You are on page 1of 5

REFERENCES

1. INTRODUCTION

1. D.M.J. Devasirvatham, Time delay spread measurements of wideband


radio signals within a building, Electron Lett 20 (1984), 950 951.
2. D.M.J. Devasirvatham, Time delay spread and signal level measurements of 850 MHz radio waves in building environments, IEEE Trans
Antennas Propagat AP-34, pp. 1330 1335, Nov. 1986.
3. A.A.M. Saleh and R.A. Valenzuela, A statistical model for indoor
multipath propagation, IEEE J Select Areas Commun SAC-5 (1987),
138 145.
4. D.M.J. Devasirvatham, Multipath time delay spread in the digital portable radio environment, IEEE Commun Mag 25 (1987), 1321.
5. T.S. Rappaport, Characterization of UHF multipath radio propagation
inside factory buildings, IEEE Trans Antennas Propagat AP-37 (1989),
1058 1069.
6. Y.P. Zhang, Y. Hwang, and P.C. Ching, Wideband UHF radio propagation characteristics in a tunnel environment, Int J Wireless Personal
Commun 8 (1998), 291299.
7. K. Tai, R.J. Fischer, K.W. Wang, S.N.G. Chu, and A.Y. Cho, Frequency
response and path loss measurements of indoor channel, Electron Lett
27 (1991), 10211022.

Research about estimation of directions of arrival (DOA) has


undergone great development in recent years. These works have
been focused on many aspects: methods of estimation, statistical
performance of these methods, solutions for narrow- and broadband signals, results for considering mutual coupling between
antennas, et cetera. One important topic in this context is the
derivation of the explicit analytic CramerRao bound (CRB). The
CRB provides a lower bound on the variance of any unbiased
estimator and it is well known that, under mild regularity conditions, it is the asymptotic covariance matrix of the maximumlikelihood estimator.
Previous articles have shown that the mutual coupling affects
the estimation of arrival directions [19], and that these effects can
be corrected [5 8]. The CRB considering mutual coupling can be
straightforwardly obtained with the use of numerical methods
[13] or calibration methods [9 12] to model mutual coupling
effects. However, in these previous works, in which the electromagnetic (EM) coupling with the platform is neglected, the CRB
presents only minor differences when it is compared to the free
coupling case.
In this Letter the goal is to obtain the CRB considering both
mutual coupling between array elements and also electromagnetic
coupling with the mounting platform of the antenna. A 3D method
of moments (MoM) [13, 14] has been used to model both coupling
effects. Starting with the MoM matrix formulations, closed expressions for the CRB have been obtained in a direct way. Alternative
simpler expressions could have been obtained with the use of
calibration matrices (which implies consideration of only mutual
coupling effects with the EM coupling with the platform neglected).
However, the formulation proposed here is much more accurate.

2002 Wiley Periodicals, Inc.

THE CRAMERRAO BOUND FOR THE


ESTIMATION OF ANGLES OF ARRIVAL
IN ON-BOARD ARRAY ANTENNAS
I. T. Castro,1 L. Landesa,2 J. M. Taboada,3 and F. Obelleiro3
1
Dpto. Matema ticas
Facultad de Ciencias
Universidad de Extremadura
Avda Elvas s/n 06071
Badajoz, Spain
2
Dpto. Informa tica
Escuela Polite cnica de Ca ceres
Universidad de Extremadura
Avda de la Universidad s/n
10071 Ca ceres, Spain
3
Dpto. Tecnoloxas das Comunicacions
Universidade de Vigo
ETSE Telecomunicacion
Campus Universitario s/n
36200 Vigo, Spain

2. STATEMENT OF THE PROBLEM

The problem of finding the direction of arrival (DOA) of N plane


waves using a narrowband array of N a sensors can be reduced to
the estimation of the parameters in the following model:
yt A xt nt,

t 1, 2, . . . N s,

(1)

where y(t) N a1 are the vectors of array outputs, x(t)


N 1 are the source signals, n(t) N a1 is a noise process, N s
denotes the number of data samples, and is the vector of arrival
directions to be estimated:

Received 10 October 2001

1, 2, . . . , N T,
ABSTRACT: Array antennas mounted on aircraft, ships, or any other
complex platform suffer from mutual coupling and coupling with the
platform in the transmission or reception of electromagnetic waves.
These coupling effects affect the estimation of the direction of arrival
(DOA) of plane waves performed with on-board antennas. With the use
of the method of moments (MoM) to model these electromagnetic (EM)
coupling terms, closed-matrix expressions for the CramerRao bound
(CRB) are obtained for on-board array antennas considering both mutual coupling and platform coupling effects. Some significant differences
in the CRB have been found when this bound is compared with the same
bound calculated in an ideal free coupling environment. Numerical results are shown to illustrate these effects. 2002 Wiley Periodicals,
Inc. Microwave Opt Technol Lett 33: 119 123, 2002; Published online
in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.
10249

Key words: CramerRao bound (CRB); directions of arrival (DOA);


method of moments (MoM); mutual coupling; on-board antennas

(2)

where the superscript T indicates the transpose of the matrix.


