for Miniaturized Fractal Slot RFID Antennas Particle Swarm Optimization

Transcription

1 A Design Methodology for Miniaturized Fractal Slot RFID s Using Particle Swarm Optimization 1 D. K. Naji, 2 J. S. Aziz, 3 R. S. Fyath 1,2 Department of Electronic and Communications Engineering, College of Engineering, Alnahrain University, Baghdad, Iraq 3 Department of Computer Engineering, College of Engineering, Alnahrain University, Baghdad, Iraq 1 dknaji73@yahoo.com, 2 jsaziz53@yahoo.com, 3 rsfyath@yahoo.com ABSTRACT An optimization-based methodology for the design of miniaturized microstrip-fed fractal slot RFID antenna is introduced. This methodology gives an automated process for generation of antenna geometrical parameters and mainly associated with Particle Swarm Optimization (PSO). Two software programs are used in parallel, and in a coupled synchronized fashion. The first one is the electromagnetic simulator CST Microwave Studio which gives instantaneous estimation of the antenna performance. The second is MATLAB which is used to run the PSO algorithm in order to optimize the geometry of the antenna after considering two objective functions: return loss and antenna size. The design methodology is applied to a third-order Minkowski fractal antenna which is optimized using two different approaches, the direct one which is based on the conventional (non-fractal) antenna, and the sequential one which is based on the optimized previous-order fractal antennas. The simulated results show excellent performance requirements with less than 40 of return loss, stable gain and radiation pattern, and more than 90% reduction in overall size in comparison with the conventional reference antenna. Keywords: Minkowski fractal, microstrip slot antenna, particle swarm optimization, sequential and direct optimization. 1. INTRODUCTION Recently there is increasing interest in RFID systems operating at microwave frequency (5.8 ) to achieve higher range with higher data transfer rate. In this circumstance, printed microstrip slot antennas are a very attractive choice because of their well-known advantages of low profile, light weight, easy integration with monolithic microwave integrated circuits, low cost, easy fabrication, and stable radiation patterns [1]-[4]. Therefore, great interests in various wide-slot antennas with different feed methods for 5.8 RFID applications have been reported in the literature [5]-[9]. However, miniaturization is still of main concern with the design process of such RFID antennas. The growing need for miniaturization, not only requires small devices, but also small-sized radiators. Fractals, owing to their geometrical properties, can be used successfully in antenna miniaturization and recently some interesting applications have been studied and presented in the scientific literature [10]. Also, the space filling property defined by fractal geometry and applied to a slot antenna element leads to reduce the total area occupied by the antenna [11]-[13]. Usually, antenna miniaturization design involves many geometrical or material parameters. These parameters may be discrete, and often include constraints in allowable values. Optimizing such antennas to closely approximate desired criterion performance is similar to searching the global solution from a multidimensional solution space. So far, many stochastic evolutionary search techniques, such as Simulated Annealing (SA) and Genetic Algorithms (GAs) have been employed successfully in antenna design [14]. Particle Swarm Optimization (PSO), based on the simulation of a simplified sociological behavior associated with swarm such as bees, bird flocking, and fish schooling, is an alternative optimization algorithm first proposed by Kennedy and Eberhart in 1995 for solving multidimensional discontinuous problems [15]. The PSO algorithm has been shown to be an effective alternative to other evolutionary algorithms in handling certain kinds of optimization problems. Compared to GA and SA, the PSO algorithm is much easier to implement and apply to design problems with continues or discontinuous parameters. Especially, the PSO, used in conjunction with the numerical electromagnetic solver, is found to be a revolutionary new approach to antenna design