Anda di halaman 1dari 7

376 IEEE TRANSACTIONS ON -WAS AND PROPAGATION, VOL. AP-23, NO.

3, MAY 1975

[5] E. N. Fox, “ n e diffraction of sound pulses by an iniinitely long plane,” Quart. Appl. Math., vol. 25, pp. 193-204, 1967.
strip,” Philosophical Transactions of the Royal Society of London, [13] N. Marcuvitz and J. Schwinger, “On the representation of the
VOI. A241, pp. 71-103, 1948. electric and magnetic fields produced by currents and discon-
[6] -, “The diffraction of two-dimensional sound pulses incident tinuities in wave guides, I,” J . Appl. Phys., vol. 22, pp. 806-819,
on an infinite uniform slit in a perfectly reflecting screen,” Phil. 1951.
Trans. Roy. SOC.London, vol. A242, pp. 1-32, 1949. [14]P. E. Mayes, “The equivalence of electric and magnetic sources,” ,
[7] W.E. Ackerknecht, 111, “The scattering of pulses by screens,” IRE Trans. Antennas Propagat. (Commun.), vol. AP-6, pp.
Ph.D. dissertation, Purdue Univ., W. Lafayette, Ind., 1972. 295-296, July 1958.
[8] J. B. Keller and A. Blank, “Dlffractlon and reflection of pulses by [15] S . N. Karp and J. B. Keller, “Multiple diffraction by an aperture
wedges and corners,” Commun. Pure Appl. Maflz., vol. 4, pp. in a hard screen,” Opt. Acta, vol. 8, pp. 61-72.
75-94~ 1971.
. - - _, 1161 J. B. Keller and E. B. Hansen. “Survev of the theorv of diffraction
J. W. Miles, “On the diffraction of an acoustic pulse by a wedge,” of short waves by edges,“ A&a Phys: Pol., vol. 27,pp. 217-234,
Proc. Roy. Soc. London, vol. A212, pp. 543-547, 1952. 1965.
-,“On the diffraction of an electromagnetic pulse by a wedge,” P. J. Davis and P. Rabinowitz, h’lrmerical Integration. Waltham,
Proc. Roy. Soc. London, vol. A212, pp. 547-551, 1952. Mass. : Blaisdell, 1967.
W. L. Weeks, ElecfromagneficTheory for Engineering Applica- V. I. Krylov e f al., Tables for the h‘lrmerical Integration of Func-
Zions. New York: Wiley, 1964, pp. 345-350. tions with Power Singularifies. Minsk, USSR: Izd-vo Akademii
G. E. Barr, “On the diffraction of a cylindrical pulse by a half- Nauk, Belorusskoi, 1963.

Sinusoidal Reaction Formulation for Radiation


and Scattering from Conducting Surfaces
NAN N. WANG, MEMBER, IEEE, JACK H. RICHMOND, IEEE,
SENIOR ~ M B E R , AND MELVIN C. GILREATH

Absrruct-A piecewise-sinusoidal reaction technique is developed for overlapping sinusoidal bases inthe direction of current
Scattering and radiation from perfectly conducting bodies of arbitrary flow, and rectangular-pulse bases in the transverse direction
shape. This paper presentsthe theory and numerical results for scattering
patterns of rectangular plates and radiation patterns of corner-reflector
as in Fig. 3. This arrangement satisfies Kirchhoff’s current
antennas. In all cases,experimentalmeasurements are included for law withoutintroducing line charges. Via an application
comparison with the calculated data. of Galerkin’s method,theintegralequationformulated
with the zero-reaction concept [6] is reduced to a matrix
equation. From a physical viewpoint, thisvariational
I. INTRODUCTION solution follows from enforcement of reaction tests with an
array of sinusoidal electric test sources. The current dis-

