z
zyxw
Integrating Antenna Modeling Codes in
Web-Based Visualization Environments
zyxwv
zyxwvuts
Antonis I. Kostaridis, Christos G. Biniaris, Andy J. Marsh, Hristos T. Anastassiu, and
Dimitra 1. Kaklamani
Department of Electrical and Computer Engineering
National Technical University of Athens
9. lroon Polytechniou Str., GR-15780 Zografos. Athens, Greece
Tel: +30210 7722277; Fax: +30210 7723557; E-mail: dkaklam@cc.ece.ntua.gr
Abstract
The development of an interactive online asynchronous environment for advanced antenna modeling is presented. Electrcmagnetic (EM) codes are coupled with the World Wide Web (WWW), and the Virtual Reality Modeling Language (VRML) is
used for the creation of a flexible visual pre- and post-processor. The adoption of object-oriented methodologies, in conjuno
tion with the employment of platform independent standards, contributes to the flexibility of the developed application. It is
generic enough so that various numerical codes for antenna modeling can be easily integrated under a common multi-disciplinary environment. A case study regarding the modeling of conformal microstrip arrays is presented.
Keywords: Visualization; modeling; antenna radiation patterns; conformal antennas; microstrip antennas; microstrip arrays;
client-server systems; object oriented programming; WWW; Java servlets; VRML; antenna modeling
1. Introduction
R
esearch in antenna modeling has been greatly favored by
ecent advances in computational electromagnetics. The
improvement of the diverse numerical techniques and their
hybrids, in conjunction with the availability of cheap processing
power, has given rise to the simulation of more realistic scenarios.
The increase in complexity of the simulated scenarios has also led
to the need for development of sophisticated visualization tools
that permit the users to conceptualize all the involved parameters
in an efficient way. For example, the ability to visualize two- and
three-dimensional radiation patterns is a key to the understanding
of basic antenna principles. Additionally, antenna analysis and
design software packages can he made accessible through standard
browsers, due to the unprecedented opportunities offered by current Web technologies, removing otien-complicated interfaces.
Several issues on this topic have been addressed in the literature
[l-61.
Our objective is to introduce the potential possibilities of
using the WWW and associated technologies as the interface environment and the communication mechanism to integrate EM
codes. By coupling Web technology with visualization languages
that offer interactivity and are suited to a Web environment - such
as the Vimal Reality Modeling Language (VRML) we can
achieve the development of realistic visualization applications,.
which provide a number of unique features over traditional visualization techniques.
~
In this context, we present the development of an interactive,
online, asynchronous application for antenna modeling. The preprocessing and post-processing phases, along with the numerical
IEEEAntennas and Propagation Magazine. Vol. 45, No. 4, August 2003
analysis, are integrated under a unified environment that revolves
around a Web server. This environment utilizes server-side components that provide the necessary functionality for the Web integration, and the VRML to visualize two- and three-dimensional
objects. The system acts as a front-end to external numerical
codes. Preliminary work on the aforementioned application is
described in [7]. As a case study, we present the use of the interactive environment for the modeling of conformal microstrip antenna
arrays, using the Modified Method of Auxiliary Sources (MMAS)
[8-121.
zyxwvuts
In Section 2 we propose a generic client-server architecture,
and in Section 3 we describe the enabling Web technologies. Section 4 focuses on the implementation of components residing on
the Web server that provide the core functionality, realizing the
Web integration ofthe EM codes. Finally, the prototype interactive
environment for our case study is demonstrated in Section 5 .
2. The Client-Server Architecture
The main component of the proposed architecture is a Web
server where the service'logic is implemented (Figure I). To
access the remote facilities, a prospective user makes an HlTP
(Hyper Text Transport Protocol) request to the applications-password-guarded home page. He or she enters a password and posts it.
At the other side, an HTTP-request-handler component checks the
validity of the password, and returns the page containing a fill-in
form for the input parameters of the application. The user then
completes the form and posts it. This request is handled by another
server component, which is responsible for wrapping the EM code,
zyxwvuts
11
providing the necessary execution environment, and activating the
native executable.
Since complex simulations require significant CPU resources
and may take several minutes or even longer to complete, the client’s Web-browser timeout setting will quite oAen intempt the
process. Thus, a job-submission mechanism is mandatory. Namely,
the user submits his or her input processing by additionally providing an e-mail address. After posting the values of the input
parameters, the user may close the browser, and wait for the simulation to be completed.
