See discussions, stats, and author profiles for this publication at: https://www.researchgate.


Design of optimum gain pyramidal horn with
improved formulas using particle swarm

Article in International Journal of RF and Microwave Computer-Aided Engineering · September 2007
DOI: 10.1002/mmce.20245


17 788

3 authors, including:

Nihad Dib
Jordan University of Science and Technology


All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Nihad Dib
letting you access and read them immediately. Retrieved on: 02 November 2016

These formulas are iterative techniques. respectively. Several methods [1–8] have been gives more accurate results than the previous meth- presented and used to design a pyramidal horn with ods.Design of Optimum Gain Pyramidal Horn with Improved Formulas Using Particle Swarm Optimization Yahya Najjar. expansion) for the path length error that gives rise to In this article. 2007. V C 2007 Wiley Periodicals. Inc. The method proposed in [6] plane slant lengths. the Optimum gain pyramidal horns are widely used as horn problem was formulated in terms of a fourth standard gain antennas since their gains can be calcu. The methods des- [1. high-gain and lengths are derived using the more accurate gain horns. Int J RF and Microwave CAE 17: 505–511. improved formulas for the relation- aperture quadratic phase error [1]. Nihad Dib Electrical Engineering Department. The method ship between the optimum gain horn aperture size described in [1] is only valid for long. accepted 29 August 2006 ABSTRACT: In this article. expression given in [9]. Jordan University of Science and Technology.1002/mmce. in limitations. Also. the actual gain results do not agree well with the desired gain results.1 dB using their known dimensions the narrow aperture dimension. and is not restricted to high-gain then used along with the particle swarm optimization technique [10–13] to find directly the physical dimen- Correspondence to: N. but it needs a large number of inputs and outputs optimum gain. Keywords: horn antenna.O. particle swarm optimization I. This method is based on improving existing design equations along with the use of particle swarm optimization technique. One com.20245 Published online 22 June 2007 in Wiley InterScience (www. which does not involve the The method proposed in [2] avoids the application of approximate path length error.wiley. and computational complexity. Improved horn dimensions. a new method to design an optimum gain horn antenna is pre- sented. are obtained which give the desired gain almost exactly. Inc. give the desired gain exactly. The obtained results are more accurate than those available in the literature since no path length error approximation is involved. VC 2007 Wiley Periodicals. and uses trial and error (iterative technique). to develop and train a neural network. the differential evolution technique and the mon feature in [1–8] is the use of the expressions that particle swarm optimization method were used to give optimum gain that are derived from an approxi. Each method has its own accuracy. 8]. However. Jordan Received 5 June 2006. Box 3030. e-mail: nihad@just. Dib. degree polynomial that was analytically solved for lated to within in [2] using curve fitting. obtained using other techniques [1–8] and shown to interscience. for different 15 designs. Irbid. INTRODUCTION horn design. 2].jo sions of the horn. [7. 505 . derive new expressions for the narrow and wide aper- mate formula (the first two terms of binomial series ture dimensions. P. An optimum gain pyramidal horn antenna is cribed in [3–5] improved the design equations given one that produces maximum gain for given E.and H. Our results are compared to those DOI 10. Mohammad Moneer.

25 k and 0.and H-planes. B1. Such an approximation was done to simplify the derivation.506 Najjar. The other parameters (B1. (a) Pyramidal horn antenna (b) E-plane view It is interesting to note that the above expression (c) H-plane view. PE) and H-plane (A. In [9]. an expression for the gain of pyramidal horn antenna was derived without approximating the path length error. Moneer. However. and Dib II. A horn is said to be optimum when the aperture ure 1. Here. respectively [1]. and PH) are then determined using a specific with the E-plane (B. the wavelength k. was used in [2–8] to find the gain of the designed International Journal of RF and Microwave Computer-Aided Engineering DOI 10. According to [1]. PH) cross-sectional views as shown in Fig. PE. design method. and the inner given slant lengths in the E. A1.375 k. The expression is given by: 32A1 B1 G¼ ðLe ðSÞÞðLh ðTÞÞ ð3Þ pk2  B =2 2  Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2   2 Le ðSÞ ¼  ejKL1 1þðy=L1 Þ 1 dy ð4Þ  1 B  0  A =2 2  Z1   pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2   p px Lh ðTÞ ¼  cos ejKL2 1þðy=L2 Þ 1 dx  A1 A1  0 ð5Þ Figure 1. A1. LE. LE. which gives rise to the quadratic phase error. (1) and (2) were not exact since the path length error. A1) that give the maximum gain. Consider the geometry of a pyramidal horn antenna L2. In most cases. L2. it should be emphasized that the expressions used to derive eqs. B1) and (L2.and H- planes are 0.1002/mmce . The two equations above were obtained by means of curve fitting for (L1. PYRAMIDAL HORN DESIGN dimensions A and B of the feeding rectangular wave- guide. To design a pyramidal horn. LH. LH. The above approxi- mate relations were used in all the design methods presented in [1–8]. this occurs when the following relations are satisfied: pffiffiffiffiffiffiffiffiffiffiffi B1 ¼ 2k L1 ð1Þ pffiffiffiffiffiffiffiffiffiffiffi A1 ¼ 3k L2 ð2Þ This means that optimum gain is obtained when the maximum aperture phase deviations in the E. L1. L1. new improved formulas will be presented as explained below. it yields to very good results for long high gain horns. which led to closed form expressions for the electric and magnetic fields in terms of cosine and sine Fres- nel integrals. was approxi- mated by the first two terms of a binomial series expansion of the exact expression [1]. one usually knows dimensions are adjusted to give maximum gain for the desired gain G.

