Title

Author(s)

Citation

Issue Date

Finite-Element Formulation in Terms of the Electric-Field

Vector for Electromagnetic Waveguide Problems
Koshiba, M.; Hayata, K.; Suzuki, M.
IEEE Transactions on Microwave Theory and Techniques,
33(10): 900-905
1985-10

DOI

Doc URL

http://hdl.handle.net/2115/6037

Right

1985 IEEE. Personal use of this material is permitted.

However, permission to reprint/republish this material for
advertising or promotional purposes or for creating new
collective works for resale or redistribution to servers or lists,
or to reuse any copyrighted component of this work in other
works must be obtained from the IEEE.
IEEE, IEEE Transactions on Microwave Theory and
Techniques, 33(10), 1985, p900-905

Type

article

Additional
Information
File
Information

ITMTT33_10.pdf

Instructions for use

Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP

900

IEEE TRANSACTIONS ON MICROWAV E THEORY AND TECHNIQUES, VOL. MTI-3 3, NO. la, OCT08ER

Finite-Element Formulation in Terms of the

Electric-Field Vector for Electromagnetic
Waveguide Problems
MASANORI KOSHIBA,

SENIOR t-tEMBER, IEEE,

AND MICHIO SUZUKI, SEN IOR

Ab.ftruct - A veclo r fin ite-element method for the analysis of anisotropic

\\'1lveguides wilh off-diagonal elements in Ihe permeability tensor is formulated in te rm s of all three COmponents of Ihe electric field. In this
approach, spurious, nonphyslcal solutions do rIOl appear anywhere above
the "air-line." The appli cation of this finite-ele ment method to waveguides
with an abrupt discontinuity In the permittivjty is discussed. In particular,
we discuss bow to use the boundary conditions of the electric field al the
[nt.,.rface betw~n two medi a with different pennlttivilies. To show the
validity and usefulness of this formulation, examples are computed for
die lectric-loaded waveguides Ilnd fernie-loaded waveguides.

I.

INTRODUCTION

H E VECTOR finite-element method is widely used

either in an axial-component (E, - H z) formulation
[1)- (4) or in a three-component (either the electric field E
or the magnetic field H ) formulation (5), [6), which enables
one to compute accurately the mode spectrum of an electromagnetic waveguide with arbitrary cross section: The
most serious difficuhy in using the vector finite-element
analysis is the appearance of spurious, nonphysical solutions [1)-[6]. Hano [7) has presented a three-component
finite-element formul ation with rectangular elements. In
his formulation , spurious solutions, except for zero eigenvalues, do not appear, but a diagonal permittivity tensor
and a diagonal permeability tensor are assumed. Recently,
an improved finite-element method with triangular ~le
ments has been formulated for the analysis of anisotropic
dielectric waveguides in terms of all three components of H
[8) - [l1J. In dielectric waveguides, the permeability is always assumed to be that of free space. Therefore, each
component of II is continuous in the whole region and it is
more advantageous to solve for H than for E [12]. In this
improved H-field formulation, no spurious solutions appear anywhere above the "air-li ne" corresponding to /3/ ko
= 1 in a f3/ k o versus k o diagram [11), where k o is the
wavenumber of free space and /3 is the phase constant in
the z-direction. The appearance of spurious solutions is
limited to the region /3/ ko <1 and these solutions are
equivalent to the TE modes of "hollow" waveguides [11).
The H-field formulation is valid for general anisotropic
waveguides with a nondiagonal permittivi ty tensor. How-

Manuscript received February 1: 1985; revised May 22, 1985.

