A Novel Optimization Technique - AKK

A Novel Optimization Technique - AKK
of 4
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Proceedings of Asia-Pacific Microwave Conference 2007 A Novel Optimization Technique for a Stacked Patch Antenna V. Devaraj, K.K. Ajayan and M. R. Baiju Dept. of Electronics and Communication College of Engineering, Trivandrum, Kerala, India mrbaiju@gmail.com Abstract-In this paper a noveloptimization technique for a stacked antenna is presented. The stacked patchantenna consists of two layers of E-shaped patches. The optimization strategy for this E-shaped stacked patchantenna is basedonGeneticAlgorithm. In the proposed strategy, optimization is done in arecursive manner. The optimizedantenna exhibits across-pol maximum atleast 16dB below theco-pol maximum with -13.5dB returnloss at 2.4GHz. The antenna exhibits an impedance bandwidth of around 10 . The proposed antennacan beused forwireless communication applications in 802.11bband. The optimization is performed in MATLAB and the simulations are verified in Microwave Office. Keywords- E-Shaped Patch antenna, Stacked antenna,Genetic Algorithm I. INTRODUCTION Optimization of the characteristics of a patch antenna by having its structure modified byremoving metal sections in squares from the rectangular patch is described in [1]-[2]. Over years evolutionaryoptimization strategies such as Genetic Algorithmhave become popular in the electromagnetic design and is presently used in a wide variety of electromagnetic applications [2]-[3]. Genetic Algorithm based optimizationtechnique utilizes therecursive nature of GA in arriving at an optimal solution fromrandomly generated various possibilities. The bandwidthof the rectangular patch antennacan be improved by using E-shaped patch antenna [4]. E shaped patch antenna [4] has the attractive features as simplicity and small size compared to other conventional wideband antennas. Probe fedstacked patch antennas of regular shapes (rectangular circular or triangular) have achieved wide attention over years [5]-[14]. Stacking improves the characteristics of patch antennas such as gain, bandwidth, efficiency, and input impedance and reduces the phenomenon of cross polarization [12]-[13]. In this work a novel recursive optimization technique for a stacked antenna consisting of two layers of E shaped patches (a Fed/Driven patch and Parasitic patch) is presented. For optimization first the stacked antenna is optimized by keeping the fed patch intact and modifying the structure of theparasitic patch. Later the structure of the fed patch is optimized keeping the previous configuration of theparasitic patch unchanged. The antenna is optimized over aresonant frequency of 2.4GHz which can beused for WiFi applications. The resulting antenna exhibited good bandwidth over 2.4GHz, with-13.5dB returnloss and good polarization characteristics. The fitness function for optimization in GA is taken as the sum of the input match and thepolarization errors [1]. The total fitness error is to be minimized to achieve thedesired characteristicsfor the antenna. The GA optimization searches for an optimal structure from a large number of candidate solution structures that satisfies the design goals such as minimum return loss and fine polarization characteristics within the desired frequency range. The optimization is performed in MATLAB and the simulations are verified using EMSight, the nativesimulator in Microwave Office  MWO) DesignEnvironment ofApplied Wave Research  AWR) Inc. II. GENETIC ALGORITHMBASED OPTIMIZATION OF PATCH ANTENNA. Inorder to optimize the patch antenna using Genetic Algorithm[1], the E-shaped patch is divided in to 42 rectangular patch sections. This can be directly mapped to a chromosomes of a GA, which are simple binary strings of ones and zeros. A  0 represents absenceof metal and  1 represents the presence of metal in that section [1]. Metalremoval is constrained to be in square (or rectangular)patches. GA willarrive at a shape for the E-shaped patch antenna which satisfies both criteria of low return loss and good polarization characteristics. Each rectangular patch is of size 10mm x 5mm. The probe feed is in the symmetry plane. For the probe feed, a 10mm xlOmm square area is left free. Alltheother cells can take a value of either one or zero. So totallythere are 240 candidate solutions to the problem. In order to reduce the cross polarization component, E-Plane field symmetry is imposed. Hence only half of the actual patch needs to beencoded for the GA. Hence the number of bits required in the chromosome is reduced 20. The overall fitness function is calculated as the sum of the two polarization errors. ETOt S = EP S +EM_S The total fitness error is to beminimized to achieve the desired characteristics for the antenna. 