Upon completion of the job, the system e-mails a message to
the user containing the URL of the visualization home page. This
page requires a password and an e-mail address in order to access
the user-specific directory containing the results of the last simulation. After posting this information, a last component is activated
that performs the transformation of the output data into the VRML
format, and constructs the post-processor page containing links to
the VRML files.
zyxwvutsrq
zyxwvutsrqp
zyxwvutsrq
zyxwvutsrqponm
zyxwvutsrq
zyxwvuts
The client-server architecture utilized provides the possibility
for multiple users to interact with the system simultaneously. Furthermore, accessing the stored simulated results through the Web
enhances collaboration, as engineers can interact with their colleagues and review their work. Moreover, by adopting an objectoriented approach and by standardizing the format of the EM
codes’ output data, a generic and extensible framework can be
developed so that established engineering analysis and design tools
can he embedded into the system with minor additions andor
modifications.
3. Web Technologies
As has been previously mentioned, the integration of EM
codes in a Web environment requires an adequate server-side
mechanism. The most common mechanism is provided by the
Common Gateway Interface (CGI) scripts [14], written in either
the Per/ or C languages. Nevertheless, during the last few years,
significant developments toward Java servlets [I51 have been
achieved. Servlets are an efficient platform-independent integration technology for server-centric development and seamless
deployment of back-end systems and modules. They are pieces of
Java source code that add functionality to a Web server in a manner similar to the way applets add functionality to a Web browser.
Servlets are designed to support a requestlresponse computational
model that is commonly used in Web browsers. Unlike CGI
scripts, a servlet does not run in a separate process, and remains in
memory between requests. On the other hand, when a server
receives a CGI request, it must create a new process, allow this
process to start and terminate, and then return the resulting text to
the Web browser. As one can imagine, sending many simultaneous
CGI requests would very quickly bring a dramatic slowdown in the
performance of the processor hosting the Web sewer. On the contrary, when a servlet is invoked as a thread by an instance of the
Java Virtual Machine, multiple requests for the same servlet do not
result in multiple instances of the Java interpreter running simultaneously.
Figure 2. A typical servlet implementation.
Under the examined framework, servlets can act as wrappers
for the native EM code, with primary responsibilities: namely, the
verification of the integrity constraints on input data that a user
posts, the activation of the EM code, and, finally, the transformation of the code’s results according to VRML [16]. VRML specifies a file format (.wrl) for describing three-dimensional interactive
worlds and objects. It may be used to create three-dimensional representations of complex scenes, such as illustrations, product definition, and virtual-reality (VR) presentations. VRML is capable of
representing static and animated objects, and can have hyperlinks
to other media, such as sound, movies, and images. Interpreters
(browsers) for VRML [17, 181 are widely available for many different platforms, as are authoring tools for the creation of VRML
files.
4. Implementing Java Servlets
The proposed architecture mainly defines four types of servlets, residing on a servlet-capable Web server. The first one handles the user-authentication stage. This servlet interacts with userprofile data, determining the EM codes that a user is authorized to
access and the position where simulation results should he stored.
The second servlet handles the application-profile management,
guiding the user in the data-input process, and performing the validation according to the selected application. The third servlet acts
as a wrapper for the EM code, enabling the allocation, configuration, and remote execution of the selected application. This servlet
monitors the simulation process, and is responsible for informing
the user after execution has been completed. Finally, the fourth
servlet generates the VRML objects that represent the simulation
results (e.g., radiation pattems, current distributions, etc.) and
zyxwvutsrqp
The main feature of a servlet is the capability of dynamic
HTML code generation. In addition, being Java source code, the
servlets inherit the advantages of the object-oriented and platformindependent Java language, and can access the Application Programming Interfaces (APls), as do any other programs written in
Java. Thus, the functionality they can provide is significant.
12
IEEEAntennas and Propagation Magazine, Vol. 45, No. 4, August 2003
zyxwvutsrqponml
zyxwvutsrqponmlkji
zyxwvutsrqpo
P a 5 m d Reqvest
Form
EM nwlt
3
Graphics1user
m c e
zyxwvutsrqponmlkjih
Figure 1. The proposed architecture.
Figure Sc. Visualization of the array: an expanded view of the
upper-right target frame in Figure Sa.
Figure Sa. Visualization of the results.
