# Firefly Algorithm: A Brief Review

of the Expanding Literature
Iztok Fister, Xin-She Yang, Dušan Fister and Iztok Fister Jr.

Abstract Firefly algorithm (FA) was developed by Xin-She Yang in 2008 and it has
become an important tool for solving the hardest optimization problems in almost
all areas of optimization as well as engineering practice. The literature has expanded
significantly in the last few years. Various FA variants have been developed to suit
different applications. This chapter provides a brief review of this expanding and
state-of-the-art literature on this dynamic and rapidly evolving domain of swarm
intelligence.
Keywords Firefly algorithm · Discrete firefly algorithm · Nature-inspired algorithm ·
Scheduling · Combinatorial optimization · Engineering optimization

1 Introduction
Among swarm-intelligence-based algorithms, firefly algorithm (FA) is now one of
the most widely used. Firefly algorithm was developed by Xin-She Yang in 2008 [1],
based on inspiration from the natural behavior of tropical fireflies. FA tries to mimic
the flashing pattern and attraction behaviour of fireflies. The purpose of these flashI. Fister (B) · D. Fister · I. Fister Jr.
Faculty of Electrical Engineering and Computer Science, University of Maribor,
Maribor, Slovenia
e-mail: iztok.fister@uni-mb.si
D. Fister
e-mail: dusan.fister@uni-mb.si
I. Fister Jr.
e-mail: iztok.fister2@uni-mb.si
X.-S. Yang
School of Science and Technology, Middlesex University, North London, UK
e-mail: x.yang@mdx.ac.uk
X.-S. Yang (ed.), Cuckoo Search and Firefly Algorithm,
Studies in Computational Intelligence 516, DOI: 10.1007/978-3-319-02141-6_17,
© Springer International Publishing Switzerland 2014

347

• The brightness of a firefly is determined by the landscape of the objective function. The movement of a firefly i is attracted to another more attractive (brighter) firefly j is determined by xit+1 = xit + β0 e−γ ri j (x tj − xit ) + α εit . and provide a good snapshot of the current literature. In essence. Fister et al. and εit is a vector of random numbers drawn from a Gaussian distribution at time t. Fister et al. and we can set α0 = O(1) without losing generality. • The attractiveness is proportional to the brightness and they both decrease as their distance increases. where t is the index of iterations/generations. provided a comprehensive review of the current literature of the firefly algorithm and its variants [6]. 2 (1) where β0 is the attractiveness at r = 0. As a firefly’s attractiveness is proportional to the light intensity seen by adjacent fireflies. 2 (2) where the second term is due to the attraction.348 I. it would be useful to summarize the key works/papers that we can get hold of and highlight the main and representative results. however. research works. Obviously. a quick Google scholar search using firefly algorithm as the keyword returned 625 hits at the time of writing this chapter in July 2013. and one way to achieve this is to use (3) α = α0 θ t . it seems impossible to review every single piece of research work concerning firefly algorithms. FA uses the following three idealized rules [1]: • Fireflies are unisex so that one firefly will be attracted to other fireflies regardless of their sex. Studies have shown that FA is very efficient [2–5]. 1). including physical laws. ing lights are twofold: to attract mating partners and to warn potential predators. If there is no brighter one than a particular firefly. It is worth pointing out that γ controls the scaling. Therefore. Here α0 is the initial randomness factor. Thus for any two flashing fireflies. A similar search using Scirus gave 658 hits with 158 peer-reviewed journal papers. Since then. about 30 more journal papers published in the last a few months alone. it will move randomly. Therefore. the main aim of this chapter is to briefly introduce the readers the state-of-the-art developments so as to provide classifications of variants. In fact. these flashing light and its intensity can obey some rules. while α controls the randomness. Other studies also use the randomization εit can easily be extended to other distributions such as Lévy flights. we can now define the variation of attractiveness β with the distance r by β = β0 e−γ r . the less brighter one will move towards the brighter one. randomness should be gradually reduced. The rest of chapter . The third term is randomization with α being the randomization parameter. θ ∈ (0. For the algorithm to convergence properly.

