Vous êtes sur la page 1sur 5

AIRBORNE GROUND CLUTTER MEASUREMENT SYSTEM DESIGN CONSIDERATIONS

Michael Wicks Rome Air Development Center Ferrel Stremler University of Wisconsin Stephen Anthony Environmental Research Institute of Michigan

ABSTRACT
For the large angle bistatic geometries selected for this experiment, the isorange and isodoppler contours are aligned nearly parallel, thereby precluding normal processing for resolution. This causes the clutter cells to be large and elongated and impacts the achievable calibration accuracy for a clutter measurement system. In order to achieve an accuracy of approximately f 4 dB, careful measurements of system parameters are required. This paper discusses the geometric limits, calibration considerations, and error analysis for such a system.

INTRODUCTION
This paper addresses calibration considerations and the performance analysis of an L-Band airborne clutter measurement system designed to collect simultaneous bistatic and monostatic clutter measurements in support of the Hybrid Bistatic Radar (HBR) System. Table 1 highlights key system parameters. The measurements, which will be performed over both sea and terrain using horizontal and vertical polarizations, will provide a data base for the development of bistatic clutter models for HBR and other bistatic spaceborne and airborne radar concepts. In the design and development of the collection system, numerous physical and geometric limits on performance were identified and quantified. In bistatic radar, particularly at the large bistatic angles of interest in this program, various intersection geometries occur, which cause a large variation in the resolution cell shape and size. The analysis techniques used to define the

experimental geometry and resulting resolution cells, are discussed, as well as calibration considerations, and their impact on clutter radar cross section calculations. The geometric limit is only one of the actual limits to the experimental accuracy. Other important factors include antenna pattern measurement accuracy, aircraft position measurement accuracy, and calibration reference accuracy. The calibration techniques involve the transfer of data from optimal geometries for calibration, to the geometries of interest in this program. The paper includes a description of the implementation of this technique in a clutter measurement system, as well as other calibration equipment, including ground based navigation beacons, and an active radar calibrator (calibrated target placed in the clutter cell of interest).

Table 1 Key Clutter Measurement System Parameters


Bistatic Pulse Width Monostatic Pulse Width Transmitted Waveforms Transmitter Power Transmitter PRF Antenna Polarizations Total Channels Samples Per PRI Sample Word Length Instantaneous Dynamic Range Adjustable Dynamic Range Coherence Interval System Timing Accuracy
125 nsec
4 Usec

Pulsed Carrier
5 kW; L-Band

2000 Hz Each Source V, H Selectable 4 Bistatic and 4 Monostatic 40 61&6Q 29 dB (w/lO dB CNR) >40 dB additional 100 msec
1:10*

GEOMETRIC CONSIDERATIONS
The simultaneous collection is accomplished by interleaving samples sequentially bistatic and monostatic, on a pulse to pulse basis.

Presented at the IEEE 1988 National Radar Conference 0885-8985/88/10004027 $1.00 0 1988 IEEE

The bistatic collection geometry selected for this experiment involves the use of two aircraft flying a fixed formation, on parallel flight paths at the same velocity. This configuration allows the aircraft to maintain the same geometry over an extended period of time. It was determined that this

IEEE AES Magazine, October 1988

27

collection configuration was required in order to collect a statistically relevant amount of data with sample diversity. Collection passes would be held for five minutes, collecting data over a number of miles. The general geometry associated with this configuration is shown in Figure 1. In this figure, the distances labeled wich an "R" are actual slant range, while the associated distance labeled with a "D", is the ground projection of the slant range. The subscript "T" refers to the transmitter, while the subscript "R" refers to the receiver.

system, though, the isodoppler contours are determined by the velocity components of both of the aircraft. For the case where the aircraft are directly across from each other (the aircraft and clutter patch are coplanar, with the plane perpendicular to the velocity vectors), the isodopplers are generally radial on the sides of the aircraft to the outside, and somewhat elliptical in the region between the aircraft. The isodoppler contours become skewed when the aircraft and clutter patch are not coplanar, but generally retain the same shape. The implication of this is that, for the region of interest between the aircraft, the isorange and isodoppler contours are essentially parallel. No second axis exists for the improvement of the resolution. The patch then becomes the region bounded by the isorange or iscdoppler contours on two sides and the convolved antenna patterns for the other sides. Since the grazing angles of interest are very shallow (1-30 degrees), the patterns become very long in the cross track direction. The resulting clutter patches, then, are very long and narrow, encompassing a large amount of area (hundreds of thousands of square meters).

REFERENEE POINT

'

Assume Parallel Flight Lines Actual LOS Ranges Convenllon R D - Ground Projectionsof 3D Lines

CALIBRATION CONCEPTS
The general concept being utilized in this experiment is one of transfer calibration. By this, we mean that the system will be calibrated at one geometry, and the critical parameters will be monitored while the system is then used to measure the clutter of interest. The calibration is to be performed by using an active radar calibrator during the calibration pass. In order to determine the quality of the calibration, however, an analysis is required of the various elements in the system. In bistatic clutter measurements systems, the received power can be related to the bistatic radar cross section, uc, and to the bistatic clutter reflection coefficient, uo, which is the radar cross section per unit area. The recorded signal is assumed to be proportional to the received power. Therefore the relationship between the recorded signal power, Pd, and the bistatic clutter reflection coefficent, uo is:
Pd =
KCI'
0 BORE
0 h h

Figure 1
This geometry has a number of problems. (See Figure 2 for the plots of the isorange, isodoppler, and convolved antenna patterns as described). The parallel flight paths and large bistatic angles produce an unusual effect. The range contours are defined by the intersection of an ellipsoid of revolution with the two aircraft at the focii, and the ground surface (represented by a plane). When the two aircraft are at the same altitude, these intersections are ellipses, symmetrical about both the line connecting the nadir points under the aircraft, and the perpendicular bisector of that line. When the aircraft are at different altitudes, however, the ellipses become skewed toward the lower aircraft, thereby retaining the symmetry only about the line connecting the nadir points.

( 4 xl3L

PtA2Hr

1
D

Gl(x.y)

G2(x.y)
oodxdY
(1)

Rl(x,y)2 R2(x.yI2

IRHNS

RtI. -36-1006
206-3000 TIME

Where:

P UOI'PLER

Figure 2. Bistatic Radar Ground Footprint


The other axis normally used to determine resolution is the doppler frequency. In a monostatic system, such as an imaging synthetic aperture radar, the isodoppler contours are generally radial, being determined by the inteeections of the constant doppler comes with the ground plane. In a bistatic

Pt = transmitter power, A = wavelength of the transmitted signal, Hr = receiver transfer (power) gain, L = system losses, G G = antenna transfer pattern gains " of the transmitter and receiver, respectively, evaluated a the clutter cell location ( x , y ) , R 1 , R2 = range distances between a clutter cell located a t (x.y) and the transmitter and receiver. respectively D = region of the clutter, defined by the combined antenna patterns, the range gate, and the doppler f i l t e r .
IEEE AES Magazine, October 1988

28

We can rewrite Eq. (1) as:


