Professional Documents
Culture Documents
2019-2047
7-11 January 2019, San Diego, California
AIAA Scitech 2019 Forum
Carina Marcus∗ † ‡
Saab Aeronautics, Bröderna Ugglas Gata, SE-581 88 Linköping, Sweden
When designing fighter aircraft it is necessary to be able to balance abilities against costs
and penalties. An often recurring trade-off is how improved sensor performance degrades the
signature of the aircraft. In this paper, properties of a passive radar warning interferometer
are modeled and tested in a simple tactical situation. The methodology is intended for use
in design space exploration in a process where the aircraft and sensor system are developed
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
I. Introduction
esigning stealthy fighter aircraft with sensor capability entails balancing the increased situation awareness obtained
D from information gathered by the sensor system against the increased risk of detection brought on by the added
signature of the sensors. A model of the sensor properties and the signature of the sensor and the aircraft can be used in
a tactical context in order to evaluate the performance. It can also be used as one of the models in a Design Space
Exploration (DSE) of the entire aircraft and its systems during the concept design. This approach differs from the
integration of a sensor system onto a separately developed aircraft as more parameters are included in the trade-off
process. This is a way to expand the design space from being confined to the vehicle itself [1] to include sensor systems
and treat them as any other property of the aircraft during the design phase [2, 3].
A sensor model appropriate for use in DSE also includes how the sensor affects the aircraft with regard to e.g. mass,
power and cooling budget along with the signatures. There is often a need to balance different types of signatures, [4] in
different scenarios to maximize the stealthiness. Activities that include balancing sensor properties, signatures, the
requirements on the aircraft’s infrastructure and considering the impact on the flight performance in scenarios is a truly
complex Multidisciplinary Design Optimization (MDO) which can be applied in several ways, [5].
The purpose of this paper is to establish a methodology where basic models of a radar interferometer with respect to
detection range, accuracy of the direction of arrival measurement and signature are used in a tactical situation to analyze
the properties of different antenna configurations.
The Radar Cross Section (RCS) of an aircraft is directly linked to its detectability and by extension, its survivability.
It was initially assumed that aircraft with a very low RCS would provide air superiority, but the increasing abundance of
low-frequency radars offset [6, 7] most of the RCS reductions. Extreme signature adaption is likely to increase the
operational and maintenance costs. Also, sensor functions may also have to be omitted or restricted since their presence
compromise the stealth capability, either by emissions or by the RCS contributions from the associated antennas.
Instead of aiming at further reducing the RCS, the advantages of moderately reduced RCS can be balanced against
other properties, such as action radius and maneuverability which are important to the survivability. To assess the
results of this balancing process, integrated testing in scenarios should be used when choosing which aircraft concepts
to develop further.
∗ Systems Engineer, Saab Aeronautics, Bröderna Ugglas Gata, SE-581 88 Linköping, Sweden, Member AIAA
† Linköping University, Department of Physics, Chemistry and Biology, SE-581 83 Linköping, Sweden
‡ Swedish Defence University, Drottning Kristinas väg 37, Stockholm, Sweden
Copyright © 2019 by Carina Marcus. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
This paper is organized as follows. In Section II a methodology for balancing sensor performance vs. signature
is described. Two versions of a radar warning interferometer mounted on a wing tip rail structure, are presented in
Section III. In Section IV the antennas are modeled and compared with measurements, while their RCS contribution
is analyzed in Section V. In Section VI the properties of the interferometer are analyzed in a simple tactical context.
The results are presented in Section VII, the methodology is discussed in Section VIII and conclusions are drawn in
Section IX.
Fig. 1 Four antenna assemblies provide coverage in the horizontal plane of the aircraft. The normal directions
of the antenna assemblies are indicated.
2
A. Original antenna assembly design
The initial design, a faceted structure with five spiral antennas in each assembly, is shown in Fig. 2 where position of
the antennas are represented by darker numbered areas. The antennas have a bandwidth of several GHz, but the data for
f =10 GHz is used throughout the paper.
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
Fig. 2 The original array structure in the global coordinate system. Height 104 mm, width 120 mm. The
antenna circles have a radius of 10 mm.
The facets supporting the antenna elements are oriented in such a way that one assembly provides coverage in a
cone with the half angle 50◦ from the normal, n̂.
The normal
qdirections of the antenna elements,
q expressed inthe global coordinate system, are given in Table 1 with
φ = arcsin x/ x 2 + y 2 and θ = arccos z/ x 2 + y 2 + z2 , c.f. Fig. 2.
