Progress In Electromagnetics Research, PIER 30, 157178, 2001

ANTENNA MEASUREMENTS IN THE PRESENCE OF

MULTIPATH WAVES
H.-J. Li, T.-Y. Liu, and J.-L. Leou
Department of Electical Engineering &
Graduate Institute of Communication Engineering
National Taiwan University
Taipei, Taiwan, R.O.C.
AbstractIn this paper antenna measurements in the presence of multipath waves are discussed. Methods are proposed to diagnose the
degradation sources of a compact range measurement system operating at a frequency much lower than the designed frequency range. A
range-gating technique is employed to improve the capability of the
compact range measurement system. With this technique, the eld responses over a bandwidth for each rotation angle are coherently measured, and the fast Fourier transform is applied to obtain the range
prole. A suitable window function is applied to extract the desired
path and eliminate all other undesired paths. The antenna pattern is
plotted according to the ltered response of the desired path. We have
expressed the receiving voltage of a testing antenna in terms of its gain
pattern and input impedance of the testing antenna operating in multipath environment. If the bandwidth of the testing antenna is very
narrow and the mismatch problem is very serious over the required
bandwidth, an algorithm is proposed to compensate the mismatch effect so that the obtained radiation pattern and the antenna gain are
more accurate. Numerical and experimental results have veried the
eectiveness of our method.

1.
2.
3.

Introduction
Diagnosis of Degradation Sources for a Compact-Range
Measurement System
Antenna Pattern Measurement in the Presence of
Multipaths

158

Li et al.

4.

Antenna Gain Measurement in the Presence of

Multipaths
5. Numerical and Experimental Results
5.1 Example 1
5.2 Example 2
5.3 Example 3
6. Discussion and Conclusion
References
1. INTRODUCTION
For antenna pattern or radar cross section (RCS) measurements, the
antenna under test or the object should be illuminated by a plane
wave. To obtain a plane wave illumination, the separation between
the testing antenna or object and the illumination source should satisfy the far-eld condition, 2D2 / , where D is the dimension of the
testing antenna or the object and is the operating wavelength. To
reduce the separation requirement, compact range systems are usually
implemented in an anechoic chamber [15]. The quiet zone, dened as
the area which is nearly uniformly illuminated, is used to characterize
the permissible largest dimension of the testing object.
There are several types of compact range systems. The compact
range system in the EM Lab of National Taiwan University is a Tworeector type and was built by the Chung-Shan Institute of Science
and Technology Taipei, Taiwan, ROC. This system is designed for antenna measurements for frequencies higher than 8 GHz . However, our
anechoic chamber is not only used for antenna measurements but also
used for scattering measurements, and the operating frequency can be
below 2 GHz . If the compact range system is not carefully operated
its presence in the anechoic chamber will introduce undesirable scattering and degrade the chamber performance. We will show how this
compact range system designed for antenna pattern measurements can
be used for scattering measurements and how to extend the operating
frequency to below 2 GHz .
In this paper we will examine how the measurement performance
is degraded when the system is used for scattering measurements and
when the operating frequency exceeds its designed range. We will employ microwave imaging technique [6, 7] to diagnose the cause of degradation and propose methods to improve the measurement capability
of the compact range system.

Antenna measurements in the presence of multipath waves

159

In Section 2 the basic properties of a compact range measurement

system and practical limitations will be described. The performance of
the compact range system when it is used for antenna and scattering
measurements will be compared. How the quiet zone is degraded when
the operating frequency is beyond the designed range will be demonstrated. The sources causing degradation will be diagnosed. In Section
3 we will discuss antenna measurement techniques in the presence of
multipath interference. We will express the received voltage response
in terms of the antenna input impedance and radiation pattern. In
Section 4 we will propose methods to dierentiate dierent multipaths
and propose algorithms to obtain the radiation pattern and antenna
gain of a testing antenna. Some numerical and experimental examples
will be given in Section 5. Discussion and conclusion will be given in
Section 6.
2. DIAGNOSIS OF DEGRADATION SOURCES FOR A
COMPACT-RANGE MEASUREMENT SYSTEM
The purpose of a compact-range system is to obtain uniform illumination with a distance much shorter than that required by the far-eld
condition. There are several dierent types of compact range measurement systems [13]. A sketch of the compact range system at the
EM Lab of the National Taiwan University is shown in Fig. 1. A
horn feeder is used to illuminate the rst reector. The wave is reected to the prime reector and then reected back to illuminate the
testing antenna. This compact-range system is designed to operate
at frequencies higher than 8 GHz . To examine the performance of
the compact-range system a waveguide-type probe is placed on a linear scanner to measure the magnitude and phase of the received eld.
Amplitude and phase of the elds received along a straight line are
shown by the solid curves in Fig. 2(a) and (b) respectively. It is seen
that the phase curve in Fig. 2(b) is almost linearly proportional to the
scanner position because the direction of wave arrival is slightly oset
to the normal of the scanner. If the linear variation is subtracted from
Fig. 2(b), the result is shown in Fig. 2(c). A quiet zone is dened as
the region where it is uniformly illuminated. From Fig. 2(a) and 2(c),
one can see that the quiet zone is in the center part with a range of
about 75 cm . In this region the amplitude variation is smaller than
1 dB and the phase variation is smaller than 5 .

160

Li et al.

Figure 1. Compact range system at the EM Lab of N.T.U.

(a)
Figure 2. Amplitude and phase distributions alone a linear scanner
for three dierent feeder positions (a) Amplitude.

Antenna measurements in the presence of multipath waves

161

(b)

(c)
Figure 2. Amplitude and phase distributions alone a linear scanner
for three dierent feeder positions (b) Phase, (c) Phase when the linear
variation of (b) is substrated.

162

Li et al.

If the compact range system is used for RCS measurement, the testing antenna is replaced by a testing object. To measure the backscattered eld, an additional horn antenna should be placed by the side of
the illuminating horn. In other words, two antennas are placed side
by side, one for transmitting and the other for receiving, and both are
oset to the original center position as shown in Fig. 1. With this oset
illumination, we again measure the magnitude and phase distribution
along the same straight line, and the results are also shown by the
dashed curves in Fig. 2. It is seen that with such a small oset in the
feeding position, the deviations in the magnitude and phase distribution are small. Therefore we conclude that the compact range system
can also be used for scattering measurement.
It is noted that the scanning of a probe along a linear line can be
viewed as a synthetic linear array. The direction of the wave arrival
can be obtained by taking the Fourier transform of the complex eld
received along the linear scanner. The derived directions of wave arrivals when the feeding horn is mounted at the center, at the left oset,
and at the right oset are respectively shown in Fig. 3. It is seen that
the directions of wave arrivals change accordingly.
If we reduce the operating frequency and repeat the eld measurement, the amplitude distributions along the line for several dierent
frequencies are shown in Fig. 4. It is seen that the lower the operating
frequency is, the larger the amplitude variation is. Notice that the
periodic variation in Fig. 4 is the result of interference of multipaths.
To see this, we can take the Fourier transform of the eld received
along the synthetic array for each frequency and obtain the direction
of arrival of each path. The results for several dierent frequencies are
shown in Fig. 5. It can be seen that the smaller the frequency, the
stronger the multipath components. A strong component comes from
the direction of 35 , which is in the direction of the feeding antenna.
In the above we determine the direction of wave arrival by taking
the Fourier transform of the elds received along the synthetic array.
We can further determine the distances of dierent paths by measuring the frequency response at a given location and then applying the
Fourier transform to the frequency response to obtain the range prole
[6, 7]. Fig. 6 shows the range prole obtained at the center frequency
2.5 GHz . The transmitting antenna is a double ridged wideband antenna, and the receiving antenna is a log-periodic antenna pointing

Antenna measurements in the presence of multipath waves

163

Figure 3. Directions of wave arrivals for three dierent feeding

positions.

Figure 4. The amplitude distributions for dierent operating

frequencies.

164

Li et al.

Figure 5. The direction of wave arrival for dierent frequencies.

Figure 6. Range proles with center frequency equal to 2.985 GHz .

Antenna measurements in the presence of multipath waves

165

Figure 7. Sketch of multipath of the compact range system for explanation of the range prole of Fig. 6.
to the main reector. The range prole provides the source of each
multipath. These multipaths are sketched in Fig. 7.
3. ANTENNA PATTERN MEASUREMENT IN THE
PRESENCE OF MULTIPATHS
In the last section we observed that the presence of multipaths would
limit the measurement capability and degrade the performance. In this
section we will rst express the received voltage in terms of the gain
pattern and input impedance of the receiving antenna in the presence
of multipaths and then propose a method to dierentiate multipaths
so that the antenna pattern and the gain can be accurately obtained.
Consider a receiving antenna with input impedance response Zin (k)
being connected through a transmission line (with characteristic
impedance Zo and length  ) to a receiver with load impedance ZL .
The equivalent circuit is shown in Fig. 8, where Voc (k) is the open
circuit of the receiving antenna. Assume ZL = Zo = Ro , the voltage

166

Li et al.

