MALMSTRÖM: ON THE REACTION THEOREM FOR SCATTERED FIELDS 1

On the Reaction Theorem for Scattered Fields

Johan Malmström

Abstract— ε(r), µ(r)

Two novel formulations of the reaction are derived. The formu- Antenna 1 Antenna 2
lations decompose the electromagnetic fields in multiple scattered
components. It is shown that some of the scattering components (a) ∼hI11 E1 , H1 S2 b V21
do not contribute to the reaction. The novel formulations of the -
reaction are derived by excluding the non-contributing terms n̂2
from the traditional reaction formulations. The correctness of
one of the formulations is verified with a numerical example. The ε(r), µ(r)
novel formulations are useful for estimating the errors introduced Antenna 1 Antenna 2
if multiple scattering are neglected in a scattering environment.
They are also important for a conceptual understanding of the b V12
reaction. (b) 2
E , H2 S2 ∼hI22
Index Terms—Electromagnetic scattering, electromagnetic the- n̂2
ory, antennas, mutual impedance, mutual coupling.
Fig. 1. The original reaction (1); (a) Antenna 1 transmits with a current I11
while Antenna 2 is open circuit, and (b) Antenna 2 transmits with a current
I22 while Antenna 1 is open circuit. The media is the same in (a) and (b).
I. I NTRODUCTION
The mutual impedance between antennas, and the closely
ε(r), µ(r)
related mutual coupling, is of big importance when installing Antenna 1 Antenna 2
antennas for simultaneous transmission on platforms [1]. A
high mutual coupling means that much of the transmitted (a) ∼hI11 E1 , H1 S2 b V21
energy leaks into the other antenna [2], something that can -
n̂2
severely degrade system performance.
The reaction is a measure of the interaction between two
pairs of electric and magnetic sources, carried by their corre- ε0 (r), µ0 (r)
Antenna 2
sponding electromagnetic (EM) fields. The reaction, denoted
with h2, 1i, is defined as [3]
I (b) 2
E , H2 S2 ∼hI22
0

h2, 1i = (E2 × H1 − E1 × H2 ) · n̂2 dS, (1) n̂2

S2
where the fields E1 , H1 are generated by Antenna 1 and
E2 , H2 by Antenna 2, as in Fig. 1. The integration surfaces
S2 , with normal n̂2 , must completely enclose one and only one
of the antennas (in this case Antenna 2). To be able to handle
inhomogeneous regions, such as a platform or other structure,
the complex valued ε(r), µ(r) depend on the position r.
The reaction h2, 1i does not depend on the environment
outside the integration surface S2 , when generating E2 , H2
[4], which is used in the generalized reaction [4],
I the two sources or antennas involved,
h2, 1i = (E20 × H1 − E1 × H20 ) · n̂2 dS, (2)
S2
where unprimed variables are from the original environment
and primed variables are from an alternative environment, as
depicted in Fig. 2. The structure outside the integration surface
S2 , described by ε0 (r), µ0 (r), can be changed between the
original and alternative environments, but structure enclosed
by S2 must remain unchanged.
Manuscript received September 30, 2017; revised February 4, 2018; ac-
cepted Xxx xx, 2017.
J. Malmström is with Saab Surveillance, Stockholm, Sweden, and with the
Electromagnetic Engineering Laboratory, KTH Royal Institute of Technology,
Stockholm, Sweden.
The work was supported by Saab.
Fig. 2. The generalized reaction (2); (a) Antenna 1 transmits with a current
I11 in the original environment with Antenna 2 open circuit, and (b) Antenna 2
0 in an alternative environment. The media inside
transmits with a current I22
S2 must be the same in the two environments but can differ outside S2 .
The reaction theorem relates the field quantities in (1)–(2)
with circuit quantities, i.e. currents and voltages defined on the
terminal of antennas, see Fig. 1–2. For example, it can relate
the reaction h2, 1i with the mutual impedance Z21 between
h2, 1i
Z21 = − . (3)
I11 I22
The terminal current I21 in Antenna 2, when Antenna 1
transmits, is assumed to be zero, i.e. an open circuit. The
terminal current I22 is replaced with I22 0
in the generalized
reaction theorem (2).
The reaction theorem has been used for calculating mutual
impedance in several works recently, see e.g. [5]–[8]. It is
observed that multiple scattering is neglected in these works,
to make the reaction theorem applicable. It is hence interesting
to investigate the effects of multiple scattering on the reaction
theorem, which is a goal with this paper.
MALMSTRÖM: ON THE REACTION THEOREM FOR SCATTERED FIELDS 2

SR
E (0) , H (0) VR \V2
Ed, H d
E (1) , H (1)
V−
E (2) , H (2) S2


V+ r V2

+

...
...
| {z } | {z
E+, H + E−, H −
Fig. 3. Decomposition of fields E , H generated from an antenna based on
source origin into the fields E + , H + and E − , H − . Scattering components,
E (n) , E (n) , are denoted with their scattering order n, i.e. the number of
reflection they have been under. The direct components, E d = E (0) , H d =
H (0) , is included in E + , H + .
II. T HE R EACTION F OR S CATTERED F IELDS
A. Field Decomposition
We consider an antenna generating the fields E, H that are
scattered on some objects, as in Fig. 3. We use a fictitious
surface to separate the transmitting antenna (if necessary, also
including parts of the platform) and scatterers in two disjoint
regions, V + and V − , as in Fig. 3.
The fields from the antenna will induce a current J on
the platform and all scattering objects. We note that, from
the method of moment approach, given the induced current J
and material parameters ε(r), µ(r), the fields E, H can be
uniquely determined, see e.g. [9]. Clearly, E, H are linearly
dependent of J . The current J is decomposed into two parts,
J = J + + J −,
based on the origin of the currents, so that
J+ = J , in V + ,
− +
J = 0, in V ,
J+ = 0, in V − ,
− −
J = J , in V .
Based on the superposition principle, we consider the fields
E + , H + to be generated by the sources J + and the fields
}
Fig. 4. The surfaces SR , S2 , enclose the volumes VR and V2 , respectively.
The volume VR \V2 is between the two surfaces SR and S2 .
First, we will focus on the second of these four integrals, i.e.
(12) in the above expression. The integral (12) depends on the
fields E2+ , H2+ from Antenna 2, and the fields E1− , H1− from
Antenna 1 that are scattered on objects enclosed by S2 .
B. The Radiation Condition
We let a sphere SR enclose a volume VR that include both
antennas and the surface S2 , see Fig. 4, and calculate the
reaction (1) over the surface SR . When the radius r of the
sphere goes to infinity r → ∞, the radiated fields decay as
r−1 and the cross products as r−2 . Since the area of integration
goes as r2 , the integral will not obviously converge. We have
to convince ourselves that the reaction over the large sphere
SR with the decomposed fields is zero.
We assume that the medium far away from the antennas is
homogeneous and isotropic, and that the sources are localized.
With these assumptions, the EM fields will propagate outwards
(4) from the antennas and, asymptotically (i.e. infinitely far away),
form a locally plane wave.
For large but finite distances from the source, the wave
(5) will not be perfectly plane. We consider the Silver-Müller
radiation condition, which gives information about the asymp-
(6)
totic behavior of the fields far from the sources. In [10]
(7)
the Silver-Müller radiation condition is given as (with a
(8) omitted/normalized impedance Z = 1)
lim r (H × n̂ − E) = 0. (15)
r→∞

E − , H − by the sources J − , so that Using the little-o notation1 , the radiation condition (15) for
a large, but finite, r and a general impedance Z can be
E = E+ + E−, (9)
formulated as a plane wave including a correction term as
H = H + + H −. (10)
E = Z(H × n̂) + o(r−1 ), (16)
Consequently, the fields E1− , H1−
are source free in V and +

the fields E2+ , H2+ are source free in V − . in which it is understood that each field component goes as
We apply the decomposition (9)–(10) to E1 , H1 and o(r)−1 . Using (16), we can write the integrand in the reaction
E2 , H2 in (1). We let the surface that separates V + and integral (1) as,
V − coincide with the integration surface S2 in Fig. 1–2. The E2+ × H1− − E1− ×H2+ = (17)
reaction (1) separates into four terms,
I = Z(H2+ × n̂) + o(r −1
)) × H1−
E2+ × H1+ − E1+ × H2+ · n̂2 dS (11) −(Z(H1− × n̂) + o(r −1
)) ×H2+ =


h2, 1i =
I S2 H2+ (H1− · n̂) − H1− (H2+ · n̂) + (H1− − H2+ )o(r−1 )