1-4244-0749-4/07/ 20.00 @2007 IEEE.  Fig.6. Blockdiagram indicating the flowof the proposed Genetic Algorithm   GA) based recursive optimization algorithm. The parasitic patch and the fed patch are optimized in arecursive way one after theother alternatively. Genetic algorithm is called two times one for theoptimization of the parasitic patch and theother for the fedpatch successively. III. THE PROPOSED RECURSIVE OPTIMIZATION TECHNIQUE FORTHE STACKED CONFIGURATION. First an E-shaped patchantenna is designed for 2.4GHz [11]. The dimensions for 2.4GHz are Lengthof the ground plane, Lg = 10Omm, Width of the plane, Wg = 80mm, Lengthof thepatch, L = 42mm, Width of thepatch, W = 65mm, DielectricPermittivity er= 2.2, Heightof the dielectric substrate = 10.575 mm, Position of the slot Ps =7mm, Width of the slot W, = 5mm andLength of the slot, Ls = 34mm. The E-shaped patch antenna designed is then stacked over two layers. By considering the design issues [12] related to the bandwidthand considering the first order modes of thepatches,the dielectric constant of the lower and upper layers are chosen as 1.07 and 2.2 respectively. The overall optimization procedure falls in the following steps. 1. Design an E-shaped stacked patch antenna as explained above. 2. The E-shaped patch antenna of the lower layer(Fed patch) is remained as such and the upper layer (parasitic patch) is selected for optimization. 3. Generate a random population of 6 chromosomes per population and 20 bits per chromosomes. Each bit represents asquare metal subsection in thepatch. Synthesize six patches corresponding to the six  Fig.2. Structure of the optimized Fed patch of the proposed stacked E- shaped patch antenna as figured in Microwave Office. The via port is shown by the white block. ~~~~~~~Bj h   b Fig. 4. Return loss versus frequency for the proposed optimized B-shaped stacked patch after 150 generations, simulated in M\WO. chromosomes. Hence now we have 6stacked units in hand. 4. Calculate fitness function (polarization component +return loss using Method of Moments  MoM) of each stacked unit and select the winnersof the current population. 5. By applyingmutation and crossover, generate a new populationof six chromosomes and consequently six new stacked units are synthesized. 6. If the number of generations is less than N return to step4, otherwise retain the stacked unit with minimum fitness value of the Nth generation.(Total Number of gens. N = 150). 7. The stacked unit retained is now one side optimized. The stacked antenna inthis configuration obviously shows improved results compared to the unoptimized E-shaped stacked patch. The parasitic patch now is in its optimum configuration, and the fed patch in its unoptimized configuration. 8. The optimized structure of thetop layer is left free as such, by maintaining the optimized configuration and the bottom layer (Fed patch) is selected for optimization. 9. Repeat steps 3 ,4,5,6, 7. Fig. 3. Structure of the optimized parasitic patch of the proposed stacked E- shaped patch antenna as figured in Microwave Office. The shade of theviaport is seen. u50 0 5 Anqle theta Degrees (b) Fig. 5. Normalized E Plane farfield patternsfor the optimizedStacked E shaped stacked antenna for 2.4GHz. 10. The stacked unit we have now is the optimized one. IV. RESULTS   DISCUSSIONS The optimized configurations of the parasitic patch and the fed patch of the Probe fed E-shaped Stacked patch antenna after the optimization are shown in theFig. 2-3. Both configurations are the structures that arrived at after evolving through 150 generations each, one after the other in Genetic Algorithm. TABLE I indicatesthe various chromosomes that evolved through successive generations in Genetic Algorithm. Since in the first stage of optimization, the structure of thefed patch is kept unchanged, the chromosomes are all ones At the sametime parasitic patch chromosomes undergo evolution throughout the 150 generations. The 150th generation evolved a binary string   11100101011110001110'. This is followed by the optimization of thefed patch, with parasitic patch chromosomes being maintained at the previously optimized binary string, '11100101011110001110'. TheFed patch after another 150 generations reached the optimized binary string '11001111111011110001'. Fig. 4 shows the returnloss ofthe optimized stacked E-shaped patch. The optimized configuration exhibits -13.5dB returnloss at 2.4GHz. The  TABLE I THE PROPOSED RECURSIVEOPTIMIZATION TECHNIQUE Gen. Parasitic Patch Fed Patch I 1101111111001100101125 11110000101010111110 No Metal is 50 00100111100101001001 Removed 100 01111101011001000111 Always in 150 1110010101111000111011111111111111111111 GenFed Patch Parasitic Patch 1 11100101011110001110 No Metal is 25 11100101011110001110 Removed 50 11100101011110001110 Maintained in the 100 11100101011110001110 Configuration. 