Figure Sd. Visualization of the array: an expanded view of the
lower-left target frame in Figure Sa.
zyxwvutsrq
Figure Sh. Visualization of the array: an expanded view of the
upper-left target frame in Figure Sa.
/€€€Antennas and Propagation Magazine, Vol. 45, No. 4. August 2003
Figure Se. Visualization of the array: an expanded view of the
lower-right target frame in Figure Sa.
13
zyxwvutsr
zyxwvuts
zyxwvutsrq
zyxwvutsrqpo
zyxwvutsrq
responds with the post-processing HTML pages containing hyperlinks to these VRML objects.
From the developer's perspective, the aforementioned sewlets should extend the HttpServlet class of Sun's Servlet API.
The respective compiled classes have to be placed in appropriate
directories inside the Web server's installation path [19, 201. Each
servlet interacts using two objects, HttpServletRequest and
HttpServletResponse [15]. A new request invokes the
servlet doPost ( ) method, which should be overridden so that
parameters of the request can he accessed and processed according
to the given sewlet call. The example code in Figure 2 presents a
typical implementation of a servlet class, and some indicative
methods that some of the servlets should implement. Within the
doPost method, the getparameter ( 1 method gets the sewlet's expected argument(s) and processes them. As a response to
the client's request, the example doPost method uses a Writer
from the HttpServletRespOnSe object to return the HTML
data. Before accessing the Writer, the example sets the contenttype header. At the end of the doPost method and after the
response has been set, the writer is closed.
lnstnuteof Communicationsand Computer Systems
-
MICIOW~M~
and Fibel Optics Laboratory
-
Virtvol A m e m Derign Simulotw
Figure 4a. The introductory pre-processing Web page.
zyxwvutsrqpon
Concerning the method that calls the EM code, it must be
pointed out that Java provides two options for executing an external native code from the context of a Java program. The first one
is provided through the Java Native Interface ( N I ) API, which
gives a Java program the possibility to call native methods [15].
The second option - which is simpler and more adequate for hatchtype simulations - is provided through the use of the exec method
ofthe Runtime class o f the standard Java API. As it can be seen
from the code fragment of Figure 3, a new Process object must
be instantiated that is used to execute a specific command. When
Java forks a new process, it redirects stdin, stdout, and
stderr. For this reason, in order to communicate with the program one must make use of the methods getInputStream(),
getOutputstream0, and getErrorStream0 of the
Process object. This provides full control of the new executing
process. It is important, though, to underline that Java applications
that use the exec method lose their portability. However, it is
possible to identify the underlying operating system from a
Properties object, and to adapt the command to be executed
I'
ProcessEM_code:
Figure 3. A code fragment for the exeeution of an external
numerical program from fava.
14
Figure 4b. The introductory pre-processing Web page for
entering geometrical parameters.
zyxw
Figure 4c. The introductory pre-processing Web page for
entering MMAS parameters.
IEEE Antennas and Propagation Magazine, Vol. 45. No.4, August 2003
zyxwvutsrq
Figure 6. A 2 x 3 antenna-array geometry
Figure 8. Combining the antenna geometry and the radiation
pattern.
Figure 9a. A sample output VRML file.
Figure 9b. A sample output VRML file.
zyxwvutsrqp
zyxwvutsrqpo
zyxwvutsr
Figure 9c. A sample output VRML file.
IEEEAnfennas and Propagation Magazine. Vol. 45, No. 4, August 2003
Figure 9d. A sample output VRML file.
15
zyxwvutsrqpo
zyxwvutsrqpo
zyxwv
zyxwv
zyxwvutsr
zyxwvutsrq
zyxwvuts
accordingly. The code in Figure 3 presents an example of how the
execution differs on Windows 98 and 2000 platforms.
5. A Case Study: Modeling Conformal
Microstrip Antenna Arrays
5.1 The Interactive Environment
In our case study, Java servlets interface with a code wrinen
in FORTRAN90 to provide a Web-based front-end. This code
analyzes conformal patch arrays with microstrip excitation, printed
on a cylindrical surface. The numerical technique employed is the
Modified Method of Auxiliary Sources (MMAS) [8-12]. The main
concept of MMAS is given in Subsection 5.2.
a m y of patch antennas, fed by microstrip lines. Each antenna consists of a radiating element (conducting patch), mounted on a
dielectric substrate.
