, Vol. 6, No. 2, pp. 224-237.
Tilahun, S. L. and Ong, H. C. (2013). Vector optimisation using fuzzy
Vol. 21, No. 2 / February 2017 545
preference in evolutionary strategy based firefly algorithm.
International Journal of Operational Research, Vol. 16, No. 1, pp.
81-95.
Tilahun, S. L. and Ong, H. C. (2015). Prey predator algorithm: A new
metaheuristic optimization algorithm. International Journal of In-
formation Technology & Decision Making, Vol. 14, No. 6, pp. 1331-
1352, DOI: 10.1142/S021962201450031X.
Tilahun, S. L., Kassa, S. M., and Ong, H. C. (2012). A new algorithm
for multi-level optimization problems using evolutionary strategy,
inspired by natural adaptation. In Pacific Rim International Conference
on Artificial Intelligence, pp. 577-588. Springer.
Upadhyay, P., Kar, R., Mandal, D., and Ghoshal, S. (2016). A new
design method based on firefly algorithm for IIR system identification
problem. Journal of King Saud University-Engineering Sciences,
Vol. 28, No. 2, pp. 174-198.
Yang, X.-S. (2008). Nature-Inspired Metaheuristic Algorithm, 2nd ed..
England: Luniver press.
Yang, X.-S. (2009). Firefly algorithms for multimodal optimization.
In Inter-national Symposium on Stochastic Algorithms, pp. 169-178.
Springer.
Yang, X.-S. (2010). Firefly algorithm, stochastic test functions and design
optimisation. International Journal of Bio-Inspired Computation, Vol.
2, No. 2, pp. 78-84.
Yang, X.-S. and He, X. (2013). Firefly algorithm: recent advances and
applications. International Journal of Swarm Intelligence, Vol. 1,
No. 1, pp. 36-50, DOI: 10.1504/IJSI.2013.055801.
Yang, Y., Mao, Y., Yang, P., and Jiang, Y. (2013). The unit commitment
problem based on an improved firefly and particle swarm optimization
hybrid algorithm. In: Chinese Automation Congress (CAC), IEEE,
7-8 Nov. 2013; Changsha, pp. 718-722.
Yuce, B., Packianather, M., Mastrocinque, E., Pham, D., and Lambiase,
A. (2013). Honey bees inspired optimization method: The bees
algorithm. Insects, Vol. 4, No. 4, pp. 646-662, DOI: 10.3390/
insects4040646.
Zhang, J., Gao, B., Chai, H., Ma, Z., and Yang, G. (2016a). Identification of
DNA- binding proteins using multi-features fusion and binary firefly
optimization algorithm. BMC bioinformatics, Vol. 17, No. 323, pp.
1-12.
Zhang, L., Liu, L., Yang, X.-S., and Dai, Y. (2016b). A novel hybrid
firefly algorithm for global optimization, PLoS One, 11, e0163230.