Table 1 Element normal vector directions for the original antenna assembly, in the global system.
Element φ [◦ ] θ [◦ ]
1 45 45
2 20 71
3 60 90
4 20 109
5 45 135
3
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
Fig. 3 Assembly A1 with the positions of antenna elements 1-6 and 1’-2’ indicated.
Structure P1 P2 P3
A1 [93; 0; 0] [72; 44; 0] [40; 72; 0]
A2 [108; 0; 0] [84; 44; 0] [46; 72; 0]
A3 [126; 0; 0] [98; 44; 0] [57; 72; 0]
A4 [147; 0; 0] [115; 44; 0] [ 63; 72; 0]
A5 [171; 0; 0] [134; 44; 0] [74; 72; 0]
4
k dm
dE φ = − j sin(θ)e−jkr (2)
4πr
where k is the wavenumber, dm = Ms lm is the infinitesimal magnetic dipole moment, r is the distance between the
center of the dipole and the field point and sin(θ) is the angle between the field vector and the direction of the dipole
moment. The total far-field from each antenna element consists of the direct radiated far field:
Õ
E®direct = d E® φ , (3)
current ring with the circumference corresponding to one wavelength was modeled as a number of magnetic dipoles:
2π
® ring = ρ̂Aring e j M m
dm for m = 1...M, (5)
where Aring is the amplitude of the propagating magnetic current on the spiral antenna.
Then, the effects of reflections from the cavity fields have been added as two TE11 modes in quadrature and expressed
in terms of a magnetic current, viz.:
® s,T E11 = d m
M ® (ρ, φ) + jd m
® (ρ, φ + π/2) . (6)
As for the current ring, the far fields are obtained using Eq. (3) and Eq. (4).
The coefficients ATE11 and Aring have been chosen so that half of the transmitted power is associated with the ring
current and the other half with the TE11 modes for an antenna element on an infinite PEC plane.
Fig. 4 The measured element normalized gain for element #4 in its antenna assembly (u,v)-space. Vertical
polarization. Contours at 3dB intervals, f =10 GHz.
First, the currents of the antenna elements are prescribed and not derived self-consistently, e.g. from a full-wave
simulation.
Second, no mutual coupling between the elements is included and the contributions from diffraction is limited to the
edges of the facet where the antenna element is located. The antenna elements are modeled to only illuminate their own
facet.
5
Fig. 5 The calculated element normalized gain for element #4 in its antenna assembly (u,v)-space. Vertical
polarization. Contours at 3dB intervals, f =10 GHz.
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
Third, the antenna that has been measured is covered with a thin layer of dielectric and the edges between the facets
are not perfectly sharp. The dielectric layer can bind the field, the analysis of which is beyond the capabilities of ordinary
UTD.
Despite these approximations, the agreement between measured and calculated data is deemed sufficient for use in
this test of the methodology as their behavior is similar, see Fig. 4 and Fig. 5.
Fig. 6 The calculated normalized gain for element #1, assembly A1, in its antenna assembly (u,v)-space. Vertical
polarization. Contours at 3dB intervals, f =10 GHz.
V. RCS analysis
The RCS was calculated in the horizontal plane for the vertical polarization using the FDTD solver TFDTD [10]. It
was verified by means of successive grid refinements that the results have converged on the lowest levels presented here.
6
In order to keep the computational effort within manageable limits, the wing tips were truncated length-wise. The
artificial scattering from the end was gated out in the time-domain.
The RCS analysis must be performed for all bands where there active radar threats are considered, whether this is in
or out of band for the antennas.
The signature and antenna performance of the assemblies are directly linked to both the electrical and mechanical
design of the antenna elements as well as the shape of the structure in which they are mounted. The RCS can be reduced
by shaping or by using absorbing materials, of which shaping should be considered first. Absorbers should be avoided
as far as possible, mainly because their need for costly maintenance but also since they degrade the performance of
some antenna types. Therefore, we consider here PEC assembly structures. There are methods for calculating the RCS
of spiral antennas such as [11, 12] but here the conservative estimate of the in-band RCS of the antennas is used.
In terms of RCS, the modified structures exhibit lower levels than the original design, see Fig. 7.
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
Fig. 7 The vertically polarized RCS in the horizontal plane for all structures.
The peak in RCS that occurs around 20◦ for the original antenna design can be traced back to the two facets with the
forward-looking elements in accordance with the directions in Table 1.