MMAS was introduced to improve the ability of MAS to
treat relatively flat three-dimensional structures. Unlike standard
MAS, where the analysis is based on elementary source fields [I 112, 23-24], in (MMAS), the building blocks of the methodology
are the source current and charge densities on the auxiliary surfaces. Both charges and currents lie at the nodes or midpoints of a
canonical grid, which serves as a discrete model of the auxiliary
surface where derivatives can be approximated by finite differences. Numerical calculations have demonstrated that this modification, apan from improving the accuracy of the technique, also
significantly reduces the computational effort, when applied to
geometries of small thickness.
Applying MMAS, and assuming an exp(-/ot)
The process of performing a simulation is initiated by a client
request that includes the interaction of several dynamically generated HTML pages controlled by servlets. More specifically, the
user interface at the pre-processing phase comprises a passwordprotected HTML page, and two pages containing fill-in forms,
where all the parameters of the simulation are defined (see Figure 4).
In the first page, the user inserts the geometrical parameters
of the structure under investigation. At the end of the page, he or
she is requested to type his or her e-mail address, in order to enable
the asynchronous communication. The form is then posted, and the
triggered servlet creates the file containing the geometrical
parameters and the VRML representation of the conformal antenna
array geometry.
:
If a parameter value is invalid or missing, the servlet returns
a page describing the error(s). Otherwise, it returns a second page,
where the user must fill or select from drop-down lists the EM
parameters and the parameters related to the implementation of
MMAS, such as the number of auxiliary sources, the amplitudes
and phases of the array elements, etc. This page also contains a
link to the VRML file of the geometry, so that the user can inspect
it and make modifications, if needed, by pressing the back button
of his or her browser.
The user automatically receives an e-mail upon completion
of the job, notifying him or her of the URL of the page containing
the links of the results. The lack of a standardized command-line email system in Windows 9YNT has been overcome by using the
JavaMail API [ZI]and the JavaBeans Activation Framework API
time
dependence, the electric field, E(r) , due to a surface current density J(r') on the auxilialy surface ,,s
,
is given in terms of the
vector potential, A, and the scalar potential, 0 ,
as
E(r) = - j o A ( r ) - V @ ( r )
,
(1)
where
together with the current continuation law
- j o p ( r ' ) = Vs* J (1').
(3)
In Equations (l)-(3), J is the unknown surface current density, and
p is the unknown surface charge density, both being due to the
auxiliary sources. Moreover, p and E are the medium permeability and permittivity, respectively, and o is the frequency. To
approximate the divergence in Equation (3) in a discrete way, an
auxiliary source grid is utilized, the planar projection of which is
shown in Figure 7. Following a particular discretization pattern,
PI.
The post-processor Web page is shown in Figure 5 . It consists of one inspector frame, on the left, and four target frames, on
the right. From the inspector kame, the user can select one or more
hyperlinks of the VRML files to he visualized on a target frame. In
this way, easy comparison of radiation pattems and independent
manipulation of the different plots are achieved.
5.2 The EM Solver
zyxwvu
zyxwvutsr
The EM solver developed implements MMAS [S-121 to ana-
16
U'
IEEE Antennas and Propagation Magazine, Vol. 45. No. 4, August 2003
zyxwvuts
zyxwvutsrqp
the charge density is sampled at the nodes of the grid, whereas the
current density is sampled at its midpoints. If the subscripts in J
denote a vector component along a given direction, the discrete
approximation of Equation (3) at a point (mAu’,nAv’)of the grid
is written as (for simplicity, Au’ and AV’ are suppressed in all
arguments)
-iw(m,n)
objective of the authors’ current activities is to utilize the Web
resources for parallelidistributed computing, and to demonstrate
the capabilities of the interaction between advanced EM solvers,
remote computing networking, and advanced visualization techniques, including stereoscopic imaging and immersive VR. At present, the results in VRML format can be used to illustrate a threedimensional far-field model. The VRML 2.0 format could also be
used to introduce a temporal dimension and WWW-supported
facilities. For example, audio could also be used to enhance the
presentation of results by using sonification.
7. Acknowledgements
The authors would like to thank the EC DGIII for the supportive funding of this work, within the MADS Project EP 28363
under the ESPRIT Program.
implying a dependency between charge and current samples.