The matrix A( ) N aN consists of N vectors (steering
vectors) as follows
A a 1, a 2, . . . , a N,

(3)

where the ith element of a( k ) corresponds to the response of the


ith sensor to a plane wave impinging from direction k . Sometimes, the matrix A( ) is denoted simply as A.
3. THE CRAMERRAO BOUND

Under the assumptions that the aforementioned signals are temporally white and Gaussian distributed and the sensor noise is both
spatially and temporally white and Gaussian distributed, the array
antenna output is also temporally white and Gaussian distributed.
Noting

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 33, No. 2, April 20 2002

119

xt N0, P,

(4)

nt N0, 2I,

(5)

yt N0, R,

(6)

where Z N MoMN MoM is the impedance matrix of the MoM


solution, i N MoM1 is a column vector containing the coefficients of the MoM basis functions, and v N MoM1 is the
excitation vector closely related to the electromagnetic fields due
to the impinging plane waves. For a given excitation v, coefficients
i are obtained as:

being

i Z 1v.
P E xt x Ht,

(7)

R E yt y t,

(8)

where the superscript H indicates the conjugate transpose of the


matrix. With the use of (1), R can be written as
R A PA H 2I.

(9)

In what follows, I represents the identity matrix of appropriate


orders. In the above expressions, the matrix P, the noise variance
2 and the set of arrival directions [ 1 , 2 , . . . , N ] are
unknown parameters. These parameters can be estimated with the
use of the maximum-likelihood method [15]. It is well known that
the asymptotic covariance matrix of the maximum-likelihood estimator leads to the CRB under some regularity conditions. It can
show that the closed-form expression for CRB() matrix is given
by [1517]:
CRB

2
ReA H A A J PAH R1 APT 1 ,
2Ns

a k

da k
.
dk

i a Ci,

(15)

C I NaNa0 N aN MoMNa.

(16)

v v 1 v 2 v N

(17)

where

Let

represent the excitation v due to N plane waves where v( i )


represents the excitation v due to a plane wave impinging from the
i direction. S( ) denotes the following matrix:
S v 1, v 2, . . . , v N,

(11)

whose columns v ( i ) represent the excitation v due to a plane


wave of magnitude 1V/m impinging from the i direction.
For the sake of simplicity, it will be assumed that the wave
impinges parallel to the XY plane. Under this assumption, the
incident electric field due to a plane wave of 1V/m magnitude
impinging from the 0 direction can be expressed as

and J represents the SchurHadamard product. Matrix A is composed of the derivatives of steering vectors with respect to the
arrival directions as follows:
A a 1, a 2, . . . , a N,

Employing a basis functions in each sensor output and ordering


the basis functions in a given orderfirst the basis corresponding
to the array antenna outputs and second the othersthe vector
output of the antenna i a N a1 can be obtained as

(10)

where A is the orthogonal projection matrix of A:


A I A A HA 1A H,

(12)

For the evaluation of the CRB, two different subsets of parameters


are needed, ones related with the signals (matrix P, noise variance
2, and number of snapshots N s ) and the rest related to the array
antenna response [A( ) and A ( )]. These last parameters can be
evaluated in on-board array antennas with the use of the method of
moments (MoM). The following two sections indicate how we can
obtain closed expressions for the evaluation of A( ) and A ( ). A
similar MoM matrices treatment has been addressed in [18, 19] in
order to include the electromagnetic coupling effects in the synthesis of on-board array antennas.

i e e jk 0 x cos 0y sin 0
E

v i k

120

w
i x, y, z e e jk 0 x cos ky sin k dri ,

(13)

(19)

(20)

ri

where r i is the domain corresponding to the ith weighting function


of the MoM and w
i ( x, y, z) corresponds to the ith weighting
function of the MoM formulation.
Then
v S xt

Considering an electromagnetic problem (e.g., an array antenna


mounted on a complex platform), the method of moments solves a
equation integral expanding current along the structure in a series
of N MoM basis functions. After some algebra involving a scalar
product with a set of weighting functions adequately chosen, the
MoM formulation is reduced to the following matrix expression
[13]:

(18)

where e is the unit polarization vector, k 0 is the wave number, and


0 is the angle between the incident direction and the X axis.
With this criterion, the ith element of v ( k ) can be obtained as

4. EXPRESSIONS OF THE MATRIX A()

Zi v,

(14)

(21)

can be expressed. With the use of Eqs. (14), (15), and (21),
i a CZ 1v CZ 1S xt

(22)

can be obtained. Therefore, the signal y(t) can be written as y(t)


i a n(t) and
yt CZ 1S xt nt A xt nt,

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 33, No. 2, April 20 2002

(23)

which can be written as


s ik jk 0 x isin k yi cos ksik ,

(30)

where x i and y i are the coordinates of the centroid of each r i


surface. With the use of the last equation, we
S S J G

(31)

G g ik

(32)

is obtained, with

Figure 1 On-board array antenna of 19 monopoles attached to the stern


of a ship

being
g ik jk 0 x isin k yi cos k ,

1 i NMoM,

1 k N .

and then the expression for the matrix A( ) in function of MoM


formulation is obtained:
A CZ 1S .

(24)

For the evaluation of the matrix A ( ) Eq. (12) needs to be


evaluated with the use of (24). In (24), matrix C and impedance
matrix Z are independent of the arrival angle. However, matrix
S( ), given by Eq. (18), depends on a nontrivial way of the arrival
angles .
The matrix S ( ) can be defined as
i 1, . . . , N MoM,

k 1, . . . , N ,
(25)

s ik

With the use of Eq. (24),


A CZ 1 S CZ 1S J G

5. EXPRESSIONS OF THE MATRIX A()

S s ik,

(33)

ds ik dv i k

,
dk
dk

(26)

(34)

is obtained.
6. NUMERICAL RESULTS

One example of an on-board array antenna is given above. How


the structure affects the CRB will be shown. To evaluate the CRB
Eqs. (10), (24), and (34) are used. For the results of this article the
RaoWiltonGlisson triangular basis functions have been chosen
for both the basis expansions and the weighting functions [14, 20,
21]. The example corresponds to a 19-element array antenna
mounted on the stern of a ship operating at a frequency of 300
MHz. Antenna elements correspond to /4 monopoles with /2
separation. Figure 1 represents the on-board array antenna. In
Figure 2, the CRB of the estimation of two arrival plane waves

where s ik represents the (i, k) element of the matrix S( ). Derivation of Eq. (20) leads to
s ik

d
dk

w
i x, y, z e e jk 0 x cos ky sin k dri .

(27)

ri

Analytic evaluation of Eq. (27) leads to:


s ik jk 0 sin k

w
i x, y, z e e jk 0 x cos ky sin k x dri

ri

jk 0cos k

w
i x, y, z e e jk 0 x cos ky sin k y dri .

(28)

ri

This expression can be calculated in a similar manner to S( ).


However, it is possible to obtain a good approximation when the
subdomain basis functions for the MoM are used. In this case the
surfaces associated with each weighting function have a negligible
variation of its coordinates and (28) can be approximated as
s ik jk 0 x isin k yi cos k

ri

w
i x, y, z e e jk 0 x cos ky sin k dri ,

(29)

Figure 2 CRB versus signal-to-noise ratio for the example of Figure 1


for two equal power uncorrelated signals. 1 2 degrees, 2 9 degrees,
N s 100 snapshots. Solid line, CRB for 1 in the on-board array antenna;
dashed line, CRB for 2 in the on-board array antenna; dash-dot line, CRB
for 1 in an isotropic conformal array antenna; dotted line, CRB for 2 in
an isotropic conformal array antenna. In this figure, the dash-dot and dotted
lines are superposed

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 33, No. 2, April 20 2002

121

Figure 3 CRB versus 2 angle for the example of Figure 1 for two
equal-power uncorrelated signals. 1 5, SNR 10 dB, N s 100
snapshots. Solid line, CRB for 1 in the on-board array antenna; dashed
line, CRB for 2 in the on-board array antenna; dash-dot line, CRB for 1
in an isotropic conformal array antenna; dotted line, CRB for 2 in an
isotropic conformal array antenna