The au th ors a re with the Department of Electronic Engineering. Hok.
kaido Universi ty , Sapporo, 060, Japan

KAZUYA HA YATA,

MEMBER ,

IEEE

ever, It IS difficu lt to apply this H-field formulation tQ:

waveguides containing anisotropic media such as ferri ~
because the tensor permeability may vary from material to:;
material. In such cases, it is advantageous to solve for f j
rather than for H .
. ~
In this paper, an improved finite-element method with._
triangular elements is formulated for the analysis of anis~c
tropic waveguides with a nondiagonal permeability tensor~
using all three components of E. II) ferrite-loaded wave-_
guides, the permittivity is assumed to be constant in each
material, but may vary from material to material. At an~.
abrupt discontinuity in the permittivity, there is an abrupt
change in E . In this work, the application of the E-field
formulation to waveguides with abrupt discontinuities in
the perrrtittivity is discussed in detaiL In particular, we
discuss how to use the boundary conditions of E at the
interface between two media with different permittivities.
In this improved E-rield formulation, no spurious solutionf'_
appear anywhere above the "air-line." The appearance of
spurious solutions is limited to the region /3/ ko < 1 and
these solutions are equivalent to th~ TM modes of " hollow"
waveguides. To show the validity and usefulness of this"
formulation, ex.amples are computed for dielectric-loaqed""
waveguides and ferrite-loaded waveguides.
II.

FUNCTIONAL FORMULATION

We consider an anisotropic waveguide with a tensa

permeability and a scalar permittivity. With a time dependence of the form exp(jwt) being implied, Maxwell'
equations are
"V

x E -- jw., [.,] H

V x H - jW(ofrE

(I)
(2)

where w is the angular frequency, 11 0 and (0 are th

permeabili ty and permittivity of free space. respectively
(I1 rJ is the relative permeability tensor, (. J denotes a matrix;
and ( r i:s the relative permittivity which is assumed to
constant in each material.
From (1) into (2), the following wave equation is derived :

0018-9480/85/1000-0900101.00 C1985 IEEE

(3)

901

" " " " el Ill. : ELECTROMAGNETIC WAVEGUIDE PROBLEM S

k~ -

(4)

", 2( 0 11 0 "

functional [12], (13) for (3) is known to be

-k ~ JL( ,E.'Edn (5)

Fig. l . Interface with an abrupt discontinuity in the permittivity.

where n represents the cross section of the waveguide and obey the following equations:
asterisk denotes complex conjugation. In the finite-ele(12a)
,E - v1/;
analysis using (5), spurious solutions appear scattered
(12b)
the propagation diagram (51-112], [141. 115]. These
('\7 2 +
-= 0 in region n
SP''';'). s solutions belong to two distinct categories [11),
'" _ 0 on perfect electric conductor (12c)
The first one (SI) can be characterized as follows:
af / an -= 0 on perfect magnetic conductor
V x E =O '\7'( , + 0
fork ~- O .
(6)

k5)'"

(12d)

,"1,, second group (S2) can be characterized as follows:

vxE*'O

'C' ( rE-.l-O

for

k5 > O.

(7)

In order to eliminate the spurious solutions SI and S2 ' we

propose the following functional according to the H-field
formulation (8]-(11):

where", is the scalar field. The electric field E of (12)

satisfies the stationary requirement 8F - 0, but the divergence o f , E is not zero. Therefore, in the finite-element
analysis using (8), spurious solutions S3' which are no t
included in (5), do appear. The solutions S3 are equivalent
to the TM modes of " hollow" waveguides (replace'" in
(12b)- (12d) with Ez ) and the appearance is limited to the
region f3 / ko < 1. They do not appear anywhere above the
"air-line."
III.

(8)
For the functional (8), the first variation aft is given by

~t -

ffo ~E'

. [" X

(["X '" X E ) -

"(,, . ,E) - k5< ,E] dO

- Ir~E'[n x ([" , r '" XE) - n(,,.,E)]dr (9)

where f represents the contour of the region 0 , n is the
outward unit normal vector to f , and the term n X
([p,, 1- 1v X E) corresponds to the tangential components
of the magnetic fi eld H on f . The stationary requirement
8ft = 0 yields

FINITE-ELEMENT DISCRETIZATION AN D
BOUNDARY CONDITIONS

Dividing the cross section n of the waveguide into a

number of second-order triangular elements as shown in
Fig. 1 and using the finite-elemenLmethod on (8), we can
write the functional fo r the whole region 0 in the form

t - Lt.

(Il)

t. -

(14)

(E );[AJ.(E ).

[AJ .- [SJ. +[UJ .- k5[TJ .

(15)

where { E } ~ is the electric-field vector corresponding to the

nodal points within each element, thy matrices ISI ~, [ TI ~,
and [UJ ~ for each element are related to the fi rst, second,
and third terms on the right-hand side o f (8), respectively,
" x ([",J- '" XE) - ,,(,,.,E) - k;. ,E - O (lOa) T , {.} , and { . f denote a transpose, a column vector, and
a row vector, respectively, and the summation L ~ extends
as the Euler equation and
over all different elements. Variation of (13) with respect to
the nodal variables leads to the eigenvalue problem.
n (v . ,E) "" 0
on perfect electric conductor
In (14), the nodal electric-field . vector { E} ~ should be
(lOb)
forced to satisfy the boundary conditions at the interface
on perfect magnetic conductor between two media with different permittivities. We con(10e) sider the interface f ' with an abrupt discontinuity in the
permittivity as shown in Fig. 1, where ( \ and ( 2 are
as natural boundary conditions, since 8E'" in (9) is arbi- permittivities of the regions 1 and 2, respectively, the unit
trary. The spurious solutions SI and S2 are not includ in vector n normal to r ' makes an angle 8 with respect to the
(8), but (8) may have other solutions than (3). This new x-axis in the xy-plane, and the elements related to f ' are
group (S), characterized by
grouped into two classes: the elements (e 1) in region 1 and
(11) the elements (e 2 ) in region 2.
for k5 > 0

902

IEEE TRANSACTIONS ON MTCROW.WE THEORY AND TECHNIQUES, VOL, MIT-33, NO. 10, OCTOBER

If the functional for

is used in its origirial form (14),

Using (17), (16) can be transformed as follows:

we should modify the functiona l for e 2 in order to satisfy

the boundary conditions of the electric field Eon f', For

F,~ (E)i!XI,(E),

e 2 the functional (14) can be rewritten as

(E,\,

F,~(E)ilAI,(E),

(16a)

(E, ),

(E, ),

(E), ~

(E,) ,
(EL ~

(E, )2
(E,,),

(19b)

(E" ),

(E,),

(16b)

(E,,),

(E,,),

(E" ),
(E,,),
[ Axx] ~

[AI, ~

[A.i:zL

lA,,)'

lA,,),
IA,,],

[A.xll

lA,,],

[A,,] ,

[A u' ],
IA,,),
[A u' )'

lA",],
lA",),
lA,,],

[A",],

[A x '>;' ] 2

[Ax'xli

IA",L
[Az'xL

lA,, ],

lA",], lA",, ],
[A" ,), ' [A ",, ),

[AX .J2

lA,,],

[Au]2

IA,,],

IA,,],

IA,,],

IA",),
[A " J,

[A x:' L

lA" ,],
[A" ,),
[A",, ),

(16c)

IA" ,],
[A" ,, ),

IA,,], [A", ),
IAp), IA",], IA" ,),
[Au'], IA",], [A",]'
[AN]' IA,,,,], [AN]'
lA",,), IA""I, IA" " I,
[A i',' ], IA",,), [A "~' ]2
[Ad],

lA,,), [A,,),
lA,,), [A",],
IA",I, IA" ,I,
lA",), [A",],

[A,J,
[A ,,'],

[ Xh~

lA,,),
IA", ],
IA",),
IA" , ],
lA",,],
lA",, ),

(19c)

where
where

the components of the {E

vector are the values of

the eiectric field Ej(l = x, y, z) at the nodal points within
the element e z except P, the components of the {Ei'h
vector are the values of E; at the nOdal points on f '
iricluded in the element e z, and the [A xxh. [Axyh,"',
and [Az-z,h are the submatric~ of the matrix (15) for ei.
The tangential components of E and the normalromponent of f.rE should be continuous at the interface r', These
boundary conditions can be written as
j }