and optimization [16]. It is worthy to note here that, to the author s knowledge, there is no published works that use optimization techniques for miniaturizing fractal RFID antennas. The related published works are concerned with conventional (non-fractal) RFID antennas, and use optimization techniques for gain enhancement [17], multiband applications [18, 19] and robust environment and performance characteristics [20, 21]. Recently, Hidetoshi et al. [22] have presented an approach to optimize UHF mender-line RFID antennas for maximizing power and minimization of antenna size. In this paper, PSO, implemented in MATLAB, is used in along with CST Microwave Studio (CST MWS), which is an electromagnetic simulator based on finite integration time domain (FITD) [23], to introduce new design methodology for miniaturizing fractal slot antenna. The methodology is applied to Minkowski fractal slot antenna as an illustrative example. The convergence of the proposed methodology is studied in terms of optimization behavior and success time rate. The originality of the work in this paper resides in analyzing the proposed algorithm in actual real time electromagnetic design by utilizing both MATLAB (as the optimizer tool) and CST MWS (as the electromagnetic tool). 477

2 2. PSO ALGORITHM Particle swarm optimization is a population based search algorithm initialized with a population of random solutions, called particles. Each particle flies throughh the search space with a velocity that is dynamically adjusted according to its own and its companion s previous behavior. This velocity consists of three parts, the social, the cognitive, and the inertia parts. The social part is the term guiding the particle to the global best position achieved by the whole swarm of particles so far, the cognitive term conducts it to the local best position achieved by itself so far, and the inertia part is the memory of its previous velocity. Figure 1 shows the flowchart of a PSO algorithm. During the PSO operation, each candidate solution is represented as a particle with position and velocity represented by and, respectively. Therefore, for N- dimensional optimization problems, the position and velocity of the ith particle can be represented as,,,,,, and,,,,,, where is the transpose operator. Each particle of must be kept changed within the allowed solution space, bounded by,,, to avoid physically invalid designs that allows an error to encountered by the fitness evaluator (CST MWS) and then an optimization process may be failed. If position of any particle exceed the limit, one of three control approach can be used, which are absorbing wall, reflecting wall, and visible wall. In this paper, an absorbing wall and reflecting wall techniques are used for their effectively pull back any particles that flies outside the boundary of solution space in that dimensions, by forcing their velocity to zero or changing the sign of their velocity, respectively. The following formulae demonstrates the updating process of positions and their velocities matrices, and, respectively, for N geometrical parameters and M particles in swarm [25] 1 2 Define the Solution Space, Fitness Function and Population Size Initialize X, V, P and G For Each Iteration For Each Particle Evaluate Fitness If Fit (X) > Fit (G) then G = X If Fit (X) > Fit (P) then P = X Solution is Final Next Particle Update Position Using (1) Update Velocity Using (2) Fig. 1: Chart showing the main steps of the PSO algorithm [24]. In (1) and (2), denotes the current iteration, and is the time interval between two consecutive iterations which is assumed to be unity. In (1), and are matrices where and are calculated and stored in each iteration. The parameters and are the cognitive and social factors whose values are set in the range [1.0, 2.0]. It has been suggested that the best choice is 2.0 makes the weights for cognition and social parts on average to be 1.0 [25]. In (1), the time varying parameter, which is called inertia factor, decreases with iteration number from a maximum value at the first iteration and goes to a minimum at the last iteration. Two statistically independent random variable and are both uniformly distributed in the interval [0, 1], are introduce to stochastically vary the relative pull of the personal and global best particles. 