Note: Descriptions are shown in the official language in which they were submitted.
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
REDUCTION OF RADAR ANTENNA AREA
FIELD OF THE INVENTION
The present invention relates to a radar signal pulse
protocol and signal processing system allowing a reduction
in the size of an antenna; the present invention is
particularly useful for space based synthetic aperture
radar systems.
BACKGROUND OF THE INVENTION
In traditional methods of synthetic aperture radar
(SAR) imaging, imaging is pe:rfonaed by processing several
radar signal pulses reflected along the length of a single
determined swath or band of 'terrain. In particular, each
radar pulse is typically reflected from a distinct terrain
area, denoted a "footprint," wherein a series of
overlapping footprints are used in determining an image for
some portion of a swath. The aperture area (hereinafter,
also simply denoted as "area") of a synthetic aperture
radar (SAR) antenna is conventionally determined by at
least one of the constraints;. (1.1) the gain required
to achieve a desired sensitivity, and
(1.2) the theoretical minimum area required to reduce
signal reflection ambiguities to acceptable levels.
For some SAR systems, the second constraint leads to larger
antennas than may be required by gain considerations of
constraint t1.1) alone. At L-band, for example, a typical
orbiting SAR system on a platform operating at 700 lan
altitude and designed to image the earth at a 55° incidence
angle heretofore required a minimum antenna area of
approximately 50 m2. Such a 7.arge antenna becomes a strong
design driver for a transporting spacecraft from a mass and
volume standpoint and additionally requires complex
procedures for deployment of such a large antenna.
When imaging, for example, the earth from an orbiting
platform, the ambiguity-driven theoretical minimum area
constraint (1.2) above arises from the desire to reduce the
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
a
ambiguities in the reception of reflected radar
transmissions simultaneously in both: (a) the directions
substantially perpendicular to the trajectory of the
orbiting SAR platform (these directions denoted hereinafter
the "range" dimension), as well as, (b) the directions
substantially coincident or parallel with the trajectory of
the platform (these directions denoted hereinafter the
"azimuth" dimension). In the range dimension, the maximal
distance along the surface: where a radar signal is
reflected is denoted the "elevation beamwidth ground
projection." For unambiguous reception of reflected
signals in the range dimension, a range ambiguity
constraint exists wherein the elevation beamwidth ground
projection must be sufficiently small so that a reflected
signal pulse from any portion of the illuminated footprint
is not received simultaneous with the reflected signal
associated with a different pulse. In the azimuth
dimension, the maximal distance along the surface where a
radar signal is reflected (i.e., the footprint) is denoted
the "azimuth beamwidth ground projection." For unambiguous
reception of reflected radar signals in the azimuth
dimension, an azimuth ambiguity constraint exists wherein
the azimuth beamwidth for the SAR antenna must be
sufficiently small that the Doppler bandwidth of a received
reflected signal can be properly sampled at the pulse
repetition frequency (PRF) of the radar. Accordingly, the
range ambiguity constraint imposes an upper limit on the
PRF while the azimuth ambiguity constraint imposes a lower
limit. Consequently, assuming the pulse sampling rates are
substantially the same in the azimuth and range directions,
equating the corresponding two limits leads to the standard
ambiguity theoretical constraint on the minimum value for
the area of the antenna, as described, for example, in (a)
Elachi, C., T. Bicknell, R.L. Jordan, and C. Wu (1982)
Spaceborne Synthetic-Aperture Imaging Radars:
CA 02318074 2000-07-12
WO 99!36797 PCT/US98/00836
:3
Applications, Techniques, and Technology. Proc. IEEE, 70,
(Oct. 1982). 1174-1209: (b) Curlander, J.C., and R.N.
