This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Research ArticleDesign of a Novel UWB Omnidirectional Antenna UsingParticle Swarm Optimization
Chengyang Yu Tanghong Xu and Changjun Liu
School of Electronics and Information Engineering Sichuan University Chengdu 610064 China
Correspondence should be addressed to Changjun Liu cjliuieeeorg
Received 23 January 2015 Revised 11 March 2015 Accepted 13 March 2015
Academic Editor Stefano Selleri
Copyright copy 2015 Chengyang Yu et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
A UWB E-plane omnidirectional microwave antenna is designed and fabricated for IEEE 80211a communication system andmicrowavemagnetron source system as a radiationmonitor A cooptimizationmethod based on particle swarmoptimization (PSO)algorithm and FDTD software is presentedThe presented PSO algorithm is useful inmany industrial microwave applications suchasmicrowavemagnetron design and other techniqueswith a high power levelThemaximummeasured relative bandwidth of 65 isachieved for the proposed antenna after a rapid and efficient optimization Furthermore the measured antenna polarization purityreaches about 20 dB at the communication C band The PSO algorithm is a powerful candidate for microwave passive componentdesign
1 Introduction
Omnidirectional antennas are widely used in wireless com-munication systems especially for high-multipath commu-nication applications based on polarization diversity tech-nique A typical polarization diversity system is composedof two orthogonally polarized antennas such as a verticallypolarizedmonopole and a horizontally polarized Alford loopantenna As an H-plane omnidirectional antenna monopolehas been widely researched However in such a situation E-plane omnidirectional antennas are also needed to investigateAlford loop antenna which is suitable at low frequencies withthe wire type was firstly reported in [1] Several improvedantennas based on Alford structure were also investigatedto generate E-plane omnidirectional radiation patterns [2ndash5] In [3] a dual-frequency Alford structure loop antenna isrealized with eight T-dipoles However broadband omnidi-rectional antennas are urgently needed for modern commu-nication systems [6 7]
In this paper an ultrawideband (UWB) characteristicis realized on the Alford structure loop antenna with E-plane omnidirectionality Such an antenna will be used as aradiation monitor at an actual microwave magnetron source
system Particle swarm optimization (PSO) algorithm isintroduced to optimize the whole structure The proposedantenna can be easily realized on a planar substrate while ithas a far-field radiation pattern similar to that of a magneticdipole In addition the optimized omnidirectional antennahas a measured impedance bandwidth from 46 to 90GHz(relative bandwidth is about 65) which covers the entire5GHz bandwidth of IEEE 80211a (515 GHzndash535GHz and5725ndash5875GHz)
2 Antenna Design
21 Antenna Structure The configuration of the proposedUWBomnidirectional antenna is shown in Figure 1 Itmainlyconsists of three identical pairs of printed half-wave dipoleradiators Each pair includes two dipoleswhichwork togetherto generate a broadband characteristic It is the distributedmicrostrip dipoles and the power combining structure thatgenerate an omnidirectional feature for the antenna Thelengths of the two dipole radiators are 2119877
1times 1205791and 2119877
2times
1205792 respectively Combining with double-sided strip lines
two parts of a dipole radiator are fabricated on the oppositesides of one substrate So the proposed dipole structure is
Hindawi Publishing CorporationInternational Journal of Antennas and PropagationVolume 2015 Article ID 303195 7 pageshttpdxdoiorg1011552015303195
2 International Journal of Antennas and Propagation
(a) (b)
Top plane
F4B
Bottom plane
Inner conductor
(c)
Y
Z
X
W
R
b a
1205792
1205791
W2
W1
R2
R1
Rg
R = R1 + a
R1 = R2 + b
Figure 1 The structure of the microwave antenna (a) top plane (b) bottom plane and (c) side view
Table 1 Range setup of optimized parameters
Parameters 1198821
1198822
119882 1205791
1205792
119886 119887 1198772
1198771
119877 119877119892
Optimization range 01mmndash4mm 12058718ndash1205873 01mmndash2mm 01mmndash6mm 8mmndash14mm 1198772+ 119887 119877
equivalent to the conventional dipole In order to form anomnidirectional radiation according to the theory of antennaarray the excitation phase of each dipole pair should be equalThe three dipole pairs are directly fed by microstrips whilethe common ground plane of microstrips is a circular patchwith a constant radius 119877
119892 The center of the ring structure
is soldered with an SMA connector It is obvious that all thedipole pairs are fed with not only an equal excitation phasebut also an equal excitation amplitude
22 Antenna Optimization and Fitness Function Based onthe proposed structure the final goal of our work is to obtaina planar antenna with omnidirectional radiation and lowreturn loss over WLAN operation in the 5GHz bands How-ever due to the narrow impedance bandwidth of conven-tional dipolemassive optimizations on radiators and connec-tion structures among them are needed In order to improveoptimization accuracy and velocity a cooptimizationmethodbased on PSO and FDTD simulator is introduced in this
paper The PSO algorithm and cooptimization processes willbe detailed later
The omnidirectional antenna is realized on a F4B sub-strate with a dielectric constant of 265 and thickness of 1mmThe specific optimized parameters of the proposed antennaas shown in Figure 1 are listed in Table 1 Parameters ldquo119886rdquoand ldquo119887rdquo are selected to match the restricted condition amongkinds of radiuses of the antenna structure
According to design targets of the proposed antennaespecially used on the entire 5GHz bandwidth of IEEE80211a the fitness function can be defined as
Fitness = 05 times BW + 119860 + 119861 (1)
where BW indicates the desired antenna impedance band-width expressed in terms of upper frequency 119891
119880and lower
frequency 119891119871 The upper and lower frequencies are the
boundary points of antenna bandwidth with dB(11987811) lt
minus10 dB 119860 and 119861 represent the weight factor of reflection
International Journal of Antennas and Propagation 3
Determine the solution space according to antenna variable ranges
For each particle
Utilize VBA to invoke FDTD software and the file containing antenna variables
Calculate the fitness function eq (1) with simulated S parameters
Update velocity with eq (3)
Update position with eq (6)
No
Yes
Randomly initialize particle with velocity Vi
and position Xi
Number of iterations T = T + 1
Gbest = Xi(T)
Pi = Xi(T)
T lt maximum iteration
Final antenna variables = Gbest
(1 ⩽ i ⩽ M)
If fitness of the particle gt fitness of Gbest
If fitness of the particle gt fitness of Pi
Figure 2 Flow chart of the proposed antenna cooptimization method
coefficient to optimize on the antenna at 52 GHz and58GHz respectively These factors can be expressed by
BW =
119891119880minus 119891119871
1GHz 119891119880
gt 6GHz 119891119871lt 5GHz
0 others
119860 =
1 dB (11987811)10038161003816100381610038161003816 119891=52GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=52GHz gt minus10 dB
119861 =
1 dB (11987811)10038161003816100381610038161003816 119891=58GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=58GHz gt minus10 dB
(2)
23 PSO Algorithm and Cooptimization with FDTD Soft-ware As an evolutionary computation technique based onthe movement and intelligence of particle swarm PSO ispresented by Kennedy et al [8] Each particle in the swarmrepresents a possible solution to the specific optimizationevent There are 119872 particles to search an 119873 dimensionssolution space respectively So the velocity position and thepersonal best position are expressed by119872times119873matrixes Theposition of particle 119894 at a fixed iteration 119879 is usually expressedas a vector X
119894(119879) = [119883
1198941(119879) 119883
1198942(119879) 119883
119894119873(119879)] where 119894
satisfies 1 le 119894 le 119872 This particle adjusts its position withvelocity V
119894(119879) = [119881
1198941(119879) 119881
1198942(119879) 119881
119894119873(119879)] through the
solution space According to the fitness function calculationthe personal best particle and global best particle are involvedin P119894= [1198751198941 1198751198942 119875
119894119873] and Gbest = [119866
1 1198662 119866
119873]
4 International Journal of Antennas and Propagation
Figure 3 The fabricated UWB omnidirectional microwave antenna (top view and bottom view)
Clerc and Kennedy have introduced a constriction factor[9] 119870 which is used to constrain and control velocities forPSO In [10] Eberhart and Shi concluded that the PSO usinga constriction factor 119870 is the best approach while limitingthe maximum velocity 119881max to the dynamic range of variable119883max on each dimension compared with performance usingan inertia weight The velocity function of PSO used in thispaper is
119881119894119895
(119879 + 1) = 119870 times [119881119894119895
(119879) + 1206011rand ( ) times (119875
119894119895(119879) minus 119883
119894119895(119879))
+ 1206012rand ( ) times (119866
119895(119879) minus 119883
119894119895(119879))]
(3)
where the constriction factor119870 is computed as
119870 =2
10038161003816100381610038161003816100381610038162 minus 120601 minus radic120601
2minus 4120601
1003816100381610038161003816100381610038161003816
120601 = 1206011+ 1206012gt 4
(4)
We tested different groups of the cognitive and socialcomponent values of the PSO (120601
1and 120601
2) with Griewank
function and Sphere function The standard value settingsin [10] (120601
1= 1206012
= 205) and those in [11] (1206011
= 28 and1206012
= 13) result in a better optimization accuracy and abetter convergence rate respectively In this paper aiming at acompromise on performances improved cognitive and socialcomponent values are used for the PSO Cognitive and socialrates vary from28 to 205 and from 13 to 205 synchronouslyThe variation is linear to iteration times Tested results showthat the proposed settings of (120601
1= 28sim205 and120601
2= 13sim205)
result in the best performance on optimization accuracy anda good convergence rate for PSO
Reflecting boundary condition [12] is used to limit theparticle velocity andpositionwhen it hits the boundary in oneof the dimensions The particle velocity and position beyondthe boundary can be expressed by
Figure 4 Simulated and measured results of reflection coefficient
where 119898119895is determined by the distance 119889 from particle
position to the boundary
119898119895=
119889
119883max119895
minus 119883min119895
119889 le 119883max119895
minus 119883min119895
119883max119895
minus 119883min119895
119889119889 gt 119883
max119895
minus 119883min119895
(7)
The steps of cooptimization with the proposed PSO andFDTD software (CST) are described in Figure 2
Step 1 Determine the antenna variables and ranges to beoptimized Randomly initialize 119872 particles with velocity V
119894
and position X119894in the solution space
Step 2 Write the variables into atxt file at fixed positionInvoke the file and CST software automatically by using VBA(a macro language of Visual Basic) Use the simulated 119878-parameters to calculate the fitness of each particle accordingto (1) Record the personal particles and global best particleaccording to the fitness function value
International Journal of Antennas and Propagation 5
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30
(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
f = 48 f = 48
(a) At 48 GHz
f = 52 f = 52
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(b) At 52 GHz
f = 58 f = 58
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50
(dB)
(c) At 58 GHz
Figure 5 Continued
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
equivalent to the conventional dipole In order to form anomnidirectional radiation according to the theory of antennaarray the excitation phase of each dipole pair should be equalThe three dipole pairs are directly fed by microstrips whilethe common ground plane of microstrips is a circular patchwith a constant radius 119877
119892 The center of the ring structure
is soldered with an SMA connector It is obvious that all thedipole pairs are fed with not only an equal excitation phasebut also an equal excitation amplitude
22 Antenna Optimization and Fitness Function Based onthe proposed structure the final goal of our work is to obtaina planar antenna with omnidirectional radiation and lowreturn loss over WLAN operation in the 5GHz bands How-ever due to the narrow impedance bandwidth of conven-tional dipolemassive optimizations on radiators and connec-tion structures among them are needed In order to improveoptimization accuracy and velocity a cooptimizationmethodbased on PSO and FDTD simulator is introduced in this
paper The PSO algorithm and cooptimization processes willbe detailed later
The omnidirectional antenna is realized on a F4B sub-strate with a dielectric constant of 265 and thickness of 1mmThe specific optimized parameters of the proposed antennaas shown in Figure 1 are listed in Table 1 Parameters ldquo119886rdquoand ldquo119887rdquo are selected to match the restricted condition amongkinds of radiuses of the antenna structure
According to design targets of the proposed antennaespecially used on the entire 5GHz bandwidth of IEEE80211a the fitness function can be defined as
Fitness = 05 times BW + 119860 + 119861 (1)
where BW indicates the desired antenna impedance band-width expressed in terms of upper frequency 119891
119880and lower
frequency 119891119871 The upper and lower frequencies are the
boundary points of antenna bandwidth with dB(11987811) lt
minus10 dB 119860 and 119861 represent the weight factor of reflection
International Journal of Antennas and Propagation 3
Determine the solution space according to antenna variable ranges
For each particle
Utilize VBA to invoke FDTD software and the file containing antenna variables
Calculate the fitness function eq (1) with simulated S parameters
Update velocity with eq (3)
Update position with eq (6)
No
Yes
Randomly initialize particle with velocity Vi
and position Xi
Number of iterations T = T + 1
Gbest = Xi(T)
Pi = Xi(T)
T lt maximum iteration
Final antenna variables = Gbest
(1 ⩽ i ⩽ M)
If fitness of the particle gt fitness of Gbest
If fitness of the particle gt fitness of Pi
Figure 2 Flow chart of the proposed antenna cooptimization method
coefficient to optimize on the antenna at 52 GHz and58GHz respectively These factors can be expressed by
BW =
119891119880minus 119891119871
1GHz 119891119880
gt 6GHz 119891119871lt 5GHz
0 others
119860 =
1 dB (11987811)10038161003816100381610038161003816 119891=52GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=52GHz gt minus10 dB
119861 =
1 dB (11987811)10038161003816100381610038161003816 119891=58GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=58GHz gt minus10 dB
(2)
23 PSO Algorithm and Cooptimization with FDTD Soft-ware As an evolutionary computation technique based onthe movement and intelligence of particle swarm PSO ispresented by Kennedy et al [8] Each particle in the swarmrepresents a possible solution to the specific optimizationevent There are 119872 particles to search an 119873 dimensionssolution space respectively So the velocity position and thepersonal best position are expressed by119872times119873matrixes Theposition of particle 119894 at a fixed iteration 119879 is usually expressedas a vector X
119894(119879) = [119883
1198941(119879) 119883
1198942(119879) 119883
119894119873(119879)] where 119894
satisfies 1 le 119894 le 119872 This particle adjusts its position withvelocity V
119894(119879) = [119881
1198941(119879) 119881
1198942(119879) 119881
119894119873(119879)] through the
solution space According to the fitness function calculationthe personal best particle and global best particle are involvedin P119894= [1198751198941 1198751198942 119875
119894119873] and Gbest = [119866
1 1198662 119866
119873]
4 International Journal of Antennas and Propagation
Figure 3 The fabricated UWB omnidirectional microwave antenna (top view and bottom view)
Clerc and Kennedy have introduced a constriction factor[9] 119870 which is used to constrain and control velocities forPSO In [10] Eberhart and Shi concluded that the PSO usinga constriction factor 119870 is the best approach while limitingthe maximum velocity 119881max to the dynamic range of variable119883max on each dimension compared with performance usingan inertia weight The velocity function of PSO used in thispaper is
119881119894119895
(119879 + 1) = 119870 times [119881119894119895
(119879) + 1206011rand ( ) times (119875
119894119895(119879) minus 119883
119894119895(119879))
+ 1206012rand ( ) times (119866
119895(119879) minus 119883
119894119895(119879))]
(3)
where the constriction factor119870 is computed as
119870 =2
10038161003816100381610038161003816100381610038162 minus 120601 minus radic120601
2minus 4120601
1003816100381610038161003816100381610038161003816
120601 = 1206011+ 1206012gt 4
(4)
We tested different groups of the cognitive and socialcomponent values of the PSO (120601
1and 120601
2) with Griewank
function and Sphere function The standard value settingsin [10] (120601
1= 1206012
= 205) and those in [11] (1206011
= 28 and1206012
= 13) result in a better optimization accuracy and abetter convergence rate respectively In this paper aiming at acompromise on performances improved cognitive and socialcomponent values are used for the PSO Cognitive and socialrates vary from28 to 205 and from 13 to 205 synchronouslyThe variation is linear to iteration times Tested results showthat the proposed settings of (120601
1= 28sim205 and120601
2= 13sim205)
result in the best performance on optimization accuracy anda good convergence rate for PSO
Reflecting boundary condition [12] is used to limit theparticle velocity andpositionwhen it hits the boundary in oneof the dimensions The particle velocity and position beyondthe boundary can be expressed by
Figure 4 Simulated and measured results of reflection coefficient
where 119898119895is determined by the distance 119889 from particle
position to the boundary
119898119895=
119889
119883max119895
minus 119883min119895
119889 le 119883max119895
minus 119883min119895
119883max119895
minus 119883min119895
119889119889 gt 119883
max119895
minus 119883min119895
(7)
The steps of cooptimization with the proposed PSO andFDTD software (CST) are described in Figure 2
Step 1 Determine the antenna variables and ranges to beoptimized Randomly initialize 119872 particles with velocity V
119894
and position X119894in the solution space
Step 2 Write the variables into atxt file at fixed positionInvoke the file and CST software automatically by using VBA(a macro language of Visual Basic) Use the simulated 119878-parameters to calculate the fitness of each particle accordingto (1) Record the personal particles and global best particleaccording to the fitness function value
International Journal of Antennas and Propagation 5
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30
(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
f = 48 f = 48
(a) At 48 GHz
f = 52 f = 52
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(b) At 52 GHz
f = 58 f = 58
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50
(dB)
(c) At 58 GHz
Figure 5 Continued
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
International Journal of Antennas and Propagation 3
Determine the solution space according to antenna variable ranges
For each particle
Utilize VBA to invoke FDTD software and the file containing antenna variables
Calculate the fitness function eq (1) with simulated S parameters
Update velocity with eq (3)
Update position with eq (6)
No
Yes
Randomly initialize particle with velocity Vi
and position Xi
Number of iterations T = T + 1
Gbest = Xi(T)
Pi = Xi(T)
T lt maximum iteration
Final antenna variables = Gbest
(1 ⩽ i ⩽ M)
If fitness of the particle gt fitness of Gbest
If fitness of the particle gt fitness of Pi
Figure 2 Flow chart of the proposed antenna cooptimization method
coefficient to optimize on the antenna at 52 GHz and58GHz respectively These factors can be expressed by
BW =
119891119880minus 119891119871
1GHz 119891119880
gt 6GHz 119891119871lt 5GHz
0 others
119860 =
1 dB (11987811)10038161003816100381610038161003816 119891=52GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=52GHz gt minus10 dB
119861 =
1 dB (11987811)10038161003816100381610038161003816 119891=58GHz le minus10 dB
0 dB (11987811)10038161003816100381610038161003816 119891=58GHz gt minus10 dB
(2)
23 PSO Algorithm and Cooptimization with FDTD Soft-ware As an evolutionary computation technique based onthe movement and intelligence of particle swarm PSO ispresented by Kennedy et al [8] Each particle in the swarmrepresents a possible solution to the specific optimizationevent There are 119872 particles to search an 119873 dimensionssolution space respectively So the velocity position and thepersonal best position are expressed by119872times119873matrixes Theposition of particle 119894 at a fixed iteration 119879 is usually expressedas a vector X
119894(119879) = [119883
1198941(119879) 119883
1198942(119879) 119883
119894119873(119879)] where 119894
satisfies 1 le 119894 le 119872 This particle adjusts its position withvelocity V
119894(119879) = [119881
1198941(119879) 119881
1198942(119879) 119881
119894119873(119879)] through the
solution space According to the fitness function calculationthe personal best particle and global best particle are involvedin P119894= [1198751198941 1198751198942 119875
119894119873] and Gbest = [119866
1 1198662 119866
119873]
4 International Journal of Antennas and Propagation
Figure 3 The fabricated UWB omnidirectional microwave antenna (top view and bottom view)
Clerc and Kennedy have introduced a constriction factor[9] 119870 which is used to constrain and control velocities forPSO In [10] Eberhart and Shi concluded that the PSO usinga constriction factor 119870 is the best approach while limitingthe maximum velocity 119881max to the dynamic range of variable119883max on each dimension compared with performance usingan inertia weight The velocity function of PSO used in thispaper is
119881119894119895
(119879 + 1) = 119870 times [119881119894119895
(119879) + 1206011rand ( ) times (119875
119894119895(119879) minus 119883
119894119895(119879))
+ 1206012rand ( ) times (119866
119895(119879) minus 119883
119894119895(119879))]
(3)
where the constriction factor119870 is computed as
119870 =2
10038161003816100381610038161003816100381610038162 minus 120601 minus radic120601
2minus 4120601
1003816100381610038161003816100381610038161003816
120601 = 1206011+ 1206012gt 4
(4)
We tested different groups of the cognitive and socialcomponent values of the PSO (120601
1and 120601
2) with Griewank
function and Sphere function The standard value settingsin [10] (120601
1= 1206012
= 205) and those in [11] (1206011
= 28 and1206012
= 13) result in a better optimization accuracy and abetter convergence rate respectively In this paper aiming at acompromise on performances improved cognitive and socialcomponent values are used for the PSO Cognitive and socialrates vary from28 to 205 and from 13 to 205 synchronouslyThe variation is linear to iteration times Tested results showthat the proposed settings of (120601
1= 28sim205 and120601
2= 13sim205)
result in the best performance on optimization accuracy anda good convergence rate for PSO
Reflecting boundary condition [12] is used to limit theparticle velocity andpositionwhen it hits the boundary in oneof the dimensions The particle velocity and position beyondthe boundary can be expressed by
Figure 4 Simulated and measured results of reflection coefficient
where 119898119895is determined by the distance 119889 from particle
position to the boundary
119898119895=
119889
119883max119895
minus 119883min119895
119889 le 119883max119895
minus 119883min119895
119883max119895
minus 119883min119895
119889119889 gt 119883
max119895
minus 119883min119895
(7)
The steps of cooptimization with the proposed PSO andFDTD software (CST) are described in Figure 2
Step 1 Determine the antenna variables and ranges to beoptimized Randomly initialize 119872 particles with velocity V
119894
and position X119894in the solution space
Step 2 Write the variables into atxt file at fixed positionInvoke the file and CST software automatically by using VBA(a macro language of Visual Basic) Use the simulated 119878-parameters to calculate the fitness of each particle accordingto (1) Record the personal particles and global best particleaccording to the fitness function value
International Journal of Antennas and Propagation 5
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30
(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
f = 48 f = 48
(a) At 48 GHz
f = 52 f = 52
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(b) At 52 GHz
f = 58 f = 58
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50
(dB)
(c) At 58 GHz
Figure 5 Continued
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
4 International Journal of Antennas and Propagation
Figure 3 The fabricated UWB omnidirectional microwave antenna (top view and bottom view)
Clerc and Kennedy have introduced a constriction factor[9] 119870 which is used to constrain and control velocities forPSO In [10] Eberhart and Shi concluded that the PSO usinga constriction factor 119870 is the best approach while limitingthe maximum velocity 119881max to the dynamic range of variable119883max on each dimension compared with performance usingan inertia weight The velocity function of PSO used in thispaper is
119881119894119895
(119879 + 1) = 119870 times [119881119894119895
(119879) + 1206011rand ( ) times (119875
119894119895(119879) minus 119883
119894119895(119879))
+ 1206012rand ( ) times (119866
119895(119879) minus 119883
119894119895(119879))]
(3)
where the constriction factor119870 is computed as
119870 =2
10038161003816100381610038161003816100381610038162 minus 120601 minus radic120601
2minus 4120601
1003816100381610038161003816100381610038161003816
120601 = 1206011+ 1206012gt 4
(4)
We tested different groups of the cognitive and socialcomponent values of the PSO (120601
1and 120601
2) with Griewank
function and Sphere function The standard value settingsin [10] (120601
1= 1206012
= 205) and those in [11] (1206011
= 28 and1206012
= 13) result in a better optimization accuracy and abetter convergence rate respectively In this paper aiming at acompromise on performances improved cognitive and socialcomponent values are used for the PSO Cognitive and socialrates vary from28 to 205 and from 13 to 205 synchronouslyThe variation is linear to iteration times Tested results showthat the proposed settings of (120601
1= 28sim205 and120601
2= 13sim205)
result in the best performance on optimization accuracy anda good convergence rate for PSO
Reflecting boundary condition [12] is used to limit theparticle velocity andpositionwhen it hits the boundary in oneof the dimensions The particle velocity and position beyondthe boundary can be expressed by
Figure 4 Simulated and measured results of reflection coefficient
where 119898119895is determined by the distance 119889 from particle
position to the boundary
119898119895=
119889
119883max119895
minus 119883min119895
119889 le 119883max119895
minus 119883min119895
119883max119895
minus 119883min119895
119889119889 gt 119883
max119895
minus 119883min119895
(7)
The steps of cooptimization with the proposed PSO andFDTD software (CST) are described in Figure 2
Step 1 Determine the antenna variables and ranges to beoptimized Randomly initialize 119872 particles with velocity V
119894
and position X119894in the solution space
Step 2 Write the variables into atxt file at fixed positionInvoke the file and CST software automatically by using VBA(a macro language of Visual Basic) Use the simulated 119878-parameters to calculate the fitness of each particle accordingto (1) Record the personal particles and global best particleaccording to the fitness function value
International Journal of Antennas and Propagation 5
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30
(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
f = 48 f = 48
(a) At 48 GHz
f = 52 f = 52
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(b) At 52 GHz
f = 58 f = 58
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50
(dB)
(c) At 58 GHz
Figure 5 Continued
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
International Journal of Antennas and Propagation 5
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30
(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
f = 48 f = 48
(a) At 48 GHz
f = 52 f = 52
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(b) At 52 GHz
f = 58 f = 58
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50
(dB)
(c) At 58 GHz
Figure 5 Continued
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
6 International Journal of Antennas and Propagation
f = 65 f = 65
Measured HMeasured Hx
Simulated HSimulated Hx
Measured EMeasured Ex
Simulated ESimulated Ex
0
30
60
90
120
150
180
210
240
270
300
3300
minus5
minus10
minus15
minus20
minus25
minus30
minus35
0
minus5
minus10
minus15
minus20
minus25
minus30(dB)
0
30
60
90
120
150
180
210
240
270
300
3300
minus10
minus20
minus30
minus40
0
minus10
minus20
minus30
minus40
minus50(dB)
(d) At 65 GHz
Figure 5 Simulated and measured microwave radiation patterns at (a) 48GHz (b) 52 GHz (c) 58 GHz and (d) 65 GHz
Step 3 Update the velocity and position of each particleaccording to (3) and (6)
Step 4 Calculate the fitness of each particle again Update thepersonal particles
Step 5 Read the personal best particle If its correspondingfitness function value is better than that of global best particleupdate the record of the global best particle
Step 6 Repeat Steps 3 4 and 5 until the maximum iterationnumber is reached
3 Fabrication and Measurements
The omnidirectional antenna mentioned above is optimizedwith the proposed optimization method based on PSO andFDTD software The operation frequency of this antenna isthe entire 5GHz wideband of IEEE 80211a especially forfrequency bands of 52 GHz and 58GHz For obtaining afine UWB characteristic 20 particles and 150 iteration timesare employed The final optimized geometric parameters are119886 = 088mm 119887 = 390mm 119877
2= 965mm 120579
1= 030
1205792= 074119882 = 030mm119882
1= 239mm and119882
2= 175mm
Figure 3 shows the fabricatedUWBomnidirectional antennaThe diameter of this antenna is 36mm
The reflection coefficient was measured using an AgilentN5230A vector network analyzer As shown in Figure 4 thesimulated results matched well with the measured resultsThis indicates that the cooptimization method based on PSOand FDTD software is effective for antenna design Fur-thermore the improved Alford structure antenna withoutany additional matching circuits definitely has an UWBcharacteristic from 46GHz to 90GHz