Equation (4) infers that it is advantageous to consider the currentdensity samples as primary unknowns that exclusively define the
degrees of freedom in the problem, and the charge-density samples
as secondary (dependent) unknowns. Thus, the position of the
“auxiliary sources” in MMAS is identical to the sampling location
of the current densities (grid midpoints). Moreover, at each sampling site, a single component of the current density is defined
(either along U‘ or v’ ,but never along both directions),
8. References
zyxwvutsrq
zyxwvutsrqpo
zyxwvutsrqpo
zyxwv
By expanding the auxiliary source currents in terms of
weighted basis functions, the unknown weights are determined as a
solution to a linear system of equations, constructed through
enforcement of the boundary conditions on the physical surfaces of
the structure. More mathematical details about MMAS are given in
1131.
5.3 Post Processing
The results of the analysis include characteristic antenna
parameters of the antenna array, which may be either single-Valmued, such as the input impedance, or functions of the co-ordinates,
such as the antenna near field at a specified distance, and the far
field. Furthermore, it is advantageous to the code user to visualize
the far-field pattern superimposed on the actual antenna geometry.
In this way, the user can have a global perception of how the geometrical features of the antenna affect the radiation pattern. As
shown in Figure 8, the geometry (leA-hand side) and far-field pattern (right-hand side) which can be gyrated independently are
combined onto a single screen. Each one of them is presented in its
own three-dimensional coordinate space, with a pointer depicting
an angle of 6 and (. The user can manipulate the pointers by
adjusting the slide-bar values for 8 and @ .The user can also visualize each plot in a separate window by clicking on it. A list of
sample outputs, consisting of two- and three-dimensional patterns,
is depicted in Figure 9.
~
~
I. E. S. Fisher and E. Michielssen, “Mathematica Assisted WebBased Antenna Education,” IEEE Transactions on Education, 41,
4, November 1998, pp. 353-368.
2. D. P. Economou, A. Marsh, and D. I. Kaklamani,
“P(MoM/MAS): An Interactive Environment, Coupling WWW
Technology and Parallel Processing, to Solve Large Size Electromagnetic Problems,” IEEE Antennas and Propagation Magazine,
41, 1, February 1999, pp. 130-137.
3. A. Marsh, D. 1. Kaklamani, H. T. Anastassiu, and P. Shuhitidze,
“The P(MMAS) Interactive Environment: An Intemet Service
Coupling Web Technology and HPCN to Analyze Conformal
Microstrip Arrays,” Proceedings of the I 0th Mediterranean Elecrrotechnical Conference (MeleCon ‘2000), Cyprus, May 19-21,
2000,pp.411-414.
4. C. Bajaj and S. Cutchin, “Web Based Collaboration Visualization of Distributed and Parallel Simulation,” Proceedings of the
I999 IEEE Symposium on ParaNel Visualization and Graphics,
1999, San Francisco, California, United States, pp. 47-54.
5. R. Muralidhar, S. Kaur, and M. Parashar, “Architecture for
Web-Based Interaction and Steering,” Proceedings of Euro-Par
2000, August 2000, pp. 1332-1339.
7. A. I. Kostaridis, C. G. Biniaris, A. Marsh, H. T. Anastassiu, and
D. I. Kaklamani, “Advanced Visualization Techniques in Antenna
Modelling Using VRML and Web Technologies,” List of Abstracts
of the 8th International Conference on Advances in Communications and Control (COMCON S’, Crete, Greece July 25-29, 2001,
p. 16.
6. Future Work
This paper concentrates on using the WWW and the associated technologies as a remote service to solve EM problems. The
IEEEAnlennas and Propagation Magazine, Vol. 45, No. 4, August 2003
zyxwv
zyxw
zyxw
6 . R. D. Kriz, R. T. Levensalor, and S. D. Parikh, “Interactive Scientific Visual Data Analysis Using Java, PV-Wave, and IMSL,”
Visualization Development Environments 2000 Proceedings,
Princeton, New Jersey, April 27-28, 2000.
8. P. Shubitidze, D. I. Kaklamani, and H. T. Anastassiu, “Modified
Method of Auxiliary Sources Applied to the Analysis of Planar
and Cylindrically Shaped Microstrip Antennas,” Proceedings of
the Intu‘mational Conference on Electromagnetics in Advanced
17
zyxwvutsrqpo
zyxwvut
zyxwvutsr
zyxwvutsrq
zyxwvutsrqp
zy
zyxwvutsrq
zyxwvutsrqp
Applications (lCEAA99), Torino, Italy, September 13-11, 1999, pp.