McDonough (1992) Synthetic Aperture Radar, John Wiley and
Sons, New York, 1991, 154-171 and (c) Hanger, R.O. (1970)
Synthetic Aperture Radar Systems, Academic Press, New York,
1970, these publications being incorporated herein by
reference. Accordingly, when transmitting and processing
radar signals in accordance with the assumptions for the
standard ambiguity theoretical constraint, an antenna area
smaller than the minimum theoretical value for a SAR system
has heretofore resulted i:n ambiguities in detecting
reflections of radar transmissions wherein such ambiguities
produce unacceptable distortions or noise in resulting
images. Moreover, as one: skilled in the art will
understand, this theoretical minimum antenna area is
proportional to sin6/cosz6, where 8 is the incidence angle
of a swath being illuminated by the radar transmissions.
Thus, increasingly larger antennas are required as the
incidence angle between the antenna and the swath increases
toward 90°. Thus, the antenna minimum area constraint
becomes particularly problematic for radars designed to
image at large incidence ang:Les.
Accordingly, it would be advantageous to have a method
and system for relaxing or mitigating the theoretical
antenna minimum area constraint without generating
unacceptable signal ambiguity as described above.
SUMMARY OF '.SHE INVENTION
The present invention is a method and apparatus for
allowing the minimum antenna area constraint to be offset
or relaxed by: (a) transmitting signals with a novel format
and (b) processing the reflections of the transmitted
signals differently from conventional SAR signal processing
systems. The present invention partitions a fundamental
radar pulse period (herein also denoted the "fundamental
CA 02318074 2000-07-12
WO 99136797 PCT/US98/00836
4
period" and/or the "pulse interval") into N sub-pulse
intervals with each such sub-pulse interval transmitting a
signal whose reflection is distinguishable from the
transmitted signals in the other sub-pulse intervals of the
fundamental period. Such partitioning by the present
invention effectively increases the PRF of the radar along
the azimuth dimension. However, since the sub-pulses are
also provided in a manner making at least one such sub-
pulse reflection in each fundamental period distinguishable
from sub-pulse reflections _Ln other fundamental periods,
the range ambiguity constraint is still determined by the
fundamental pulse period rather than the sub-pulse period.
Thus, use of sub-pulses in this manner provides for a
reduction in the antenna area below that of the standard
minimum theoretical antenna area constraint. In fact, by
using N sub-pulses in each fundamental period, a reduction
in antenna area by a factor of N is achievable while
providing images of acceptable quality.
Accordingly, the present invention may be utilized for
generating and processing data derived from such sub-pulse
reflections when these reflections are received by a SAR
sensor system residing on, for example, a satellite or
platform orbiting the earth.. In particular, the derived
data is typically generated on board the satellite or
platform and subsequently transferred to a ground based SAR
processing system for performing sampling and ambiguity
reduction on the derived dlata. Accordingly, it is an
aspect of the present invention for the SAR processing
system (or any other system with similar functionality) to
sample the derived data from (and ordered according to the
reception of) each sub-pulse reflection at two different
sampling rates in order to reduce signal reception
ambiguity in the derived data. That is, the derived data
is sampled at a first sample rate that is at least as
frequent as frequent as a first repetition frequency
CA 02318074 2000-07-12
WO 99136797 PCT/US98/00836
determined for reducing ambiguity along the direction of
the trajectory of the platform due, substantially, to
Doppler bandwidth increases :in the sub-pulse reflections.
Additionally, the derived data is sampled at a second rate
5 that is less than the first repetition frequency for
reducing ambiguity, wherein this second rate is utilized to
reduce ambiguity in the derived data resulting from an
extended time over which a sub-pulse reflection may be
detected in the direction traverse to the trajectory of the
platform (i.e., the range di:rection). More particularly,
the second sampling rate is no greater than the repetition
frequency of the fundamental period or pulse interval.
Accordingly, in one embodiment of the present invention,
the first derived data sampling rate is the sub-pulse
repetition frequency and the second derived data sampling
rate is the repetition frequency for the fundamental period
or the pulse interval.
In one embodiment of the present invention, the sub
pulses within a fundamental period are orthogonal to one
another in that each sub-pulse signal transmission is
within a bandwidth not overlapping the transmission
bandwidth of any other sub-pulse within the fundamental
period. In particular, for linear FM (chirp) radars, the
radar bandwidth is increased by N (beyond that required to
achieve the desired range resolution) and N sub-chirps are
generated in each fundamental period wherein each sub-chirp
is offset in center frequency from the other sub-chirps.