Pd =

ERROR ANALYSIS
H ,
mo

where Hs is the total radar system.transfer function. This transfer function relates the recorded power, Pd, to the effective value of o , and includes the receiver transfer gain, Hr. Absolute calibration of the clutter measurements requires a knowledge of the total radar system transfer function Hs, to relate the measured value of 0, to the terrain scattering effects. For the clutter measurements system to be calibrated, we require the satisfaction of two basic requirements. First, each parameter in Eq. (1) must be known as accurately as possible. Second, provisions must be made to relate the recorded value, Pd, to the radar cross section. The error analysis that follows addresses the question of how accurately each parameter in the system can be determined. We divide the parameters in Eq. (1) into two categories, those that can be considered invariant over relatively long periods of time compared to one data recording pass, and those that require continual monitoring during a time comparable to a data recording pass. In the former category are such parameters as the wavelength, the system losses, and the response of each antenna when referred to its own angular space (i.e., azimuth and elevation). System parameters that require regular monitoring are the transmitter power, Pt, and the receiver transfer gain, Hr. The transmitter power measurement will be entered into the auxiliary data record during the data measurements portions of the flight. A receiver transfer gain calibration will be performed immediately following each data measurements portion of a flight. Flight parameters that must be monitored closely during data measurements are the aircraft position (x,y,z), alongtrack and cross-track velocities, aircraft roll, pitch and yaw angles, and antenna pointing angles. Aircraft roll, pitch and yaw angles, and antennas azimuth and elevation angles are needed to determine the true pointing angle and ground boresight point of the antenna. Because the aircraft navigation parameters are so crucial to the overall system calibration, they will be determined as accurately as possible and recorded. The preceding parameters are needed to accurately measure the net system transfer function. Now we turn to consideration of a calibration reference to relate the received power to the clutter cross section. A s a result of the relatively large clutter region and the fact that near-simultaneous monostatic and bistatic measurements are desired, active radar calibrators should be used for the calibration reference. As a result of the large background clutter expected in some geometries, the active radar calibrator should have a radar cross section of 50-55 dBsm and include both a time delay (400 nsec) and a doppler offset (0-250 Hz). Thus, the receiver power from the active radar calibrator can be placed in different time-doppler cells than that of the predominant background clutter. Calibration passes will be flown at higher grazing angles to minimize multipath problems. The system is then carefully monitored as the collections are made at lower angles.
IEEE AES Magazine, October I988

The parameters that contribute most to the total error in the bistatic clutter measurements are those uncertainties associated with a determination of clutter cell location and area. This directly affects the value of the integration, and therefore the value of the measured clutter coefficient. For purposes of analysis, we have chosen the following definitions which, taken simultaneously, define the clutter cell location: (1) the -10dB contours of the combined antenna patterns: (2) a range distance beginning at the boresight aim point and extending in distance to correspond to a range gate duration of 125 nsec; and (3) a doppler bandwidth of 10 Hz, centered at the doppler frequency of the boresight aim point. A second important error source in the clutter measurements is that of the antenna gain patterns. The antenna gains not only help define the effective illuminated clutter area, but also serve as weighting functions over the clutter area. While this variation is more smoothly varying than that of a determination of the clutter cell location, it directly affects the determined value of the measured clutter coefficient. Antenna weighting errors incurred in the antenna pattern calibration procedures also enter here, and can be minimized by use of carefully calibrated antenna patterns and carefully measured site locations. The next major error source is that of the radar cross section calibration. The active radar calibrators must themselves be carefully calibrated, checked and maintained. They have some angle dependency, and thus care must be exercised in the accurate pointing of these devices. Multipath effects, particularly for low grazing angles, can be a problem and proper siting and analysis are necessary. The remaining factors in the determination of the clutter coefficient which are a major source of error are those associated with the radar instrumentation. These can be minimized with careful design choices and monitoring. In cases in which the parameters are relatively stable (e.g., microwave components), the subsystems will be checked and measured carefully during assembly and testing. In cases in which some drifts can occur during flights (e.g., transmitter power levels), provisions are made to monitor the parameters and record them during the flights. The method followed for evaluating the effects of the error sources on the estimated values of the clutter coefficient is that of conventional first-order analysis. Suppose that a parameter Y is related to the parameters X i , X2, . . . , Xn by means of the equation:
Y = f (XI,

xp,

..., X).

(3)

This approach assumes a small input error so that, to a first approximation, the output error is related to the input errors by means of a first order Taylor series expansion of the form:
(4)

where the partial derivatives are evaluated at the nominal operating point. We assume the errors to have zero mean and are uncorrelated. The uncorrelated assumption may not 29

be strictly true for all cases, and corrections can be made in these situations. However, the general approach remains the same. Because the geometric considerations become very involved for the bistatic case, the use of a digital computer simulation is enlisted to determine the partial derivatives evaluated at the nominal operating point. A complete analysis of the various parameters was performed and the position errors were found to be of the highest concern. As a result, the following discussion will focus on the analysis of the position errors. Before going to computer results to determine the error sensitivities in clutter cell location due to position errors, we introduce errors in the y (i.e., cross-track) and z (altitude) positions of one aircraft to obtain graphs of the range gate locations. An example of this output is shown in Figure 3. The vertical distance
ms "N O
.SMM