The RCS for the vertical polarization, is then averaged (i) over the frequency interval 9 - 10 GHz and (ii) over the the
so-called forward threat sector, here 0◦ -34◦ at zero elevation, see Table 5.
Assembly σ [dBm2 ]
A1 -23.1
A2 -25.9
A3 -27.7
A4 -30.5
A5 -31.5
In order to find the assembly where the desired feature of reduced signature is useful, but the draw-back in terms of
reduced gain is acceptable, the RCS management process is assumed to have resulted in an allocation of the assembly
RCS amounting to -20 dB of the total aircraft RCS. The levels are consistent with a reduced RCS-class aircraft, but the
process would be similar for any class of RCS reduction.
A. Tactical situation
The antenna performance and the signature are factors in a situation when a (red) radar is searching for our (blue)
aircraft in an angular sector. Assuming that the blue and the red aircraft are flying at 0.5 km altitude, the radar horizon
limits the radar range between the two aircraft according to [13]:
p p
Rhor = 4.12 hrr + hobs , (7)
7
where hrr is the height of the radar, hobs the observation height which are both given in [m]. Using the constant 4.12
yields the result Rhor in km. For hrr = 500 m and hobs = 500 m, Rhor is 184 km, which covers the detection ranges
obtained in this paper.
The idea is that the blue system is able to detect the red signal not only during the short time when the main lobe
illuminates the blue aircraft, but also to some given level of sidelobe illumination. This in turn puts requirements on the
signal detection ability of the blue system, i.e. its SNR. Including information on relative speed between the aircraft,
here 0.6 km/s, makes it possible to calculate how the different choices of the blue ESM system translates into time to act.
The distance at which red detects blue is obtained from the radar equation [14]:
s
4 Pt x,r Gt x,r Gr x,r σb λ2
Rr = (8)
(4π)3 SNRr Lr N Fr k B T0 /τr )
C. ESM system
1. Detection range
The blue ESM system parameters are given in Table 7.
8
Blue must be able to detect the signal through most of the side lobes of the red antenna in order to establish tracking-like
conditions with as short interrupts as possible. The transmit sidelobes of red’s Active Electronically Scanned Array
(AESA) radar follow those of a uniformly excited circular disc. The sidelobe threshold level was set to (SLLr =-36 dB)
above which blue is able to detect the red sidelobe emissions for at least 45% of the red radar sweep time.
Both Rr and Rb have been corrected for the atmospheric attenuation, here 1.5 dB per 100 km.
g1 = c1 + n1 , (10)
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
g2 = c2 e jξ
+ n2 . (11)
In the following, we will assume that the transmission lines are perfectly calibrated, i.e. c1 = c2 = c. The coherence
between the two signals Eq. (10) and Eq. (11) is:
where < · > represents averaging. As the signal and the noise are uncorrelated we obtain:
|c| 2
γ= , (13)
|c| 2 + |n| 2
and then, using:
|c| 2
SNR = , (14)
|n| 2
the coherence parameter becomes:
1
γ= . (15)
1 + SNR−1
We proceed from the Cramer-Rao bound as an approximation of the error σξ for the ML estimate of the phase difference:
s
1 − γ2
σξ = , (16)
2Nγ 2
where N is the number of samples used in the ML estimate. In order for for Eq. (16) to hold, one must ensure that N and
SNR are sufficiently large. These conditions are assumed to be fulfilled.
Taking the noise power PN to be constant, the SNR will be low when the incident signal enters the antenna from a
direction where the directive gain is small, and vice versa, viz.:
where
|Si | λ2
a0 = . (18)
PN 4π
Here, Si is the Poynting vector of the incident wave and λ its wavelength. See Fig. 8.
The phase difference error σξ translates to an error in the estimate of the elevation σε close to the horizon through:
where k is the wave number. It should be clear that minimizing the estimation error can be achieved by maximizing
the gain, i.e. by orienting the antennas so that their directions of maximum gain lie in the direction towards the emitter.
9
Fig. 8 Schematic geometry of antenna elements and an incoming signal.
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
In summary, from Eq. (16), Eq. (15) and Eq. (17) our expression for the estimation error becomes:
s
1 1 − γ2
σε = , (20)
k h 2Nγ 2
VII. Results
A. Detection distances
The detection distances for the blue and the red systems are shown for each assembly, A1 through A5, in Fig 9.
