Hybrid Neural Network Optimization for Feed Point Determination in Antenna Design Umut Özkaya Levent Seyfi Department of Electrical and Electronics Engineering Department of Electrical and Electronics Engineering Selçuk University Selçuk University Konya, Turkey Konya, Turkey e-mail: uozkaya@selcuk.edu.tr e-mail: leventseyfi@selcuk.edu.tr Abstract—In this paper, coaxially feed rectangular microstrip method. The inputs of the artificial neural network are antenna is designed for WIFI communication in accordance with dielectric constant, height, TE and TM modes of the material; IEEE 802.11a standard between 5.15 GHz and 5.725 GHz. on the other hand, resonant frequency is the output of neural Feeding position of coaxial probe significantly affected antenna network [2]. Microstrip patch antenna was designed for wide characteristics. Optimum feeding point should be selected in 2-D band applications with dielectric material thickness of 2 mm. patch plane on the purpose of better antenna characteristics. The The operating frequency of the microstrip patch antenna with model is used to solve the optimization problem. It has three 10.5 to 12 GHz bandwidth was tried to be calculated by the input variables which are antenna parameters as resonance genetic algorithm [3]. Hybrid artificial neural networks frequency, bandwidth and return loss; on the otherhand, two techniques and fuzzy logic methods were used to calculate the output such as x and y coordinates of feeding position. Also, error function is updated by proposed artificial intelligence algorithms. operating frequencies of microstrip antennas. The hybrid Unlike conventional methods, contemporary artificial intelligent method based on the least squares method with the algorithms have been proposed for the antenna design. Genetic backpropagation algorithm was performed for dimension Algorithm (GA), Spider Monkey Optimization (SMO) and Grey optimization of square, circle and triangle microstrip patch Wolf Optimizer (GWO) are preferred for optimization. antennas [4]. Operating frequency of the microstrip patch According to comparison of these results, optimal antenna for antenna with coaxial feed was determined by artificial neural WIFI Protocol is designed. network methods. The inputs of the artificial neural network were the antenna's width and height, and the output was the Keywords— Microstrip Antenna, WIFI Communication, operating frequencies in the dual band [5]. The bandwidth of Artificial Neural Network, Artificial Intelligence Algorithm, the microstrip patch antenna with rectangular geometry was Optimization. tried to be optimized by using artificial neural network. Error rates of simulation and network results were compared with I. INTRODUCTION each other [6]. Forward feedback propagation algorithm is used to optimize and microstrip antenna parameters with square and In the recent years with the development of the technology, rectangular pitch are provided. The dimensions of the the use of microstrip patch antenna has gradually increased in microstrip antenna were designed as network inputs and the spacecraft, doppler and navigation radar, satellite operating frequency was optimized as the network output [7]. communication, mobile radio and guided missiles. Microstrip A particle swarm optimization algorithm was used to determine antennas have several advantages over many known the patch sizes of a multi-layer microstrip antenna that can conventional antennas. These advantages are low profile, operate in the X and Ku band [8]. A microstrip patch antenna changeable polarization with feeding position, integrated with with many slots for wireless communication was designed by solid-state equipment and compatible with co-planar surface multilayer perceptron artificial neural network model [9]. [1]. Hybrid artificial neural networks, which is a combination of Great strides in the electronics sector provide more radial based function and backpropagation algorithm, was used functionality and size reduction for especially communication for design of acoupled microstrip antenna. The performance of devices. Besides, demands for multiple applications (GPS, the