Figure 8. The equivalent circuit of the receiver.

at the receiver VL (k) can be expressed as
VL (k) =

Voc (k) Ro ej2kl

Zin (k) + Ro

(1)

When the antenna is illuminated by a eld incident from direction

(, ) , the open circuit voltage of the receiving antenna can be expressed as [8]


2 Rin (k)G(k, , )
 i (k, , )
(2)
Voc (k) = j
et (k, , ) E
1
k

where is the characteristic impedance of free space, Rin (k) is the

real part of Zin (k), G(k, , ) is the gain pattern of the receiving
antenna, e(k, , ) is the polarization unit vector of the antenna when
 i (k, , ) is the incident
it is used as a transmitting antenna, and E
eld at the antenna terminal.
If the antenna is illuminated by multipath waves the open circuit
voltage can be expressed as



2 Rin (k)G(k, m , m )
 i (k, m , m )
j
et (k, m , m ) E
Voc (k) =
m
k

m
(3)
where (m , m ) is the arrival direction of the mth component and
i (k, , ) is the mth illuminating eld at the terminal.
m
E
m
m

Antenna measurements in the presence of multipath waves

167

Let m be the distance of the mth path, and assume that the input
impedance, the radiation pattern, and the polarization unit vector are
nearly constant over the bandwidth of interest (which is much smaller
than the center frequency). Then Voc (k) can be approximated by


Rin (ko )G(ko , m , m )

am ejkrm

(4)
i

where am = et (ko , m , m ) Em (ko , m , m ) .
To dierentiate the multipath, we can measure VL (k) over a bandwidth and then apply the Fourier transform to obtain the range prole.
The bandwidth should be wide enough so that the desired path can be
resolved. With this procedure, the contributions by dierent paths or
the desired paths can be obtained by searching the peak positions and
peak values of the range prole.
If the antenna is rotated, the open circuit voltage will be

Voc (k) = Voc (ko + k)

=

 2
Voc (k, , ) j
ko

2
j
ko

Rin (ko )G(ko , +m , +m )

am (, )ejkrm

(5)
i

where am (, ) = et (ko , + m , + m ) Em (ko , + m , + m ) .
If we measure the frequency response VL (k, , ) over a bandwidth
for each rotation angle and take the Fourier transform, we obtain the
range prole. We can identify each path component by searching the
corresponding peak of each range prole. Note that amplitude of each
peak is proportional to the antenna pattern. By plotting the peak
value of a certain path versus the rotation angle we can obtain the
radiation pattern.
In the above derivation we have assumed that the radiation pattern
and the input impedance are independent of frequency. If the assumption is not valid, then the obtained radiation pattern is the mean
pattern averaged over the bandwidth. For example, the null position
of a large array or a large aperture antenna is a function of frequency.
If the bandwidth used in the above procedure is too large, the null
position obtained will not be as deep as that obtained at a single frequency. In that case we can apply following algorithm to obtain the
radiation pattern for each certain frequency:

168

Li et al.

1. Choose a suitable bandwidth so that we can dierentiate the

desirable path from the other paths.
2. For each rotation angle, measure the frequency response VL (k)
over the bandwidth.
3. Take the Fourier transform of VL (k) to obtain the range prole
at each rotation angle.
4. Multiply the range prole with a suitable window function centered at the peak position of the desired path so that only the
desired path can pass through and all other multipaths are attenuated.
5. Take the Inverse Fourier transform of the range prole of the
desired path to obtain the ltered voltage response VL (k, , )
for each rotation angle.
6. The radiation pattern for certain frequency k can be obtained
by plotting VL (k, , ) versus rotation angle.
4. ANTENNA GAIN MEASUREMENT IN THE
PRESENCE OF MULTIPATHS
The antenna gain of a testing antenna is usually obtained by comparing
the voltages received by the testing antenna and by a standard antenna
with a known gain value. In the presence of multipaths, the antenna
gain can be obtained by the following procedure.
1. Replace the testing antenna by a standard antenna with a known
gain value.
2. Measure the frequency response. The bandwidth and the frequency points must be the same as those used in measuring the
testing antenna. Apply the inverse Fourier transform on the frequency response to obtain the range prole.
3. Record the peak value of the desired path.
4. The gain of the testing antenna is obtained by taking the ratio of
the peak value obtained in the case of testing antenna over the
peak value obtained in the case of standard antenna.
5. Or apply the same window function to retain the desired path
and eliminate all other paths, and then take the inverse Fourier
transform to obtain the ltered frequency response. The gain of
the testing antenna at a certain frequency is obtained by comparing the ratio of the two ltered responses at that frequency.