'

(E" h ~ q,,( E,,), + q,,( E,,},

(E,,), -' q.,( E,,), + q,,( E,,},
(E,,),~

(E,,),

(17a)
(17b)

+q';y[Ayy,12 (20a)
[ A X'Y'] 2 = qxxqxy (A x'x']2 + qxxqyy [ A X'y']2

+ q';y [Ay'x' 12 + qXyqyy [A yy12 (20b)

[A Y 'x,j2 = q;rxqxy[Ax'x, h+ qxxqyA A Y'x 12
+ q';y [ Ax'y' 12+ qXyqyy[ A Y'Y' ]2 (20c)

IA""I, ~ q;,[A N ), + q"q,,(1 A",,], + IA",,] ,)

+q;y [A Y 'y' ]2 (20d)

(17c)

where the components of the {Ei'} 1 vector are the values

of E; at the nodal points on f' included in the element e 1,
and qxx' qxy' and qyy are given by
qxx ~ sin28 +(fl/~2)COS28
(18a)

qxy== [(f1/f 2)-1]sin8cos8

[Ax'x' b = q';x[A X'X']2+ qxAxy([ A x,y12+ [A y;x' ]2)

(18b)

IAy ], ~ q,,1 Av ],+ q,,1 AI"]',

[ Ay I, ~ q" IAi"]' + q" lAy]"
IAd, ~ q,,1 Ad ,+ q,,1 A"/]"
[A"/]' ~ q,,1 A"/]' + q,, 1Ad"

}=x,y,Z,t

(200)

'
, (20r)
}=x,y,z,z

}=x,y,z,z (20g)
'
,
}=x,y,z,z. (20h)

""

~ KOS HIBA el al.: ELECTROMAGNETIC W AVEGUIDE PROBLEMS

903

b ,.!;t"QI2

( ". 1.5

"

,.

,,,
,,

"
Fig. 3.

,,

"

100

_,0 15-0

p-depcndence for the spurious solutions~.

becomes F. For (21), (12) is reduced to

(p2V' 1+ k~)1f _ 0
Fig. 2.

Dispersion characteristics of a half-fi!1ed dielectric waveguide.

By using the original functional (14) for e 1 and 'the mod-

ified functional (19) for e 2 the boundary conditions of the

electric field E at the interface with an abrupt discontinuity in the permittivity' are satisfied.
IV.

NUMERICAL RESULTS

A. Dielectric- Loaded Waveguides

First, let us consider a rectangular waveguide half-fil led

wfth a dielectric of permittivity (\ (relative permittivity
f ,\ -

(1/( 0 )'

We subdivide one half of the cross section into secondorder triangular elements as shown in the insert in Fig. 2,
where ( ,1 - 1.5, the plane of symmetry is assumed to b~ a
perfect magnetic conductor, 36 elements (N) are used,
and the number of the nodal points (Np ) is 91. Computed
results (solid lines) for the LSM mn and LSEm~ modes agree
well with the exact results [16J. Spurious solutions S, and
S2 ' which are included in (5), do not appear. Spurious
solutions S3 (dashed lines) corresponding to the solutions
of (12) appear only in the region Il/ko < 1. The solutions
S3 with cutoff frequenci es koQ =,ff 'IT and IS'IT are equivalent to the TMII and TM I2 modes of a "hollow" waveguide of square cross section, respectively.
One can control the solutions S3 by changing the functional (8) as follows [10], [15]:

Where p is a positive number. If p is set equal to 1, Fp

If -

in region 0

(22a)

on perfect electric conductor

(22b)
on perfect magnetic conductor.
(22c)