T WO METHODSare available for electromagnetic


modeling of continuous conducting surfaces with
arbitrary shape: the wire-grid model [l] and the surface-
tribution over the conducting surface is determined via
matrix inversion, and finally the scattered field is obtained
by integrating the surface currents. The “reactiontechnique”
current model [2], [3] using rectangular-pulse bases. Both presented here is a moment method based on the electric-
methods have similar limitations with the maximum cell field integral equation.
width restricted to approximately A/lO. Unless the conduct- The remaining text presents the theory of the sinusoidal
ing body is symmetric or is a figure of revolution, computer reaction formulation for radiationand scattering problems.
storage requirements have limited the moment-method The time dependence e’”‘ is understood and suppressed.
application to bodies with surface area not exceeding one Numerical results are presented for the scattering patterns
or two square wavelengths. of rectangular plates and corner reflectors and theradiation
In this paper, the piecewise-sinusoidal reaction technique patterns of corner-reflector antennas.
applied by Richmond [4], [SI to thin-wire antennas is
extended to scattering by conducting bodies of arbitrary 11. THE REACTION TECHNIQUE
shape. The continuous conducting surface is divided into Thereaction concept and itsapplicationhave been
cells andthe surface-current distribution is expanded in discussed by Rumsey [6], Cohen [7], Harrington [SI, and
Richmond [SI.
Consider the exterior scattering problem illustrated in
Manuscript received May 1, 1974; revised November 2, 1974. This Fig. l(a). In the presence of a dielectric or conducting body,
work was supported in part by the NASA Langley Research Center
under Grant NGL 36-008-138 to the Ohio State University Research the impressed electric and magnetic currents (Ji,Mi)
Foundation. generate the electric and magnetic field intensities (E,H).
N. N. Wang and J. H. Richmond are with the ElectroScience
Laboratory, Department of Electrical Engineering, Ohio State For simplicity, let the exterior medium be free space.
University, Columbus, Ohio 43212. From the surface-equivalence theorem of Schelkunoff
M. C. Gilreath is with the NASA Langley Research Center, Hampton,
Va. 23665. [lo], the interior field will vanish (without disturbing the
WANG et a[.: SINUSOIDAL REACTION FORMGLATION 377

0 JiM i
SOURCE

FREE SPACE
0 Ji,Mi
SOURCE

Fig. 2. Electric test source J, is positioned in interior of scattering


regon.

where (Et,Hr)is the free-space field of the test source. In

0 J ; *Mi
SOURCE
words, (5) states thatthe interior test source has zero
reaction with the other sources. This "zero-reaction
theorem'' was developed by Rumsey [ 6 ] .
Equation (5) is the integral equation for the scattering
problem, andour objective is to use thisequation to
,/-
(b)
\
determine the surface-current distributions J, and Ms. To
\
I \ (E..Hs) accomplish this, we expand these functions in finite series
' \
\(-&kJ') \ so there will be a finite number N of unknown expansion
\ \
\ \
constants. Next we obtain N simultaneous linear equations
\ \ to permit a solution for these constants. One such equation
( J,,M,), I
is obtained from (5) each time we set up a new test source.
I\
FREE SPACE,'
\
'.-/ 0
The magnetic current M, vanishes if the scatterer is a
perfect conductor. We assume a finite conductivity and
use the impedance boundary condition
(4
Fig. 1. (a) Source (&,Mi) generates field (E&) with scatterer. (b) M, = Z,J, X A (6)
Interior field vanishes when currents (J,,M,) are introduced on
surface of scatterer. (c) Exterior scattered field may be generated by
(J,,M,) in free space. where 2, denotes the surface impedance. If M i vanishes,
(5) and (6) yield

exterior field) if we introduce the following surface-current


densities
J,=A X H (1)
-
S
J, . [E, - ( A x H,)Z,] ds =
IS J i . E, ds (7)

where (E,,,,H,) denotes the free-space field of test-source in.


M,=ExA (2) We represent the electric current distribution as follows:
N
on the closed surface S of the scatterer. (The unit vector
ii is directed outward on S.) In thissituation, illustrated
in Fig. I(b), we may replace the scatterer with free space where the complex constants I, are samples of the function
without disturbing the field anywhere. J,. The vector functions 4, are known as basis functions,
By definition, the incident field (Ei,Hi)-is generated by subsectional bases, expansion functions or dipole modes.
(Ji,Mi)in free space, and the scattered field is We employ expansion functions J, and test sources J,
E, = E - Ei (3) with unit current density at the terminals.
From (7) and (8) we obtainthe simultaneous linear
H, = H - Hi. (4) equations
When the surface current (J,,MA radiates in free space, it N
generates the field (E,,H,) in the exterior and (- Ei,-Hi) Z
C
,, = A,, with m = 1,2.3; . .,AT (9)
n=l
in the interior region. This result, illustrated in Fig. l(c),
is deduced from Fig. l(b) and the superposition theorem. where
With the scatterer replaced by free space, we have noted C,, = - f/
in Fig. I(b) thatthe interior region hasa null field. As
shown in Fig. 2, we place an electric test source J, in this n
J, . [E, - ( A x H,)Z,] ds = -
m
J,, * E,, ds