hybrid network was compared with other types of network GSM, WIFI) in a single device has caused designers to focus results [10]. Conjugate gradient artificial neural network model more on microstrip antenna. Many applications of artificial was used for determination of the operating frequency in neural network and artificial intelligence algorithms exist in circular microstrip patch antenna [11]. The truncated edges of literature. The results obtained by using the feedback square microstrip antenna was designed with artificial neural multilayer perceptron network for the design of the equilateral network. Levenberg-Marquardt algorithm with three hidden triangle microstrip antenna were compared with conventional layers was chosen as an artificial neural network method [12]. Artificial neural networks were used for fractal antenna design. Copyright held by the author(s). 28 Analysis of the error rates of the network results were carried  c   out with the Generalized Regression Neural Networks (GRNN) W [13]. Variables such as operating frequencies, gains, r 1 2f 0 directionality, antenna efficiency and radiation efficiency in the 2 dual band were assigned to the inputs of the different artificial neural network based algorithms and height of gap between the ground plane and the dielectric material was computed [14].  Leff  c    The design of microstrip antennas with rectangular and circular 2f 0  eff patches for wireless communication applications was implemented with particle swarm optimization technique. The inputs of the optimization technique are the operating frequency, the dielectric constant and the height of the  ( eff  0.3)( w / h  0.264)    L  0.412h dielectric material. [15]. Sathi et al. used to be processed ( reff  0.258)( w / h  0.8) moments matrices optimally with the genetic algorithm, ensuring that the simulation and test results are in harmony with one another in the antenna design [16]. Antony et al.  L  Leff  2L    proposed PSO as a design tool in a microstrip antenna array created by a parasitic method. With their design, IEEE 802.11a WLAN achieves a multi-directional radiation pattern and reflection coefficient of <-10 dB in the 5-6 GHz band [17].  Lg  6h  L    Amir et al. calculated resonance frequency and bandwidth in the rectangle microstrip antenna design. Faster and more accurate results were obtained by optimizing the moments method with Bacterial Search Optimization [18]. Arunava et al.  Wg  6h  W    applied Cuckoo Search algorithm to increase the bandwidth of the microstrip patch antenna running in the X band [19]. W×L has been computed as 21.56 × 17.46 mm2 for Metaheuristic optimization method have shown a great patch and on top of that Wg×Lg has been calculated as 41.8 × development and its applications has been carried out in many 35.6 mm2 for ground plane. The antenna designed can be fields for the last twenty years. Genetic, Differential Evolution, excited by coaxial feed as Figure 1. It can be at any position to Gravitational Search and Teaching-learning based optimization match with input impedance. algorithms are given examples for some commonly used metaheuristic optimization algorithms [20]. In this study, It is introduced that how to design microstrip antenna in Section II. Proposed Artıfıcıal Neural Network algorithm is described in Section III. Genetic Algorithm, Spider Monkey Optimization and Grey Wolf Optimizer, which are based on swarm intelligence and are known as a novel optimization algorithms, are used to determine coaxial feeding position of microstrip antenna for Wi-Fi protocols in Sections IV. The experimental results drawn in Sections V. Section VI inserts the conclusion part. II. MICROSTRIP ANTENNA DESIGN Basic microstrip patch antenna design consists of the three main parts. These are patch plane, dielectric substrate and ground plane. FR4 material (ɛr=2.2) is used in dielectric substrate of proposed microstrip antenna for Wi-Fi protocol. The thickness of this material is determined to be 1.58 mm. The thickness of the copper patches on the dielectric substrate is 38 μm. The operating frequency of the antenna is designed around 5.38 GHz and