Antenna measurements in the presence of multipath waves

169

If the testing antenna is a narrowband antenna, there will be a mismatch between the antenna and the receiver over the bandwidth. It
is also noted that in Eq. (1) and Eq. (2) the received voltage VL (k)
is a function of the antenna input impedance Zin (k) . A mismatch
in impedance will reduce the load voltage. If we apply the Fourier
transform to the frequency response VL (k) to obtain the range prole, the range resolution will become poorer and the peak value will
decrease. If the bandwidth is too narrow, it may not be able to resolve
the desired path and the derived antenna gain value can be inaccurate.
The reection coecient |(k)| of a testing antenna usually can be
measured. It is known that the power delivered to the receiver will be
reduced by a factor of 1 |(k)|2 due to a mismatched impedance [3].
1/2

If we correct the measured VL (k) with a factor of 1/ 1 |(k)|2
and then follow the procedure described in the previous section, we
can obtain a more accurate measurement of the range prole and the
antenna pattern and the gam value.
5. NUMERICAL AND EXPERIMENTAL RESULTS
In this section we will show some numerical and experimental examples
to demonstrate the eectiveness of the proposed method.
In our anechoic chamber we mount a directional antenna on a pedestal which can be rotated, and measure the frequency response over
a frequency bandwidth for each rotation angle. We then apply the
Fourier transform to each frequency response and use the window
method to isolate each path component, and then plot the receiving
pattern of each path component. From the pattern plots we can obtain the amplitude, the time delay, and the arrival direction of each
path component for the imperfect compact range measurement system.
The parameters obtained for center frequency 2.985 GHz are shown
in Table 1.
5.1 Example 1
These parameters are then being used for the following simulations.
Assume that the testing antenna is a linear broadside antenna array
with spacing d and a total of N + 1 elements. The array factor can
be expressed as
AF (k, ) =

sin(kN d cos )
kd cos

(6)

170

Li et al.

Table. 1 Path parameters of F c = 2.985 Ghz .

Paths

Time-Delay (nsec)

Amplitude

Arrival-Angle (deg.)

1
2
3
4

122
133
151
187

1
0.092
0.92
0.0366

40
-20
0
190

Assume that the input impedance of the antenna array is matched to

that of the receiver over the frequency bandwidth. The eld received
at frequency k and rotation angle can be expressed as [9]
E(k, ) =

4

i=1

ai

sin[kN d cos( i )]
kd cos( i )

(7)

The theoretical value of AF (k, ) versus for f = 2.985 GHz, d =

5 cm and N = 21 is shown in Fig. 9(a). The plot of E(k, ) versus
for f = 2.985 GHz is shown if Fig. 9(b). It is seen that the existence
of multipath distorts the radiation pattern. For each rotation angle we
calculate the frequency response over a bandwidth of 500 MHz , apply
the inverse Fourier transform to obtain the range prole, and nd the
peak value of the desired path. We plot the peak value versus rotation
angle and the resultant antenna pattern is shown in Fig. 9(c). It is
seen that the nulls of Fig. 9(c) are not as deep as those in Fig. 9(a) due
to the averaging eect. We conclude that a bandwidth of 500 MHz
(1/6 of the center frequency) is too broad for obtaining such an array
pattern. If we reduce the bandwidth to 100 MHz (it is wide enough
to resolve all path components in the example), the result is shown in
Fig. 9(b). It is seen that the pattern of Fig. 9(b) is much closer to that
of Fig. 9(a).
5.2 Example 2
Next we demonstrate the eect of impedance mismatch on the antenna measurements. Fig. 10(a) shows the return loss of a real microstrip antenna element. The corresponding input impedance responses (real part and imaginary part) are shown in Fig. 10(b). We
assume that the antenna array of the previous example has the same
input impedance response as that of Fig. 10(b), where the bandwidth

Antenna measurements in the presence of multipath waves

171

(a)

(b)
Figure 9. Antenna patterns of a linear array (a) Theoretical pattern,
(b) Pattern obtained in the presence of multipath.

172

Li et al.

(c)

(d)
Figure 9. Antenna patterns of a linear array (c) Pattern obtained
after range gating with B.W. = 500 MHz , (d) Pattern obtained after
gating with B.W. = 100 MHz .

Antenna measurements in the presence of multipath waves

173

(a)
Figure 10a. The return loss of a microstrip antenna element.

(b)
Figure 10b. The input impedance (real part and imaginary part) of
a microstrip antenna element.

174

Li et al.

Figure 11. Pattern obtained after range gating with input impedance
of Fig. 10.