Firefly Algorithm: A Brief Review of the Expanding Literature 349 Fig. 1). 5. conclusions are drawn briefly and the directions for future work are discussed in Sect. Therefore. and thus may be richer in terms of dynamical characteristics. it is no surprise that FA can efficiently deal with a diverse range of optimization problems. Using similar classification as proposed in [6]. • FA can include PSO. 2. and this attraction can speed up the convergence. FA can deal with multimodal optimization naturally. Finally. 2 Classifications of Firefly Algorithms The standard firefly algorithm has been proved very efficient and it has three key advantages • Automatic subdivision of the whole population into subgroups so that each subgroup can swarm around a local mode. DE and SA as its special cases as shown in Chap. • FA has the novel attraction mechanism among its multiple agents. In total. and they have also tried to develop various variants to suit for specific types of applications with improved efficiency. a brief review of the modified and hybridized firefly algorithms is presented. Among all the local modes. Many researchers use FA to solve a diverse range of problems. there exists the global optimality. 4 focuses on the application of the firefly algorithm in engineering optimization. Therefore. Section 3 deals with the application domains where the firefly algorithms were successfully used. while Sect. 1. the variants of the firefly algorithm can be divided into modified and hybridized algorithms (Fig. In Sect. 1 Variants of the firefly algorithm is organized as follows. . The short review of research papers concerning the classical firefly algorithms can be summarized in Table 1. there are more than 20 different FA variants. The attractiveness term is nonlinear.

and . Gaussian. Various hybridizations have been applied on the firefly algorithm to seek improvements. population scheme (swarm.2 Hybrid Firefly Algorithms The firefly algorithm has been designed as a global problem solver that should obtain the good results on the all classes of optimization problems.1 Modified FA The modified firefly algorithms can be analyzed according to the setting of their algorithm-dependent parameters. classifications. machine learning techniques. The short review of the major hybrid firefly algorithms is illustrated in Table 3. Hybridization incorporates some problem-specific knowledge to the firefly algorithm. Normally. etc. in the last few years. However. including optimization. In order to overcome the limitations imposed by this theorem.350 Table 1 Classification of the firefly algorithms I. hybrid methods tend to be used to develop new variants of nature-inspired algorithms including the variants of firefly algorithms. Fister et al. As a results. In line with this. 3 Applications Since its first appearance in 2008. the existing studies concerning the modified algorithms can be presented in Table 2. elitism). travelling salesman problem. multi-swarm). the firefly algorithms are classified according to the following criteria: • • • • • representation of fireflies (binary. the firefly algorithm has been used in almost every area of sciences and engineering. image processing. chaos distributions). it was hybridized with other optimization algorithms. heuristics. scheduling. randomization of moving the fireflies (uniform. the No-FreeLaunch theorem usually poses some limitations [42]. Lévy flights. 2. Topic References The original firefly algorithm Multi-modal test functions Continuous and combinatorial optimization Review of nature-inspired meta-heuristics [1] [3] [7] [8–10] 2. determination of the best solution (non-elitism. evaluation of fitness function. real-valued).

whilst in the combinatorial optimization problems. solutions are sought from a . multi-modal. multi-objective. combinatorial. classification and engineering designs. 34] [35–37] [15. constraint. 38] [39] [40] [41] engineering designs. Some application domains are more theoretical. 27] [28] Topic (with which the firefly algorithm hybridizes) References Eagle strategy using Lévy walk Genetic algorithms Differential evolution Memetic algorithm Neural network Cellular learning automata Ant colony Simulated annealing Evolutionary strategies [29] [15. The taxonomy of firefly algorithm applications can be seen in Fig. Continuous optimization problems often concern a set of real numbers or functions. 3. dynamic and noisy. 18] [4. 30] [31.1 Optimization The firefly algorithm has been applied into the following classes of problems: • • • • • • continuous. 32] [33. 2 where we have focused on three major areas of applications: optimization.Firefly Algorithm: A Brief Review of the Expanding Literature Table 2 Modified firefly algorithms Table 3 Hybrid firefly algorithms 351 Topic References Elitist firefly algorithm Binary represented firefly algorithm Gaussian randomized firefly algorithm Lévy flights randomized firefly algorithm Chaos randomized firefly algorithm Parallel firefly algorithm Multi-population Harmonic clustering Quaternion firefly algorithm [11] [12–16] [17. whilst others have focused on solving the real-world problems. 18. 24] [25] [26. 19] [20–22] [23.