If the values of B1 (in maximum gain in the E-plane and H-plane of a py- k) which correspond to the maximum gains in Figure ramidal horn antenna were given as 0.interscience. 2 are plotted versus their corresponding values of L1 respectively. Normalized H-plane gain of pyramidal horn as a function of A1 for different lengths of L2. Optimum Gain Pyramidal Horn Design 507 Figure 2. The normalized E-plane gain Gn This corresponds to maximum phase deviation in the is defined as: H-plane of 0. as A1 ¼ 3:1831kL2 ð9Þ shown in Figure 2. the following equa- the use of the approximate eqs.] pyramidal horns. Our results obtained using eqs. (1) and (7) a function of B1 for different lengths of L1. To our knowl- curve fitting for 25 [Color figure can can be viewed in the online issue. (2). edge. (2) and (9) as compared to those Gn ¼ ð6Þ obtained directly from eq. E-plane normalized gain of pyramidal horn as Figure 3. the phases that correspond to the Gauss/Lobatto quadrature rule. although the design itself involved corresponding values of L2 (in k).2633 k. tion can be obtained using curve fitting for 25 points: Now. Here. Figure 5 shows the differences Gpk2 between eqs. (7) gives better results for B1 than eq. (3) using the peaks in Figure k. numerically within an error of 109 using an adaptive Recently.] www. (1) and (2). (3). It can be seen that eq.interscience. (3) using the peaks in Fig- 32A1 Lh ðTÞ ure 4. the H-plane normalized gain Gn is defined as: Gpk2 Gn ¼ ð8Þ 32B1 Le ðSÞ Figure 4. in [14].] International Journal of RF and Microwave Computer-Aided Engineering DOI 10. It can be seen that eq. the following relation can be obtained using (9) agree very well with these values.26 k and 0. the E-plane normalized gain pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Gn is plotted versus B1 for different lengths of L1.1002/mmce . [Color figure If the values of A1 (in k) which correspond to the can be viewed in the online issue.wiley. [Color figure compared to values obtained from (3). B1 ¼ 2:1060kL1 ð7Þ This corresponds to a maximum phase deviation in the E-plane of using eq. (1) and (7) as compared to those obtained directly from eq. (3). (9) gives better results The integrals in eqs. B1 versus L1 obtained using eqs. Figure 3 shows the differences between eqs. the improved formulas (7) and (9) have not pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi appeared in the literature before. which is available at maximum gains in Figure 4 are plotted versus their www. which is available at www.wiley. (7) and (in k). Similarly.interscience. (1). Figure 4 shows the H-plane normalized gain versus A1 for different lengths of L2 as obtained using eq. which is available at be viewed in the online issue. (4) and (5) are evaluated for B1 than eq.4 k.

