Professional Documents
Culture Documents
8, 213241, 2008
L. Liu and K. Li
The Electromagnetics Academy
Zhejiang University
Hangzhou, Zhejiang 310058, China
W. Y. Pan
China Research Institute of Radiowave Propagation
19 Qingda 1 Road, Qindao, Shandong 266071, China
1. INTRODUCTION
z
Region 0: k 0
( , , z)
h Vertical electric
Region 1: k 1
dipole at (0, 0, d)
x, y
0
l Region 2: k 2
Region 3: k 3
C4 e C5 ei2 l
= C4 e + C5 ei2 l
(23)
k22 k32
then,
2 3
2 2
k k3
C4 ei2 l = 22 i2 l
3 C5 e . (24)
+
k22 k32
With (17) and (18), we have
1 0 i1 (hd)
C1 ei1 h C2 ei1 h e
k12 21
0 0 i1 (hd)
= 2 C1 ei1 h + C2 ei1 h e (25)
k0 21
0 1 1 0 1 0 0 i1 (hd)
2 + 2 C2 ei1 h = 2 2 C1 ei1 h 2 2 e . (26)
k0 k1 k1 k0 k1 k0 21
Multiplying ei2 l to both sides of (19) leads to
0 i1 d i2 l
C1 C2 + e e = C4 e C5 e
k12 21 k22
3
2
2 k22 k32 i2 l i2 l
= 2 e e C5 . (28)
k2 2 + 3
k22 k32
Then, we write
1 3 1 k22 3 2
C1 i tan 2 l + 2 i 2 tan 2 l
k12 k32 k12 2 k3 k2
0 i1 d 1 3 1 k22 2 3
+ e 2 i 2 2 tan 2 l + i 2 tan 2 l 2
21 k1 k3 k1 2 k2 k3
1 3 1 k 2 2 3
= C2 2 i 2 2 2 tan 2 l + i 2 tan 2 l 2 . (30)
k1 k3 k1 2 k2 k3
where
1 0
2 2 (m cos 1 d in sin 1 d) (1 i tan 1 h)
k k
P = 1 0 . (38)
0 1 1 0
m i tan 1 h + n i tan 1 h
k02 k12 k12 k02
Then, we have
1x = 0 i1 z 0 i21 h i1 z 0 i1 |zd|
B Qe e Pe e
21 21 21
0 i1 z
= Qe + P ei1 (2hz) + ei1 |zd| . (39)
21
From the relations in (8)(12), we have
1y = 0 i1 z
B Qe + P ei1 (2hz) + ei1 |zd| (40)
21
1z = 0
B (41)
1x = 0 Qei1 z P ei1 (2hz) ei1 |zd|
E (42)
2k12
1y = E
E 1x = 0 Qei1 z P ei1 (2hz) ei1 |zd| (43)
2
2k1
d 2 1x
E1z = + k1 B
k12 dz 2
0 2 i1 z
= 2 Qe + P ei1 (2hz) + ei1 |zd| . (44)
21 k1
It is now convenient to express the eld components in the cylindrical
coordinates , , z with the relations
x = cos , y = sin (45)
= cos , = sin (46)
and the integrated representations of the Bessel functions, viz.,
2
in
Jn () = ei( cos +n) d. (47)
2 0
From (39)(44), using the Fourier integrals like (5) and the following
relations
E1 = E1x cos + E1y sin (48)
B1 = B1x sin + B1y cos (49)
220 Liu, Li, and Pan
where the upper sign in (50) is for the region z d, and the lower
sign for 0 z d. In order to see useful physical insights, and
(1) (2)
taking into account the relationship Hn () = Hn ()(1)n+1 , it
is convenient to rewrite the integrated formulas in the following forms.
(1) (2) (3)
E1 = E1 + E1 + E1 (53)
(1) (2) (3)
E1z = E1z + E1z + E1z (54)
(1) (2) (3)
B1 = B1 + B1 + B1 (55)
where
i
ei1 |zd| H1 ()2 d
(1) (1)
E1 = (56)
81
(2) i (1)
E1 = Qei1 z H1 ()2 d (57)
81
(3) i (1)
E1 = P ei1 (2hz) H1 ()2 d (58)
81
1
ei1 |zd| 11 H0 ()3 d
(1) (1)
E1z = (59)
81
1
Qei1 z 11 H0 ()3 d
(2) (1)
E1z = (60)
81
1
P ei1 (2hz) 11 H0 ()3 d
(3) (1)
E1z = (61)
81
Progress In Electromagnetics Research B, Vol. 8, 2008 221
(1) i0 i1 |zd| 1 (1)
B1 = e 1 H1 ()2 d (62)
8
i0
Qei1 z 11 H1 ()2 d
(2) (1)
B1 = (63)
8
i0
P ei1 (2hz) 11 H1 ()2 d.