150 11oo0o1o1011ooo100o10 o10010101o10001o10 crosspol, co-pol characteristics of the optimizedpatch are shown in the Fig. 5. The optimized E-shaped patch antenna is found to exhibitexcellent cross-pol/co-pol characteristics compared to the unoptimized E patch. The optimized configuration has the co-pol maximum at about 16dB above the cross-pol maximum. The visualizations of the optimized stacked E-shaped patch antenna in MWO is shown in Fig 6. The   parameters employed for the optimizationhere are the following. 150 Generations, 6 Chromosomes per Generation, 20 bitsin a single Chromosome.The choice of150 generations is based on thefinding that   reached an acceptable convergence around 140   150 generations when it was put to runover a number of sample generations. Selection type is binary tournament with probability of crossover, 0.7 and probability of mutation, 0.05. 150 generations is taken as the convergence criteria. With this parameter set the approximate time elapsed to run this 150 generations on a P-IV processor with 2.8 GHz speedhaving 1.5GB RAM is (6 chrom./gen.) x (3min./chrom.) x (150 gen.) = 45 hrs, approximately two days. We use the same GA parameters for the optimization of both Fed patch and parasitic patch. Hence the total time elapsed for the optimization ofboth the Fed patchand theparasitic patch is approximately90 hours. Since the optimization is offline and a one time activity, the above run times are not critical and donot impose any constraints. V. CONCLUSION In this paper a Genetic algorithm based recursive optimization technique is proposed to a widebandE-shaped stacked patch antenna. The optimization is aimed at achieving suitablecharacteristics of the patchsuch as such as low return loss and fine polarization characteristics. The optimizedpatch Fig. 6. Structure of the proposed optimized E-shaped stacked patch antenna as represented in MWO. The optimized configurations ofboth thefed patch and parasitic patch antenna exhibited reasonably low return loss -13.5dB,and sufficient marginbetween the crosspol maximum and copol maximum, approximately 16dB withan impedancebandwidthof around 10 . REFERENCES [1] Frank J.Villegas, Tom Cwik, and YahyaRahmat Samii,  A Parallel electromagnetic Genetic Algorithm Optimization (EGO) Application for Patch Antenna Design , IEEE Trans. onAntennas and Propagat., Vol. 52, No.9, Sep. 2004., pp. 2424- 2435. [2] J. M. Johnsonand Y. Rahmat-Samii,  Genetic algorithms and method of moments  GA/MOM) for the design of integratedantennas, IEEE Trans. Antennas Propagat., vol. 47, pp. 1606-1614, Oct. 1999. [3] J. M. Johnsonand Y. Rahmat-Samii,  Genetic algorithm optimization for aerospace electromagneticdesign and analysis, in Proc. IEEE Aerospace Applications Conf   Feb. 1996, pp. 87-102. [4] F. Yang, X. Zhang, X. Ye, and Y. Rahmat-Samii,  Wide-band e-shaped patch antennas for wireless communications, IEEE Trans. Antennas Propagat vol. 49,pp. 1094-1100, July 2001. [5] A.Sabban,  Anew broadband stacked two layerMicrostripantenna . IEEE AP-S, Int. Sym.. Digest, 1983, 63-66. [6] C. H. Chen, A. Tulintseff, and R. N. Sorbello, Broadband Two layer Microstrip antennas , IEEE AP-S, Int. Sym..Digest, 1984, 251-254. [7] P. S. Bhatnagar, et. al.  ExperimentalStudyon stacked triangular Microstrip antenna ,Electron. Lett. 22, 1986, 864-865. [8] R. Q. Lee and K. F. Lee,  Two layer electromagnetically coupled rectangular patch antenna . IEEE AP-S, Int. Sym..Digest, 1988, 948- 951. [9] R. Q Leeand K. F. Lee,  Experimental study of the two-layer electromagnetically coupled rectangular patch antenna , IEEE Trans.Antennas Propagat., vol. 38,pp. 1298-1302, Aug. 1990. [10] Egashira, S., andNishiyama, E,:  Stacked microstrip antennawith wide bandwidthand high gain , IEEE Trans. Antennas Propag.,1996,44, (ll),pp. 1533-1534 [11] S. D. Targonski, R.B. Waterhouse  Designof Wide Band ApertureStackedPatch Microstrip Antennas , IEEE Trans. Antennas and Propagat., vol. -46,pp. 1245-1251, September 1998. [12] Rod B. Waterhouse,  DesignofProbe-Fed Stacked Patches , IEEE Trans. Antennas Propagat., vol. 47, pp. 1780-1784, Dec. 1999. [13] Rod B. Waterhouse, Stacked patches using high and low dielectric constant material combinations IEEE Trans. Antennas Propagat., vol. 47, pp. 1767-1771, Dec. 1999. [14] Nishiyama E, AikawaM, Egashira S,  Three-Element Stacked Microstrip Antenna with Wide-Band and High-Gain Performances , inIEEE International Symposium on AntennasPropagation, Vol. 2, pp. 900   903, Jun. 2003.
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks