1056 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 3, MARCH 2014

MEMS Reconfigurable Optimized E-Shaped PatchAntenna Design for Cognitive Radio

Harish Rajagopalan, Member, IEEE, Joshua M. Kovitz, Student Member, IEEE, andYahya Rahmat-Samii, Fellow, IEEE

AbstractReconfigurable antennas offer attractive potentialsolutions to solve the challenging antenna problems related to cog-nitive radio systems using the ability to switch patterns, frequency,and polarization. In this paper, a novel frequency reconfigurableE-shaped patch design is proposed for possible applications incognitive radio systems. This paper provides a methodology todesign reconfigurable antennas with radio frequency microelec-tromechanical system (RF-MEMS) switches using particle swarmoptimization, a nature-inspired optimization technique. By addingRF-MEMS switches to dynamically change the slot dimensions,one can achieve wide bandwidth which is nearly double the orig-inal E-shaped patch bandwidth. Utilizing an appropriate fitnessfunction, an optimized design which works in the frequency rangefrom 2 GHz to 3.2 GHz (50% impedance bandwidth at 2.4 GHz)is obtained. RF-MEMS switch circuit models are incorporatedinto the optimization as they more effectively represent the ac-tual switch effects. A prototype of the final optimized design isdeveloped and measurements demonstrate good agreement withsimulations.

Index TermsCognitive radio, E-shaped patch, frequency re-configurable, Particle Swarm Optimization, radio frequency mi-croelectromechanical system (RF-MEMS), wideband antenna.

I. INTRODUCTION

T HE communication link between two antennas typicallysuffers from multipath, interference, and fading. Thesevarious phenomena can severely restrict the performance ofpresent-day wireless communication systems. In recent years,numerous techniques and solutions to enhance communicationlinks have been devised [1][4]. The improvements involvingantennas include the usage of antenna diversity to overcomelimitations imposed on the system performance with a singlereceiving antenna [1]; antenna reconfigurability to enhancea single antenna by adding additional functionalities [2], [3];and multiple-input multiple-output (MIMO) antenna systems toexploit rich multipath environments for a given frequency band[4]. In these methods, the frequency band is typically assumed

Manuscript received November 08, 2012; revised August 03, 2013; acceptedOctober 30, 2013. Date of publication November 25, 2013; date of current ver-sion February 27, 2014. The work of J. M. Kovitz was supported by the Depart-ment of Defense (DoD) through the National Defense Science and EngineeringGraduate Fellowship (NDSEG) Program.H. Rajagopalan was with the Department of Electrical Engineering, Univer-

sity of California, Los Angeles, Los Angeles, CA 90095 USA. He is now withApple Inc., Cupertino, CA 95014 USA.J. M. Kovitz and Y. Rahmat-Samii are with the Department of Electrical En-

gineering, University of California, Los Angeles, Los Angeles, CA 90095 USA(e-mail: jmkovitz@ucla.du; rahmat@ee.ucla.edu).Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.Digital Object Identifier 10.1109/TAP.2013.2292531

to be set, and the goal is to improve overall system performancefor that given band by exploiting spatial features of the wire-less environment. Another technique that is currently gainingmomentum is the use of software defined radio in the contextof dynamic spectrum access, also known as cognitive radio[5]. In general, cognitive radios refer to full communicationsystem architectures that are able to sense the environmentfor primary (licensed) users and utilize available spectrumnot currently being used [5]. In contrast to the previouslymentioned techniques, cognitive radio takes advantage of thefrequency and time aspects of the wireless environment. Atany given location, time, and direction the frequency spectrummay not be fully utilized as shown in the inset figure in Fig. 1,where less than 6% occupancy is observed for a representativescenario [6]. Thus, enabling dynamic spectrum access offersmany benefits to wireless systems, including the opportunity tocombat fading and possibly improve channel capacity throughwider bandwidths.The required features of cognitive radio systems provide

many unique challenges to antenna designers. Some of thesechallenges and requirements are detailed in [7]. The antenna re-quirements for cognitive radio systems can also depend upon thenetwork architecture. Some possible network architectures canbe seen in Fig. 1, where a base station infrastructure is depictedas well as an ad hoc network with no pre-existing infrastructure[6], [8]. In the architecture with infrastructure, base stationsmake up the backbone of the network, and often directivearrays are implemented for these systems to provide sectoralcoverage. However, in ad hoc networks the backbone is madeup of select terminals, and hence omnidirectional coverage isoften desired. Previous work concerning antenna designs forthese systems have often targeted scenarios requiring omni-directional coverage and extremely wide bandwidth designsusing UWB-class antennas as well as reconfigurable antennas[9][14]. Patch antennas form another class of antennas whichhave been widely used in many wireless applications such aslaptops [15] and base stations, but the effort to investigate theiruse in cognitive radio systems has been limited, primarily dueto their narrow bandwidth. However their bandwidth could beextended through novel patch topologies, such as the E-shapedpatch, as well as frequency reconfigurability.In this paper, a novel frequency reconfigurable E-shaped

patch antenna (FR-ESPA) design is presented as a new wide-band patch antenna for possible use in cognitive radio systems.This design could be used in large terminals such as laptops[15] or as an array element in highly directive base stationantennas, as depicted in Fig. 1. In particular, this reconfigurable

0018-926X 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

RAJAGOPALAN et al.: MEMS RECONFIGURABLE OPTIMIZED E-SHAPED PATCH ANTENNA DESIGN 1057

Fig. 1. Two potential network architectures in cognitive radio. The frequency reconfigurable E-shaped patch antenna proposed in this paper could be used as anarray element for base station applications for future cognitive radio paradigms. Spectral plot in upper right is adapted from [6].

E-shaped patch design offers a simple single-layer single-feedstructure which is straightforward to manufacture. It also canprovide a wide instantaneous bandwidth in comparison to othertuned narrowband systems. To accomplish this, a methodologyfor the design and optimization of reconfigurable antennas withRF-MEMS switches for cognitive radio systems is presented.The design is analyzed through full-wave electromagneticsolvers, optimized through nature-inspired optimization tech-niques, and fully fabricated with RF-MEMS switches.The paper is organized as follows. Section II discusses the

frequency reconfigurable concept and introduces our newlyproposed design. Section III discusses the application ofParticle Swarm Optimization (PSO) [16], a nature inspiredoptimization technique, in conjunction with the full-waveelectromagnetic solver, HFSS, for the design of a frequencyreconfigurable E-shaped patch antenna design. The resultingdesign from the optimization is then fabricated using idealswitches and demonstrated through antenna measurements.Next, Section IV details the different switch models that can beused for modeling the switch (RF-MEMS switch) within HFSSto realize a design which can be implemented using RF-MEMS.Section V describes the final optimized E-shaped RF-MEMSreconfigurable antenna through optimizations, simulations,and measurements. Section VI provides the final design withRF-MEMS switches along with bias lines for switch activation.Both impedance matching and radiation pattern measurementsare provided and some observations are made. Section VIIprovides some concluding remarks.

II. FREQUENCY RECONFIGURABLE E-SHAPEDPATCH ANTENNA CONCEPT

Some of the more popular techniques in literature for in-creasing the bandwidth of probe-fed patch antennas include thestacked patch [17], L-shaped probe-fed patch [18], U-slottedpatch [19], and the E-shaped patch [20]. The E-shaped patch an-tenna is advantageous due to its single-layer, single-feed struc-ture, and the accessibility provided by the slots for the bias

Fig. 2. Frequency reconfigurable E-shaped patch design concept. (a) Currentstravelling along the patch length. (b) Currents travelling around the slots.(c) Effect on current by changing slot length.

lines to control the switches. In the past, the E-shaped patchantenna has been optimized for dual band and wideband de-signs; however, frequency reconfigurability was not incorpo-rated [21]. In a recent conference paper, the authors briefly intro-duced frequency reconfigurability into the E-shaped patch an-tenna design [22], and this paper presents the comprehensiveand complete study of the FR-ESPA using MEMS.Typically, the E-shaped patch antenna has dual resonance due

to the slots introduced into the patch topology [20]. These slotswhich create the E-shape allow another mode to resonate ata lower frequency relative to the typical patch mode. This isdue to the currents resonating over a longer geometrical pathas shown in Fig. 2(b). This mode has strong dependence on theslot geometry [Fig. 2(b)], while the normal patch mode dependsprimarily on the patch resonant length [Fig. 2(a)]. This patchresonant length is often given by , as shown in Fig. 2(a).Changing the slot dimensions strongly controls the resonantmodes of the E-shaped patch, and therefore they can be alteredto provide a desired impedance matching performance. The slotlength in Fig. 2(c) is shortened in comparison to Fig. 2(b), andconsequently the current has a smaller distance to travel aroundthe slots, giving rise to a higher resonant frequency. By im-plementing RF switches such as PIN diodes or MEMS, one

1058 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 3, MARCH 2014

Fig. 3. Frequency reconfigurable E-shaped patch design schematic using RFswitches with optimization parameters listed.

can alter the dimensions of the slots. Ultimately, these resonantmodes can be manipulated by turning the switches ON and OFF.Many switches can be incorporated in the slot for various reso-nant mode excitations but we use only two RF MEMS switchesfor basic proof of concept.

III. DESIGN IMPLEMENTATION USING PARTICLESWARM OPTIMIZATION

In this section, the frequency reconfigurability concept is re-alized through the use of Particle Swarm Optimization. Theschematic of the FR-ESPA is shown in Fig. 3. A multilayer de-sign was used with a Rogers RT Duroid 5880 substrate with

and 1.574-mm thickness on top of a foam substratewith and 10-mm thickness. The duroid layer was em-ployed to allow the antenna topology to be etched onto the sub-strate through photolithography, thus providing satisfactory fab-rication accuracy. The foam substrate was used to increase thebandwidth and have an overall effective substrate permittivityclose to 1.In this case, there are seven variables whose values must

be chosen. The complexity of the antenna design optimizationproblem increases drastically with higher parameter space di-mensionalities, making this a difficult design problem to solve.Parametric studies for problems of this nature are complicatedto quantitatively estimate the effect of each design parameteron the antenna performance. Therefore, the PSO technique wasapplied to this problem due to its robust convergence for prob-lems that are multimodal, non-differentiable, discontinuous,nonlinear, non-convex, and highly dimensional. PSO is a globaloptimization technique and is also well known for its simplealgorithm based on the social and cognitive mechanisms of beeswarms searching for food [16], [23]. More references on PSOand other nature-inspired optimization techniques can be foundlisted in [23].In Fig. 3, is the length of the patch, is the width of the

patch, is the slot length, is the slot width, is the slot

TABLE ISUMMARY OF THE FREQUENCY RECONFIGURABLE E-SHAPED PATCH

ANTENNA OPTIMIZATION (ALL DIMENSIONS IN mm)

position, is the position of the feed and is the position ofthe MEMS switch bars. Table I lists the fixed parameters, op-timization parameters, swarm size, number of iterations, andthe different boundaries and constraints used in this FR-ESPAimplementation. The swarm size was chosen to be double thenumber of parameters to be optimized based on the implemen-tation in [21]; however, a larger population can always be used.The constraints are formed in order to avoid designs which donot maintain the E-shape. The boundaries and the constraintsdefine the solution space and feasible space and thus accountfor all the geometrical aspects of the optimization.Here, two simulations are needed to evaluate the given set of

parameters. One simulation outputs the OFF state characteristicsand the other one outputs the ON state features. For this antennadesign problem, our main objective is to obtain good impedancematching dB over two specified frequency bands:

(1)

In this equation, is the fitness as a function of the designparameters . This equation can bedecomposed in the following manner. The first two terms rep-resent a form of a minimax optimization problem, where theworst (dB) in the frequency regions andare minimized. We chose to set 2.4 GHz as the center frequencyfor this design due to many popular bands for wireless commu-nications located near this band. Previous designs have shownroughly 30% bandwidth for a typical E-shaped patch of similarheight [21], and therefore we set GHz based onprevious observations. Similarly, the ON state frequency rangewas set to GHz. The difference term in (1) isadded in the fitness function to ensure convergence in both thefrequency bands. A penalty function is also incorporated intothe fitness function by simply adding a large number to

RAJAGOPALAN et al.: MEMS RECONFIGURABLE OPTIMIZED E-SHAPED PATCH ANTENNA DESIGN 1059

Fig. 4. (a) Prototype using ideal switch model. (b) Simulated and measuredperformance of the prototype with ideal switches.

the fitness if the design parameter set does not satisfy the con-straint equations.PSO was linked with High Frequency Structure Simulator

(HFSS) in order to simulate the performance, and the portdata fromHFSSwas extracted and processed by the fitness func-tion given in (1). Ideal switch models were used to represent theRF switches as a first-pass proof of concept and to reduce simu-lation and optimization time, as discussed in Section IV. In thismodel, we assume that the OFF state can be represented by asimple open circuit, while the ON state can be represented bya short circuit. The termination criterion utilized for our opti-mization runs was a maximum number of iterations, which wasset to 500 iterations. The PSO-HFSS program convergence re-sults showed that the average fitness approaches the global bestvalue, which is typically a good indication that the optimizationrun has converged, and no significant improvements are to beexpected. This can also indicate that the design is tolerable topossible design and fabrication errors if encountered.Fig. 4(a) shows the fabricated prototypes representing the

ideal switch model. Fig. 4(b) shows the simulated and measuredperformance for the optimized antenna and the fabricated

prototype. Table II shows the final design parameters for theideal switch case. The simulated return loss shows that the de-sign satisfies the criteria of 10 dB in both bands. In order toreplicate the ON state, copper tape was placed across the bars torepresent RF switches in ON state. The comparison betweenthe simulated and measured data shows very good agreement.The typical E-shaped antenna behavior is clearly seen in the OFFstate ranging from 2 to 2.7 GHz. By shorting the bridges, thefrequency shifts to a higher value thus covering the band from2.7 to 3.2 GHz due to the change in E-shaped slot dimensions.The slight differences between simulations and measurementcan be attributed to air gaps between layers and variations infoam thickness.

TABLE IIFINAL DESIGN PARAMETERS FOR THE IDEALSWITCH CASE (ALL DIMENSIONS IN mm)

Fig. 5. FR-ESPA OFF state response comparison between the simulatedMEMS switch models and measurement with ideal switches [Fig. 4(a)] andmeasurement with wirebonded MEMS. Major differences can be observed be-tween the ideal switch (open circuit) and the wirebonded MEMS.

IV. RF SWITCH MODELINGRF-MEMS switches are chosen as the switching elements

for antenna reconfiguration due to their satisfactory RF prop-erties including low insertion loss, excellent linearity, goodimpedance matching, and high isolation [24][27]. In thispaper, we used Radant MEMS RMSW100HP switches due totheir availability and good switch performance. RF-MEMSswitches were placed and wirebonded on the same fabricatedprototype from the previous section in order to test the per-formance with these switches. Fig. 5 provides a comparisonof the measured results between the wirebonded MEMSmeasurement when the MEMS switches are not actuated (OFFstate) and the ideal switch model case which assumes an opencircuit. The ON state only showed minor changes and thereforewas excluded from the plot. Clearly, significant differencesexist between the ideal switch model and the measurement withwirebonded MEMS due to the differences in the models, whichwas previously reported in [25]. This implies that the predicteddesign performance from the ideal switch model may not bevery accurate. From this, one can conclude that the ideal switchmodel may be insufficient for a final implementation.Accurate and fast switch modeling is critical for optimiza-

tion and final implementation. For global optimization runs, itis common to test thousands of different designs before arrivingat the global optimum, and therefore minimizing the simulationtime is critical to obtaining a design solution in a reasonableamount of time. Thus, there is a need to utilize both accurateand simple models for rapid optimization and final implemen-tation. Next, we briefly discuss three popular models often usedto model switches and decide which is most appropriate.

A. Ideal Switch ModelThe ideal switch model uses an open circuit to represent the

OFF case. The ON case is represented by a short circuit usually

1060 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 3, MARCH 2014

Fig. 6. Different switch models that can be used to implement RF-MEMSswitches in simulations. (a) Ideal switch model. (b) Circuit model forthe Radant RMSW100HP switch. (c) Full-CAD model for the RadantRMSW100HP switch.

realized by a galvanic connection between the two switch nodes.This model implementation is shown in Fig. 6(a). Internal ca-pacitances/reactances of these switches are not considered inthis model and in most cases this model has the fastest simu-lation time.

B. Circuit ModelCircuit models using lumped elements have been created in

an effort to model some of the observed MEMS switch featureswith minimal overhead. With the circuit model of the MEMSswitches, the OFF case is simulated by using a lumped capac-itor between the two nodes. The ON case is then simulated asa lumped resistor between the two terminals due to the contactresistance of the cantilever beam. This model implementationis shown in Fig. 6(b). The circuit model provides more detail toproperly model the MEMS switch and slight increases are ob-served in the relative simulation time in comparison to the idealswitch model. Also note that our model incorporates wirebondeffects by utilizing wirebonds connected to the RLC impedancesurface in HFSS.

C. Full-CAD ModelOne can simulate the full electromagnetic model of the

MEMS switch by incorporating all (or most) of its geometricalfeatures. This includes a high-loss silicon substrate as well assome representative transmission lines. In the OFF position,the transmission lines form an open connection. In the ONposition the transmission lines are connected metallically. ThisFull-CAD model is based on the model shown in [25][27].This model implementation is shown in Fig. 6(c). The fine

TABLE IIISUMMARY OF RELATIVE SIMULATION TIMES FOR THE SWITCH MODELS

COMPARED TO THE IDEAL SWITCH MODEL

geometrical features often require a long simulation time,which significantly prolongs the optimization run.As shown in Fig. 5, we observe good agreement between the

circuit model, Full-CAD model, and the wirebonded MEMSmeasurement when the MEMS switches are not actuated (OFFstate). This implies that both the Full-CADmodel and the circuitmodel are fairly comparable in predicting the performanceof the design. Some relative benchmark simulation times arealso shown in Table III to compare themodels. In the benchmarktests, a computer with two quad-core 2.5-GHz Intel Xeon E5420processor with 32 GB of RAM was used. The numbers given inthe table are all relative to the ideal switch model, which showthat the Full-CAD simulation model is much more time-con-suming than the ideal switch or circuit models.For this particular paper and application, we had to find the

best balance between accuracy and simulation speed for opti-mization. While more accurate models exist in literature [28],the circuit model shown in Fig. 6(b) seemed to provide the bestbalance in speed, accuracy, and simplicity, as evident in thecomparison to the measurements. Therefore, we proceeded touse the circuit model due to its rapid simulation time and abilityto accurately compute the performance of the design.

V. OPTIMIZED E-SHAPED PATCH ANTENNAWITH RF-MEMS SWITCHES

With the FR-ESPA concept proven through simulations andmeasurements of the optimized design with ideal switchmodels,the final step was the optimization of the E-shaped antenna usingthe circuit model. Thus, we incorporated the circuit model ofthe switch in the HFSS simulations, allowing us to proceed di-rectly from optimization to implementation. The same optimiza-tion methodology was applied to this design to obtain a finaloptimized design with accurate switch properties incorporated.Similarly, we chose to set this design to GHzfor the OFF state and GHz for the ON state asthe final implementation. In our studies, it was observed thatthe OFF state capacitive effects slightly decreased the OFF statebandwidth, and therefore we chose to use 2.6 GHz.Fig. 7(a) shows the fabricated final optimized prototype,

where MEMS switches were used to test the antenna perfor-mance. This was done so that radiation patterns for both thestates can be measured independently in the UCLA spher-ical near-field chamber. Fig. 7(b) shows the comparisonbetween the circuit model and the measured results for thefinal optimized FR-EPSA. Good agreement between the mea-surement and the simulation can be observed, and good

RAJAGOPALAN et al.: MEMS RECONFIGURABLE OPTIMIZED E-SHAPED PATCH ANTENNA DESIGN 1061

Fig. 7. (a) Fabricated final optimized FR-ESPA which incorporates MEMSswitches as shown in the inset figure. (b) Comparison between the simulatedcircuit model, Full-CAD model, and measured antenna for the final optimizedFR-ESPA.

TABLE IVFINAL DESIGN PARAMETERS FOR THE MEMSSWITCH CASE (ALL DIMENSIONS IN mm)

performance ( 10 dB) can be seen for 22.6 GHz and 2.63.2GHz in the OFF and ON states, respectively. This is a fairlylarge bandwidth achieved with an antenna of this size, and thefractional bandwidth is roughly 50%, which nearly doublesthe original bandwidth given by the original E-shaped patchantenna. The agreement once again validates the circuit modelfor MEMS switches used for optimization. Table IV shows thefinal design parameters for the circuit model case. Fig. 7 alsoshows the comparison between the circuit model and Full-CADModel for the final RF-MEMS optimized design. This resultalso validates that the circuit model is very similar to actualMEMS model.The radiation patterns of all different states are given in

Figs. 8 and 9 with the orientation of the E-shaped antennaprovided in Fig. 3. Fig. 8(a) and (b) shows the E- and H-planefor the OFF state at 2.05 GHz. This frequency correspondsto the mode where the currents travel around the slots forthe E-shaped patch. Cross- polarization is observed in theH-plane due to structural asymmetry created by the slots.