is suitable for IEEE 802.11a protocol. Patch and ground plane size are designed in accordance with the following formulas in which L, W and Lg, Wg represents patch and ground dimension respectively: 0.5 r 1 r 1  h   eff   1  12  (1) 2 2  W Fig. 1. Top and Side View of Coaxial Feed Microstrip Patch Antenna 29 Pseudo code of GA is given in Fig. 2. CP represents crossing-over point and should be smaller than chromosome Typically, even if cos2(πy0/h) formula suggests matching size. Mutation rate is abbreviated as MR. It is important for with 50 ohm impedance, there will be some feed position to diversity in new generation. obtain greater return losses at operation frequency in two- dimensional plane. Hence, utilized hybrid optimization technique is composed of artificial neural network with Genetic Algorithm, Spider Monkey Optimization and Grey Wolf Optimizer in order to estimate feed position. III. ARTIFICIAL NEURAL NETWORK Artificial Neural Network (ANN) has ability to learn, use memory the knowledge about the system. Moreover, it can search, reproduce and discover new knowledge without any help. A neural network is a natural propensity for storing experiential knowledge. Also, it can prepares to use when they are needed [21]. ANN is a computer program that simulates biological neural networks. With these features, it can offer effective solutions for optimization, classification, prediction, pattern recognition, memory management and control Fig. 2. Pseudo-code of GA problems. In GA process, chromosomes are randomly generated for initial population. At following iterations, crossing-over and IV. ARTIFICIAL INTELLIGENCE ALGORITHMS mutation process are performed to obtain a new generation. Crossing-over operation is used to make better solutions from combination of different parts in others. Thanks to random Metaheuristic algorithms are divided into three main changes in character string, a copy of individuals in the groups: evolutionary, physics based and swarm intelligence previous generation prevent the transfer to the next generation algorithm. Evolutionary algorithm is an adaptation of evolution via mutation process. Then, fitness values are determined for events in nature for optimization algorithm. In 1992, Holland all chromosomes by comparison with each other. The better proposed genetic algorithm which is the most popular and first new ones have fitness value, the more they have chance of algorithm in this branch. Then, Differential Evolution, survival in the next generation. Biogeography-Based Optimizer, Genetic Programming and Evolution Strategy are some examples of evolution algorithm [22-23]. Physics-based technique is the second subclass of B. Spider Monkey Optimization metaheuristic algorithms. This kind of algorithms based on imitate physical rules in nature. Gravitational Search, Charged Spider monkeys have fission-fusion social network which system search, Artificial Chemical Reaction, Black Hole, Ray, includes temporary subgroups for larger communication Small-World, Galaxy-based Search and Curved-spaced structure. Also, fission-fusion social mechanism provides food algorithm are well-known optimization methods [24-31]. competition among members of smaller foraging groups. It is Finally, the third main branch of meta-heuristics is the swarm necessary to divide into smaller groups in case adequate food intelligence method. These type of algorithms usually mimics supply could not be found by female monkey leader. Although the social behavior of swarm in natural atmosphere. Particle the number of individuals in main monkey groups can be up to swarm, Ant colony, Artificial Bee Colony and Bat inspired 50, the size may be reduced to 3 or 4 [34-37]. The members in optimization are the main examples of swarm intelligence all subgroups interact with internal and external individuals method [32]. about food availability and territorial boundaries. A. Genetic Algorithm In fission-fusion social structure, the foraging of spider monkeys have four steps. At the first, monkey groups try to search and find the food foraging. In the second stage, groups’ Genetic algorithm (GA) is a heuristics