The particle swarm optimization (PSO) method is chosen to solve the problem since. (2) and (9) compared to values obtained from (3). In [8]. This is done so Aþ A2  6:0458ðBB1  B1 Þ A1 ¼ ð12Þ that we have a fair comparison since (1) and (2) 2 were used in [6]. Our results for the wider dimension B1 along with recently. an optimization problem is introduced in Fitness ¼ Equationð3Þ  Gdesired ð15Þ which the feeding waveguide dimensions (A and B).com. Specifically. for details of the PSO algorithm. The parameters L1.interscience. [Color figure can using the PSO. be viewed in the online issue. PSO is similar in some ways to genetic algorithms. while in our approach the gain formula (3) is directly optimized Figure 5. In [6]. a closed form expression. Table I shows 15 different horn design examples. we have been interested in the application those reported in [6] are shown in Table II. while III. the following equa- Equation (12) is obtained by enforcing the fact that. our optimization problem here is a one-dimen. tions for L1. B1 that gives www. for the wider dimen- sion A1 was derived using the PSO. of PSO in different electromagentics and microwave a neural network was trained to find the value of circuits problems [12. which is available at In our implementation of the PSO.508 Najjar. the following relations for PE and B1 2 L1 ¼ ð16Þ PH can be easily derived: 2k L1 ðB1  BÞ PE ¼ ð13Þ A1 2 B1 L2 ¼ ð17Þ 3k L2 ðA1  AÞ PH ¼ ð14Þ A1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Aþ A2  6ðBB1  B1 Þ So. (3) has been solved using PSO while using the approximate formulas (1) and (2) instead of the pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi improved formulas (7) and (9). 2:1060k First. 10–13]. From the geometry of the horn in Figure 1. including the fact that the basic algorithm is very easy to understand and implement. which involves 11 constants. using PSO. The interested reader can refer to [8. to check the validity of our design approach and to compare with the results presented A1 2 L2 ¼ ð11Þ in [6] (since they are the most accurate among [1– 3:1831k 8]). It should be noted that our applica- tion of the PSO differs completely from that pre- sented in [8]. and Dib 11]. L2.wiley. while it is found directly here been successfully applied to antenna design [8. and the results proved that this method is powerful and effective for optimization problems. and A1 are used instead of equa- for a pyramidal horn to be physically realizable PE tions (10)–(12): and PH must be equal. and the desired optimum gain (G) are given as input parameters. The PSO technique has the dimension B1. but requires less computational bookkeeping and generally fewer lines of code.1002/mmce . A1 versus L2 obtained using eqs. ing equations: which have been used to compare our PSO results to those obtained using the previous design methods B1 2 L1 ¼ ð10Þ [1–8]. A1 ¼ ð18Þ 2 sional problem. 10. and the references therein. eq. L2.] the optimum gain is found using the following fitness function: Now. Moneer. the frequency (f). 13]. RESULTS equation (3) is used to find the dimension B1. and A1 are given by the follow. The desired values for B1 are also taken International Journal of RF and Microwave Computer-Aided Engineering DOI 10.

20728011748883 7.5 cm for the lowest iterative technique.356 from [6].1736 3 109 0.163 16.214 3. were obtained by using an highest gain example to around 1.3075 3 1010 0.3144154246448677 0.50 2.34 22.8573 3 1010 0.968 19.08969475353841 3.03 7.08969475492360850 1.24966814122088 5.636 4.3199 3 107 0.60 0.27745690986449739 4.014 8 R70 WR137 6.30 0.40 8.404 5 R40 WR229 4.75 3.341876692568690 3 1014 International Journal of RF and Microwave Computer-Aided Engineering DOI 10.40 23. TABLE II. A Comparison of the Desired B1 Values with the Results Reported in [6] and Those Obtained Using the PSO with eqs.85 4.70 2.95 23.22 24.8840 3 1010 0.22140310957023 6. Desired B1 (m) B1 (m) Absolute Error B1 (m) Absolute Error 9 1 0.215 7 R58 WR159 5.1222429966826040 0.19634325240751146 6.24966901518942222 8.0537725214751637 0.580 0.1963432517473291 0.2007 3 10 0.1378958142667574 0.061 18.2496681412209338 0.3259 3 1010 0.43867213531014 4. results for A1 and B1 are given in Table III along lished formulas [3–5].463474283042615 3 1014 8 0.711 0.736 17.8752 3 1010 0. usually needed for horn antennas when used as a Having verified the design approach.2072801174889028 0.849 20.3465 3 109 0. According to [15]. then the desired B1 values.2819 3 1011 0.6018 3 1010 0.80 5. The difference ranges from 2 mm for the and the desired gains.444045380111675 3 1014 7 0.648 14 R280 WR34 27.0896947535384475 0. I EC WR (GHz) f Gain (dB) G A B 1 R14 WR650 1.95 4.1050296530375660 0.4027 3 109 0.95 5. It can be seen that the PSO gain example.318 4 R32 WR284 3.12224299668255 5.2607 3 1010 0.779 21.771561172376096 3 1014 3 0.698219673888616 3 1014 11 0.50 10.2214031095703027 0.20728011787642514 3.18805991932393937 7.735311976920229 3 1014 15 0.05377252147514 2.295 0. the standard gain antenna to determine the precise gain improved formulas (7) and (9) are now used to find of other antennas.908 6 R48 WR187 4.13789581426670 5.50 1.218048215738236 3 1014 4 0.016 11 R120 WR75 12.13789581544039639 1.461 3 R26 WR340 2. We believe that these small differen- results are in excellent agreement with the desired ces are very important since exact dimensions are B1 values. (1) and (2) Guney [6] Our Results Using PSO No.1679716957223386 0.31441542405646650 5.432 15 R320 WR28 33.851086116668512 3 1014 14 0.850 1.286 1.06303561588684 2.475586568020162 3 1014 13 0.16797169572227 7.790 13 R180 WR51 18.262 10 R100 WR90 10. Optimum Gain Pyramidal Horn Design 509 TABLE I. It can be seen that which lead to an excellent agreement between the there are small differences between the two sets of computed gains of the designed pyramidal horns results.43867213410993439 1.1002/mmce .0630356158868642 0.708455847435289 3 1013 2 0.10502965396363224 9.817 2.2774569102970922 0.679489880501706 3 1013 9 0.10502965303752 4.290 22.31441542464479 7.19634325174725 7. these desired values the wider dimension B1 using PSO.16797212771119607 4.255 2 R22 WR430 2.51 8.12224299802912712 1.20 1.22140310964312218 7.1880599185382076 0. Values of Frequencies and Inner Dimensions of Common Standard Rectangular Feed Waveguides and Desired Gains Waveguide Waveguide Standard Dimensions (cm) Example Frequency Desired No.431 15. Our improved were first obtained roughly using previously pub.24 23.27745691029704 5.772360450213455 3 1014 6 0.905 0.953 12 R140 WR62 14.18805991853804 1.3855 3 109 0.107025913275720 3 1014 5 0.7397 3 107 0.864 0.233103005432895 3 1014 10 0. with our results from Table II.4386721353106108 0.755 2.580 9 R84 WR112 8.05377252230591544 8.039 2.85 16.06303561728956132 1.275 18.485 1.85 1.150047055479945 3 1014 12 0.36 24.