375-378.
14. The Common Gateway Interface, http://hoohoo.ncsa.uiuc.edu/
cgi/overview.html.
9. D. L. Kaklamani, H. T. Anastassiu, and P. Shuhitidze, “Analysis
of Conformal Patch Arrays via a Modified Method of Auxiliary
Sources (MMAS),” Abstract Book of the Millennium Conference
on Antennas and Propagation, Davos, Switzerland, 9-14 April
2000, p. 292, and Proceedings of the Millennium Conference on
Antennas and Propagation (AP2OOO), Davos, Switzerland, April 914,2000.
15. SUN Microsystems Java Tutorial, http:/ljava.sun.com/docs/
booksltutoriall.
16. VRML 2.0, http:/l~w.vrml.org/VRML2.0/FINAL.
17. Cosmo Software, http:l/www.cai.comlcosmoihtml/player.htm.
18. OpenVRML, http://openvrml.sourceforge.neti.
10. H. T. Anastassiu, A. I. Kostaridis, C. G. Biniaris, and D. I.
Kaklamani, “Reduced Complexity Analysis of Microsnip Patch
Arrays, Conformally Mounted to a Cylindrical Conducting Surface,” 2001 IEEE International Symposium on Antennas and
Propagation and USNCAJRSI National Radio Science Meeting,
Boston, Massachusetts, July 8-13, 2001.
11. D. I. Kaklamani, “The Method of Auxiliary Sources in Computational Electromagnetics,” Proceedings of the European Congress on Computational Methods in Applied Sciences and Engineering (ECCOMAS ZOOO), Barcelona, Spain, September 11-14,
2000.
12. D. I. Kaklamani and H. T. Anastassiu, “Aspects of the Method
of Auxiliary Sources (MAS) in Computational Electromagnetics,”
IEEE Antennas and Propagation Magazine, 44, 3, June 2002, pp.
48-64.
13. F. Shuhitidge, H. T. Anastassiu, and D. 1. Kaklamani, “An
Improved Accuracy Version of the Method of Auxiliary Sources
for Computational Electromagnetics,” submitted to IEEE Transactions on Antennas and Propagation.
19. The Apache Software Foundation, http:llwww.apache.org/.
20. The vqServer Home Page, http://www.vqsoft.comJvq/server/.
21. SUN Microsystems JavaMail API, http://java.sun.com/
productsljavamaill.
22. SUN Microsystems Java Beans, http:/ljava.sun.comeans,
23. P. J. Papakanellos, D. I. Kaklamani, and C. N. Capsalis,
“Analysis of an Infinite Current Source Above a Semi-Infinite
Lossy Ground Using Fictitious Current Auxiliary Sources in Conjunction with Complex Image Theory Techniques,” IEEE Transactions on Antennas andPropagation, 49, October 2001, pp. 14911503.
24. H. T. Anastassiu, D. I. Kaklamani, D. P. Econornou, and 0.
Breinbjerg, “Electromagnetic Scattering Analysis of Coated Conductors with Edges Using the Method of Auxiliary Sources (MAS)
in Conjunction with the Standard Impedance Boundaty Condition
(SIBC),” IEEE Transactions on Antennas and Propagation, 50,
January 2002, pp. 59-66. $?
I Changes of Address
Information regarding subscription addresses is managed by
IEEE headquarters. It is not maintained, nor can it he changed, by
any member of the Magazine staff. If you are a member of the
IEEE, your subscription is sent to the address in your IEEE member record. Your record can he updated via the Web a1
http:llwww.ieee.orglmembersbip/coa.html.
This can also he done
by contacting IEEE headquarters: Member Address Records, IEEE
Headquarters, 445 Hoes Lane, Piscataway NJ 08855-1331 USA;
Tel: +I (908) 981-0060 or +I (800) 678-4333; Fax: +I (908) 9819667; E-mail: address. change@ieee.org. If you are an institutional
or other non-member subscriber, contact IEEE Customer Service
at the above address, telephone, and fax numbers; E-mail:
customer.service@ieee.org. Do not send requests to any member
of the Magazine Staff.
zyxwvuts
tEEE Antennas and Propagation Magazine, Vol. 45, No. 4, August 2003