The appearance of the solutions of (22) is limited to the

region Pl ko <l i p and the cutoff frequen cies of these
solutions vary in proportion to the value of p.
Fig. 3 shows the p-dependence for the solutions S3 in
the same waveguide as shown in Fig. 2. Solid and dashed
lines in Fig. 3 correspond to the TMIl and TM I2 modes in
Fig. 2, respectively. When p = 2, the solutions S3 appear in
the region /3l k(j< 0.5 and the cutoff frequencies of the
solutions corresponding to the TMA) and TM Il modes in
Fig. 2 become koQ - 2fi'IT and 2v'5 'IT, respectively. When
p = 0.5, the solutions S3 appear in the region Plk o < 2 and
the cutoff freq uencies of the solutions corresponding to the
TM ll and TM 12 modes in Fig. 2 become k oQ" 0.5fi'IT
and 0.51S'IT, respectively. The p-dependence is very small
for the physical solutions. For larger values of p, however,
the degree of accuracy for the physical solutions becomes
poorer. For smaller values of p. on the other hand, more
spurious solutions appear, because the cutoff frequencies
of these spurious solutions become lower. Hereafter, we use
p = 1, namely the functional (8).
Fig. 4 shows the dispersion characteristics for the fundamenta! mode of half-filled dielectric waveguides, where the
plane of symmetry is assumed to be a perfect magnetic
conductor. For both ( ,I = ].5 and 10.0, ou r results agree
well with the results of the H-field fini te-element fo rmulation [11].
In Figs. 2 and 4, the normal direction of the interface
with an abrupt change in the permittivity coincides with
the direction of a coordinate axis.
Next, let us consider a rectangular waveguide with a
diamond-shaped dielectric insert (17), as shown in Fig. 5.
In this waveguide, there are abrupt changes in the permit
tivity at the interface whose normal direction does not
coincide wi th the direction of a coordinate axis. Fig. 5
shows the dispersion characteristics for the fundamental

IEEE TRANSACTIONS ON MICIlOWA VE THEOIlY AND TECHNIQUES, VOL. MIT-B , NO. 10, <X:TOBEIl 1985

1=0,--,

"r

,
,

H_h.ld 101 ..... 1&1'011

",.36. 1'1 91

"

IIctt ~,1

1'1 . . 36. 1'1 . . 91

F ig. 6. Ferri te-loaded waveguide.

TABLE I
DI SPERSION C HARACl"ERl STICS OF FERIUTE- LoADED WAvEo u m ES

".10. 0

"

Fig. 4.

'ob

Dispersion characteristics for th e fundamen tal mode of half-filled

dielectric waveguides.

."1

..

C-.SOI ... lH

H_Ii.la lo,,,,""" ion

20-

Np50. N 121

E_ II.la l.... mulation

,
I
10~

"
.,
0
,

0,_

al"";4

Finite
element
calcul at ion

h.et
calel/I.tlon

Finitee l!'llltnt
"leI/lo t ion

Euct
calculat iOf'

0.18877

0.78816

1.45379

1. 45364

0.66538

0.665 37

1.16095

1.1 6081

0.18871

0.18876

1.39979

1 .39966

B. Ferrile -Loaded Waveguides

..

"'.50.1'1 12 1
1'1. . 12&. 1'1 289

[,,, 100

E.,.IS

,
b . 1 Sa

Fig. S.

E_li.la 1"""'.... aI IOf'

1,

[ 0. ~o

D ispersion characteristics for the fundamental mode of rec tangular waveguides with a diamond -shaped dielectric insert.

mode, where two planes of symmetry are assumed to be

perfect magnetic conductors and one quarter of the cross
section is divided into second-order triangular elements. In
Fig. 5, the results of the H-fie ld formulation with N= 50
and N" - 121 and the results of the modal approximation
techniques (1 7} are also presented. For (,I - 1.5, the results
of the E-field formulation with N - 50 and N" = 121
agree well with those of the H -field formulation. For a
larger value of relative permittivity, td - 10.0, the results of
the E-field formulation with Ne - 50 and N" = 121 deviate
from those of the H-field formulation at higher frequencies. However, the E-field finite-element solutions can be
improved by increasing the number of the elements. The
results of the E-field form ulation with Ne - 128 and N" =
289 are closer to those of the H-field formulation.
The computed results in Figs. 2, 4, and 5 prove the
validity o f (19) and (20).