The radiation patterns of the proposed antenna aremeasured and simulated at 48GHz 52 GHz 58GHz and65GHz Figure 5 shows the comparison of simulated andmeasured patterns which include the coplanar polarization(E and H) and cross polarization (Ex and Hx) of the antennaIt is obvious that the proposed antenna has an excellentomnidirectional radiation in the entire 5GHz band of IEEE80211a The measured polarization purity in the E-planereaches about 20 dB The data differences between measuredand simulated polarization purity parameters are mainlycaused by the noise background of power receiver in theantenna measurement system However 20 dB polarizationpurity is good enough to be an E-plane omnidirectionalantenna of polarization diversity system
The simulated and measured antenna gains are shown inFigure 6 The maximum measured antenna gains are 13 and10 dBi at 52 and 58GHz respectively
4 Conclusion
A novel UWB E-plane omnidirectional antenna has beenproposed for polarization diversity of IEEE 80211a com-munication system and some industrial applications Powercombining construction with three microstrip dipoles isinvestigated to form the omnidirectional radiation featureThe PSO algorithm is a powerful candidate for the design andoptimization on the proposed UWB antenna The measuredresults show that the antenna has a relative bandwidth of65 (46 to 90GHz) The good measured omnidirectionalradiation feature in the 5GHz band enables the antenna tooperate at IEEE 80211a system and monitor the radiationlevel in microwave magnetron source effectively Further-more it is experimentally demonstrated that the proposed
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004
International Journal of Antennas and Propagation 7
Frequency (GHz)
MeasuredSimulated
45 50 55 60 65 70 75 80 85 90
Gai
n (d
Bi)
15
10
05
00
Figure 6 Simulated andmeasured results of themicrowave antennagain
E-plane omnidirectional antenna is suitable for realizingpolarization diversity technique associated with an H-planeomnidirectional antenna
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was supported in part by the 973 Program2013CB328902 NSFC 0971051 and NCET-12-0383
References
[1] A Alford and A G Kandoian ldquoUltra-high frequency loopantennardquo AIEE Transactions vol 59 no 12 pp 843ndash848 1940
[2] A J Fenn ldquoArrays of horizontally polarized loop-fed slottedcylinder antennasrdquo IEEE Transactions on Antennas and Prop-agation vol 33 no 4 pp 375ndash382 1985
[3] C-H Ahn S-W Oh and K Chang ldquoA dual-frequencyomnidirectional antenna for polarization diversity of MIMOand wireless communication applicationsrdquo IEEE Antennas andWireless Propagation Letters vol 8 pp 966ndash969 2009
[4] C-C Lin L-C Kuo and H-R Chuang ldquoA horizontally polar-ized omnidirectional printed antenna for WLAN applicationsrdquoIEEE Transactions on Antennas and Propagation vol 54 no 11pp 3551ndash3556 2006
[5] H-R Chuang and L-C Kuo ldquo3-D FDTD design analysis ofa 24-GHz polarization-diversity printed dipole antenna withintegrated balun and polarization-switching circuit for WLANand wireless communication applicationsrdquo IEEE TransactionsonMicrowaveTheory and Techniques vol 51 no 2 pp 374ndash3812003
[6] T Sedghi M Jalali and T Aribi ldquoFabrication of CPW-fedfractal antenna for UWB applications with omni-directional
patternsrdquo The Scientific World Journal vol 2014 Article ID391602 5 pages 2014
[7] M N Iqbal Hamood-Ur-Rahman and S F Jilani ldquoAn ultra-wideband monopole fractal antenna with coplanar waveguidefeedrdquo International Journal of Antennas and Propagation vol2014 Article ID 510913 7 pages 2014
[8] J F Kennedy R Eberhart and Y Shi Swarm IntelligenceElsevier Science 2001
[9] M Clerc and J Kennedy ldquoThe particle swarm-explosion sta-bility and convergence in a multidimensional complex spacerdquoIEEE Transactions on Evolutionary Computation vol 6 no 1pp 58ndash73 2002
[10] R C Eberhart and Y Shi ldquoComparing inertia weights and con-striction factors in particle swarm optimizationrdquo in Proceedingsof the 2000 Congress on Evolutionary Computation vol 1 pp84ndash88 July 2000
[11] A Carlisle and G Dozier ldquoAn off-the-shelf PSOrdquo in Proceedingsof the Workshop on Particle Swarm Optimization IndianapolisInd USA 2001
[12] J Robinson andY Rahmat-Samii ldquoParticle swarmoptimizationin electromagneticsrdquo IEEE Transactions on Antennas and Prop-agation vol 52 no 2 pp 397ndash407 2004