algorithm developed members update their location and calculate the distance from for nonlinear problems [33]. GA could mimic evolutionary the food sources repeatedly in accordance with their group processes observed in nature. In the complex multi- leader. Also, it is defined as an individual group leaders who dimensional search space, it tries to search for the best solution have the best position in the group and replaced continuously via the principle of having the best life. Variables of the in order to reach a better position. Therefore, other group problem are represented as unique or group of gene in the individuals can direct in different directions to search for food. chromosome. The most important factor in deciding the Finally, global leader has ever updated its best position under success of genetic algorithms is the representation of stagnation condition. individuals for solving the problem. 30 In Spider Monkey Optimization (SMO), Global Leader Limit and Local Leader Limit are two major parameters to help for taking leader decisions. Moreover, maximum group (MG) and perturbation rate (pr) are other parameters to control amount of groups and perturbation in current iteration. SMO has a heuristic process which is based on a trial and error. It starts to initialize population of N spider monkeys SMi (i=1,2,3…. N) with D dimension vector. D denotes the number of variables in optimization structure. SMi is the ith spider monkey in the swarm and generated as: SM ij  SM min j  U (0,1)  (SM max j  SM min j ) (8) SMminj and SMmaxj are boundary for SMi of jth direction. U(0,1) is a uniformly distributed random number between 0 and 1. The following step is Local Leader phase in which each individual changes the position thanks to experience of local group leader and members. The fitness value is computed every new position of swarm member. If the fitness value of new position is greater than former one, current position is Fig. 3. Pseudo-code of SMO replaced with new one. The position update equation is described as: C. Grey Wolf Optimizer  SMnewij  SM ij  U (0,1)  ( LLkj  SM ij )  U (1,1)  (SM rj  SM ij )  Grey Wolf Optimizer (GWO) is one of the novel swarm intelligence method. It is inspired from grey wolves which have 5-12 swarm members. They have a very dominant social SMij is the ith SM member in the jth dimension, LLkj hierarchy for concept of search and hunt mechanism. denotes the kth local group leader position. SMrj represents randomly selected kth group member (r ≠ i). After the Local Grey wolf pack consists of alpha (α), beta (β), delta (δ) Leader phase, the Global Leader phase embarked on a new wolf and omega’s (ω) wolves in Figure 4. Alpha wolf with process. All of the SM’s members change their position by decisions give the pack directions to hunt. Although some using experience of Global Leader. The position update is democratic behaviors are observed, alpha has strict authority in implemented in this phase via following equation: the pack. Thanks to strong authority, the pack also has more discipline. At the second best position in the social hierarchy is beta grey wolf. In case of death or aging of the alpha wolf, beta  SMnewij  SM ij  U (0,1)  (GLj  SM ij )  U (1,1)  (SM rj  SM ij )  wolf is the best  leadership candidate for the  in the pack. It has  advisory role to alpha wolf and helps to discipline the pack. It GLj is the global leader in the jth dimension. Next, the new is primarily responsible for providing coordination and global and local leaders are necessary to be determined again. discipline between the alpha wolf and other members of wolf A member with the best fitness value is proposed for global pack. Delta wolves in the third layer of the hierarchy have the leader in entire population; on the otherhand, the best fitness task sharing for scouts, sentinels, elders, hunters and values in each group is appropriate candidate for local leaders. caretakers. They help pack management in groups for the hunting process. Moreover, they take care of the newborn and the older members in the pack. Omega group is in the lowest layer of the social hierarchy. It is required for them to become increasingly powerful and social structure. They provide candidate for the next generation of alpha, beta and gamma wolves [38]. 