3. PROPOSED DESIGN METHODOLOGY PSO algorithm has been tested by different research groups to different benchmark functions and the results show that it is an excellent global optimizer that can be used for different electromagnetic problems especially antenna miniaturization that used in this paper. To calculate the antenna fitness function associated with the PSO algorithm, a comprehensive numerical modeling must be carried out to simulate the electromagnetic (EM) performance of the antenna at each stage of optimization. The EM model should be very efficient in both speed of computation and accuracy since the geometry of the fractal antenna is relatively more complicated than the conventional counterpart and the dimensions of some structure parameters are much smaller than the operating wavelength. The required EM model features are recovered in this paper by using a commercial EM simulator namely CST MWS. This simulator uses FITD method to assign the EM properties of antennas and has been proven in the literature as a powerful and very accurate tool for this purpose. 478

3 3.1 PSO-FITD Algorithm In this work, the fractal RFID tag antenna is optimized using PSO technique while the FITD method is used in parallel with it to compute the EM part of the fitness function (see Fig. 2). The PSO technique runs under MATLAB environment and the FITD method is offered by CST MWS software package. For each generation of the PSO algorithm, the antenna geometrical parameters are updated and mapped to CST MWS to simulate the EM properties of the antenna. According to the EM simulator results that mapped back to MATLAB environment, the fitness function is evaluated by the PSO kernel. the itness function : 1, 1,2,, 4 Initial Parameters 20 log 5 PSO Parameter Update Parameters Fitness Evaluation CST MWS Return Loss In (4), refers to the Heaviside step function while and denote, respectively, the area of the nth-order and optimized 1th-order fractal antennas. In (5), and refer, respectively, to the return loss and the input impedance of the antenna at the resonance frequency, and is the characteristic impedance ( 50Ω. Note that the optimization fitness function (4) consists of two objective functions whichh are related to antenna return loss and antenna area No MATLAB Enough Iteration? Fig. 2: Flowchart of the PSO/FITD algorithm. 3.2 Optimization model Yes Optimal Design The general nonlinear global optimization problem to be solved is mathematically defined as: find the set,,, of variables that will minimize the function 3 0, 0 3 :, 1,2,, 3 where is the fitness function, is the equality constraint, is the inequality constraint, and is the vector of design variables. Also, and are the lower and upper bounds on the design variables, respectively. The goal for the electromagnetic miniaturization design considered here is to minimize the size of the fractal antenna by altering the geometrical parameters within allowed prescribed ranges while keeping the return loss below a desired threshold value at the required resonance frequency. A suitable optimization model is where 6 and where and denote the return loss and area objective functions, respectively. The objective function represents the amount of matching at the desired frequency and its value between "0" and " ". Its zero value denotes that the goal is achieved, that is to say a return loss of at least at the desired frequency is satisfied. The objective function is used to achieve a minimum area from the optimization process. The range of is between 0, 1; it is zero if the area of n-th order fractal antenna is equal to the optimized area of the 1-order counterpart, and to " 1" when the area of the nth-order fractal antenna is zero (not physically allowed). Thus the value of the total fitness function "" which is considered as multiobjective optimization problem ranges between a minimum value of " 1" to a maximum value of " " 479