Fig. 8. Comparison between simulated and measured radiation patterns ofthe final optimized FR-ESPA for the OFF state. (a) E-plane2.05 GHz.(b) H-plane2.05 GHz.. (c) E-plane2.55 GHz. (d) H-plane2.55 GHz. Theboresight directivity for 2.05 and 2.55 GHz are 8.89 and 10.46 dB, respectively.

Fig. 9. Comparison between simulated and measured radiation patternsof the final optimized FR-ESPA for the ON state. (a) E-plane2.8 GHz.(b) H-plane2.8 GHz. (c) E-plane3.1 GHz. (d) H-plane3.1 GHz. Theboresight directivity for 2.8 and 3.1 GHz are 7.79 and 5.17 dB, respectively.

Fig. 8(c) and (d) shows the E- and H-plane for the OFF state at2.55 GHz. This frequency corresponds to the patch mode asseen in Fig. 2(a). The pattern features are similar to a simplepatch antenna. Fig. 9(a) and (b) shows the E- and H-plane forthe ON state at 2.8 GHz and Fig. 9(c) and (d) shows the E- andH-plane for the ON state at 3.1 GHz. High cross-polarizationcan be observed for the ON case, and this can be attributed toan increased presence of higher order modes under the patch.These modes resonate along the width of the patch, where thewidth is roughly 1 wavelength for frequencies near 3 GHz.

1062 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 3, MARCH 2014

Fig. 10. (a) Final optimized FR-ESPA design with MEMS switches and biaslines. (b) Comparison of between the simulated circuit model and measuredantenna for the final optimized FR-ESPA including the bias lines.

The measured directivities for each frequency are also listedin Figs. 8 and 9. Future optimizations of this design mightalso incorporate the cross-polarization in order to minimizeits presence at higher frequencies. In the array environment,techniques such as element rotation can also be used to helpalleviate cross-polarized radiation.

VI. MODEL WITH BIAS NETWORKSIn order to use the switches, one must deliver a DC voltage

to actuate the switches, which necessitates the use of someform of DC bias network. For this particular antenna design,the slots offer the most accessible location to place the biaslines. However, the E-shaped patch antenna has strong fringingfields within these slots, which can be problematic due tostrong interactions between the patch and the bias lines. Aftersome extensive research and tests, it was discovered that con-ductive adhesives could provide a compact, cost-effective, andwideband design solution [29]. A PELCO isopropanol basedgraphite based paint distributed by Ted Pella, Inc. was found toprovide the required resistivity which was roughly 2400for 25 m thickness [30]. The conductive adhesive was appliedusing scotch tape to create a mask. After the mask was created,the adhesive was painted onto the Duroid substrate by brush,which approximately gives a thickness of 25 m. After theadhesive has cured, the tape mask can be easily removed,leaving behind the desired bias lines [29]. The total resistanceof the DC bias line using conductive adhesive was measuredto be approximately 180200 k for the bias lines that were

about 0.5 mm 42 mm in size. These bias lines are depictedin Fig. 10(a), where the fabricated E-shaped patch with thewirebonded MEMS switches is also shown.A 90 V driver was connected to the MEMS switches, and

the ON and OFF states were measured by applying 90 and 0 V,respectively, to the gate of the MEMS switches. This voltagewas required for the RadantMEMS switches; however, it shouldbe noted that MEMS switches with lower actuation voltagescan be used. Note that the middle bias line was connected toground in order to properly control the gate-source voltage andfor grounding purposes. The results of the measurement areshown in Fig. 10(b). Good agreement between the measurementand the simulation can be observed, and good performance( 10 dB) can be seen for 2.02.6 GHz and 2.63.2 GHz inthe OFF and ON states, respectively. The bias lines do not affectthe performance of the antenna and strictly act as an RF-DC iso-lator. Therefore, negligible effects on the radiation patterns areto be expected.

VII. CONCLUSIONCognitive radio is an emerging and promising technology that

aims to provide freedom to wireless networks by taking ad-vantage of the unused spectrum. Reconfigurable antenna tech-nology can help address many of the challenges for antenna de-signs for these systems. This paper demonstrates a novel wide-band E-shaped patch antenna with frequency reconfigurability.The proposed design can be used to progress the functionalityof larger terminals or access points using patch antennas suchas laptops or base station antennas.This paper detailed the development, design, optimization,

and implementation of this antenna. The concept of frequencyreconfigurability for E-shaped patch antennas was proposed andthe concept verified using PSO. An initial prototype using idealswitches validated the concept. Different variations of MEMSswitch models were presented, and the circuit model was chosendue to its simulation accuracy and rapid optimization time. Finaloptimized designs using these circuit models were fabricatedand the frequency reconfigurable E-shaped patch antenna con-cept was demonstrated. An impedance bandwidth of 50% wasachieved. Overall, the measurements showed good agreementwith the simulations, and the frequency reconfigurability wasable to nearly double the fractional bandwidth of the E-shapedpatch. An implementation of resistive bias lines using conduc-tive adhesives was investigated and showed minimal interfer-ence with the radiation patterns and the impedance matchingperformance.