31 C  2r2 (14) a is an algorithm component which is linearly changed between 2 and 0. r1 and r2 are random vectors in valid interval [0,1]. The best and latest three solutions are X1, X2 and X3 which are saved as the best position of α, β, and δ. The next final solution X(t+1) is defined as average of alpha, beta and delta position. These expressions are formulated as: D | C1.X -X|, D  | C2 .X  -X|, D | C3 .X -X| (15) X1  X -A1 (X ), X 2  X  -A 2 (X  ), X 3  X -A 3 (X ) (16) Fig. 4. Social Hierarcy of Grey Wolf X1  X 2  X 3  X(t  1)  (17) In the GWO, the alpha (α) is the fittest solution in the 3 search space. Then, beta (β) and delta (δ) is assigned as the In GWO algorithm, alpha, beta and delta are allowed to second and third best solutions. The other solutions are in the update their location for right position. At the right iteration, omega’s (ω) group. We consider the fittest solution as the they can attack towards the prey. With the above formulas, alpha (α). The hunting mechanism is guided by α, β, and δ. The grey wolves gradually continue to scan the global space until ω wolves follow and help them. GWO algorithm is basically they reach the optimum solutions in mathematical concept. If divided into three main stages such as encircling, hunting and A<1, candidate solutions converge towards the prey; otherwise, attacking prey. they diverge from it in Figure 6. Grey wolves are coming close and surrounding the prey before hunting. The following formula has been proposed to model mathematically the encircling process in GWO algorithm: D  C.X p (t)-X(t) (11) X(t  1)  X p (t )  A.D (12) t is the current iteration indices. A and D are indicated coefficient vectors. Xp and X are position vector of prey and a grey wolf in Figure 5. Fig. 6. Pseudo-code of GWO V. PROPOSED METHOD ANN consists of multiple neural cells with a combination. Therefore, the concept of the neural cells will help in Fig. 5. Position Updates In GWO understanding the entire network. Inputs, weights, biases and outputs are main elements for neural network. Neural networks The coefficient vectors are calculated as bellows: are separated into the layers such as input, hidden and output layers in Figure 7. Conventional ANN is modified by Artificial A  2a.r1  a (13) Intelligence Algorithms on the propose of update in weights 32 (Wi,k,j,o) and biases (BI,L1,L2,O). In this way, the hybrid network Fig. 7. Hybrid Artificial Neural Network Model structure optimizes the outputs in lower error rate and processing time. VI. RESULTS AND DISCUSSION In the proposed method, perceptron structure of artificial neural network is used as objective function in the optimization process. Weights and bias in artificial neural network structure To study the effects of changes in feeding position on the are optimized as objective function. Genetic algorithm, Grey presented geometry, the antenna characteristics are wolf optimizer and spider monkey optimization algorithms are investigated. The optimization algorithms are performed with used to update weights and biases in the network structure. The AMD FX8 AMD 3.5GHz processors. 32GB of RAM with 4 main use of these algorithms is to prevent linerization in the GB of GDDR3 GeForce supported system memory with network structure. The reason why the proposed algorithms are nVIDIA graphics card is used. Figure 8 indicates the used is that the accuracy of each algorithm is different from simulation results about antenna design of conventional that of the algorithm. According to mean error function method and proposed artificial intelligent algorithms for WIFI obtained after each iteration, weights and biases of artificial communication. neural network are updated in artificial intelligence algorithms. In order to train the proposed network, 4075 pieces of data At the beginning of each algorithm, all weights and biases are were produced in the patches according to feed points spaced randomly assigned. The parameters in optimization algorithms 0.1 mm step. Produced data includes