permutations. The dynamic and noisy problems are non-stationary. Usually. typically. and neural networks. 58–63] [64] [65–69] finite or infinite set. there are a (large) number of local modes that are better than all their neighboring solutions.2 Classifications Classification algorithms are procedures for selecting a hypothesis from a set of alternatives that best fits a set of observations or data. That is. A review of existing studies from this area can be summarized as follows: • The firefly algorithm was hybridized with the Rough Set Theory (RST) for finding a subset of features [70]. 7. 57] [5. 3. The latter class of optimization problems can also be called discrete optimization.352 I. for multimodal problems. . data mining. • The firefly algorithm was used for training the radial basis function (RBF) network [71]. Fig. 2 Taxonomy of firefly algorithm applications Table 4 Optimization applications Topic References Continuous optimization Combinatorial optimization Constrained optimization Multi-objective optimization Multi-modal optimization Dynamic and noisy environment [2. or graphs [43]. but do not have as good a fitness as the globally optimal solution [44]. 4. Fister et al. Solutions of constrained problems must obey some limitations (also known as constraints). Various studies of the firefly algorithm in this application domain can be summarized in Table 4. of integers. In the multi-objective problems. 46] [47–55] [56. objectives. sets. 9. On the other hand. 19. possibly conflicting. these algorithms are more relevant in machine learning. they change over time [45]. 18. the quality of a solution is defined by its performance in relation to several.

the rapid expansion of the research literature on the firefly algorithms in engineering optimization proves that the firefly algorithms enter in its matured phase. scheduling. 4 Engineering Optimization The firefly algorithm has become one of the most important tools for solving the design optimization problems in routine engineering practice. the firefly algorithm has become a matured optimization tool for solving a diverse of range of optimization problems such as engineering designs. In summary. As can be seen from Table 5. image processing. . The majority of studies come from engineering and industries. which means that this algorithm has become a serious tool for solving various challenging real-world problems. 119] [120] [121] [122] 22 9 5 4 3 2 2 1 1 1 • The firefly algorithm was used for clustering data objects into groups according to the values of their attributes [72]. classifications and industrial applications. after initial theoretical studies.Firefly Algorithm: A Brief Review of the Expanding Literature Table 5 Engineering applications 353 Engineering area References Total Industrial optimization Image processing Antenna design Business optimization Robotics Civil engineering Chemistry Semantic web Meteorology Wireless sensor networks [73–94] [95–103] [104–108] [109–112] [113–115] [116. flexibility. The popularity of the firefly algorithm and its variants may be due to their simplicity. travelling salesman problems. Over 20 new FA variants have been developed and new applications and studies are emerging almost daily. feature selection. versatility and superior efficiency. engineering and industry. almost every engineering domain has been covered by the applications of this algorithm. 117] [118. It is no surprise that FA has been used in almost every area of sciences. more and more applications from realworld case studies have been emerged. That is. 5 Conclusion Though with a relative short history up to now.