REFERENCES[1] W. Jakes, Microwave Mobile Communications. New York, NY,

USA: Wiley-IEEE Press, 1974.[2] J. Bernhard, Reconfigurable Antennas. San Rafael, CA, USA:

Morgan & Claypool, 2007.[3] D. Anagnostou, G. Zheng, M. Chryssomallis, J. Lyke, G. Ponchak, J.

Papapolymerou, and C. Christodoulou, Design, fabrication and mea-surements of an RFMEMS based self-similar reconfigurable antenna,IEEE Trans. Antennas Propag, vol. 54, no. 2, pp. 422432, Feb. 2006.

[4] A. Goldsmith, Wireless Communications. Cambridge, U.K.: Cam-bridge Univ. Press, 2005.

[5] J. Mitola, III, Cognitive radio for flexible mobile multimedia commu-nications, in Proc. IEEE Int. Workshop Mobile Multimedia Commun.,1999, pp. 310.

RAJAGOPALAN et al.: MEMS RECONFIGURABLE OPTIMIZED E-SHAPED PATCH ANTENNA DESIGN 1063

[6] B. Wang and K. J. R. Liu, Advances in cognitive radio networks: Asurvey, IEEE J. Sel. Topics Signal Process., vol. 5, no. 1, pp. 523,Feb. 2011.

[7] P. Hall, P. Gardner, and A. Faraone, Antenna requirements for soft-ware defined and cognitive radios, Proc. IEEE, vol. 100, no. 7, pp.22622270, Jul. 2012.

[8] P. Pawelczak, K. Nolan, L. Doyle, S.W. Oh, and D. Cabric, Cognitiveradio: Ten years of experimentation and development, IEEECommun.Mag., vol. 49, no. 3, pp. 90100, Mar. 2011.

[9] Y. Tawk and C. G. Christodoulou, A new reconfigurable antenna de-sign for cognitive radio, IEEE Antennas Wireless Propag. Lett., vol.8, pp. 13781381, 2009.

[10] H. Tarboush, S. Khan, R. Nilavalan, H. Al-Raweshidy, and D. Budimir,Reconfigurable wideband patch antenna for cognitive radio, inProc. Loughborough Antennas Propag. Conf., Nov. 2009, pp.141144.

[11] Y. Tawk, J. Costantine, K. Avery, and C. G. Christodoulou, Imple-mentation of a cognitive radio front-end using rotatable controlled re-configurable antennas, IEEE Trans. Antennas Propag, vol. 59, no. 5,pp. 17731778, May 2011.

[12] J. R. Kelly, P. Song, P. S. Hall, and A. L. Borja, Reconfigurable 460MHz to 12 GHz antenna with integrated narrowband slot, Progress inElectromagn. Res. C, vol. 24, pp. 137145, 2011.

[13] G. T. Wu, R. L. Li, S. Y. Eom, S. S. Myoung, K. Lim, J. Laskar, S. I.Jeon, and M. M. Tentzeris, Switchable quad-band antennas for cog-nitive radio base station applications, IEEE Trans. Antennas Propag.,vol. 58, no. 5, pp. 146681476, May 2010.

[14] E. Ebrahimi, J. R. Kelly, and P. S. Hall, Integrated wide-narrowbandantenna for multi-standard radio, IEEE Trans. Antennas Propag., vol.59, no. 7, pp. 26282635, Jul. 2011.

[15] J. Guterman, A.Moreira, C. Peixeiro, and Y. Rahmat-Samii, Wrappedmicrostrip antennas for laptop computers, IEEE Antennas Propag.Mag., vol. 51, no. 4, pp. 1239, Aug. 2009.

[16] J. Robinson and Y. Rahmat-Samii, Particle swarm optimization inelectromagnetics, IEEE Trans. Antennas Propag., vol. 52, no. 2, pp.397407, Feb. 2004.

[17] K. Kona, K. Bahadori, and Y. Rahmat-Samii, Stacked microstrip-patch arrays as alternative feeds for spaceborne reflector antennas,IEEE Antennas Propag. Mag., vol. 49, no. 6, pp. 1323, Dec. 2007.

[18] K. Luk, C. Mak, Y. Chow, and K. Lee, Broadband microstrip patchantenna, Electron. Lett., vol. 34, pp. 14421443, Jul. 1998.

[19] T. Huynh and K.-F. Lee, Single-layer single-patch widebandmicrostrip antenna, Electron. Lett., vol. 31, pp. 13101312, Aug.1995.

[20] F. Yang, X.-X. Zhang, X. Ye, and Y. Rahmat-Samii, Wide-bandE-shaped patch antennas for wireless communications, IEEE Trans.Antennas Propag., vol. 49, no. 7, pp. 10941100, Jul. 2001.

[21] N. Jin and Y. Rahmat-Samii, Parallel particle swarm optimization andfinite-difference time-domain (PSO/FDTD) algorithm for multibandand wide-band patch antenna designs, IEEE Trans. Antennas Propag.,vol. 53, no. 11, pp. 34593468, Nov. 2005.