The aircraft equipped with A1 does not have the ability to detect red before being detected, despite having the best
antenna properties but, consequently, also a high signature. The remaining assemblies can be studied according to the
difference between Rb and Rr . A2 and A3 are both in the 4-5 km range, which, under the assumption that blue and
red are closing on each other at 0.6 km/s, leaves roughly a 7 s window of advantage for blue. At the extreme end of
the range, A5 supports the lowest signature, but at the cost of a reduced detection capability. In fact, A4 provides the
largest window of opportunity, 16 s, compared to A5’s 12 s. The drawback of A4 is that the distance to the red aircraft is
shorter than if A2 or A3 were used instead. Both the radar and the ESM system could apply methods that allow tracking
of an object before an actual detection is established which would increase the respective distances in the analysis.
B. Direction error
The separation in distance, h, between the four front antenna elements 1, 1’, 2 and 2’, see Fig 3, is constant for all
assemblies due to integration restrictions. They are, in the vertical and horizontal planes, h H = 34 mm and hV = 35 mm,
10
respectively. Even if the DOA error is larger for A3 than A4, one must take the difference between the detection distances
into account. The transverse position error, ∆ε , is given by:
∆ε = σε R , (22)
where R is the distance between the aircraft. Results are presented in Table 8 where N = 10 is used.
Table 8 Maximum transverse errors from DOA estimates for assemblies A2 - A4.
As is evident from Table 8, A4 yields the smallest transverse errors at initial acquisition as well as the largest
window of opportunity. That A4 by the time of acquisition initiation has to be closer to the red aircraft than A3, could
be outweighed by the possibility that blue aircraft with A3 might have to wait before the errors permit further actions.
VIII. Discussion
Choosing equipment for fighter aircraft requires much more than identifying superior technical properties. As
was demonstrated in this paper, properties such as DOA determination capability and signature can be counter-acting
each other, which makes it necessary to compromise in order to reach a solution. The choice of method when
determining DOA can also put requirements on the antenna placement. Looking beyond the obvious technical results
and incorporating a simple tactical solution supports a balanced decision process since it forces a consideration of how
the equipment can be used and offers, to some degree, an understanding of how changes in the design affects the end
results.
The results in this paper are, despite including one tactical situation, not conclusive. Even though the results are
known from an electromagnetic point of view, e.g. the behavior of the RCS of a PEC facet or the radar equation,
combining them with a basic tactical situation yields results that are not trivial. That a large Rb − Rr is good is obvious,
but Rb itself is also a factor. The new antenna structure has properties that give a constant relative DOA error, but the
absolute error increases with Rb . This makes it necessary to include tactical experience and judgment to complete the
analysis.
It is important to consider the impact of the entire EW system on the aircraft, from several perspectives, e.g. sensor
performance, requirements on aircraft infrastructure and flight performance. Sensor performance has been addressed in
this paper, while the effects on the aircraft itself is left for future work. It is worth noting, that the position of the antenna
assemblies, at the wing tips, is likely to influence on the flight performance, which in turn can put requirements on how
the sensor system is used to further complicate the design process. Furthermore, an extended range of tactical situations
in which sensor performance as well as flight performance are tested should be used before deciding on the design of the
sensor system.
IX. Conclusion
A methodology for balancing the sensor capability of a system for passive detection of a radar signal against the cost
in term of higher RCS and thus detection risk has been developed and tested.
The model includes the sensor capability in terms of detection range and DOA accuracy and the radar signature
which is exploited by a hostile airborne radar. The evaluation of five different antenna configurations was performed in a
tactical situation to provide a link to testing in scenarios.
The results were not conclusive, and it became apparent that the choice of configuration could not be chosen
unless further testing in several scenarios and knowledge about tactics were factored into the decision. However, the
methodology presented provides a connection to scenarios that can be useful when considering different design options.
The model did not contain information about the sensor system’s requirements on the infrastructure of the aircraft,
nor its impact on flight performance. This is left to be studied in another paper where the results from this paper can be
one part of a DSE process.
11
Acknowledgments
The author would like to thank A. Wikström for kindly providing the measurement data and B. Leijon for data on
the original structure. They are both with Saab Surveillance. Valuable discussions with B. Andersson at the Swedish
Defence Research Agency and M. Appelgren at Saab Surveillance are gratefully acknowledged. Excellent graphics of
antennas and aircraft provided by M. Räf at Saab Aeronautics.
References
[1] Martinelli, S. K., and Braun, R. D., “Centerline heating methodology for use in preliminary design studies,” 2011 Aerospace
Conference, 2011, pp. 1–18. doi:10.1109/AERO.2011.5747498.