(3) (1)
B1 = (64)
8
It is seen that (56), (59), and (62) stand for the direct wave, which have
been evaluated in the monograph by King, Owens, and Wu [8]. When
k0 = k1 , the above formulas can be reduced to those for three-layered
case addressed in [29]. Obviously, the above integrals including the
(1)
Bessel functions Ji () or Hi () (i = 0, 1) with high oscillatory,
these integrals converge very slowly. It is necessary to evaluate the
above integrals including Q and P by using analytical techniques.
Im ( )
0 1 2
Re ( )
0
*1 *2
k0 k1 k2
In the second case with k0 < < k1 , 0 = i 2 k02 = i0 , 0 ,
1 , and 2 are positive real numbers. Then, we obtain
1 0 1 2 tan 2 l 0 2 1 2
2 2+ + 2 2 tan 2 l tanh 1 h 4 tanh 1 h = 0. (68)
k1 k0 k12 k22 k0 k2 k1
From the above analysis, it is concluded that the poles may exist in
the intervals k0 < j < k2 , which can be determined by using Newton
(2) (3)
method as addressed in [29]. Then, the integrals E1z and E1z can be
expressed as follows:
0 i1j
Q (j )e z H1 (j ) 2
(2) (1)
E1 =
4k12 j
i (1)
+ Qei1 z H0 ()2 d (70)
81 0 +1 +2
0
P (j )ei1j z H1 (j ) 2
(3) (1)
E1 = 2
4k1 j
1 (1)
P ei1 (2hz) H0 ()2 d. (71)
81 0 +1 +2
where
1j 1j
Q(j )
1j
= n cos 1j d + 2 tan 1j h sin 1j d
k12 k12 k1
0j
0j
i 2 tan 1j h cos 1j d + i 2 sin 1j d /q ( ) (72)
k k0
0
1j
0j 1j
cos 1j din sin 1j d 1i tan 1j h
k12 k02 k12
P (j ) = (73)
q (j )
1 2 0 1 n 1 0
q() = i 4 tan 1 h i 2 n tan 1 h + 2 + 2 2 (74)
k1 k0 k1 k1 k0
i
q () = 2 tan 1 h + h 1 sec 2
1 h
k14
1 2 0
+ 2 2 tan 2 l tan 1 h + tan 2 l tan 1 h
k0 k2 0 2
0 2 h
+0 l sec2 2 l tan 1 h + tan 2 l sec2 1 h
1
1 2 tan 2 l 1 2
+i 2 2 + tan 2 l + 1 l sec 2 l
k1 k2 1 2
1 0 1
2 2 + (75)
k1 k0 1 0
ij = ki2 j 2 , i = 0, 1 (76)
224 Liu, Li, and Pan
2
n = i tan 2 l. (77)
k22
Because both the integrands P () and Q() are even functions of
2 , the integrals in (70) and (71) along the branch cut 2 are zero.
Next, we will evaluate the integrals in (70) and (71) along the branch
cuts 1 and 0 .
Taking into account the conditions of k1
1 and (z+d) , the
dominant contribution of the integral along the branch line 1 comes
from the vicinity of k1 . Let = k1 (1 + i 2 ), 0 , 1 , and 2 at the
vicinity of k0 can be approximated as follows:
01 = k02 2 i k12 k02 (78)
3
11 = k12 2 2k1 ei 4 (79)
21 = k22 2 k22 k02 . (80)
Considering the case of interest that both h and d are not very large,
we arrive at the following expressions.
cos 11 d 1; tan 11 h 11 h; sin 11 d 11 d. (81)
Substituting (81) into (36), and neglecting the high-order terms of 11 ,
we have
1 i01 01 3
then, we write
i (1)
Qei1 z H1 ()2 d
81 1
i0 3
2 i(k1 3 ) k1 2 2
i 2k1 ei 4 z
= Qe e 4 e k1 i2k1 d
8k12 k1
k1 z 2
0 k1 2 i k1 3
4
+ 2
= e ( + Apk1 )
4 k1
3
2
k1 i ei 4 z
Bpk1 e 2
d. (86)
2
0 i k1 2 +z 2 + 4 1 z2 z i 3
= e Bpk1 +i + Apk1 e 4 .
