investigation into the limitations and …signal to jamming ratio (sjr). this study experiments on...
TRANSCRIPT
1
INVESTIGATION INTO THE LIMITATIONS AND EFFECT OF PARAMETERS OF
GSC ON RADAR ARRAY PERFORMANCE
Wu Weiming¹, Edrea Tan Yi Lin¹, Yeoh Wee Soon²
¹Raffles Girls’ School, 20 Anderson Road, Singapore 259978
² DSO National Laboratories, 20 Science Park Drive, Singapore 118230
ABSTRACT
The performance of radar systems can be greatly hindered with the presence of radar
interference. A wellestablished electronic countercountermeasure (ECCM) technique is
the sidelobe canceller approach. Modelled in MATLAB, the adaptive array processing
algorithm used in this study is the Generalized Sidelobe Canceller (GSC), which optimally
estimates the parameters for combining data from an array of sensors to improve the target
signaltojamming ratio (SJR). This study experiments on the limitations and ability of the
radar system when faced with different external factors, namely jammer power and angle of
jammer attack. As an extension, an internal factor is also studied, namely the spacing
between the array sensors, so as to observe how the performance of a radar system can be
improved and optimized where possible. Results are quantified in terms of the SJR of
output signal before and after the introduction of GSC and through the nulltonull
beamwidth of the array. The results of this study show that the application of GSC is an
effective ECCM technique that can be used even with substantial change in external factors. It
is also discovered that λ/2 is the ideal parameter for spacing between array elements for this
particular array.
INTRODUCTION
Radar has been employed prominently and extensively in the field of defence. Through
the transmission and receiving of radio signals in the form of electromagnetic waves, the
presence and type of objects can be detected with the use of radar [Villiard, O.G., Jr., 1976].
However, the performance of radar systems can be limited by noise and other sources of
interference. Radar jamming and deception, also known as electronic countermeasure
(ECM),
hinder its performance by saturating the receiver with noise or false information [Corcoran,
A, 2009].
To minimise the amount of interference that a jammer can introduce to a radar receiver,
electronic countercountermeasure is utilized. The application of an ECCM, such as a
general sidelobe canceller, is a technique used to reduce the effect of jammers received
through the sidelobe of a radar system. It is essentially an adaptive array capable of
automatically sensing and reducing, or even eliminating unwanted signals entering the radar’s
field of view, in particular its sidelobes, while still enhancing reception of the desired target
returns. By using the signals received through additional auxiliary antennas, the radar system is
able to cancel incoming interfering signals, hence displaying only the desired target signal
[Budge, 2007].
2
Diagram 1. An illustration of Generalized Sidelobe Canceller.
Diagram 1 illustrates the GSC system used for this study. The setup of such a radar
system consists of external aerials (auxiliary antennas), placed near the main antenna (main
channel) [Bucciarelli, 1984]. The main antenna is pointed towards the desired signal
angular location while the interfering signals can often be detected somewhere in the
sidelobes. This configuration means that the main antenna would receive returns from both
the desired and interfering signals [Taylor and Francis Group, 2009]. However, as jamming
power is generally
much stronger than the target signal, the auxiliary antenna returns are primarily from
the interfering signal.
The signals received through the auxiliary array antennas are multiplied by proper weights
and then summed obtaining an estimate of the jammer signal received through the radar
sidelobes. The jammer estimate is then subtracted from the radar output. With the feedback
circuit, the weights can be obtained through the evaluation of the correlation coefficients
3
between each auxiliary signal and the residue of cancellation [Skolnik M., 1981].
With the use of GSC as an ECCM technique, it is certain that there are limitations and
also parameters that can be altered so as to facilitate a better array performance. Hence, this
study aims to obtain a better understanding of the limitations and effectiveness of such a
technique with varying parameters of the jammer. A second portion of the research also aims to
investigate how the accuracy and efficiency of adaptive array processing can be improved
by adjusting relevant parameters of the radar’s internal system to a more optimal state to
improve the radar system’s performance.
MATERIALS AND METHODS
The codes used for this study is run on a Matrix Laboratory (MATLAB) simulator, Octave
3.8.0 to derive the results and graphs and the text editor “sublime text” is utilized. A simulation
of the elements of a radar system, including main and auxiliary arrays, as well as target and
jammer signals, is carried out.