band width, resonance also start randomly at the start. This is to prevent proposed frequency, return loss, x and y axis feed point. The constructed algorithms from reaching local minimum solutions in search hybrid artificial neural network consists of 10 hidden layers. space. The weights and biases in these hidden layers are updated with In the proposed method, the inputs of the hybrid artificial artificial intelligence algorithms instead of the gradient descent neural networks are composed of resonance frequency, band algorithm. The input 3 of the network consists of band width, width and return loss. The network output is the antenna feed resonance frequency and return loss. At the output, 2-D feed coordinates of the microstrip antenna in 2 dimensions. The point is tried to be obtained. proposed method is a supervised algorithm and has training and testing steps. Antenna parameter data was obtained by changing the antenna feed point locations designed using the HFSS program. Fig. 8. Results of Antenna Design The feeding points of the antenna were separately determined via GA, SMO and GWO algorithms in order to obtain the most appropriate return loss, operating frequency and bandwidth. Computing feeding positions were respectively (1.3230, 4.7317), (1.7770, 4.5424) and (1.8330, 4.6262) for GA, SMO and GWO. The results in Table 1 were obtained when the design parameters for the feed points were arranged and antenna parameters were examined. TABLE I. ANTENNA PARAMETERS OF EACH DESIGN 33 Resonance Return Bandwidth Design Methods Analyzing the data shown on Figure 8, some technical Frequency (GHz) Loss (dB) (MHz) parameters of the antenna such as operating frequency, return Conventional 5.3204 -11.8998 40 Method loss and bandwidth were extracted from Table 1. According to the values given in Table 1, there are significant differences GA 5.4033 -25.8998 550 between conventional and other design methods. The best GWO 5.3756 -26.6732 550 return loss and the widest bandwidth are obtained by antenna SMO 5.3756 -36.0399 560 design of SMO algorithm. If we examine the resonance frequencies obtained from the design methods, they are all very ACKNOWLEDGMENT close together. The resonance frequencies obtained from the There is no conflict of interest between the authors methods according to the order given in Table 1 are 5.32, 5.4, 5.37 and 5.37 GHz. When these frequencies are examined, it is observed that the proposed antenna is suitable for the IEEE REFERENCES 802.11a (5.15-5.725 GHz) standard. Taking into account the return loss, the conventional method has performed quite [1] C.A. Balanis, "Antenna Theory Analysis and Design",Jhon Wiely & poorly with -11.89 dB. If we think that bandwidth is calculated Sons, USA 2005. starting from -10 dB, conventional method design seems to be [2] Gopalakrishnan R. and Gunasekaran N. (2005) "Design Of Equilateral quite inadequate. For the proposed optimization algorithms Triangular Microstrip Antenna Using Artificial Neural Networks", IEEE International Workshop on Antenna Technology: Small Antennas and GA, GWO and SMO, return losses are -25.8998 dB, -26.6732 Novel Metamaterials, 0-7803-8842-9/05, pp. 246-249. dB and 36.0399 dB, respectively. Return loss of the antenna [3] Dipak K. Neog, Shyam S. Pattnaik, Dhruba. C. Panda, Swapna Devr, designs of GA and GWO algorithms are close to each other. Bonomali Khuntia ve Malaya Dutta (2005) "Design of a Wideband The SMO algorithm alone showed the highest design Microstrip Antenna and the use of Artificial Neural Networks in performance with a return loss of -36.0399 dB. The return loss Parameter Calculation", IEEE Antennas and Propagation Magazine, of the SMO algorithm is about -24 dB lower than the return ISSN 1045-9243/2005, pp. 60- 65. loss obtained by conventional methods, and is about -10 dB [4] Guney K. and Sarikaya N. (2007) "A Hybrid Method Based on Combining Artficial Neural Network and Fuzzy inference System for lower than GA and GWO. The return loss is affected by the Simultaneous Computation of Resonant Frequencies of Rectangular, feeding point changes in μm level when the feeding points of Circular, and Triangular