(10)
J-1
region and h d from the reciprocity theorem that

fiS (Js . E* - Ms 4 ) ds
A, = 11
i
Ji E, ds =
ISm
J, . Ei ds. (11)

In (10) and (11) theintegrations extend over the region


+ III(Ji * Et - Mi Ht) du = 0 ( 5 )
where the integrand is nonzero. For example, region ?I is
318 EEE TRANSACTIONS ON A h T N N A S AND PROPAGATION, MAY 1975

Fig. 4. Nonplanar surface dipole with edges at s1 and t l and terminal


Fig. 3. Electric surface dipole and its current-density distribution. at 0.

density and its slope arecontinuous across the terminal


thatportion of the surface S covered by the expansion at z = z 2 .
function J,,. Region m covers the interior test source J,. Fig. 4 illustrates a surface V-dipole. Distance along the
The reciprocity theorem relates the first and second in- dipole arms is measured by the coordinates s and t with
tegrals in (10). In the second integral, E,, is the free-space origin at the terminal 0. The surface-current density is
field generated by J, and theassociated magnetic current M,,.
Forcomputational speed and storage, it will bead-
vantageous to have asymmetricimpedancematrix Cm.
J = -$cos
(3 -, onarms

Furthermore, the test sources should be selected to yield


a well-conditioned set ofsimultaneous linear equations.
For these reasons, we employ test sources J, of the same
size, shape, and functional form as the expansion functions Having defined the sinusoidal surface dipole, one is now
J,,. Finally we position theinterior test sourcesasmall in a position to explain its relevance. The dipole current
distance 6 from surface S and take the limiting form of the distribution ((14) and (15)) will be usedas thebasis functions
integrals as 6 tends to zero. (Numerical experiments showed (J, in (8)) for expanding the unknown current distribution
better convergence with Galerkin's method than with induced on aconducting surface. Furthermore, surface
point matching, and with test sources locatedon the surface dipoles will be employed as test sources with the reaction
than with interior test sources.) For perfectly conducting concept to solve the integral equation.
bodies, the presentformulation applies equally well to
open and closed surfaces. With an open surface such as a
IV. THEIMFEDANCE MATRIX
thin plate, J, denotes the sum of the currentdensities on the Fromthe viewpointof reaction, thecomplexnumber
front and back sides. C,, in (10) represents the reaction between two sources.
The next section discusses the electric surface dipoles Let (m,n) denote the reaction between sources rn and n ,
which are employed as test sources and expansion modes. then (10) can be written as

111. THESI~XJSOIDAL
SURFACE
DIPOLFS
A planar surface dipole located on the JJZ plane is illus-
C,, = -(m,n> = -
ss
rn
-
J,,, E, ds.

It has been pointed out [6] that the reaction between two
(16)

trated in Fig. 3(a). This source is an electric surface-current


density with height a and width b. The surface-current sources is related to the circuit parameters by
density is Cmn = Vdmm (17)

J = 2 cos ( -
- for z1 s where V,, is the open-circuit terminal voltage induced at m
z 5 z2 (12)
by source n, and Immis the terminal current of source m
when it transmits.
= cos (n(Z -J :; , for z2 Iz3 (13) Although the surface dipole described in Section I11 is a
2(z3 - hypothetical source, it is useful to define
self-impedance
its
with the induced emf formulation
As illustrated in Figs. 3(b) and 3(c), thecurrent density
vanishes at the edges z = z1 and z = z3, and is uniformly - Vmm - Cmm
Zmm 2' (1 8)
distributed
the in transverse direction. The surface-current I- 4lm
WANG et al.: SINUSOIDAL REACTION FOILMULATION 379

TABLE I TABLE II
SELFIMPEDAWCE
OF CENTER-FEDPLAWAR MUTUALIMPEDAWCE
OF CENTER-FED
COPLANAR
SHOWN IN FIG.5
SURFACE DIPOLE SURFACE
DIPOWSHOWN IN FIG.6

a/x. = 0.5 b/). = 0.25 I


0.75 1.076-j 7.913 -2.925-j 5.950 -5.968+j 1.761 0.8871+j 7.019
0.2 11.48 -69.76
0.3 23.98 -35.26 II 0.50 124.43 +j 5.997 110.04 - j 9.321 I -9.568-j 6.65 I
I
-7.194 j8.971
i II
0.4 38.80 -15.36 0.25 tj55.87 29.54 - j 9.07053.10
-9.468-j19.33 -15.74 + j 6.426
52.98 - 4.52
0.0 67.09 +j13.23 139.07 -j22.49
I
II -8.659-j26.75 1-19.32
I
+j 4.263

0.0 0.25 0.50 0.75

Pa-+
Fig. 5. Surface dipole with J = 2 cos (nz/a).