= Z
E2+ × H1− − E1− × H2+ · n̂2 dS (12)


+
1 For


non-zero g(r), the little-o notation f (r) = o g(r) implies that
IS2
E2− × H1+ − E1+ × H2− · n̂2 dS (13) f (r)


+ lim = 0.
r→∞ g(r)
IS2
Additionally, the little-o notation is related to the Big-O notation. If a function
E2− × H1− − E1− × H2− · n̂2 dS. (14)


+ f (r) = o(r), it implies that f (r) = O(r) [11].
S2
MALMSTRÖM: ON THE REACTION THEOREM FOR SCATTERED FIELDS
Far from the source, the dominant parts of H1− and H2+
are orthogonal to the direction of propagation n̂. We know that
H1− , H2+ , and (H1− − H2+ ) all decay as r−1 for large r. The
radial components, H1− · n̂ and H2+ · n̂, decay faster, so that
H1− · n̂ = o(r−1 ) and H2+ · n̂ = o(r−1 ). Hence, all terms in
(17) are o(r−2 ), and so is the whole integrand, i.e.,
E2+ × H1− − E1− ×H2+ = o(r−2 ). (18)
This relation implies that the reaction (1) over the large sphere
SR goes to zero when r → ∞,
I which is valid for an empty alternative environment. Note
E2+ × H1− − E1− ×H2+ · n̂ dS = 0.


(19)
SR
C. Divergence of the Cross Product Difference
Still with the integral (12) in focus, we consider the region
VR \V2 between the surface S2 and the sphere SR , see Fig. 4,
I
E2+ × H1− − E1− ×H2+ · n̂ dS.


(20)
SR −S2
Recall that the fields E2+ , H2+ , E1− , H1− per definition (see
Fig. 3) are source free, J1− = 0 and J2+ = 0, in the volume
VR \V2 outside S2 . By considering the integral form of Lorentz
reciprocity theorem [12] in the source free region VR \V2 , we
immediately see that the integral (20) evaluates to zero,
I dipoles, with scattering order n = 1 in Fig. 6, produce left-
E2+ × H1− − E1− ×H2+ · n̂ dS = 0


(21)
SR −S2
We know from (19) that the surface integral over the large
sphere SR is zero. Combined with (21), it follows that the
integral over S2 , i.e. the integral (12), must be zero,
I
E2+ × H1− − E1− ×H2+ · n̂2 dS = 0.


(22)
S2
Changing our focus to the third integral, (13), in the
decomposed reaction, we see that the fields E2+ , H1− , E1− ,H2+
are all source free inside the integration surface S2 . We can
again consider the integral form of Lorentz reciprocity theorem
[12], to see that the integral (13) evaluates to zero,
I order produce the right-propagating fields E1+ , H1+ whereas
E2− × H1+ − E1+ × H2− · n̂2 dS = 0.


(23)
S2
D. The Resulting Reaction
Using (19), (23), the reaction h2, 1i can be formulated as
I N
E2+ × H1+ − E1+ × H2+ · n̂2 dS +


h2, 1i =
IS2
E2− × H1− − E1− × H2− · n̂2 dS. (24)