2 2 k1 22 2
(88)
Similarly, we have
11 01 11 11 01 1
2 2 2 in1 11 d 2 2 in1 d
k1 k0 k1 k1 k0 k12
P = 01 11 n1 11 01 = 01 n1 01
i 2 n1 11 h + 2 + 2 2 i 2 n1 h + 2 + 2 2
k0 k1 k1 k0 k0 k1 k1 k0
i 3 1
3 2e 4 in1 d
01 k1 ei 4 k12
= . (89)
2 01 n1 01
k1 i 2 n1 h + 2 + 2 2
k0 k1 k0 k1
226 Liu, Li, and Pan
Let
01 k1 ei 4
3
Aqk1 = 2 (90)
2k0
i 3 1
2e 4 in1 d
k12
Bqk1 = . (91)
01 n1 01
k1 i 2 n1 h + 2 + 2 2
k0 k1 k0 k1
With the similar procedures, it is obtained readily.
i (1)
P ei1 (2hz) H1 ()2 d
81 1
0 ik1 2 +(2hz)2 + 4
= e Bqk1
2 2
1 z2 z 3
+i 2
+ Aqk1 ei 4 . (92)
k1 2 2
In the next step, we consider the branch cut 0 . Let
2
k 10 20
tan l + 10
tan h
0 20 10
k12 k22 k14
Apk0
= (93)
10 20 tan 10 h tan 20 l
2
k12 k22
10
k0 ei
3
4 n0
k12
10 2
Bpk0 =
2 in0 tan 10 h
k12
10 10
2 cos 10 d + 2 tan 10 h sin 10 d
k1 k1
10
in0 tan 10 h i(sin 10 d tan 10 h cos 10 d)
k12
10 2
2 n0 i 10 tan 10 h (94)
k1 k14
10 = k12 k02 ; (95)
20 = k22 k02 ; (96)
Progress In Electromagnetics Research B, Vol. 8, 2008 227
20
n0 = i tan 20 l. (97)
k22
Then, we write
10
i(sin 10 dtan 10 h cos 10 d) n0
k12 Bpk0
Q= 10 + (98)
in0 tan 10 h Apk0
k12
i (1)
Qei1 z H1 ()2 d
81 0
i0 i10 z i(k0 3 ) 2 Bpk0
2ik03 ek0
2
= e e 4 d
8k12 k0 Apk0 ei
3
4
0 k03 i(10 z+k0 3 ) 2
= e 4 Bpk0
4k12 k0
2
Apk0 ei
3
4
k0 2
+
d .
2 e
(99)
k0
Apk0 e2 i 3
4
In terms of the variable t = k0 , and use is made of the formula
(pp.609) in [37], the result becomes
i (1)
Qei1 z H1 ()2 d
81 0
0 k03 i(10 z+k0 3 ) 2
= e 4 Bpk0
4k12 k0
3
Apk0 ei 4
3
k0 Apk0 ei 4
+
3 2
dt
k0 t2 k0 Apk0 ei 4
0 k03 i(10 z+k0 3 ) 2
= e 4 Bpk0
4k12 k0
i 5
+k0 A2pk
+ Apk0 erfc ei 4 k0 Apk0 e 4 0
. (100)
k0
Similarly, it is also obtained readily.