/ Y
Fig. 6. Coupled surface dipoles.

From (I@, (18) yields


/
X

Fig. 7. Plane wave (Ei,Hi)iiluminates electric surface dipole.

Themutualimpedance between twosurface dipoles is where (r’,Oi’,$i’) are the spherical coordinates of the source
given by and Eo is the incidentelectric field intensity at the coordinate
origin 0‘. Fig. 7 illustrates an incident plane wave illuminat-
Zm,= 5/ / J , . E, ds. (20) ing an electric surface dipole located on the y‘z‘ plane with
I m J m
m height a and width b. For the surface-current density J =
2‘ cos (nz’la);,(21) can readily be evaluated to yield
Fig. 6 illustrates a pairof center-fed coplanar surface
dipoles, andTable I1 fists themutualimpedance Z12. sin ( X i ) cos (Yi/2)
A, = (Eo * 2’)2nab Xi(Yi2 - n2) (23)
Here sz and sy specify the relative position of the dipoles.
The impedances in Table I and I1 were calculated with a where
combination of analytic and numerical integration
techniques.
Xi = 0.5kb sin Oil sin q5i‘

Yi= ka cos Bi’.


V. THEEXCITATIOX COLUMN
The complex quantities A , in (1 1) form the excitation VI. FAR-FJELD RADIATIONAND SCATTERING
column in the matrixequation C,,I, = A,. Physically, The field scattered by a perfectly conducting body may be
A , is the reaction between the impressed source and test generated by the electric surface-current density J, in free
dipole nz. From (]I), Am is given by space. To obtainthetotal field oneaddsthe free-space
field (Ei,Hi)of the electric source Ji.
A, = / / J , , , Ei ds. Consider an electric surface dipole with current density
m J, = 2‘ cos (nz’la) located on the y’z’ plane as shown in
380 IEEE TRANSACXIONS ON ANTEhWAS AND PROPAGATION, MAY 1975

100

a-? 1.0
a
0
t
0
W

L
THICKNESS 0.0317 x
0.0 I
0
1 I
0.2
I
0.4
I
0.6 0.8
I I
1.0
I
1.2 1.4
EDGE LENGTH /x'
8 IDEGREES)
Fig. 8. Backscatter cross section ofperfectly conducting square
plate for broadside aspect. Fig. 12. Normalized backscatter cross section in y z plane of rectan-
6) = 26.25 dB at (90",90").
g u l a r plate. oeeS(B,

Ill[
'6

II

Fig. 9. Electromagnetic modeling of plate.

Fig. 13. Normalized backscatter cross section in xy plane of rectan-


01 I I I gular plate. uss(8,d) = 26.25 dB at (90',90").
-0.5 -0.25 0 0.25 0.5

Fig. 10. Amplitude of surface-current density induced on perfectly-


Conducting rectangular plate for plane wave incident at broadside.

-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 I 0


e (DEGREES]
Fig. 14. Normalized backscatter cross section in yz plane of corner
reflector. ooo(B,d) = 10.62 dB at (90',90").
Fig. 11. Phase of surface-current density induced on rectangular plate
for plane wave incident at broadside.
WANG et al. : mwsomAL REACTION FOILMULATION 381