S2
Note that there is no interaction between the fields (·)+ and
(·)− from the different antennas.
The generalized reaction (2) can also be separated into
scattering components. It becomes especially simple if the
alternative environment where Antenna 2 transmits, depicted
in Fig. 2(b), is chosen to be homogeneous (e.g. vacuum)
3
outside the integration surface S2 . With this choice, there
will be no scattering from outside S2 , and, hence, E2+ =
E2d , H2+ = H2d , E2− = 0, H2− = 0 (using the notation
introduced in Fig. 3). The fields E2d , H2d , E1−, H1−, have
their sources within S2 and are source free outside S2 . With
this restriction, the same derivations can be repeated for the
generalized reaction, resulting in the following expression,
I
E2d × H1+ − E1+ × H2d · n̂ dS,


h2, 1i = (25)
S2
that the components E1− , H1− generated by Antenna 1 and
scattered inside the surface S2 do not contribute to the reaction.
III. V ERIFICATION
We verify the results for a pair of Hertzian dipoles between
infinite ground planes, as depicted in Fig. 5. The configuration
will give strong multiple scattering. The EM fields can be
determined using the method of mirroring and the analytically
known fields from a Hertzian dipole in free space [12].
Starting with Dipole 1, we determine the fields after mir-
roring in the left ground plane. The resulting two Hertzian
dipoles in Fig. 6 with scattering order n = 0 produce right-
propagating fields in the region between the ground planes.
Next, the two Hertzian dipoles for scattering order n = 0 is
mirrored in the right ground plane. The resulting two Hertzian
propagating fields.
By continuing the alternating mirroring on the two ground
planes, we get mirrored dipoles for higher scattering orders n,
as illustrated in Fig. 6. The sources will be farther away for
each mirroring and give weaker contribution to the fields in
the region between the ground planes. Referring to Fig. 6, we
can convince ourselves that summing the fields with scattering
order n = 0, n = [1, 2], n = [3, 4], . . . produce null-fields
on the left ground plane, whereas summing the fields with
scattering order n = [0, 1], n = [2, 3], . . . produce null-fields
on the right ground plane.
We let the integration surface S2 enclose the half-space
x > 0. Summing scattered components with even scattering
summing components with odd scattering order produce the
left-propagating fields E1− , H1− , i.e.,
N
(n)
X
E1+ = E1 , (26)
n=0,2,4,...
(n)
X
E1− = E1 , (27)
n=1,3,5,...
and equivalently for the magnetic fields H1+ , H1− .
The same procedure is repeated with Dipole 2, to decom-
pose the fields into E2+ , H2+ and E2− , H2− .
Based on the decomposed fields, E1+ , H1+ , E1− , H1− , E2+ ,
H2+ , E2− , H2− , we evaluate two formulations of the reaction;
the original reaction (1), and the reactions for scattered fields
(24) derived in this letter. We use a dipole separation distance
D = 1.8 m, with each dipole placed a = 1.1 m from
 MALMSTRÖM: ON THE REACTION THEOREM FOR SCATTERED FIELDS 4

Mutual impedance |Z21 | (/Ω)

z
Dipole 1 Dipole 2
-x
6

a 
- D - a-

Fig. 5. Two high-scattering antennas, each consisting of a Hertzian dipole in

front of an infinite ground plane, facing each other.

Antenna separation distance D (/m)

Scattering order n

Fig. 8. The mutual impedance magnitude |Z21 | calculated with the reaction
h2, 1i as in (1), and (24). For illustration, the integrals (22) and (23) are also
plotted. We see that the agreement of (1) and (24) is excellent and that (22)
and (23) are both very close to zero.

on one side of the integration surface only interact with fields

x (/m)
from sources on the opposite side of the surface.
Fig. 6. Mirroring of Dipole 1 with positive (up-pointing triangle) and negative The resulting reaction between sources must not change
(down-pointing triangle) Hertzian dipoles between two ground-planes at x = depending on the chosen formulation of the reaction. The clas-
±2 m. The vertical axis denotes scattering order n.
sical formulations of the reaction [3], [4] include combinations
of scattering components that do not contribute to the reaction.
Total field E1 Component E1+ Component E1− V/m
4 4 4 4 4 4 4 4 4 4 4 4 If neglecting multiple scattering, as in e.g. [5]–[8], the novel
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
3 3 3 3 3 3 3 3 3 3 3 3 formulations of the reaction makes it easier to estimate the
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
2 2 2 2 2 2 2 2 2 2 2 2
2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 error introduced due to the neglected scattering components.
(/m)

(/m)

(/m)