27745691029704 4 0. 249–250.22140310957023 6 0. Selvan.17916414178191 0.85000000000000 23. Artificial neural networks 13 23.50 6. 8 21. Moneer.15784776775620 0.50000000000000 22.30000000000000 24.21228861617687 0.10502965303752 13 0.24889199855866 0. It can be New improved formulas for the design of optimum seen that there is a difference between the desired gain horn antennas have been proposed. 2 16. The particle swarm optimization tech- gain is obtained exactly.06729446076815 0.12525307711374 0. Improved formulas for nar- 9 22.03000000000000 18.24504175271475 0. nique has been used to find the wider dimension directly from the exact gain formula.50 3.07637655171835 24.60 157–163.50000000000000 23.28391860822711 0.97793385306501 23.06917393039352 0. (1) and (2). New York.27751657819921 23.75000000000000 21.32160260813165 0. (1) and (2) Dimensions Gain Wiley.23855032644736 0. CONCLUSIONS puted using eq.05510271231507 0. 5 18. (3) and the improved formulas (7) and (9).53096300646700 21.41897752185742 0.85 for the narrow aperture dimension calculation of opti- 14 24.95 pyramidal horns. and Dib TABLE III. Guney. Int J RF Microwave Comput-Aided 4 17. 239–245.20111475647621 0.08115133942015 0.20728011748883 7 0. and dimensions obtained using eqs.31441542464479 3 0.40 pyramidal horns. Antenna theory: Analysis and design.58745939957790 0.25560936927822 0.50 row and wide aperture dimensions of optimum gain 10 22.50000000000000 16. Guney and N.19266814169802 0.28348102783180 0.05377252147514 Using the dimensions in Table III.510 Najjar.20 Eng 13 (2003). 12 23.65225377151011 15.73616544011127 19. 497– 7 20. No.85000000000000 15.16317793605405 0.14129224072333 0.43867213531014 2 0.63332003114509 20.19774296438053 17.1002/mmce .59070601292597 18.06459365453271 0. K.T.40904904300191 0.06303561588684 15 0. Hancer.75 5.11560862074174 0. Electron Lett 35 (1999). International Journal of RF and Microwave Computer-Aided Engineering DOI 10.27939335994748 22.03 Eng 11 (2001). These for- and the computed gains when using the aperture mulas do not approximate the path length error.16797169572227 10 0.13185422799275 0. C.A. 188–193.35840552867341 0.44853806855461 0.29088895881357 0. Int J RF Microwave Comput-Aided 11 22.20000000000000 23.13789581426670 11 0.60000000000000 24. 2.36748291047856 0. IV.12224299668255 12 0.62681934326002 23. 15 24.11248356493265 0. Guney.30503868563279 16. Balanis. Using the thus.15360236582933 0. the desired given gain. Improved design method for optimum gain 3 17.85 pyramidal horns.26089097456624 0. they give more accurate horn dimensions for a improved aperture dimensions A1 and B1.20747207152755 0. K.21260597834625 0.82432108495824 18.57361597193408 0. The results are shown in Table IV. Computed Gains Using (3) Along obtained for various design examples.19634325174725 8 0.13550390047196 0. The Aperture Dimensions (in meters) for the 15 Experiments Given in Table I PSO with Improved Formulas PSO with Formulas (1) and (2) Aperture Aperture Aperture Aperture No Dimension A1 Dimension B1 Dimension A1 Dimension B1 1 0.85000000000000 20. 1997. K.32252114892473 0.47893636158694 22.10761884737834 0. A new design method for optimum gain 6 19.70000000000000 22. Electromagnetics 21 (2001). Electrical Eng 86 (2004). the gain is com.07895527928804 0.17437783238506 0.26486821720179 0.40000000000000 17. Guney and H.80000000000000 18. Accurate design method for optimum gain 1 15. with (7) and (9) Gain Computed Using the Aperture Gain Computed REFERENCES Dimensions Using the Obtained Using Improved Aperture Desired 1.70 pyramidal horn.22667311271466 0.08969475353841 14 0.37570568796014 24. Sarikaya. K.95000000000000 19.85 505.24966814122088 5 0.18805991853804 9 0.27191636671072 0. Better accuracy with respect to previous design methods has been TABLE IV.33732363443676 0.09190862517665 0.80 4. K.30 mum gain pyramidal horns.