Three variables were tested separately so as to see how each of them affected the behaviour
and effectiveness of the radar system. The variables of each tests are stated as follows:
● Test 1: Changing the jamming power via jammingtosignal ratio (JSR)
● Test 2: Changing the angle of attack of the jammer (j_aoa)
● Test 3: Changing the spacing between array sensor elements (d)
Two performance metrics are used to qualify the simulation results, and they are elaborated
as follows. The ability to suppress interference signals determines a radar system’s
signaltojammer ratio, which compares the level of desired signal against the level of
interference signal at the array output. SJR of the output signal leaving the radar system is
one such measurement employed to determine the effectiveness of radar interference
reduction and suppression techniques. Such unwanted interference can enter a radar system
through its antenna and hamper the signal processing of the radar’s signal processor
[Douglas A., 2003]. The formula is presented as follows in decibels [Fourikis, 1996]:
( 1 )
Nulltonull beamwidth is also used to measure the angular width of the main beam. A
small nulltonull beamwidth is desirable as it is an indication that more power is radiated at
the angle of arrival of the target instead of being distributed at the sidelobes. The following
equation is utilized in this study [Toomay J., 2004]:
θnn = 115λ
D
degrees ( 2 )
4
Where
θnn = antenna nulltonull beamwidth for
a uniform current distribution,
radian or degrees,
λ = wavelength in meters,
D = length of the array in meters.
In the simulations, the default setting of values are adjusted such that j_aoa= 0°, JSR = 50
dB and d=lambda/2 (i.e., λ/2), and rectangular windowing is employed. Number of
targets=1 is
standard for all tests. The number of elements in the main channel is set at Nm=16, while
the number of elements in the auxiliary array is set at Na=4. Lastly, the angle of arrival of the
target at t_aoa=20°. The following tables provide a clearer picture of the variations in
parameters used in the respective tests.
Table 1.1: Variables in Test 1
Test 1.1 Test 1.2 Test 1.3 Test 1.4 Test 1.5
j_aoa (°) 0
JSR (dB) 0 10 20 30 50
Spacing
between array
sensors (d)
lambda/2
Table 2.1: Variables in Test 2
Test 2.1.1 Test 2.1.2 Test 2.1.3 Test 2.1.4 Test 2.1.5
j_aoa (°) 80 40 0 40 80
JSR (dB) 50
Spacing
between array
sensors (d)
lambda/2
5
Table 3.1: Variables in Test 3
Test 3.1.1 Test 3.1.2 Test 3.1.3 Test 3.1.4 Test 3.1.5
j_aoa (°) 0
JSR (dB) 50
Spacing
between array
sensors (d)
lambda/8 lambda/4 lambda/2 lambda 2*lambda
RESULTS
Each test produces results that are presented into three figures covering different aspects so
that more alternatives are available when analyzing the results. The figures of all the tests
can be found attached in the Appendix. A brief description of each figure can be found as
follows:
Figure 1: Main array response in decibels against the angle of arrival of the target in degrees.
Figure 2: Response of the array over time with the presence of a jammer and radar noise under
different settings, “noisy”, as well as in the absence of it, labelled “noiseless”.
Figure 3: Comparison of noiseless array output against the final array output.
The results of Test 1 and 2 would be quantified by SJR output before and after the application
of the GSC so as to give a clear indication of the limitations and effectiveness of the GSC
under different parameters. Test 3 would employ the use of final SJR as well as nulltonull
beamwidth to study the change in array response caused by a change in spacing between
array sensor elements.
Test 1: Changing the jamming power via the jammertosignal ratio, JSR (dB)
The JSR of the jammer is increased from 0 dB to 50 dB and the SJR output before and after
the application of the GSC are calculated respectively. The results are recorded as follows in
Table 1.2.
6
Table 1.2: Calculations from Test 1 done on Matlab Simulator 3.8.0
JSR (dB) 0 10 20 30 50
SJR output
without GSC
(dB)
11.974 8.709 0.657 9.113 29.097
SJR output
after GSC
(dB)
11.475 11.544 11.578 11.589 11.594
Test 2: Changing the angle of attack of the jammer, j_aoa (°).
The angle of attack of the jammer is changed by intervals of 40° from 80° to 80° and the same
calculations are made for the SJR outputs. The results are recorded as follows in Table 2.2.
Table 2.2: Calculations from Test 2 done on Matlab Simulator 3.8.0
J_aoa (°) 80 40 0 40 80
SJR output
without GSC
(dB)
26.761 17.032 29.097 32.683 19.848
SJR output
after GSC
(dB)
11.562 11.558 11.594 11.217 11.620
Test 3: Changing the spacing between the array sensors (d)
The final SJR output and nulltonull beamwidth is calculated each time the distance of
array sensor elements is doubled in intervals from d=lambda/8 to d=2*lambda. The results
are recorded as follows in Table 3.2.