4 3.3. Design via Miniaturization Methodology (Sequential and Direct) An antenna miniaturization methodology for fractal based microstrip antenna is investigated in this section. Two optimization approaches, sequential and direct optimization approaches, (SOA) and (DOA), respectively, are considered in this work. In the SOA, the following steps are performed (see Fig. 3) i) Design a conventional (non-fractal) microstrip antenna using CST MWS. The dimensions of the structure parameters are scanned to achieve the design requirements ( ) at the desired resonance frequency. This antenna will be considered as a reference antenna for the next optimization stage and its area is denoted by. ii) Apply the PSO/FITD algorithm described in subsection 3.2 to the reference antenna. This procedure yields the optimized zero-order fractal antennaa (i.e., optimized reference antenna) of are. Equation 4 is used here to describe the fitness function after replacing by and by. iii) Design an optimized first-order fractal antenna using the PSO-FITD algorithm. The input parameters used to start the PSO technique are the values of the structure parameters associated with the optimized version of the previous-order antenna (zero-order fractal antenna). Equation (4) is used to calculated the fitness function after replacing by and by. iv) Repeat step (iii) to design the optimized version of the next-order fractal antennas. v) Stop when the desired order of the fractal antenna,, is achieved. In the direct optimization approach, the desired-order fractal antenna ( ) is optimized directly with respect to the reference antenna using one-shot optimization technique: Hence (4) is used after replacing by and by. where denotes here an area of the desired-order antenna. 4. DESIGN OF MINKOWSKII ANTENNA FRACTAL SOA Yes Design Reference at Resonance Frequency Select Optimization Approach, SOA or DOA SOA or DOA Fractal Order, 0 PSO/FIT Algorithm Using eqn. (4) Optimum Designed - at Order Fractal with Area Fractal Order, 1 No Optimum Designed - order Fractal at with Area DOA Fractal Order, PSO/FIT Algorithm Using eqn. (4) To validate the ability of this methodology in antenna design, a third-order Minkowski fractal slot antenna (MFSA) is chosen as an example to miniaturize it at microwave ISM band of 5.8. Two optimization approaches, sequential and direct, are applied for this requirement in order to check if one can get the same performance and area for the designed antennas (i.e., a global optimizer approach) ). The three orders of Minkowski fractal structure is shown in Fig. 4. The geometry of the conventional microstrip-fed slot antenna is illustrated in Fig. 5. A rectangular slot with length and width is printed at the centre on the ground side of a dielectric substrate of length and width. The antenna is fed with the main 50Ω stripline followed by an op- Fig 3: Flowchart showing the main steps of the proposed Methodology. -en-circuited microstrip line printed on the other side of the substrate. To reduce costs, the antenna is printed on commercial available dielectric substrate with a permittivity of 4.3 and loss tangent ( tan 0.02). In order to achieve very thin antenna requirement for RFID applications, a substrate height 0.8 is used. The stripline width of the main 50Ω-stripline is designed through CST MWS and found to be

5 TABLE I RANGES OF THE DESIGN PARAMETERS FOR THE MINIATURIZED FRACTAL-SLOT ANTENNA. Parameter Ranges Parameter Ranges (mm) 3 ~ ~ 0.1 (mm) 3 ~ ~ ~ ~ ~ ~ ~ ~ ~ 0.4 ; 7 ; ; 7 Fig. 4: Scheme of the Minkowski fractal slot antenna orders structure. (a) (MFSA0). (b) (MFSA1). (c) (MFSA2) (d) (MFSA3). The fractal antennas (MFSA1-MFSA3) have additional structure parameters whose number equal to four times the antenna fractal order. Thus the numbers of additional parameters is 4, 8, and 12 for 1st-, 2nd-, and 3rd-order fractal antennas, respectively, (see Fig. 4). It is clear from the above discussion that the number of geometrical parameters increases with the order of the fractal. Thus the computation time of the optimization technique increases as the order of the fractal antenna increases. To solve this limitation, four scaling factors are introduced here for any fractal order. These four scaling factors are,,, and defined as follows 8 8 8c 8 Fig. 5: Geometry of the reference microstrip fed-slot antenna. (a) Feed side. (b) Ground side. (c) Bottom side. It is clear from Figs. 4 and 5 that, there are eight geometrical parameters common for the four fractal antennas (MFSA0 MFSA3). Four of these parameters are related to the ground side (length, width, slot length, and slot width ) and the other four parameters are for the feed side (stripline lengths and and widths and ). The stripline width is set to 1.56 for a 50Ωthickness as stated microstrip feed-line and 0.8 -substrate before. This leaves seven common parameters to be adapted during the optimization process. In this work, five of the common parameters enter the optimization process as scaling factors with respect to the other two main parameters, ground length and width. This is useful to ensure that physically invalid structures to be not constructed and consequently prevent total failure of the optimization process. These scale factors are where,,,,,,,,, 9,,, 9 Table I lists the constraints applied to antenna geometrical parameters during the optimization process. The design methodology starts with design of the reference slot antenna which represents the 0th-order structure MFSA0. The reference antennaa design starts by selecting the operating frequency (5.8 ), substrate with required permittivity (4, 4.3) and the thickness of the substrate ( 0.8 ) which is very thin suitable for RFID tag applications. For exciting the dimensions of ground and slot the following equations [26] 1, 2,, 9 9 operating frequency at, the can be roughly designed with 481

6 Here (, is the free space wave length, is the speed of light in vacuum, is the effective relative permittivity of the dielectric substrate, is the ground width, is the ground length, is the ground slot width, is the ground slot length and is the stripline width. Using (10a) and (10b), the reference slot antenna is designed using the following initial parameters The designed slot antenna is simulated using CST MWS by sweeping the ground and slot parameters within the limit 1 5%, % 2 The simulation takes into account other structure parameters associated with the stripline parameters, strip lengths, and strip width. Table II shows the final antenna geometrical parameters for the reference slot antenna (see Fig. 5) RESULTS AND DISCUSSION 5.1. PSO Behavior and Optimized Geometrical Parameters Illustrative results related to the design of a third-order Minkowski fractal antenna are given here. The threshold value of,, used in the fitness function is 25. Figure 6 reveals the progress of the PSO algorithm as a function of iteration number for both sequential and direct approaches. Return loss and area objectivee functions as well total fitness function are illustrated in this figure. The behavior of the total fitness function is characterized by two representations here. The first representation marks the best value of the total fitness function up to the current iteration. Such representation is useful to record the performance progress that occurs during the history of the optimization process. The best value of the fitness function up to the last iteration gives the required solution. The second representation marks the best fitness function at each PSO iteration and illustrates the instantaneous variation of the fitness function during the optimization process. The results in Fig. 6 reveals that the two approaches of optimization reach nearly the same value of fitness of 0.9 for the 3rd-order fractal antenna (MFSA3). Further, for the initial population, the return loss objective function of the best individual is greater than 10 with total fitness values greater than 12. However, the results are improved quickly at iterations 12, 20, 19, 12, and 8 for sequential approach related to MFSA0, MFSA1, MFSA2, MFSA3 and direct approach MFSA3, respectively. In PSO technique, the selection of the particle number and the maximum iteration depends on the dimension of the solution space and the fitness function. The number of particles should be comparable to the dimension of the solution space to obtain a good convergence. A reasonably large number of iterations are also necessary for the particles to get converged. In this paper, the best number of particles used is chosen between 3 to 5, where is the number of antenna geometrical parameters that are associated with the optimization technique. Thus, a 28-particle-swarm is used for 0th-order fractal antenna and 33-particle