stochastic test functions and design optimisation. 2325–2336 (2011) 3. 428–432. H. X.. Comput. Swarm and Evolutionary Computation. Nasiri. X. Int... S. Yang. K.org/10. Int. theoretical analysis is still very limited. and this is also true for many other nature-inspired algorithms. In: Evolutionary Computation (CEC). (2013) http://dx. A. but it is possible to use theories such as dynamical systems. (eds. Molla. there is still room for improvements. Berlin (2010) 5. 2(2). 78–84 (2010) 8.: Review of meta-heuristics and generalised evolutionary walk algorithm. 89(23–24). Power Energy Syst. pp. Intel.doi.. 209–218. Zang.swevo.H.. Struct. S.: Fault identification with binary adaptive fireflies in parallel and distributed systems. BioInspired Comput. IEEE (2011) . Electr.doi.) Research and Development in Intelligent Systems XXVI. References 1. IEEE (2011) 15. In: Watanabe. 79–90. Chandrasekaran. Bio-Inspired Comput. 1–16 (2011) 9. J. A. Lopes. levy flights and global optimization.354 I.: A review of nature-inspired algorithms.. Luniver Press. 169–178.-S. (2013 In press) 7... Nayak.A. 7. Yang. Cambridge (2008) 2. Springer.. Artif. Int. Gandomi.S.2013. However.. M. J. pp. 1359–1366. X. 2012. Kule.A. the current research communities strive to find better and smarter algorithms. K.1016/j. X.06. Farahani.S. Ong. Sci.06. pp. A. It can be expected that the firefly algorithm and its variants may be extended and further developed into some sort of self-evolving and truly intelligent algorithms. 921–932 (2012) 14. X. Springer (2009) 4. Bret. S.. X.2013. J. J. J. Inf. Firstly. Simon.. K. R. Abshouri. Fister. pp.. Zeugmann. Bionic Eng. Meybodi. J. N. In: Computer and Communication Technology (ICCCT). BioInspired Comput. 175–184 (2013) 6. X.S.: Multiobjective firefly algorithm for continuous optimization. Khanra. Almeida. 2011 IEEE Congress on. Nature-Inspired Metaheuristic Algorithms. M. parameter tuning and parameter control can be a very useful area for further research. Markov chains and statistics to gain insights into the working mechanisms of various variants. J.001.Jr.ins. Eng. Luleseged Tilahun.: Binary real coded firefly algorithm for solving unit commitment problem..1016/j. 97–117 (2012) 16.P. Computers 29. Int. I. T. 232–237 (2010) 11.: Firefly algorithm. Yang. Yang. Alavi. S.P.S.: Mixed variable structural optimization using firefly algorithm. Chandrasekaran. Thirdly. I.S. S. Yang. Finally. Fister. B.: Firefly algorithms for multimodal optimization. A. Sinha. Yang. X..: Some hybrid models to improve firefly algorithm performance.: A comprehensive review of firefly algorithms..R. pp. Simon. O.: Firefly algorithm.N.: Network and reliability constrained unit commitment problem using binary real coded firefly algorithm. Padhy.. In: Proceeding of the Conference on Stochastic Algorithms: Foundations and Applications. Appl.org/10. Parpinelli. R. H. Int.H. M. 12 (2012) 12. A.M. J. 8(12). Mathematical analysis is challenging... 43(1).C. more applications should focus on large-scale real-world applications. 2011 2nd International Conference on. Palit. http://dx. S.. Falcon. 77–84 (2011) 10. Math.: A cryptanalytic attack on the knapsack cryptosystem using binary firefly algorithm.P.. Zhang. Fister et al. S. Yang. S. 3(1). Hapeshi.: New inspirations in swarm intelligence: a survey. Yang. M.022 13. 3(2).: Firefly algorithm (chapter 8). H.: Modified firefly algorithm. S. Secondly. S.S.