Figure 12. Input impedance with narrowed down bandwidth of

Fig. 10.

Antenna measurements in the presence of multipath waves

175

Figure 13. Pattern obtained after range gating with input impedance
of Fig. 12.
(dened as the region with return loss 10 dB) is about 60 MHz .
We then calculate the frequency response VL (k) over a bandwidth of
100 MHz , and apply the window algorithm to nd the radiation pattern and the gain value. The resultant pattern is shown in Fig. 11. The
ratio of the two maximum values of Fig. 9(b) and Fig. 11, corresponding to the impedance matching and mismatching cases respectively, is
0.9724. It is seen that an impedance response like Fig. 10(b) doesnt
aect the measurement performance.
If we articially reduce the bandwidth to 1/4 of the original value
and assume that the impedance response is as shown in Fig. 12, then
repeating the same procedure we obtain the radiation pattern shown in
Fig. 13 and the ratio of the two maximum values of Fig. 9(b) and Fig. 13
becomes 0.6833. It is seen the gain value decreases by a factor of 3.3 dB
due to averaging eect of the mismatch loss. If we compensate the

1/2
voltage response VL (f ) by multiplying a factor of 1/ 1 |(k)|2
for each sampled frequency point, the resultant ratio becomes 0.937,
which is closer to the ideal value of 1.

176

Li et al.

(a)

(b)
Figure 14. (a) Photo of a single-ridge horn. (b) The E-plane radiation
patterns of (a) obtained without and with range gating at 2.5 GHz .

Antenna measurements in the presence of multipath waves

177

5.3 Example 3
Finally we demonstrate an experimental example. Fig. 14(a) shows
a ridge horn antenna manufactured by the Chung-Shan Institute of
Science and Technology. The measured patterns obtained without
and with range gating for the E-plane with center frequency equal
to 2.5 GHz are shown in Fig. 14(b). It is seen that after windowing,
the multipath eect has been eliminated.
6. DISCUSSION AND CONCLUSION
In this paper we have studied antenna measurements in the presence
of multipath waves. We propose methods to diagnose the degradation
sources of our compact range measurement system when it is operated
in frequency much lower than the designed frequency range. We then
propose a window technique to improve the measurement capability
and show that the operating frequency of the compact range system
can be extended to below 2 GHz without having to change the feeder
horn and satisfactory performance can still be obtained.
With this technique, the eld responses over a bandwidth for each
rotation angle are coherently measured, and the fast Fourier transform
is applied to obtain the range prole. A suitable window function
is then applied to extract the desired path and eliminate all other
paths. The antenna pattern is nally plotted according to the ltered
response of the desired path. It is noted that the required bandwidth
depends on the separation between the desired path and its closest
adjacent path. We have expressed the receiving voltage in terms of the
gain pattern and input impedance of the testing antenna. However, if
the bandwidth of the testing antenna is very narrow and the mismatch
problem is very serious over the required bandwidth, then the radiation
pattern and the antenna gain obtained can be inaccurate. We have
proposed a compensation algorithm so that this mismatch eect can
be reduced and the radiation pattern and the antenna gain can be more
accurately obtained. Numerical and experimental results have veried
the eectiveness of our method.

178

Li et al.

REFERENCES
1. Evans, G. E., Antenna Measurement Techniques, Artech House,
Inc., 1990.
2. Hollis, J. S., T. J. Lyon, and L. Clayton, Jr., eds., Microwave
Antenna Measurements, Scientic Atlanta, Atlanta, 1970.
3. LO, Y. T., and S. W. Lee, Antenna Handbook, Van Nostrand
Reinhold Company, New York, 1988.
4. Johnson, R. C., ed., Compact ranges, Joint Report of TMAT
and IEEE A.P.S., Philadelphia, June 1986.
5. Burnside, W. D., Dual chamber design reduces quiet zone ripple/taper errors, Microwave and RF, 168174, May 1987.
6. Li, H. J., N. H. Farhat, Y. Shen, and C. L. Werner, Image understanding and interpretation in microwave diversity imaging,
IEEE Trans. Antennas Propagate, Vol. AP-37, No. 8, 10481057,
1989.
7. Li, H. J., N. H. Farhat, and Y. Shen, Image interpretation and
prediction in microwave diversity imaging, IEEE Trans. on Geoscience and Remote Sensing, 27, 98101, 1989.
8. Kiang, Y. W., Antenna receiving characteristics, Private note,
Department of Electrical Engineering, National Taiwan University.
9. Elliott, R. S., Antenna theory and design, Prentice-Hall, Inc.,
1981.