We consider a ferrite-loaded waveguide as shown in Fig.

6. The ferrite material is characterized by the relative
permeability tensor
0
[ 30
(23)
le,1~ 0
1.0
- JO.8
0
3.0
and a relative permittivity of 2.0 (5}. Here, [l1r } is independent of frequency, although this assumption is not valid for
ferrites in genera1 [5}. Table I shows the dispersion characteristics for the fundamental mode, where a :'" 2b, N e "
64, and Np - 153. For both 0 1 - 0 (completely fiUed) and
0 1 - 0 / 4 , agreement between our results and the exact
results {5], (1 8) is good. In the case of 0 1 '" 0 / 4 , the value
of k ob for fib = - 1 is differen t from that for fib = 1. This
fact implies that when ko is given, the modes propagati ng
in this structure (partially filled) in the opposite directions
have different phase constants, namely these modes are
nonreciprocal (18}. The modes propagating in the completely filled waveguide in the opposite directions, on the
o ther hand, have the same phase constants, and therefore,
these modes are reciprocal (18).

J~8 l

V.

CONCLUSION

The finite-element method was formulated for the analysis of anisotropic waveguides with a nondiagona1 permeability tensor in terms of all three components of the electric
field E . In this approach, spurious solutions do not appear
anywhere above the "air-line." The application of this
E-fi eld formulation to waveguides with an abrupt discontinuity in the permittivity was discussed in detail."
REFERENC ES

[I [ P. Daly. " Hybrid-mode analysis o f microstrip by finile-element

me thods," IEEE Trans. MiuowutJe Throry T~h., vol. MTT1 9, pp.
19-25, Jan. 1971 .

t'r

[6]

[7]
[8]

[9]

[IO[
[ll]

[12J

[13J
[14]

[15]

[16J
,
[17]

[18J

al,:

ELECTROMAGNETIC WAVEGUIDE PROIJLEMS

M. Ikeuchi, H. Sawami, and H. Nih "Analysis of open-type

dielectric waveguides by the finiteelement iterative method," IEEE
Trans, Microwave Til/wry Tech" vol. MTI-29, pp. 234- 239, Mar.
Inl.
N . Mabaya, P. E. Lagasse, and P. Vandenbukke, "Finite element
analysis of optical waveguides," IEEE Trans. Microwave Theory
Tech., vol. MIT-29, pp. 600- 605, June 1981.
K. Oyamada and T. Okoshi, "Two-dimensional finite-element
calculation of propagation characteristics of axially nonsymmetrical
optical fibers," Radio Sci., vol. 17, pp. 109- 116, Jan.-Feb. 1982.
A. Konrad, "High-order triangular finite elements for electromagnetic waves in anisotropic media," IEEE Trans. Miuowaue Theory
Tech., vol. MIT-25, pp. 353- 360. May 1977.
B. M. A. Rahman and J. B. Davies, "Finite-element analysis of
optical and microwave waveguide problems," IEEE Tram. Microwave Theory Tech., vol. MIT-n, pp. 20- 28, Jan. 1984.
M. Hano, "Finite-element analysis of dielectric-loaded waveguides,"
IEEE Trans. Microwave Theory Tech .. vol. MIT-32, pp. 1275-1279,
. Oct. 1984.
M. Kosh iba, K. Hayata, and M. Suzuki, " Vectorial finite-element
formulation without spurious modes for dieleclric waveguides,"
Trans. lnsl. Eleclron. Commull . Eng. Japall, vol. E67, pp.I91 - 196,
Apr. 1984.
M. Koshiba, K. Hayata, and M. Suwki, "Vectorial finite-element
method without spurious solutions for dielectric waveguide problems," Elec/ron. Lell., vol. 20, pp. 409- 410, May 1984.
B. M. A. Rahman and J. B. Davies, "Penalty function improvement
of waveguide solution by finite elements," IEEE Truns. Mi{'rQwuve
TheaI)' Te,h., voL MIT-n, pp. 922- 928, J\ug. 1984,
M. Koshiba, K. Hayata, and M. Suzuki, "Improved finite-clement
formulation in terms of the magnetic-field vector for dielectric
waveguides," IEEE Tram. Microwave Theory Tech. , vol. MTT-33,
pp. 227- 233, Mar. 1985.
A. Konrad, " Vector variational formulation of electromagnetic
fields in anisotropic media," IEEE Tral/-$. Microwave Theory Tech.,
vol. MIT;24, pp. 553-559, Sept. 1976.
A. D. Berk, "Variational principle for electromagnetic resonators
and waveguides," IRE Tran$. Amellnas Propaga/., vol. AP-4, pp.
104-111, Apr. 1956.
J. B. Davies, F. A. Fernandez, and G. Y. Philippou, "Finile element
analysis of aU modes in cavities with circular symmetry," IEEE
Trall$. Microwave Theory Tech .. voL MTT-30, pp. 1975- 1980, Nov.
1982.
M. Hara. T. Wada, T. Fukasawa, and F. Kikuchi, "A three dimensional analysis of RF electromagnetic fields by the finite element
method," IEEE Trall$. MagI!., vol. MAG19, pp. 2417-2420, Nov.
1983.
N. Marcuvitz, Wuveguide Halldbook. New York: McGraw-Hill,
1951.
Z. J. Cscndcs and P. Silvester, "Numerical solution of dielectric
loaded waveguides: II - Modal approximation technique," IEEE
Trans. Microwave Theory Tech., vol. MTT-19, pp. 504- 509, June
1971.
B. Lax and K. J. Button, "Theory of new ferrite modes in rectangular wave guide," J. Appl. Phys., vol. 26, pp. 1184- 1185, Sept. 1955.