[22] H. Rajagopalan, J. M. Kovitz, and Y. Rahmat-Samii, Frequency re-configurable wideband E-shaped patch antenna: Design, optimization,and measurements, in Proc. IEEE Antennas Propag. Soc. Int. Symp.(APSURSI), Jul. 814, 2012, pp. 12.

[23] Y. Rahmat-Samii, J. M. Kovitz, and H. Rajagopalan, Nature-inspiredoptimization techniques in communication antenna designs, Proc.IEEE, vol. 100, no. 7, pp. 21322144, Jul. 2012.

[24] G. M. Rebeiz, RF MEMS: Theory, Design and Technology. NewYork, NY, USA: Wiley, 2003.

[25] H. Rajagopalan, Y. Rahmat-Samii, and W. Imbriale, RF MEMS Ac-tuated reconfigurable reflectarray patch-slot element, IEEE Trans. An-tennas Propag., vol. 56, no. 12, pp. 36893699, Dec. 2008.

[26] G. Huff and J. Bernhard, Integration of packaged RF MEMSswitches with radiation pattern reconfigurable square spiral microstripantennas, IEEE Trans. Antennas Propag., vol. 54, no. 2, pp. 464469,Feb. 2006.

[27] I. Kim and Y. Rahmat-Samii, RF MEMS Switchable slot patch an-tenna integrated with bias network, IEEE Trans. Antennas Propag.,vol. 59, no. 12, pp. 48114815, Dec. 2011.

[28] K. Entesari, K. Obeidat, A. R. Brown, and G. M. Rebeiz, A2575-MHz RF MEMS Tunable filter, IEEE Trans. Microw. TheoryTech., vol. 55, no. 11, pp. 23992405, Nov. 2007.

[29] J. M. Kovitz, H. Rajagopalan, and Y. Rahmat-Samii, Practical andcost-effective bias line implementations for reconfigurable antennas,IEEE Antennas Wireless Propag. Lett., vol. 11, pp. 15521555, 2012.

[30] Tech. Notes, PELCO Conductive Graphite Isopropanol Based.[Online]. Available: http://www.tedpella.com/technote_html/16053%20TN.pdf

Harish Rajagopalan (S03M11) received the B.E.degree in electronics and telecommunications fromthe Government College of Engineering, Pune, India,in August 2002, the M.S. degree in electrical engi-neering from Auburn University, Auburn, AL, USA,in June 2005, and the Ph.D. degree from the Uni-versity of California, Los Angeles (UCLA), Los An-geles, CA, USA, in June 2011.He was a Post-Doc Scholar at the UCLA Antenna

Laboratory under the supervision of Prof. YahyaRahmat-Samii and a Part-Time Lecturer at UCLA

from June 2011 to June 2012. He is currently working at Apple Inc., Cupertino,CA, USA. His research interests include characterization of reflectarray ele-ments, reflectarray antennas and diagnostics, reconfigurable antenna designs,bio-electromagnetics, ingestible bio-medical devices, RFID tags and systemsand nano-electromagnetics for solar power. He has published over 30 peer-re-viewed journal papers and conference papers. He is an active reviewer for IEEETRANSACTIONS ON ANTENNAS AND PROPAGATION and several other journals.Dr. Rajagopalan was awarded the Young Scientist award at the 2011 URSI

General Assembly, Istanbul, Turkey. He was the recipient of Electrical Engi-neering 20102011 Outstanding Ph.D. Student Award. In 2011, he won theBest Student Paper Award at the Antenna Measurement Techniques Associa-tion (AMTA) Conference. In 2009, he received the Best Poster Award at theUCLA Tech Forum. In 2008, he received the Best Student Paper Award at TheUnited States National Committee for the International Union of Radio Science(USNC-URSI) Conference. He was the recipient of the UCLADean Fellowshipaward (2006) and UCLA Dissertation Fellowship award (2010).

Joshua M. Kovitz received the B.S. degree in elec-trical engineering (summa cum laude) from the Uni-versity of Houston (UH), Houston, TX, USA, in 2010and theM.S. degree in electrical engineering from theUniversity of California Los Angeles (UCLA), LosAngeles, CA, USA, in 2012. He is currently workingtowards the Ph.D. degree at UCLA under the super-vision of Prof. Yahya Rahmat-Samii in the AntennaResearch, Analysis, and Measurement Laboratory.While at UH, he was involved with the Applied

Electromagnetics Laboratory and conducted researchin computational electromagnetics for geophysical applications. He also partic-ipated in the NSF REU program, where he worked on Structural Health Mon-itoring (SHM) with wireless sensors in the Wireless System Research Group(WiSeR) under Prof. Rong Zheng. Currently at UCLA, his primary research fo-cuses on practical antenna system design for cognitive radio applications. Hisprimary research interests include reconfigurable antennas, microstrip patch an-tennas, applied electromagnetics, nature-inspired optimization techniques, ve-hicular antennas, cognitive radio, wireless communications systems, andMIMOantenna systems.Mr. Kovitz has received several awards and is actively involved with student

and professional groups. At UH, he was awarded the Outstanding Electrical En-gineering Senior of the Year 2010 and the Outstanding Junior of the Year 2009for academic excellence and student group involvement. He was also honoredto participate as the Banner Bearer for the Cullen College of Engineering duringthe 2010 graduation ceremony. During his time at UCLA, he was awarded theUCLA Electrical Engineering Deans Fellowship as well as the UCLA Grad-uate Division Fellowship. He was the highest ranked student in the 2012 UCLAElectrical Engineering Ph.D. Preliminary Exam within the Physical and WaveElectronics Area and was awarded a University Fellowship. In 2012, he wasthe recipient of the Distinguished Masters Thesis Award in Physical and WaveElectronics for his research in nature-inspired optimization techniques appliedto antenna designs. He was also awarded a prestigious National Defense Scienceand Engineering Graduate (NDSEG) Fellowship. He was awarded the EdwardK. Rice Outstanding Masters Student for 2012. He is a member Tau Beta Pi,Eta Kappa Nu, and Phi Kappa Phi.