for antennas MFSA1, MFSA2 and MFSA3. Furthermore, a stop criterion is chosen such that 50 PSO iterations are reached or the fitness function remains unchanged with less than 2% error for at least 20 successive iterations. TABLE II OPTIMIZED ANTENNA GEOMETRICAL PARAMETERS TO ACHIEVE 5.8 FOR THE REFERENCE SLOT ANTENNA. Parameter Value (mm) Parameter Value (mm) The optimized antennas are simulated by using a laptop (hp pavilion dv6, 2.54, 4-core CPU and 4 RAM). In the sequential approach, the simulation times required to optimize 0th-order, 1st-order, 2nd-order, and 3rd-order fractal antennas are 53.50, 65.42, and hour. In the direct approach, hours is required to optimize the 3rd-order fractal antenna. Thus, the average time of each optimization run is about 60 hours. It is worth to mention here that one of the main limitations of the proposed methodology is the relatively long computational time and therefore it requires an efficient PC system to handle it successively without run failure. The PSO algorithm creates an optimal miniaturized antenna size in the two optimization approaches. Table III summarizes the final optimized geometrical parameters. One can observe from Table III that both sequential and direct optimized 3rd-order fractal antennas have almost the same geometrical and performance parameters. This proves that the proposed methodology has the ability to reach the same optimized dimensional results for the miniaturized antenna, i.e., global optimization results. 482

7 Fig. 6: Variation of return loss and area objectives as well total fitness with PSO iteration number for different fractal order. (a) 0th- order (b) 1st-order (c) 2nd-order (d) 3rd-order (sequential) and (e) 3rd-order (direct). 483

8 Fig.6: Continued. TABLE III OPTIMIZED GEOMETRICAL PARAMETERS OF MINKOWSKI FRACTAL SLOT ANTENNA Optimization Approach Type Geometrical Parameters SA RSA MFSA MFSA MFSA MFSA NA NA NA NA NA NA NA NA DA MFSA Legend: SA= Sequential Approach; DA=Direct Approach; RSA=Reference Slot ; MFSA=Minkowski Fractal Slot ; NA=Not TABLE IV SIMULATION RESULTS OF MINKOWSKI FRACTAL SLOT ANTENNA. Optimization Approach Type Performance Parameters % % SA RSA MFSA MFSA MFSA MFSA DA MFSA Legend: SA= Sequential Approach; DA= Direct Approach; = -10 db lower frequency; = -10 db higher frequency. 484

9 5.2. Performance Results of the Optimized MFSA The electromagnetic properties of the optimized antennas are simulated using CST MWS. Table IV lists some of the simulation results, namely, return loss, antenna gain, total antenna efficiency, bandwidth, and size reduction. The size reduction is computed as 13 where and is, respectively, the area of the antenna under observation and reference antenna. Fig. 7: Simulated return losses of the optimized antennas. A result related to the reference antenna is given for comparison purposes. Note further that the optimized reference antenna offers 58.5% area reduction with respect to the conventional reference antenna. Introducing the 2nd-order fractal geometry yields 73.26% area reduction with respect to the optimized reference antenna. The results in Table IV also highlight the following findings 1) A return loss less than 40, the gain greater than 2, and efficiency greater than 87% are obtained for all the designed antennas. 2) Bandwidth enhancement is obtained when the antenna is designed with fractal order more the one. The 2nd, 3rd (sequential), and 3rd (direct) fractals offer 1.46, 1.89, and 1.69 bandwidth, respectively, while the reference antenna gives 0.73 bandwidth. 3) Both sequential and direct optimized 3rd-order fractal antennas have almost the same performance parameters. The return losses of the optimized antennas are shown in Fig. 7. A result related to the reference antenna is also given for comparison purposes. It is clear from this figure and Table 4 that the 10 lower frequency is approximately the same for all antennas as compared with the 10 higher frequency which increases with increasing of fractal order. Also one can notice that a second resonance frequency appears as the order of the fractal is exceeds 1. Thus the fractal geometry enhances the antenna bandwidth and opens the possibility to design the structure for two-band operation. Figure 8a