Microstrip Antennas", IEEE Transactions on the mm level obtained from the algorithms are taken into Antennas and Propagation, 0018-926X/2007, pp. 296-296. consideration. Finally, when bandwidths are examined, [5] Vandana V.T., Singhal P. and Vivek K. (2008) "Calculation of conventional method has the lowest bandwidth at 40 MHz. The Frequency of a Rectangular Microstrip Antenna Using Artificial Neural highest bandwidth belongs to SMO algorithm with 560 MHz. Network" International Conference on Microwave and Millimeter Wave Technology, 978-1-4244-1880-0/08, pp. 1243-1245. Furthermore, the results of the operating frequency for all [6] Vandan V. T., Kamya D. and Singhal P. (2008) "Calculation Microstrip designs are appropriate for IEEE 802.11a WIFI standard with Antenna Bandwidth using Artficial Neural Network", IEEE International 5.15 GHz and 5.725 GHz. On the otherhand, SMO RF and Microwave Conference, 978-1-4244-2867-0/08, pp. 404-406. optimization design with optimal antenna feeding point [7] Bhagile V.D., Mehrotra S.C., Mishra A., Nandgaonker A.B. and Patil location is determined as 1.8330 mm in x axis and 4.6262 mm P.M. (2009) "Design of Square and Rectangular Microstrip Antenna in y axis. with the use of FFBP algorithmof Artificial Neural Network", Applied Electromagnetics Conference, 978-1-4244-4819-7/09, pp. 1-4. [8] S. K. Jain, A. Patnaik and S.N. Sinha (2011) "Neural Network Based VII. CONCLUSION Particle Swarm Optimizer for design of Dual Resonance X/Ku Band Stacked Patch Antenna" IEEE International Symposium on Antennas and Propagation, 978-1-4244-9561-0/11. In this study, the hybrid artificial neural network models are [9] Araujo W.C., d’Assunçao A.G. and Mandonça L.M. (2011) “Artificial proposed to be designed antenna structure. The proposed Neural Networks for Multi-Slot Microstrip Patch Antennas” IEEE-APS models are more effective compared to conventional methods. Topical Conference on Antennas and Propagation in Wireless Communications, 978-1-4577-0048-4/11, pp. 532-535. Artificial intelligence algorithms with neural network are [10] Bose T. and Gupta N. (2011) “Design Of An Aperture-Coupled performed in feed position determination. Artificial intelligence Microstrip Antenna Using A Hybrid Neural Network” IET Microw. methods are used for minimization mean squared error between Antennas Propag. 2012, Vol. 6, Iss. 4, pp. 470–474. computing and reference output value in training process. [11] Janvale Ganesh, Mishra Abhilasha, Patil A.J. and Pawar B.V. (2011) Therefore, appropriate weights and biases are tried to be “The Design of Circular Microstrip Patch Antenna by using Conjugate obtained for test data set. Finally, desired parameters are Gradient Algorithm” IEEE Applied Electromagnetics Conference, 978- defined by suitable algorithms. When results are analyzed, it is 1-4577-1099-5/11, pp. 1-4. observed that hybrid algorithms have better return loss and [12] Fong Shaojun, Liu Qiang Wang Hongmei and Wang Zhangbao (2012) wider bandwidth for WIFI communication than conventional "An ANN-Based Synthesis Model for the Single feed Circularly- Polarized Square Microstrip Antenna with Truncated Corners", IEEE method. Especially, Spider Monkey Optimization has the Transactions on Antennas and Propagation, Vol. 60, No. 12, December greatest performance on antenna characteristics. In future 2012, pp. 5989-5992. studies, this type of hybrid neural network model can be used [13] Singh D. B. and Pattraik S. S. (2012) "Performance Evaluation of in full antenna design in order to obtain desired antenna Partical Neural Networks in Microstrip fractal Antenna Parameter performance. Estimation", IEEE International Conference on Communication Systems, 978-1-4673-2054-2/12. 