[2] Papageorgiou, A., Ölvander, J., and Amadori, K., “Development of a multidisciplinary design optimization framework applied
on UAV design by considering models for mission, surveillance, and stealth performance,” 18th AIAA/ISSMO Multidisciplinary
Analysis and Optimization Conference, 2017, 2017. doi:10.2514/6.2017-4151.
Downloaded by IOWA STATE UNIVERSITY on January 19, 2019 | http://arc.aiaa.org | DOI: 10.2514/6.2019-2047
[3] Yifeng, T., Hu, L., and Jun, H., “Design space exploration in aircraft conceptual design phase based on system-of-systems
simulation,” International Journal of Aeronautical and Space Sciences, Vol. 16, No. 4, 2015, pp. 624–635. doi:10.5139/IJASS.
2015.16.4.624.
[4] Marcus, C., Andersson, K., and Åkerlind, C., “Balancing the radar and long wavelength infrared signature properties in
concept analysis of combat aircraft – A proof of concept,” Aerospace Science and Technology, Vol. 710, 2017, pp. 733 – 741.
doi:10.1016/j.ast.2017.10.022.
[5] Martins, J. R. R. A., and Lambe, A. B., “Multidisciplinary design optimization: A survey of architectures,” AIAA Journal,
Vol. 51, No. 9, 2013, pp. 2049–2075. doi:10.2514/1.J051895.
[6] Kuschel, H., “VHF/UHF radar. 1. Characteristics,” Electronics Communication Engineering Journal, Vol. 14, No. 2, 2002, pp.
61–72. doi:10.1049/ecej:20020203.
[7] Kuschel, H., Heckenbach, J., Müller, S., and Appel, R., “Countering stealth with passive, multi-static, low frequency radars,”
Aerospace and Electronic Systems Magazine, IEEE, Vol. 25, No. 9, 2010, pp. 11–17. doi:10.1109/MAES.2010.5592986.
[8] Kouyoumjian, R., and Pathak, P., “A Uniform Geometrical Theory of Diffraction for an Edge in a Perfectly Conducting Surface,”
IEEE, Vol. 62, 1974, pp. 1448–1461. doi:10.1109/PROC.1974.9651.
[9] Balanis, C. A., Antenna Theory: Analysis and Design, Wiley, 1982.
[10] Martin, T., Broadband electromagnetic scattering and shielding analysis using the finite-difference time-domain ..., Univ.,
Linköping, 2001. Diss. (sammanfattning) Linköping : Univ., 2001.
[11] Penney, C. W., and Luebbers, R. J., “Input impedance, radiation pattern, and radar cross section of spiral antennas using FDTD,”
IEEE Transactions on Antennas and Propagation, Vol. 42, No. 9, 1994, pp. 1328–1332. doi:10.1109/8.318663.
[12] Volakis, J., Gong, J., Alexanian, A., and Woo, A., “Electromagnetic Scattering and Radiation from Microstrip Patch Antennas
and Spirals Residing in a Cavity,” Tech. rep., 1992. URL https://ntrs.nasa.gov/search.jsp?R=19930004166.
[13] Gerdle, P., Pettersson, C., and Åkerblom, G., Lärobok i telekrigföring för luftvärnet : radar och radartaktik, 2nd ed.,
Försvarsmakten i samarbete med Autotech teknikinformation i Stockholm, Stockholm, 2008.
[14] Skolnik, M. I., Radar handbook, 3rd ed., McGraw-Hill Professional, Maidenhead, 2008.
[15] Capon, J., “High-resolution frequency-wavenumber spectrum analysis,” Proceedings of the IEEE, Vol. 57, No. 8, 1969, pp.
1408–1418. doi:10.1109/PROC.1969.7278.
[16] Ziskind, I., and Wax, M., “Maximum likelihood localization of multiple sources by alternating projection,” Acoustics, Speech
and Signal Processing, IEEE Transactions on, Vol. 36, No. 10, 1988, pp. 1553–1560. doi:10.1109/29.7543.
[17] Tough, R., Blacknell, D., and Quegan, S., “A statistical description of polarimetric and interferometric synthetic aperture radar
data.” Proceedings of the Royal Society of London, Series A (Mathematical and Physical Sciences), Vol. 449, No. 1937, 1995,
pp. 567 – 589. doi:10.1098/rspa.1995.0059.
[18] Zebker, H., and Villasenor, J., “Decorrelation in interferometric radar echoes,” Geoscience and Remote Sensing, IEEE
Transactions on, Vol. 30, No. 5, 1992, pp. 950–959. doi:10.1109/36.175330.
12