and 8b show, respectively, the gain and efficiency for the designed antennas. It is clear from these figures that the third-order fractal has almost similar spectral behavior for both optimization approaches. Also a stable gain of 2 over is obtained for the third-order fractal in both approaches. The radiation characteristics of the optimized antennas are also studied and the results are depicted in Fig. 9 which shows the 2D radiation pattern in the elevation direction 90 ) and 90 ) planes and azimutal direction 0 ) plane at 5.8 for the optimized antennas. It is appears that antenna radiates a nearly omindirectionally component in the -plane when the order of fractal is more than 1, but the radiation patterns show two nulls for component at 0 and 180 for - (a) (b) Fig. 8: (a) Gain of the optimized antennas. (b) Efficiency of the optimized antennas. A result related to the reference antenna is given for comparison purposes. -plane and two nulls for component at 90 and 270 for -planes. Note further that the antennas have relatively strong cross-polarized radiation 18 below co-polarized radiation). It is nteresting to notice the strong similarity between patterns for components for all antennas at the three planes, and. Table V shows the maximum electric field components and for the three antenna planes to clarify the similarity between the patterns. The 3D radiation characteristics of the optimized antennas are depicted in Figs. 10a-d where the 3D radiation pattern are plotted for zero-, 1st-, 2nd-, and 3rd-order fractal antennas, respectively. One can notice from these figures that omnidirectional radiation is achieved when the fractal order increases beyond 1 485

10 . -plane -plane -plane Fig. 9: Radiation Patterns for the optimized antennas; (a) Zero-order (b) 1st-order (c) 2nd-order (d) 3rd-order fractal. 486

11 Fig. 10: 3D Radiation Patterns for the threee fractal order iteration of Minkowski fractal slot antenna; (a) Zero-order (b) 1st-Order (c) 2nd- Order (d) 3rd- Orderr It is worth to note here the following findings which are common for all the optimized antennas 1) The electric field components and represent, respectively, a co-polar component (i.e.. maximum) and a cross-polar (i.e. minimum) in and -plane. 2) The electric field components and represent, respectively, a co-polar component and a cross-polar in plane. 3) The ratio of co-polar to cross-polar components is greater than 58.4 for -plane and greaterr than 17.8 for -plane and more than 22.8 for -plane. Table VI presents a performance comparison between the optimized 3rd-order fractal antenna selected from this work and some RFID antennas reported in the literature for 5.8 GHz applications. It s worth to emphasize here that the antenna designed in this work is characterized by the smallest area compared with other structures. The area of this antenna corresponds to 0.52 of the area of the antenna reported in Ref. [9] which has the smallest area among previous antennas listed in Table VI. Note further that the designed antenna offers the lowest gain and highest bandwidth compared with others. TABLE V MAXIMUM RADIATION ELECTRIC FIELD VALUES (IN /) FOR OPTIMIZED ANTENNAS. 1st-order 2nd-order 3rd-order 3rd-order (direct) Radiation plane, , ,

12 TABLE VI PERFORMANCE COMPARISON BETWEEN THE 3RD-ORDER FRACTAL ANTENNA REPORTED IN THIS WORK AND RFID REPORTED ANTENNAS IN THE LITERATURE AT 5.8 GHZ ISM BAND. Ref Structure Aim Laminate % This work Stripline-fed 3rdorder Fractal Slot Miniaturization [9] CPW-fed dual folded-strip monopole antenna Miniaturization [8] CPW-fed shorted F- shaped monopole antenna Miniaturization [5] CPW-fed Slot Broadband [7] CPW-fed Folded Slot Dual-band (2.45, 5.8) GHz [6] CPW-fed Slot Miniaturization [18] CPW-fed Slot antenna Tri-Band (0.95, 2.45, 5..8) GHz CONCLUSION An efficient methodology of designing miniaturized fractal antennas for RFID applications has been proposed. The antenna geometrical parameters are optimized using PSO algorithm which runs on MATLAB environments and synchronously coupled to full wave electromagnetic simulator implemented using CST Microwave Studio software. The used optimization objective functions reflex both return loss and antenna size. The proposed design methodology has been