Table 3.2: Calculations from Test 3 done on Matlab Simulator 3.8.0
d (m) lambda/8 lambda/4 lambda/2 lambda 2*lambda
SJR output
after GSC
(dB)
9.5601 10.884 11.594 11.370 11.127
NulltoNull
beamwidth
(°)
59.08 33.79 16.72 8.30 4.21
7
DISCUSSION AND CONCLUSION
For Tests 1 and 2, by studying the change in the quality of the signals before and after
the application of the GSC, links can be drawn between the performance of the array
and the sidelobe canceller, as shown through qualitative and quantitative analysis.
In Test 1 with reference to Figure 2, there is a trend in the absence of the GSC whereby
the “noisy” signal received by the array becomes significantly stronger as the JSR ratio
increases to the point whereby the SJR output becomes negative when JSR is set to a value
larger than 20 dB. The jamming signal has overpowered the target signal and rendered the
signal received unusable. However, after the implementation of GSC, it can be seen from
Figure 3 that the final output of the radar systems met with different jamming powers
seemed similar and comparable to one another. Through the calculation of the final SJR
output, it is discovered that the test with the highest initial JSR of 50 dB unexpectedly had
the highest SJR output by a small margin. Even when met with high jamming power, the
performance of the radar system is not compromised, giving a positive indication that the
GSC is indeed capable of effectively cancelling strong jammer signals and retaining the
quality of the desired target signal.
In Test 2, it is observed that in Figure 2, the “noisy” signal received by the radar
becomes gradually stronger when the angle of attack of the jammer approaches the angle of
arrival of the target at 20°. This is met with the sole exception of the test whereby the “noisy”
signal increased significantly with the change in location from 40° to 80°. Upon closer study
of Figure 1, it is possible that the anomaly is due to the difference in the array response of
the sidelobes. Examining the final SJR output with reference to table 2.2, the quantitative
results also confirm the observations made that radar performance is slightly poorer when
the jammer is located closer to the target. Although there is a slight but definite correlation
between the angle of attack of the jammer and the ability of the GSC to reduce the interference
signal, the GSC is extremely successful in reducing the jammer signal regardless of the wide
range of the angles of attack. This is evident from the similar final SJR output of all the
tests, whereby there is only a slight difference of 0.403 dB between the highest and lowest SJR
output.
With reference to Table 3.2 for Test 3, it is observed that an increase of spacing between
array elements (d) from lambda/8 to lambda/2 yields a remarkable increase in the SJR of the
output signal from 9.5601 dB to 11.594 dB when GSC is implemented. A further increase
in d to 2*lambda causes the final SJR output to decrease slightly from 11.594 dB to 11.127 dB.
Aside from SJR output, other factors are put into consideration when analyzing and
evaluating radar performance with the change of array spacing. Referring to Figure 1, it is
observable that as d increases from lambda/8 to 2*lambda, the nulltonull beamwidth
decreases significantly from
59.08° to 4.21°, which is an advantageous effect. However, it is also seen that more than
one mainlobe appears at various steering angles when the value of d exceeds lambda/2. Due
to the presence of undesirable multiple main lobes, the angle of arrival of the target signal
would be unknown. Putting these factors into account, it is hence evident that lambda/2 is the
8
best spacing distance for main array sensor elements due to its high SJR output, small
nulltonull beamwidth and presence of only one main lobe.
Through the results from the first two tests, it is evident that the the GSC is able to
effectively maintain a similar final SJR output, even with a wide range of varying jammer
parameters. Test 3 demonstrated how the change in the spacing of array sensors affects overall
radar performance and hence can be employed to effectively reduce the effects of a jammer and
complement the use of a particular array. Through this study, a better understanding of the
abilities of the GSC has been gained and it is concluded that the GSC is an excellent SLC
technique that can be employed so as to facilitate satisfactory array performance with the
presence of jammers.
9
ACKNOWLEDGEMENTS
Our embarkment and completion of this DSO project could not be possible without the
guidance and help received from many individuals and organisations. We would like to
express our gratitude to everyone who has helped us through the way through this exciting
project.
First of all, we would like to sincerely give our thanks to our mentor, Dr Yeoh Wee Soon,
who has supported us through the year. Under his supervision, we were able to have a much
better and in depth understanding of our research topic, which would not be possible if not for
his kind guidance.
We would also like to extend our gratitude towards DSO National Laboratories for the
provision of resources which helped our project greatly, and also to the staff who have kindly
ensured that we were able to make good progress on this research study.
Last but not least, we would also like to thank our teacherincharge, Mr Shaun De Souza,
for giving us the opportunity to participate in this project and allow us to expand our learning
out of the school curriculum.
10
REFERENCES
[1] Skolnik M. 1981. Introduction to Radar Systems, Second Edition. Cerra, F. j. (ed.)
McGrawHill Book Company, U.S.
[2] Fourikis, N. 1996. Phased ArrayBased Systems and Applications. In K. Chang (Ed.), Wiley
series in microwave and optical engineering. John Wiley & Sons, Inc.
[3] Budge, M. 2007. ST: Radar Waveform and Signal Processing. Sidelobe cancellation.
[4] Federation of American Scientists Military Analysis Network 1998. AN/PPS5B Ground
Surveillance Radar Set.
[5] Villiard, O.G., Jr. 1976. The Ionospheric Sounder and Its Place in the History of Radio
Science.
[6] Bucciarelli, T. 1984. The GramSchmidt Sidelobe Canceller.
[7] Merv, B. 2011. Radar waveforms and signal processing. Retrieved from
http://www.ece.uah.edu/courses/material/EE710Merv/.
[8] Taylor and Francis Group. 2009. Adaptive Array Processing.
[9] Douglas, A. G. 2003. Radar Signal Processing and its Applications.
[10] Toomay, J. 2004. Radar Principles for the Nonspecialist. Cavanaugh Editorial Services
(Ed.), SciTech Publishing Inc., U.S.
11
APPENDIX
Test 1 (variable: change in jamming power via JSR)
Test 1: figure 1
Figure 1: The main array response remains the same with change in jamming power via JSR.
Test 1: figure 2
Figure 2 (Test 1.1: JSR = 0)
12
Figure 2 (Test 1.2: JSR = 10)
Figure 2 (Test 1.3: JSR=20)
13
Figure 2 (Test 1.4: JSR=30)
Figure 2 (Test 1.5: JSR 50)
14
Test 1: figure 3
Figure 3 (Test 1.1: JSR = 0)
Figure 3 (Test 1.2: JSR = 10)
15
Figure 3 (Test 1.3: JSR=20)
Figure 3 (Test 1.4: JSR=30)
16
Figure 3 (Test 1.5: JSR 50)
Test 2 (variable: change in jammer angle of attack)
Test 2: figure 1
Figure 1: The main array response remains the same with change in jammer angle of attack.
Test 2: figure 2
17
Figure 2 (Test 2.1: j_aoa = 80 degrees)
Figure 2 (Test 2.2: j_aoa = 40 degrees)
18
Figure 2 (Test 2.3: j_aoa = 0 degrees)
Figure 2 (Test 2.4: j_aoa = 40 degrees)
19
Figure 2 (Test 2.5: j_aoa = 80 degrees)
Test 2: figure 3
Figure 3 (Test 2.1: j_aoa = 80 degrees)
20
Figure 3 (Test 2.2: j_aoa = 40 degrees)
Figure 3 (Test 2.3: j_aoa = 0 degrees)
21
Figure 3 (Test 2.4: j_aoa = 40 degrees)
Figure 3 (Test 2.5: j_aoa = 80 degrees)
22
Test 3 (variable: spacing between array sensors (d))
Test 3: figure 1
Figure 1 (Test 3.1: d = lambda/8)
Figure 1 (Test 3.2: d = lambda/4)
23
Figure 1 (Test 3.3: d = lambda/2)
Figure 1 (Test 3.4: d = lambda)
24
Figure 1 (Test 3.5: d = 2*lambda)
Test 3: figure 2
Figure 2 (Test 3.1: d = lambda/8)
25
Figure 2 (Test 3.2: d = lambda/4)
Figure 2 (Test 3.3: d = lambda/2)
26
Figure 2 (Test 3.4: d = lambda)
Figure 2 (Test 3.5: d = 2*lambda)
27
Test 3: figure 3
Figure 3 (Test 3.1: d = lambda/8)
Figure 3 (Test 3.2: d = lambda/4)
28
Figure 3 (Test 3.3: d = lambda/2)
Figure 3 (Test 3.4: d = lambda)
29
Figure 3 (Test 3.5: d = 2*lambda)