Figure 4 CRB versus correlation coefficient magnitude for the example


of Figure 1 for two equal-power correlated signals with 0. 1 2
and 2 5, SNR 10 dB, N s 100 snapshots. Solid line, CRB for 1
in the on-board array antenna; dashed line, CRB for 2 in the on-board
array antenna; dash-dot line, CRB for 1 in an isotropic conformal array
antenna; dotted line, CRB for 2 in an isotropic conformal array antenna

impinging from 1 2 and 2 9 degrees is represented with


respect to the signal-to-noise ratio. z polarization is assumed; N s
100 snapshots for the estimation and uncorrelated sources for the
two plane waves. In the same figure, the CRB of 1 and 2 is
compared with the same CRB for an ideal conformal array antenna
of isotropic elements located in the same origins as Figure 1,
without considering either mutual coupling or coupling with the
platform effects. Figure 3 considers the same example for an
impinging wave of 1 5 degrees when the angle of the second
plane wave varies. The signal-to-noise ratio was fixed to 10 dB. In
Figures 2 and 3, it can be seen that the effect of the platform and
the mutual coupling leads to an increment of the CRB for the
on-board array antenna. This increment produces a variance
greater to the predicted variance when the inherent electromagnetic effects in on-board array antennas are not considered. The
increment is not drastic, but it is not negligible.
Nevertheless, it is not possible to state that the platform coupling effects increment the CramerRao bound. In some cases,
when the signals impinging to the array antenna are correlated, it
can be shown that the CRB may be lower than in the free electromagnetic coupling case. Figure 4 represents the CRB of the
estimation of two arrival plane waves impinging from 1 2 and
2 5 degrees with respect to the correlation-coefficient magnitude between the two signals. It can be seen that for a large
correlation-coefficient magnitude the CRB of the on-board array
antenna is lower than the ideal free coupling case. Figure 5 shows
the same example for the case of correlation coefficient magnitude
1 with the correlation-coefficient phase varied; as in the
previous example it was found that the CRB of the on-board
antenna presents lower variation than in the free coupling case.
Finally, Figure 6 represents the CRB of the estimation of a
plane wave impinging from 1 5 degrees with respect to the
angle of a second plane wave impinging from the 2 direction
when the two signals are fully correlated ( 1). Comparison of
this result with Figure 3 shows two important effects:

2. When the arrival directions are closed, the CRB of correlated signals increases compared to the uncorrelated case,
but if the arrival directions are separated enough, there are
minor differences between CRB for correlated and uncorrelated signals.

1. The CRB of the on-board array antenna is lower if the


directions of arrivals are near.

122

7. CONCLUSIONS

The stochastic CramerRao bound for on-board array antennas


have been derived in a closed form. Terms related to the steering
vectors and its derivatives are determined with the use of matrices
related to the method of moments. This formulation permits the
CRB of an arbitrary array antenna configuration mounted on an

Figure 5 CRB versus correlation coefficient phase for the example of


Figure 1 for two equal-power correlated signals with correlation coefficient
magnitude of 1. 1 2 and 2 5, SNR 10 dB, N s 100
snapshots. Solid line, CRB for 1 in the on-board array antenna; dashed
line, CRB for 2 in the on-board array antenna; dash-dot line, CRB for 1
in an isotropic conformal array antenna; dotted line, CRB for 2 in an
isotropic conformal array antenna

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 33, No. 2, April 20 2002

Figure 6 CRB versus 2 angle for the example of Figure 1 for two
equal-power correlated signals with correlation coefficient 1. 1 5,
SNR 10 dB, N s 100 snapshots. Solid line, CRB for 1 in the on-board
array antenna; dashed line, CRB for 1 in an isotropic conformal array
antenna

arbitrary platform to be obtained, with both the mutual coupling


between array antennas and their interaction with the mounting
platform taken into account. Results show that these electromagnetic effects cause nonnegligible differences in the CRB. In general, CRBs for mounted antennas are greater than for ideal antennas, but in some cases (such as large correlation between signals)
CRB may be lower.