884–889. New narrow aperture 11. Nihad Dib and Jehad Ababneh. Particle swarm 14. K. He is currently studying for in EE from Kuwait University in 1985 B. antennas. Synthesis of sion formula obtained by using a particle swarm opti. M. A. Ann Arbor. optimization in electromagnetics. International Journal of RF and Microwave Computer-Aided Engineering DOI 10. 2nd ed. 8. and M. nas Propag 52 (2004).1002/mmce . IEEE J Electromagn Waves Appl 18 (2004). 397–407. Ababneh. His fields of interest include and Microwaves) in 1992 from University mobile communications systems. In September 1995. interdigital capacitors based on particle swarm optimi- mization for optimum gain pyramidal horns. in electrical engineering at August 2006. square spiral inductor.Sc. Then.S. Guney. J. in EE (major in Electromagnetics Jordan.A. wave Comput-Aided Eng 16 (2006). 343. Modern antenna design. and microwave systems. PSO. Karaboga. He is currently study. Trans Antennas Propag 53 (2005). Jordan. T. 2005.Sc. His fields of interest include EM and microwave communica- tions. magnetics and modeling of planar circuits. Simon. IEEE Trans and particle swarm design of coplanar waveguide Antennas Propag 41 (1993). Akdagli and K. 12. 321–339.Sc. 1201–1205.D. Int J Model Simul (in press). BIOGRAPHIES Yahya Najjar was born in Kuwait in Nihad Dib obtained his B. Int J RF Micro- wave Opt Technol Lett 48 (2006). K. p. He obtained his University of Science and Technology. IEEE Trans Anten. Robinson and Y. Wiley.G. J. 2674–2679.Sc. M. personal communication. in electrical engineering at Jordan and 1987. he joined the Mohammad Moneer was born in Kuwait EE Department at Jordan University of Science and Technology in December 1983. and became a full professor in ing for B. Maybell and P. Dib.J. GSM systems. 10. 322–330. Khodair. Christodoulou. and com- puter networks. Ph. respectively. Pyramidal horn gain 13. Linear array dimension expressions obtained using differential evo. M. 15. New wide-aperture-dimen. of Michigan. 9. Rahmat-Samii.. March 1984. Milligan. and N. Khodier and C. Guney. geometry synthesis with minimum sidelobe level and lution algorithm for optimum gain pyramidal horns. null control using particle swarm optimization. zation and artificial neural networks. nology. he antennas. Guney and D. His research interests are in computational electro- Jordan University of Science and Tech.M. Physical modeling calculation with improved accuracy. Micro. Optimum Gain Pyramidal Horn Design 511 7. worked as an assistant research scientist in the radiation laboratory at the same school. (JUST) as an assistant professor.