A.: An evolutionary firefly algorithm for the estimation of nonlinear biological model parameters. Theor.. A. Rodrigues. Srivastava. M.-S.: Hybrid firefly based simultaneous gene selection and cancer classification using support vector machines and random forests. S. (2013) 26. (eds. Ranjan Senapati. I. Das. 82–91 (2013) 27. Springer. V. Talatahari. X. pp.K. J..) Distributed Computing and.C. Abshouri.. S. 89–98 (2013) 22.P. F. Hassanzadeh. N. pp.: A multipopulation firefly algorithm for correlated data routing in underwater wireless sensor networks.. pp.-S. Abdullah. Nasiri. Abdullah. Deris. 63–67. IEEE (2012) 36.H.. In: Latest Advances in Information Science and Applications.. Lima.A. Yang. M. Seyfi. X. P. 8(3). M. Pal.C. Sarkar.06.. X. Nanosci. J. X.H. J.A. A. 2011 IEEE Congress on. In: Evolutionary Computation (CEC).A. Chakrabarti. Artificial Intelligence. Machine Learn. Yang. S. V. 2011 World Congress on.: Modified firefly algoirthm using quaternion representation.. T. Deris. S. X. Mariani. volume 6630Springer Verlag.. Anwar.: A new hybrid firefly algorithm for complex and nonlinear problem.S. 21–32.. Comput. Proceedings of Seventh International Conference on Bio-Inspired Computing: Theories and Applications (BIC-TA 2012).2013. Adaniya. C. Springer (2013) 35.-S. J. Hashim. Berlin (2011) 19. 2(4). 18(1).. Yang. Lecture notes in computer science. In. pp.: Parallelization of the firefly algorithm for unconstrained optimization problems. S. e56310 (2013) 33.. Berlin (2012) 32. pp. Fister. pp. 1–8 (2013) . M. 2012 IEEE International Conference on. arXiv. G.: Firefly algorithm with chaos.. M. Yang. A. 101–111 (2010) 30. S. L. de Andrade Bernert. (2013) 29. 151. pp.: Anomaly detection using dsns and firefly harmonic clustering algorithm. Expert Systems with Applications.. Arjunan.S. Dash.: Chaos-enhanced firefly algorithm with automatic parameter tuning. Yang. In: Omatu.. 189–198 (2012) 20.: Metaheuristic optimization: algorithm analysis and open problems.. Report CSTN-141 (2012) 24. pp.org/10.. Tuba.eswa.K.P.. Netw.. 070. pp.: Eagle strategy using levy walk and firefly algorithms for stochastic optimization. Neural Comput.V.: Local linear wavelet neural network based breast tumor classification using firefly algorithm. Gandomi. J. Brest.V. Das... vol. In: Biomedical Engineering (ICoBE). Meybodi.: A hybrid firefly algorithm using genetic operators for the cryptanalysis of a monoalphabetic substitution cipher.. Commun. S. L. In: Communications (ICC). Simul. D. Husselmann. 75–86..H. Int.) Experimental Algorithms. S. Technical. M.: A gaussian firefly algorithm. Alavi. M. Subutic. Swarm Intell. I. S. J. S. 2012 International Conference on. In: Nature Inspired Cooperative Strategies for Optimization (NICSO 2010). A.: A speech recognition system based on structure equivalent fuzzy neural network trained by firefly algorithm. http://dx. A. 1(5). 89–98. IEEE (2013) 21.H.C. 8(1). Res.. Ghosh..M.1016/j. S.: Parallel parametric optimisation with firefly algorithms on graphical processing units.. Sens..: Memetic firefly algorithm for combinatorial optimization.Firefly Algorithm: A Brief Review of the Expanding Literature 355 17. Rebennack.. 673–680. Netw. Appl. Int. A. 202–206. PloS one.. 1183–1187. IEEE (2011) 31. Liu. 1–11 (2011) 23.. S. Nandy. Faez. J. Adaniya.L.M. Fister.. dos Santos Coelho.K. K. A. Abrao. 30. preprint arXiv:1206. vol. pp. X. G. Deb. T. Yang. X. S. S.. Proenca. Hawick. Yang. A. I. J.. et al.-S. Luthra.doi. S..Jr. P.: Analysis of a nature inspired firefly algorithm based backpropagation neural network training. Fister. Jayaraman. Distrib. In: Pardalos. (eds. Comput.5360 (2012) 37. Brest. J.J. 18. IEEE (2012) 28.. 485–494. 9(2).. K.N. 264–269 (2012) 25. B. Stanarevic.. In: Information and Communication Technologies (WICT).R. Numer.M.: Efficiency analysis of swarm intelligence and randomization techniques... 448–454 (2011) 18. I. Farahani. M. S. et al. Fister.. Mohamad. Nonlinear Sci. P.: Anomaly detection using metaheuristic firefly harmonic clustering.: A chaotic firefly algorithm applied to reliability-redundancy optimization.Jr. Int. Jožef Stefan Institute (2012) 34. pp.