ble, and 2) set up a detailed error sensitivity problem using a digital computer simulation program. We have pursued both approaches. For the simulation, we use a program that computes the return signal power from the effective illuminated area for conditions expected in the clutter measurements program. Antenna patterns are assumed to be of the sin x/x type, both range gating and doppler filtering are assumed to be ideal. The received power is computed in the simulation (assuming a constant bistatic coefficient across a cell). Then an error in platform position is introduced and the received power is again computed. The ratio of the latter to the former is termed the fractional error. This error was plotted graphically, an example of which is shown in Figure 4. This plot shows the normalized error for cross track posi-

Y ~XLSS-NLI)

10 I. 17 1 .
16
I O .

14

13

la
11

0
0

2
1

ID

e e ;

,':

: , I

: ,':

,':

Figure 4. Uncorrelated Errors


-&m
D
-4m

-3m

-am

-im

W I

am
x

JOQ

d m

MOO

I=

am

"N O

p m Y. A am

a m

a . - .

Figure 3. Range Gate Migration


between plots for a given error in y in this graph (or z) represents the width of a given range gate as projected on the ground plane. Note that the locations "migrate" appreciably with relatively small variations in aircraft location. If we assume homogeneous clutter, this migration is not judged to be serious. If the bistatic coefficient changes with these shifts, then this could be a source of error. However, the major concern here is the variation' in cell dimension, signifying a variation in area. If we assume that the cell length is fixed (say, by the isodoppler and antenna contours) and that the antenna pattern weighting does not change appreciably, and variation in range gate width translates directly into cell area. From Figure 3, we observe that a relatively small error in y causes an appreciable change in the width of the first range gate. Variations in successive range gates are less severe, but all exhibit some change. Variations due to an error in z (altitude) are less severe, but are also present. As mentioned above, the effects of antenna pattern variations will also enter this problem. However, for the conditions of the case being considered, use of some trigonometry shows that distances of zero to 20,000 feet in the ground plane corresponds to zero to 5 degrees of elevation angle in the antenna pattern. Because the antenna elevation beamwidth is 60 degrees, the variation over this 5 degrees is not a predominant factor to be considered here. Having established that there is an error sensitivity problem in cell location and area, this indicates that we should 1) obtain the most accurate positioning system possi-

tion errors for an in-plane test case. These plots clearly show the severity of the position accuracy requirements. The overall results of the analysis are shown in Table 2. It is expected that the accuracy of the resulting data can be maintained to f5.0 dB with careful design and operating procedures, accurate antenna measurements, and proper navigation. An adequate number of active radar calibrators (6-8) with careful and accurate siting are also necessary to achieve these goals. Note that although the last nine error sources are only estimates, they have minimal impact on system error due to the dominance of the first four terms.

Table 2 Estimated Errors in Bistatic 'Clutter Measurements


Antenna Pattern, Transmitter Antenna Pattern, Receiver Cell Location and Pointing* Active Radar Calibrator Transmitter Power Measurement Losses in Microwave Components and Radomes Receiver Calibration Gen. Power Meas. Measurement of Receiver Transfer Gain Receiver Gain Drifts During Data Recording Receiver and Recorder Nonlinearities Receiver and Recorder Noise Receiver I and Q Channel ImbalanceTiming Inaccuracies (jitter, etc.) Overall (rms) Estimated (min.) Error:

fl.O dB f 1 0 dB . f1.8 dB fl.O dB f0.5 dB


f0.7 dB 50.5 dB
50.5 dB

k0.5 dB f0.5 dB k0.5 dB k0.5 dB f0.5 dB f3.9 dB

T h e cell location and pointing errors are derived from the platform position errors. The position error is assumed to be 100-150 feet CEP horizontally, and 300 feet CEP vertically, with one degree pointing accuracy in elevation and one-half degree in azimuth.