i (1) 0 k03 i[10 (2hz)+k0 3 ]
P ei1 (2hz) H1 ()2 d = e 4
81 0 4k12
228 Liu, Li, and Pan
2
i 5
+k0 A2qk
Bqk0 +Aqk0 erfc ei 4 k0 Aqk0 e 4 0
(101)
k0 k0
where
2
10 20 tan 20 l 10
k + tan h
0 10
k22 k12 k14
Aqk0
= (102)
10 20
2 2 tan 10 h tan 20 l
k12 k2
10
Bqk0 = k0 ei
3
4 cos 10 d in0 sin 10 d (1 i tan 10 h)
k12
2
10 10 10 2
10
i n 0 tan 10 h + n 0 i tan 10 h
k14 k12 k12 k14
10 2 . (103)
2 in0 tan 10 h
k12
i k1 2 +z 2 + 4
0 0 e
Q(j )ei1j z H0 (j )j 2 +
(2) (1)
E1 = 2
4k1 j 2 2
1 z2 z 3
Bpk1 +i 2
+ Apk1 ei 4
k1 2 2
0 k03 i(10 z+k0 3 ) 2
e 4 Bpk0
4k12 k0 k0
i 3 ik0 A2pk i 4
+iApk0 e 4e 0 erfc e k0 Apk0 . (104)
where
n1 k1 ei 4
3
Apk1 = (111)
2
i 3 1 01 01
2e 4 +i 2 di 2 h
k12 k0 k0
Bpk1 = (112)
n1 01 01
k1 + 2 2 i 2 n1 h
k12 k1 k0 k0
5
With the change of the variable t = z ei 4 ,
2
it becomes
i0 (1)() 2 1 i0 k12 1 ik1 2 +z 2
Qei1 z H1 1 d = e
8 1 4 k1
z
t + ei 4 Apk1 Bpk1 ek1 t dt
5 2
2
i0 k1 z i 5
2 2
= Bpk1 e 4 Apk1 eik1 +z . (113)
4 2
Obviously, we can also get
1 1
(1)
dQei1 z H0 ()3 01
81 1
i0 z 5
2 2
= Bpk1 ei 4 Apk1 eik1 +z . (114)
4 2
The integrals including the factor P in (107) and (109) along the
branch cut 1 can be evaluated readily. They are
i0
P ei1 (2hz) H1 ()2 11 d
(1)
8 1
i0 k1 z i 5 2 2
= Bqk1 e 4 Aqk1 eik1 +(2hz) (115)
4 2
1
dP ei1 (2hz) H0 ()3 11
(1)
81 1
i0 z i 5 2 2
= Bqk1 e 4 Aqk1 eik1 +(2hz) (116)
4 2
where
01 k1 ei
3
4
Aqk1 = (117)
k02 2
Progress In Electromagnetics Research B, Vol. 8, 2008 231
1 3
2ei in1 d
4
k12
Bqk1 = . (118)
01 n1 01
k1 i 2 n1 h + 2 + 2 2
k0 k1 k1 k0
With a similar manner in the evaluations of the integrals (70) and
(71) along the branch cut 0 , we can obtain
i0
Qei1 z H1 ()2 11 d
(1)
8 0
0 k03 i(10 z+k0 3 ) 1 2
= e 4
10 Bpk0
4 k0
i 5
+k0 A2pk
+ Apk0 erfc ei 4 k0 Apk0 e 4 0
(119)
k0
i0
P ei1 (2hz) H1 ()2 11 d
(1)
8 0
0 k03 1 i[10 (2hz)+k0 3 ] 2
= e 4 Bqk0
4 10 k0
i 5
+k0 A2qk
+ Aqk0 erfc ei 4 k0 Aqk0 e 4 0
(120)
k0
1
Qei1 z H0 ()3 11 d
(1)
81 0
1
0 k04 10 3 2
= ei(10 z+k0 4 ) Bpk0
4k12 k0
i 5
+k0 A2pk
+ Apk0 erfc ei 4 k0 Apk0 e 4 0
(121)
k0
1
P ei1 (2hz) H0 ()3 11 d
(1)
81 0
0 k04 1 i[10 (2hz)+k0 3 ] 2
= e 4 Bqk0
4k12 10 k0
i 5
+k0 A2qk
+ Aqk0 erfc ei 4 k0 Aqk0 e 4 0
. (122)
k0
Here, Apk0 , Aqk0 , Bpk0 , and Bqk0 are dened by (93), (102), (94),
and (103), respectively. Substituting (113)(116) and (119)(122) into
232 Liu, Li, and Pan
(106)(109), we have
0
1
Q(j )ei1j z H0 (j )j 3 1j
(2) (1)
E1z = i 2
4k1 j
i0 z 5
2 2
Bpk1 ei 4 Apk1 eik1 +z
4 2
1
0 k04 10 3 2
2 ei(10 z+k0 4 ) Bpk0
4k1 k0
i 5 +k0 A2
+ Apk0 erfc ei 4 k0 Apk0 e 4 pk0
(123)
k0
0
1
P (j )ei1j z H0 (j )j 3 1j
(3) (1)
E1z = i 2
4k1 j
i0 z 5
2 2
Bqk1 ei 4 Aqk1 eik1 +(2hz)
4 2
0 k04 1 i[10 (2hz)+k0 3 ] 2
+ e 4 Bqk0
4k12 10 k0
i 5
+k0 A2qk
+ Aqk0 erfc ei 4 k0 Aqk0 e 4 0
(124)
k0
0
1
Q(j )ei1j (2hz) H1 (j )j 2 1j
(2) (1)
B1 =
4 j
i0 k1 z 5
2 2
+ Bpk1 ei 4 Apk1 eik1 +z
4 2
0 k03 i(10 z+k0 3 ) 1 2
e 4
10 Bpk0
4 k0
i 5
+k0 A2pk
+ Apk0 erfc ei 4 k0 Apk0 e 4 0
(125)
k0
0
1
P (j )ei1j (2hz) H1 (j )j 2 1j
(3) (1)
B1 =
4 j
i0 k1 z 5
2 2
+ Bqk1 ei 4 Aqk1 eik1 +(2hz)
4 2
0 k03 1 i[10 (2hz)+k0 3 ] 2
e 4 Bqk0
4 10 k0
Progress In Electromagnetics Research B, Vol. 8, 2008 233
i 5
+k0 A2qk
+Aqk0 erfc ei 4 k0 Aqk0 e 4 0
. (126)
k0
Using the above derivations and the results for the direct eld
addressed in [8], the nal completed formulas for the three components
are obtained readily. They are
0 ik1 0 zd ik1 3 3i
E1 (, , z) = e 3
4k1 r1 r1 r1 r1 k1 r13
(2) (3)
+E1 + E1 (127)
2
0 ik1 r1 ik1 1 i zd
E1z (, , z) = e 2
4k1 r1 r1 k1 r13 r1
ik1 3 3i (2) (3)
2 3 + E1z + E1z (128)
r1 r1 k1 r
1
0 ik1 r1 ik1 1 (2) (3)
B1 (, , z) = e 2 + B1 + B1 . (129)
4 r1 r1 r1
2
10
|Ez| in V/m
1
10
0
10
10 20 30 40 50 60 70 80 90 100
in m
zero and the problem will reduces to that of the three-layered case.
For conveniences in evaluating the integrals including the reection
coecients Q and P , in this paper the terms of the ideal reected
wave are not separated with those of the lateral wave. Obviously, the
integrations along the branch cuts 1 and 0 includes the terms of
the ideal reected wave and the lateral wave. When the conditions
k1
1 and z + d are satised, the lateral waves with the wave
numbers being k0 and k1 can be excited eciently. Evidently, it is seen
that the lateral waves propagate in Region 0 along the boundary z = 0
and propagate in Region 1 along the boundary z = h.
In Figs. 35, for the components E1z , the total eld, the trapped
surface wave, and the DRL waves, which include the direct wave,
the reected wave, and the lateral wave, are computed and shown
in three cases of z = d = 0, k1 z = k1 d = 0.5, and k1 z = k1 d = 0.75,
respectively. In Figs. 79, the similar results for the components E1
are computed and shown, respectively. In Fig. 6, the total eld for the
component E1z is computed and shown in three cases of z = d = 0,
k1 z = k1 d = 0.5, and k1 z = k1 d = 0.75, respectively. Similar graphs
for the components E1 are shown in Fig. 10. Computations show that
there is a signicant contribution from the trapped surface wave for the
total eld in the four-layered region when both the dipole point and
the observation point are located in the upper dielectric layer under
the air.
Progress In Electromagnetics Research B, Vol. 8, 2008 235
3
10
The total field
The trapped surface wave
The DRL waves
2
10
|Ez| in V/m
1
10
0
10
10 20 30 40 50 60 70 80 90 100
in m
3
10
The total field
The trapped surface wave
The DRL waves
2
10
|Ez| in V/m
1
10
0
10
10 20 30 40 50 60 70 80 90 100
in m
|Ez| in V/m
2
10
1
10
10 20 30 40 50 60 70 80 90 100
in m
3
10
The total field
The trapped surface wave
The DRL waves
2
10
|E | in V/m
1
10
0
10
-1
10
10 20 30 40 50 60 70 80 90 100
in m
2
10
|E | in V/m
1
10
0
10
-1
10
10 20 30 40 50 60 70 80 90 100
in m
3
10
The total field
The trapped surface wave
The DRL waves
2
10
|E | in V/m
1
10
0
10
-1
10
10 20 30 40 50 60 70 80 90 100
in m
k z=k d=0
1 1
2.9 k1z=k1d=0.5
10
k z=k d=0.75
1 1
2.8
10
2.7
10
|E | in V/m
2.6
10
2.5
10
2.4
10
2.3
10
2.2
10
10 20 30 40 50 60 70 80 90 100
in m
Figure 10. The total elds E in V/m with f = 1 GHz, 1r = 2.65,
2r = 4, k1 h = 5, k2 l = 2 at three cases of z = d = 0, k1 z = k1 d = 0.5,
and k1 z = k1 d = 0.75.
5. CONCLUSIONS
REFERENCES