TABLE III
- REACTION
MEASUR
S=0.75X (1=12O0
=2.75X W = X OF MOD= USEDISCALCULATIONS
NUMBER

Number of Modes

10,ll 33
12 55
13 75
14 38
15.16 61

was found that good results can also be obtained. with a


much smaller number of modes.) For comparison, Fig. 8
also shows the experimental measurements of Kouyoumjian
[111.
The magnitude and phase of the current density induced
Fig. 15. Relative gain in E-plane of corner-reflector antenna. G(B,4) =
4.05 dB at (90”,90”). on a perfectly conducting rectangular plate are illustrated
in Figs. 10 and 11. This is the sum of the current densities
on the front and back surfaces of the plate. Figs.12 and
13 show the normalizedbackscatter cross section of a
rectangular plate. Fig. 14 shows the normalized backscatter
cross section of a corner reflector. The title of each figure
gives the echo areaatthe broadsideaspectinterms of
dB = 10 log (~/i.’).
Figs. 15 and 16 showthe gainof a comer-reflector
antennaintheE-planeandH-plane, respectively. For
comparison, these figures includeexperimentalmeasure-
mentsobtained at NASA LangleyResearchCenter. In
the measurements the receiving antenna was linearly
polarized in the theta direction. Similarly, the calculated
gain is based on E,. The dipole length is i./2 and the radius
is 0.005i..
Table I11 lists the number of modes used to obtain the
/ / I I \ \ results given in Figs. 10-16. In each case, the matrix size
Fig. 16. Relative gain ni H-plane of corner-reflectorantenna. G(B,+4)= is equal to the number ofmodes. In Fig. 14, the vertical
4.05 dB at (90”,90’). current J, was expanded in 18modes and the horizontal
current in 20 modes. For convenience, thehorizontal
currents were neglected in the other calculations.
Fig. 7. From reciprocity, the free-space electric field
generated by this source at a distant point ( ~ ’ , 6 ~ ’ , r $is~ ) VIII. CONCLUSIONS
o p ab sin ( X , ) cos ( K / 2 ) e - j k r ’ The reaction concept and Galerkin’s method are used to
E” = 8,’j- ’

sin 0,‘ (24) develop a new formulation for perfectly conducting anten-
X, ( K 2 - 2 ) r’
~

2
nas and scatterers. Numerical results are presented for
where scattering and radiation from rectangular plates and corner
X, = 0.5kb sin eS’ sin reflectors. The results show general
agreement
with
measurements.
Y, = ka cos 6;.
The techniques can be applied to surfaces with finite
The 8- and $-components of the scattered field with respect conductivity and arbitrary shape.For arbitrary polarization
tothe reference coordinate system 0 can be obtained and aspect, 12 modes per square wavelength are adequate.
easily via an appropriate coordinate transformation. If this performance is maintainedforcumedconducting
bodies, it willrepresent a significant improvementover
VII. NUMERICAL RESULTS previous techniques.
Fig. 8 presents the backscattering echo area of a square
plate with perfect conductivity forthebroadside aspect. REFERENCES
In the reaction calculation, the plate is divided into cells, [l] J. H. Richmond, “A wire-grid model for scattering by conducting
bodies,” IEEE Trans. Antennas Propagar., vol. AP-14, pp. 782-
and overlapping currentmodes were employedas illustrated 786, Nov. 1966.
in Fig. 9. In this case thetransversecurrent is neglected [2] F. K. Oshiro, “Source distribution technique for the solution of
general electromagnetic scattering problems,” in Proc. First
and 45 modes wereused for the currentdistribution. (It GISATSymp., vol. 1, part 1, Mitre Corporation, 1965.
382 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATTON,VOL. AP-23, XO. 3, MAY 1975

[3] D. L. Knepp, “Numerical analysis of electromagnetic radiation [8] R. F. Hamngton, Time-Harmonic Electromagnetic Fields.
propenies of smooth conducting bodies of arbitrary shape in New York: McGraw-Hill, 1961, pp. 340-345.
the presence of known external sou~ces,” Ph.D. dissertation, 191 J. H. Richmond, “A reaction theorem and its application to
University of Pennsylvania, Philadelphia, 1971. antenna impedance calculation,“ IRE Trans. Antennas Propagat.,
[4] J. H. Richmond, “Admittance matrix of coupled V antennas,” V O ~ .AP-9, pp. 515-520, NOV.1961.
ZEEE Trans. Antennas Propagat. (Comun.)vol. , AP-18, pp. [IO] S . A. Schelkun$T, “On diffraction and radiation of electro-
820-821, Nov. 1970. magnetlc waves, Phys. Rm., vol. 56, Aug. 15, 1939.
[5] P. K. Agrawal, G. A. Richards, G . A. Thiele, and J. H.Richmond, [l 13 R. G. Kouyoumjian, “The calculation of the echo area of perfectly
“Analysis and design of TEM-line antennas,” ZEEE Trans. conducting objects by the variational method,” Ph.D. dissertation,
Antennas Propagat., vol. AP-20, pp. 561-568, Sept. 1972. Ohio State University, Columbus, 1953.
[6] V. H. Rumsey, “Reaction concept in electromagnetic theory,’‘ [12] J. H. Richmond, “Computer analysis of three-dimensional wire
Phys. Rec., vol. 94, pp. 1483-1491, June 1954. antennas,” Ohio State University ElectroScience Laboratory,
[7] M. H. C o p n , “Application of the reaction concept to scattering Dep. Elec. Eng., Rep. 2708-4, Dec. 22, 1969; prepared under
problems, IRE Truns. Antennas Propagat., vol. “-3, pp. 193- Contract DAADO5-69-C-0031 for Aberdeen Proving Ground,
199, Oct. 1955. Maryland.