34 [14] Asok D., Taimoor K.and Moin U. (2013) "Prediction of slot Size and [30] H. Shah-Hosseini Principal components analysis by the galaxy-based Inserted Air-Gap for Improving the Performance of Rectangular search algorithm: a novel metaheuristic for continuous optimisation Int J Microstrip Antennas Using Artificial Neural Network", IEEE Antennas Comput Sci Eng, 6 (2011), pp. 132–140. and Wireless Propagation Letters, Vol. 12, pp. 1367-1371. [31] Moghaddam FF, Moghaddam RF, Cheriet M. Curved space [15] Marijan B., Niksa B. and Ivan V. (2013) "Microstrip antenna Design optimization: a random search based on general relativity theory. arXiv, Using Neural Networks Optimized by PSO", 21st International preprint arXiv:1208.2214; 2012. Conference on Applied Electromagnetics and Communications, pp. 1-4. [32] Yang X-S. A new metaheuristic bat-inspired algorithm. In: Nature [16] Sathi, V., et al. “Optimisation of multi-Frequency microstrip antenna inspired cooperative strategies for optimization (NICSO 2010), ed.: using genetic algorithm coupled with method of moments.” IET Springer; 2010. p. 65–74. Microwaves, Antennas & Propagation, vol. 4, no. 4, 2010, p. 477., [33] M. Gen, R. Cheng, “Genetic Algorithms and Engineering Design”, doi:10.1049/iet-map.2009.0020. Wiley Interscience, 2001. [17] Minasian, Anthony A., and Trevor S. Bird. “Particle Swarm [34] Simmen B, Sabatier D (1996) Diets of some french guianan primates: Optimization of Microstrip Antennas for Wireless Communication food composition and food choices. Int J Primatol 17(5):661–693. Systems.” IEEE Transactions on Antennas and Propagation, vol. 61, no. 12, 2013, pp. 6214–6217., doi:10.1109/tap.2013.2281517. [35] Storn R, Price K (1997) Differential evolution-a simple and efficient adaptive scheme for global optimization over continuous spaces. J [18] Amir, Mounir, et al. “Bacterial foraging optimisation and method of Global Optim 11:341–359. moments for modelling and optimisation of microstrip antennas.” IET [36] Suganthan PN, Hansen N, Liang JJ, Deb K, Chen YP, Auger A, Tiwari Microwaves, Antennas & Propagation, 2013, doi:10.1049/iet- map.2013.0086. S (2005) Problem definitions and evaluation criteria for the CEC 2005 special. [19] Mukhopadhyay, Arunava, et al. “Bandwidth enhancement of a microstrip patch antenna using Cuckoo Search optimization.” 2017 1st [37] Symington MMF (1990) Fission–fusion social organization inateles and pan. Int J Primatol 11(1):47–61. International Conference on Electronics, Materials Engineering and Nano-Technology (IEMENTech), 2017, [38] L.I. Wong, M.H. Sulaiman, M.R.Mohamed andM.S. HongK. Elissa, doi:10.1109/iementech.2017.8076930. “Grey Wolf Optimizer for Solving Economic Dispatch Problems” IEEE International Conference Pewer and Energy, 2014. [20] Nipotepat Muangkote, Khamron Sunat and Sirapat Chiewchanwattana, “An Improved Grey Wolf Optimizer for Training q-Gaussian Radial [39] L.I. Wong, M.H. Sulaiman, M.R.Mohamed andM.S. HongK. Elissa, Basis Functional-link Nets”, 2014 International Computer Science and “Grey Wolf Optimizer for Solving Economic Dispatch Problems” IEEE Engineering Conference (ICSEC). International Conference Pewer and Energy, 2014. [21] Blanton, H., 1997. An Introduction to Neural Networks for Technicians, [40] R. Damaševičius, C. Napoli, T. Sidekerskiene, and M. Woźniak, “IMF Engineers and Other non PhDs. 9-12 November, Proceedings of the mode demixing in EMD for jitter analysis”. Journal of Computational 1997 Artificial Neural Networks in Engineering Conference. St.Louis. Science, vol. 22 (2017), pp. 240-252. [22] R. Storn, K. Price Differential evolution – a simple and efficient heuristic for global optimization over continuous spaces J Global Optim, 11 (1997), pp. 341–359. [23] D. Simon Biogeography-based optimization Evolut Comput IEEE Trans, 12 (2008), pp. 702–713. [24] E. Rashedi, H. Nezamabadi-Pour, S. Saryazdi GSA: a gravitational search algorithm Inf Sci, 179 (2009), pp. 2232–2248. [25] [A. Kaveh, S. Talatahari A novel heuristic optimization method: charged system search Acta Mech, 213 (2010), pp. 267–289. [26] B. Alatas ACROA: artificial chemical reaction optimization algorithm for global optimization Expert Syst Appl, 38 (2011), pp. 13170–13180. [27] A. Hatamlou Black hole: a new heuristic optimization approach for data clustering Inf Sci (2012). [28] A. Kaveh, M. Khayatazad A new meta-heuristic method: ray optimization Comput Struct, 112 (2012), pp. 283–294. [29] Du H, Wu X, Zhuang J. Small-world optimization algorithm for function optimization. In: Advances in Natural Computation, ed.: Springer; 2006. p. 264–73. 35