905
Masanori Koshiba (S M'84) was born in Sapporo,
Japan, on November 23,. 1948. He received the
B.S., M.S., and Ph.D. degrees in electronic engineering from Hokkaido University, Sapporo,
Japan, in 1971, 1973, and 1976, respectively.
In 1976, he joined the Department of Electronic Engineering, Kitami Institute of Technology, Kitami. Japan. Since 1979, he has been an
Assistant Professor of Electronic Engineering at
Hokkaido University. He has been engaged in
research on surface acoustic waves, dielectric
optical waveguides, and applications of finite-element and boundary-element methods to field problem's.
Dr. Koshiba is a member of the Institute of Electronics and Communication Engineers of Japan, the Institute of Television Engineers of Japan,
the Institute of Electrical Engineers of Japan, the Japan Society for
Simulation Technology, and Japan Society for Computational Methods in
Engineering.

KazllJ"a Hayata was born in Kushiro, Japan, on

December 1,1959. He received the B.S. and M,S.
degrees in electronic engineering from Hokkaido
University, Sapporo, Japan, in 1982 and 1984,
respectively.
Since 1984, he has been a Research Assistant
of Electronic Engineering at Hokkaido University. He has been engaged in research on dieleclric optical waveguides and surface acoustic
waves.
Mr. Hayata is a member of the Institute of
Electronics and Communication Engin~rs of Japan.

Miehio Suzuki (SM'57) was born in Sapporo,

Japan, on November 14, 1923. He received the
B.S. and Ph.D. degrees in electrical engin~ring
from Hokkaido University, Sapporo, Japan, in
1946 and 1960, respectively.
From 1948 to 1962, he was an Assistant Pro. fessor of Electrical Engineering at Hokkaido
University. Since 1962, he has ~en a Professor
of Electronic Engineering at Hokkaido University. From 1956 to 1957, he was a Research
Associate at the Microwave Research Institute of
the Polytechnic Institute of Brooklyn, Brooklyn, NY.
Dr. Suzuki is a mem~r of the Institute of Electronics and Communication Engineers of Japan, the Institute of Electrical Engineers of Japan,
th e Inst itute 'of Television Engineers of Japan, the Japan Society of
Information and Communication Research, and the Japan Society for
Simulation Technology.