applied to Minkowski fractal slot antenna operated at 5.8. The results reveals that more than 90% reduction in overall antenna size can be obtained for fractal order two or more as compared with the conventional reference antenna. Further, excellent performance requirements are obtained with less 40 return loss and more than 2 gain with stable radiation pattern has been achieved. The 3rd-order fractal antenna has a just overall size of (i.e., , where is the wavelength at 5.8 GHz). REFERENCES [1] L. WeiXing, Y. YinZeng, X. WenLong, and Z. Shaoli, Compact Open-Slot With Bandwidth Enhancement, IEEE and Wireless Propag. Lett., pp , Vol. 10, [2] H.C. Richard, and L. Yi-C., Miniaturized Design of Microstrip-Fed Slot s Loaded With C-Shaped Rings, IEEE and Wireless Propag. Lett., pp , Vol. 10, [3] D. Aliakbar, I. Ali, and N. M. Mohammad, Printed Wide-Slot for Wideband Applications, IEEE Trans. on and Propag., Vol. 56, No. 10, Oct [4] J. Jen-Y., and K. Jui-C.., Novel Printed Wide-Band Rhombus-Like Slot With an Offset Microstrip- Fed Line, IEEE and Wireless Propag. Lett., Vol. 6,

13 [5] W. C. Liu and P. K. Hu, Broadbandd CPW-fed folded- Electron. slot monopole antenna for RFID application, Lett., vol. 41, no. 17, pp , Aug 18, [6] S. Y. Chen and P. Hsu, CPW-fed folded-slot antenna for 5.8 GHz RFID tags, Electron. Lett., vol. 40, no. 24, pp , July 25, [7] M.-T. Zhang, Y.-C. Jiao, Dual-band CPW-Fed Folded Monopole for RFID Application, Electronics Lett., Vol. 42, No. 21, Oct [8] W. C. Liu and C. M. Wu, CPW-fed shorted F-shaped monopole antenna for 5.8-GHz RFID applications, Microwave Opt. Technol. Lett., vol. 48, no. 3, pp , Mar [9] W. C. Liu and P. C. Kao, Compact CPW-fed dual folded-strip monopole antenna for 5.8-GHz RFID application, Microwave Opt. Technol. Lett., vol. 48, no. 8, pp , Aug [10] J. P. Gianvittorio and Y. Rahmat-Samii, Fractal antennas: A novel antenna miniaturization technique, and applications, IEEE s Propag. Mag., vol. 44, pp , Jan [11] Y. J. Sung, Bandwidth Enhancement of a Wide Slot Using Fractal-Shaped Sierpinski, IEEE Trans. on s and Propag., Vol. 59, No. 8, Aug [12] L. C. Wen, M. W. Guang, and X. Z. Chen, Bandwidth Enhancement of a Microstrip-Line-Fed Printed Wide- Slot With a Fractal-Shaped Slot, IEEE Trans. on s and Propag., Vol. 57, No. 7, July [13] D. D. Krishna, K. Gopikrishna, C. K. Anandan, P. Mohanan, and K. Vausdevan, CPW-Fed Koch Fractal Slot for WLAN/WiMAX Applications, IEEE and Wireless Propag. Lett., Vol. 7, [19] H. K. Jae, T. K. Koon, H. K. Dong, H.-B. Lee, and S. K. Hyeong, A Practical Approach to Robust design of a RFID Triple-Band PIFA Structure, IEEE Trans. on Magnetics, Vol. 46, No. 8, Aug [20] H. Chaabane, E. Perret, and Tedjini, A Methodology for Design of Frequency and Environment Robust UHF RFID Tags, IEEE Trans. on s and Propag., Vol. 59, No. 9, Sep [21] T. K. Koon, H. K. Jae, C. Kyung, and S. K. Hyeong, Robust Optimum Design of PIFA for RFID Mobile Dongle Applications, IEEE Trans. on Magnetics, Vol. 47, No. 5, May [22] M. Hidetoshi, W. Yuta, W. Kota, and I. Hajime, Evolutional Design of Small for Passive UHF- on Magnetics, Vol. 47, No. 5, Band RFID, IEEE Trans. May [23] CST Microwave Studio: [24] M. K. Majid, G. C. Christos, Linear Array Geometry Synthesis With Minimum Sidelobe Level and Null Control Using Particle Swarm Optimization, IEEE Trans. on s and Propag., Vol. 53, No. 8, Sep [25] J. Nanbo, and R-S. Yahya, Advances in particle Swarm Optimization for Designs: Real-Number, Binary, Single-Objective and Multiobjective implementation, IEEE Trans. on s and Propag., Vol. 55, No. 3, Mar [26] C. A. Balanis, Theory: Analysis and Design, 3rd ed., NJ: Wiley, [14] L. Davis, Genetic Algorithms and Simulated Annealing. London, U.K.: Pittman, [15] J. Kennedy and R. C. Eberhart, Particle Swarm Optimization, in Proc. IEEE Conf. Neural Networks IV, Piscataway, NJ, [16] J. Robinson and Y. R.-Samii, Particle Swarm Optimization in Electromagnetic, IEEE Trans. s Propag., vol. 52, pp , Feb [17] G. Marrocco, Gain-optimized Self-Resonant Meander Line s for RFID Application, IEEE and Wireless Propag. Lett., pp , Vol. 2, [18] H. H. Li, X. Q. Mou, Z. Ji, H. Yu, and L. Jiang, Miniature RFID Tri-Band CPW-Fed Optimized Using ISPO Algorithm, Electronics Letters, Vol. 47, No. 3, Feb