1 1 1 1 11 1 1 1 1 1 1 11 1 1 1 1 1 1 11 1 1 1 1 1 1
z (/m)

z (/m)
z (/m)

zz (/m)

z (/m)

z (/m)
zz (/m)

z (/m)

zz (/m)
z (/m)
z (/m)

z (/m)

For example, with the formulation (25), we see that none of

0 0 0 0 00 0 0 0 0 0 0 00 0 0 0 0 0 0 00 0 0 0 0 0 0
1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1
the odd scattering components (H1− , E1− ) contribute to the
-1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1
2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 reaction.
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2
3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3
-3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3 -3
4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4
ACKNOWLEDGMENT
02 2 -2 -2 -2 0 0-20 2 2 02 2 -2 -2 -2 0 0-20 2 2 02 2 -2 -2 -2 0 0-20 2 2 02 2
x x x x x x(/m) x xx (/m)
x x x (/m)
x x x x The author is grateful for valuable advices from B. L. G.
Fig. 7. The z component of the electric field ReE1 (left) from Dipole 1 de- Jonsson during the work with this letter.
composed in the right-propagating ReE1+ (middle), and the left-propagating
ReE1− (right), illustrated on the plane x = [−2, 2], y = 0, z = [−4, 4].
R EFERENCES
[1] T. M. Macnamara, Introduction to Antenna Placement and Installation.
the ground plane, and a wavelength λ = 1 m. Scattering John Wiley & Sons, 2010.
components up to order N = 12 are included. The field E1 [2] C. R. Paul, Introduction to Electromagnetic Compatibility, 2nd ed. John
from Dipole 1, decomposed into the right-propagating field Wiley & Sons, 2006.
[3] V. H. Rumsey, “Reaction concept in electromagnetic theory,” Phys. Rev.,
E1+ and left-propagating field E1− , is depicted in Fig. 7. vol. 94, no. 6, pp. 1483–1491, 1954.
The results, in terms of mutual impedance using (3), are [4] J. H. Richmond, “A reaction theorem and its application to antenna
depicted in Fig. 8. We see that the agreement between the impedance calculations,” Antennas Propagation, IRE Trans., vol. 9,
no. 6, pp. 515–520, 1961.
mutual impedance calculated with (1) and (24) is excellent. [5] H. Wang, et al., “Estimating radio-frequency interference to an antenna
The results are also close to the mutual impedance calculated due to near-field coupling using decomposition method based on reci-
with a full wave method. As expected, the reaction from the procity,” IEEE Trans. Electromagn. Compat., vol. 55, no. 6, pp. 1125–
1131, 2013.
scattering components (22) and (23) are very close to zero. [6] L. Li, et al., “Radiation Noise Source Modeling and Application in
The results from this strong scattering case indicate that the Near-Field Coupling Estimation,” IEEE Trans. Electromagn. Compat.,
derived relation (24) for the reaction is correct. vol. 58, no. 4, pp. 1314–1321, 2016.
[7] C. Craeye, “On the transmittance between OAM antennas,” IEEE Trans.
Antennas Propag., vol. 64, no. 1, pp. 336–339, 2016.
[8] J. Malmström and B. L. G. Jonsson, “An Effective Method for Antenna
IV. D ISCUSSION AND C ONCLUSIONS Placement on Platforms Utilizing the Reaction Theorem,” in 2017 Int.
Reaction theorems can be formulated in several ways, where Conf. Electromagn. Adv. Appl., Verona, 2017.
[9] W. C. Gibson, The Method of Moments in Electromagnetics. Chapman
the formulations by Rumsey and Richmond are well known. In & Hall/CRC, 2008.
this letter we derive two novel formulations of the reaction for [10] S. H. Schot, “Eighty years of Sommerfeld’s radiation condition,” Hist.
scattered fields; see (24) and (25). The derived formulations Math., vol. 19, no. 4, pp. 385–401, 1992.
[11] C. Stover, “Little-O Notation,” Wolfram Mathworld (http:// mathworld.
of the reaction are expressed for fields decomposed in their wolfram.com/ Little-ONotation.html).
scattered components. It is shown that not all scattering [12] C. A. Balanis, Antenna Theory Analysis and Design, 3rd ed. John
components contribute to the reaction. Fields from sources Wiley & Sons, 2005.