11. C.M.S. See, Sensor array calibration in the presence of mutual coupling and unknown sensor gains and phases, Electron Lett 30 (1994),
373374.
12. I.S.D. Solomon, D.A. Gray, Y.I. Abramovich, and S.J. Anderson,
Receiver array calibration using disparate sources, IEEE Trans Antennas Propagat AP-47 (1999), 496 505.
13. R.F. Harrington, Field computation by moment methods, IEEE Press,
New York, 1993.
14. J.M. Taboada, Estudio de compatibilidad electromagne tica e interferencias en sistemas radiantes a bordo de estructuras complejas, Ph. D
dissertation, University of Vigo, 2001.
15. P. Stoica and A. Nehorai, Performance study of conditional and
unconditional direction-of-arrival estimation, IEEE Trans Acoustics,
Speech Signal Process ASSP-38 (1999), 17831795.
16. A.J. Weiss and B. Friedlander, On the Crame rRao bound for direction
finding of correlated signals, IEEE Trans Signal Process SP-41 (1993),
495 499.
17. P. Stoica, E.G. Larsson, and A.B. Gershman, The stochastic CRB for
array processing: A textbook derivation, IEEE Signal Proc Lett SPL-8
(2000), 148 150.
18. F. Obelleiro, L. Landesa, and J.M. Taboada, A method-of-moments
based algorithm to synthesize a conformal onboard array antenna,
Microwave Opt Technol Lett 29 (2001), 324 328.
19. F. Obelleiro, L. Landesa, J.M. Taboada, and J.L. Rodrguez, Synthesis
of onboard array antennas including interaction with the mounting
platform and mutual coupling effects, IEEE Antennas Propagat Mag
43 (2001), 76 82.
20. S.M. Rao, D.R. Wilton, and A.W. Glisson, Electromagnetic scattering
by surfaces of arbitrary shape, IEEE Trans Antennas Propagat AP-30
(1982), 409 418.
21. S.U. Hwu and D.R. Wilton, Electromagnetic scattering and radiation
by arbitrary configuration of conducting bodies and wires, Tech Rep.
No. 87-17, San Diego State University, May 1998
2002 Wiley Periodicals, Inc.

REFERENCES
1. T. Svantesson, The effects of mutual coupling using a linear array of
thin dipoles of finite length, In Proc. 9th IEEE Signal Processing
Workshop on Statistical Signals and Array Processing, 1998, pp.
232235.
2. T. Svantesson, Modeling and estimation of mutual coupling in a
uniform linear array of dipoles, Proc. IEEE Int. Conference on Acoustics, Speech, and Signal Processing, 1999, pp. 29612964.
3. A.J. Weiss and B. Friedlander, Mutual coupling effects on phase-only
direction finding, IEEE Trans Antennas Propagat AP-40 (1992), 535
541.
4. B. Friedlander and A.J. Weiss, Direction finding in the presence of
mutual coupling, IEEE Trans Antennas Propagat AP-39 (1991), 273
284.
5. K.R. Dandekar, H. Ling, and G. Xu, Effect of mutual coupling on
direction finding in smart antenna applications, Electron Lett 36
(2000), 1889 1891.
6. T. Su, K.R. Dandekar, and H. Ling, Simulation of mutual coupling
effect in circular arrays for direction finding applications, Microwave
Opt Technol Lett 26 (2000), 331336.
7. T. Su and H. Ling, On modeling mutual coupling in antenna arrays
using the coupling matrix, Microwave Opt Technol Lett 28 (2001),
231237.
8. R. Adve and T. Sarkar, Compensation for the effects of mutual
coupling on direct data domain adaptive algorithms, IEEE Trans
Antennas Propagat AP-48 (2000), 86 94.
9. K.R. Dandekar, H. Ling, and G. Xu, Smart antenna array calibration
procedure including amplitude and phase mismatch and mutual coupling effects, 2000 IEEE International Conference on Personal Wireless Communications, 2000, 293297.
10. B.C. Ng and C.M.S. See, Sensor-array calibration using a maximumlikelihood approach, IEEE Trans Antennas Propagat AP-44 (1996),
827 835.

BROADBAND PROPERTY AND TIMEDOMAIN APPLICATION OF GENERAL


TRANSMISSION-LINE EQUATIONS ON
MICROSTRIP LOW-PASS FILTER
Xiaolong Zhong, Yaowu Liu, and Kenneth K. Mei
Department of Electronic Engineering
City University of Hong Kong

Received 11 October 2001

ABSTRACT: General transmission-line equations based on the theory


of differential equations [1] were introduced in [2] and [3]. The equations dispersion characteristic of a microstrip low-pass filter is studied.
The parameters L, C, , and of general transmission-line equations
are found to be almost invariant to frequency in the passband of the
microstrip low-pass filter. By restoring current and voltage distributions
with the use of equations in both frequency and time domains, the
broadband property of these equations on the microstrip low-pass filter
is evident. 2002 Wiley Periodicals, Inc. Microwave Opt Technol
Lett 33: 123126, 2002; Published online in Wiley InterScience (www.
interscience.wiley.com). DOI 10.1002/mop.10250

Key words: General transmission-line equations; MoM; FDTD; microstrip low-pass filter

MICROWAVE AND OPTICAL TECHNOLOGY LETTERS / Vol. 33, No. 2, April 20 2002

123

You might also like