1064 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 62, NO. 3, MARCH 2014

Yahya Rahmat-Samii (S73M75SM79F85)is a Distinguished Professor, holder of the NorthropGrumman Chair in Electromagnetics, member ofthe U.S. National Academy of Engineering (NAE)and past Chairman of the Electrical EngineeringDepartment, University of California, Los Angeles(UCLA). He was a Senior Research Scientist withthe National Aeronautics and Space Administration(NASA) Jet Propulsion Laboratory (JPL), CaliforniaInstitute of Technology prior to joining UCLA in1989. In summer 1986, he was a Guest Professor

with the Technical University of Denmark (TUD). He has also been a con-sultant to numerous aerospace and wireless companies. He has been Editorand Guest Editor of numerous technical journals and books. He has authoredand coauthored over 800 technical journal and conference papers and haswritten 30 book chapters. He is a coauthor of Electromagnetic Band GapStructures in Antenna Engineering (Cambridge Univ. Press, 2009), ImplantedAntennas in Medical Wireless Communications (Morgan & Claypool Pub-lishers, 2006), Electromagnetic Optimization by Genetic Algorithms (Wiley,1999), and Impedance Boundary Conditions in Electromagnetics (Taylor &Francis, 1995). He has received several patents. He has had pioneering researchcontributions in diverse areas of electromagnetics, antennas, measurementand diagnostics techniques, numerical and asymptotic methods, satellite andpersonal communications, human/antenna interactions, RFID and implantedantennas in medical applications, frequency selective surfaces, electromagneticband-gap structures, and applications of the genetic algorithms and particleswarm optimization.Dr. Rahmat-Samii is a Fellow of the Institute of Advances in Engineering

(IAE), a Fellow of Antenna Measurement Techniques Association (AMTA),a Fellow of Applied Computational Electromagnetics Society (ACES) and amember of Commissions A, B, J, and K of USNC-URSI, and a member ofSigmaXi, Eta KappaNu and the Electromagnetics Academy. He was Vice-Pres-ident and President of the IEEE Antennas and Propagation Society in 1994 and1995, respectively. He was appointed an IEEE AP-S Distinguished Lecturerand presented lectures internationally. He was a member of the Strategic Plan-ning and Review Committee (SPARC) of the IEEE. He was the IEEE AP-S

Los Angeles Chapter Chairman (19871989); his chapter won the best chapterawards in two consecutive years. He is listed in Whos Who in America, WhosWho in Frontiers of Science and Technology and Whos Who in Engineering.He has been the plenary and millennium session speaker at numerous nationaland international symposia. He has been the organizer and presenter of manysuccessful short courses worldwide. He was a Directors and Vice President ofAMTA for three years. He has been Chairman and Co-chairman of several na-tional and international symposia. He was a member of the University of Cal-ifornia at Los Angeles (UCLA) Graduate council for three years. He was thechair of USNC-URSI for the period of 20092011. For his contributions, Dr.Rahmat-Samii has received numerous NASA and JPL Certificates of Recogni-tion. In 1984, he received the Henry Booker Award from URSI, which is giventriennially to the most outstanding young radio scientist in North America. Since1987, he has been designated every three years as one of the Academy of Sci-ences Research Council Representatives to the URSI General Assemblies heldin various parts of the world. He was also invited speaker to address the URSI75th anniversary in Belgium. In 1992 and 1995, he received the Best Appli-cation Paper Prize Award (Wheeler Award) for papers published in 1991 and1993 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION. In 1999, he re-ceived the University of Illinois ECE Distinguished Alumni Award. In 2000,Prof. Rahmat-Samii received the IEEE Third MillenniumMedal and the AMTADistinguished Achievement Award. In 2001, Rahmat-Samii received an Hon-orary Doctorate in applied physics from the University of Santiago de Com-postela, Spain. In 2001, he became a Foreign Member of the Royal FlemishAcademy of Belgium for Science and the Arts. In 2002, he received the Tech-nical Excellence Award from JPL. He received the 2005 URSI Booker GoldMedal presented at the URSI General Assembly. He is the recipient of the 2007Chen-To Tai Distinguished Educator Award of the IEEE Antennas and Propa-gation Society. In 2008, he was elected to the membership of the US NationalAcademy of Engineering (NAE). In 2009, he was selected to receive the IEEEAntennas and Propagation Society Highest Award, Distinguished AchievementAward, for his outstanding career contributions. He is the recipient of the 2010UCLA School of Engineering Lockheed Martin Excellence in Teaching Award,the 2011 UCLA Distinguished Teaching Award and the 2011 IEEE Electromag-netics Award. He is the designer of the IEEE AP-S logo which is displayed onall IEEE AP-S publications.