30

IEEE AES Magazine, October 1988

CONCLUSIONS
For the bistatic geometry selected in this experiment, the range contours are defined by the intersection of ellipsoids of revolution with the two aircraft at the focii, and the ground surface. Because the isorange and isodoppler contours are essentially parallel in the region of interest, doppler can not be used to improve resolution. The resulting clutter cells are very long and narrow, encompassing a relatively large amount of area. The calibration concept being utilized in this experiment is one of transfer calibration, where the system is calibrated at tone position, (higher grazing angle geometries) and the critical parameters are monitored while the system is used to measure the clutter of interest (at lower grazing angles). The calibration is performed by using the active radar calibrator, which itself may be subject to multipath and positioning errors at lower grazing angles. The active radar calibrator is used in both the bistatic and monostatic configuration. Absolute calibration of the clutter measurements requires a knowledge of the radar system transfer function. Also, provisions must be made to relate the recorded data to the clutter radar cross section. The parameters that must be monitored fall into two categories. Those that can be considered invariant over long periods of time, such as the antenna pattern, and those that require continual monitoring, such as flight parameters.

The major sources of error in this experiment are those due to antenna pointing accuracy limitations and the active radar calibrator performance. The overall estimated RMS error in the bistatic clutter measurement is f 3 . 9 dB. This is the absolute best performance. If the antenna pattern measurement or active radar calibrator performance degrade only slightly to a value of k2.0 dB, the overall performance drops to worst than f5.0 dB. Therefore, a system accuracy goal of + 5 dB was selected considering realistic test conditions.

REFERENCES
1 . R. W. Larson, A. L. Maffett, Calibration Model for SAR System, Electromagnetics, vol. 4,pp. 277-293, 1984.

2. Bistatic Clutter Data Measurements Program, Final Technical Report, No. RADC-TR-77-389, ERIM, Ann Arbor, May 1977. 3. Measurements of Bistatic Clutter Cross Section, Final Technical Report, No. RADC-TR-79-15. May 1979.

4. D. R. Brunfeldt, F. T. Ulaby, Active Reflector for Radar Calibration, IEEE, Trans. on Geoscience and Remote Sensing, vol. GE-22, pp. 165-169, 1984. 5. R. W. Larson, A. L. Maffett, et al., Bistatic Clutter Measurements, IEEE, Trans. on Antennas and Propagation, vol. AP-24, pp. 896-899, 1976.

Michael Wicks received his BSEE degree from the Rensselaer Polytechnic Institute, Troy, NY, in 1981, and the MSEE degree from Syracuse University, Syracuse, NY, in 1985. He has been with Rome Air Development Center, Griffiss Air Force Base, Rome, NY, since 1981, where he works in the Signal Processing Branch of the Surveillance Technology Division in the Directorate of Surveillance. He is the author of several technical reports and patent applications.

Ferrel Stremler received an AB degree from Calvin College in 1957, a BSEE degree from Illinois Institute of Technology in 1959, an MSEE degree from Massachusetts Institute of Technology in 1960, and his PhD from the University of Michigan in 1967. He is a registered Professional Engineer, a member of four professional and honorary societies, and has received numerous honors and awards. He has numerous publications, including his text Introduction to Communication Systems, soon to be released in its third edition. Dr. Stremler has worked at the Environmental Research Institute of Michigan and its University of Michigan Laboratory predecessor. He currently teaches electrical engineering at the University of Wisconsin, Madison, specializing in communication theory. His affiliation with the University of Madison dates back to 1968, during which time he served as both Associate Dean of the College of Engineering and Department Chairman of the Department of Electrical Engineering. Stephen Anthony received his BSEE degree in 1975 and his MSEE in 1976, both from Michigan Technological University. From college, Mr. Anthony entered the Air Force as a research officer and continued his government service in civil service. He is currently the Director, System Programs Officer of the Environmental Research Institute of Michigan. Mr. Anthony has worked in the areas of remote sensing and image exploitation since graduation from college. He has had significant experience working with the operational elements of the Air Force and Department of Defense in the reconnaissance field. Research conducted under Mr. Anthonys guidance included the development of advanced tactical exploitation equipment utilizing automation to aid interpreters. His research and development experience spans the spectrum of sensor technologies, including infrared, optical, and radar sensor systems.
IEEE AES Magazine, October 1988

31

Vous aimerez peut-être aussi