Some NewResults on Propagation of Electro-


magnetic Waves in Strongly Turbulent Media
RONALD L. F A N E , SENIOR MEMBER, IEEE

Abstract-New results for the covariance of the intensity fluctuations starting with the solution for infinitely strong turbulence.
of a plane wave propagatingin a strongly turbulent medium are presented, For o12 = 03,where o12= 1.23k,7/6C,,2z11/6,k,is the
and the consequences of these new results on aperture averaging, angleof
arrival fluctuations, and the temporal frequency spectrumof the intensity
signal wavenumber and Cn2 is the index-of-refraction
scintillations are evaluated. The differences
between the strong turbulence structure constant, it can be seen [ 8 ] that the statistics of
and weak turbulence cases aie generally quite si&wt. the field are Gaussian, since the fi$d at (z,O) consists of the
independent contributions from very-manyoff axis eddies.
I. INTRODUCTION We emphasize
. . that the statistics are exactly Gaussian only
for o12 = co. For finite g I 2 we have shown in Appendix 1

C ONSIDERABLE EFFORT hasbeen devoted recently


to the study of thevarianceandcovariance
intensity fluctuations of a wave propagating in a random
of the
that the statistics of the field cannot be Gaussian. In fact,
for 0 Io12I 100 it appears that the field statistics are
very nearly log-normal [ 9 ] . Consequently for o12= co we
medium. The impetusfor these studies was provided by the have (see Appendix 2) from ( 1 )
experimentalmeasurement of Gracheva et al. [l],who
foundthatthe variance o12 of the intensity fluctuations
saturated for very strong turbulence, in disagreement with
theoretical predictions’ [ 2 ] , [3]. Inthelast year several where r2is the mutual coherence function andr1 = ( p l-

physical modelshave been postulatedto calculate the P 2 + P1’ - P2’)/2 and p 2 = ( P I - P 2 - P1’ + P2’)/2.
variance and covariance of the intensity fluctuations [4], [ 5 ] , If we use (2) as a lirst iteration in the integral equation
but both rely on the introductionof an arbitraryparameter derived by Tatarskii [2, eq. (72.1)] for the Fourier transform
which cannotbe specifiedexceptby comparisonwith of r4 we can show for a Kolmogorov spectrum that the
experimental data. In contrast to these other theories we covariance
have recently [6] developedanewmodel whichis free
from these limitations. To outline and amplify on our
theory let us consider the fourth moment of the electric
field‘ of the intensity fluctuations of aplane wave in homo-
geneous turbulence is given by [6]’
r4 = (e(zYpl)e(zYp2)e*(z,p,’)e*(zYp2‘)~ (1)
where z is the distance measuredalong the direction of
propagation and p is the distance measured transverse to
the propagation path.Tatarskii [ 7 ] has derived the equation
satisfiedby r4,but it has not proven possible to obtain
solutions, except inthe limit ofweak turbulence. Our
te-chnique is to obtain an iterative solution to this equation
+
2 -215
f3 [r;;;ii.j]
R

Manuscript received October 7, 1974. Note


that for simplicity of presentation we have neglected the term
The.author is with the Air Force Cambridge Research Laboratories ’ 1 1 ~ ] in [q since this contributes to Br only
( ~ , 2 , - 2 ~ 5 g [ ( u , 2 ) 3obtained
(LZP),,Hanscom AFB, Bedford, Mass; 01731. in the transition region between the fist and second t e r m in (4).

Anda mungkin juga menyukai