The result of this sub-pulse partitioning technique is a
SAR imaging system with an antenna considerably smaller
than heretofore allowed.
Other features and benefits of the present invention
will become apparent from the detailed description with the
accompanying figures contained herein.
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
6
BRIEF DESCRIPTIa~N OF THE DRAWINGS
Fig. 1 shows a satellite 10 having an antenna 22 for
imaging a swath 18 of a body 14;
Fig. 2 presents a block diagram of the SAR sensor
system 204 used for transmitting and receiving radar
signals according to the present invention from a platform
above the area about which data is being collected;
Figs. 3A and 3B are a flowchart for the SAR sensor
system 204 wherein the steps of the present flowchart
describe the processing fo:r transmitting and receiving
signals according to the present invention:
Fig. 4 is a block diagram for the present invention
illustrating the architecture for a SAR processor system
404 for processing data provided by the SAR sensor system
204 into, for example, images;
Fig. 5 is a flowchart of the SAR processor system 404,
wherein this flowchart desc~,ribes the steps performed in
processing the SAR sensor system 204 data into, for
example, images corresponding to the area from which
reflected signals have been received;
Fig. 6 presents a general representation of a pulse
format for the present invention where the sub-pulses use
a linear FM modulation. Each of the N sub-pulses have
center frequency f;, bandwidth B;, duration z;, and start time
2 5 t;;
Fig. 7A presents a timing diagram illustrating the
constraints on PRF selections from transmit interference;
Fig. 7B presents a timing diagram illustrating the
constraints on PRF selections from nadir interferences;
Fig. 8A presents an implementation of a linear FM sub-
pulse format for the present invention wherein the sub-
pulses are continuous;
Fig. 8B presents an implementation of a linear FM sub-
pulse format for the present invention wherein the sub-
pulses are separated;
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/0083b
'7
Fig. 8C presents an implementation of a linear FM sub-
pulse format for the present invention wherein the sub-
pulses are equally spaced through time
DETAILED DESCRIPTION
Fig. 1 illustrates a platform or satellite 10
traveling above a body 14 such as the earth and imaging the
body along the swath 18 using, for instance, synthetic
aperture radar (SAR). That is, the platform 10 transmits
pulsed radar signals toward the swath 18 and detects
reflections of the transmitted signals thereby generating
data relating geographical areas to detected signal values
(e. g., such reflected signal values being complex voltage
values corresponding to the complex amplitude of the
reflected electromagnetic signals and denoted for an entire
geographical area as a "radar cross section amplitude").
Accordingly, particular characteristics of the geographical
areas may be determined or imaged using such detected
signal values. In particular, the following,
characteristics for example, may be determined: positions
of natural and/or artificial objects such as roads and
buildings, soil moisture, vegetation type, mineral type,
ice/snow cover, topographical characteristics of an area,
ocean surface and sub-surface features.
The platform 10 includes an antenna 22 having a width
26 and a length 30 for transmitting the radar signals and
receiving the reflected radar images of the swath 18. That
is, the platform 10 detects reflected signals of a single
footprint 34 at a time by transmitting one or more radar
pulses toward the footprint. Accordingly, as the platform
10 travels above the body 14 in the direction of arrows 38,
reflected signals from overlapping footprints (e.g., 34 and
34') are received as the radar pulses generated by the
platform are directed toward successive footprints along
swath 18 in the direction of arrows 38.
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
8
It is important to note that the beamwidth 42 (i.e.,
the angular dimension of the widest portion of a footprint
in the directions of the double-headed arrow 40, and known
as the "azimuth beamwidth" in the art) varies inversely
with the length 30 of the antenna 22. Thus, as the length
30 increases, the beamwidth 42 decreases and vice versa.
Further, as the beamwidth 4:? increases, spectrum Doppler
shift effects become increasingly more pronounced, thereby
increasing the spectrum of Doppler bandwidth of the
reflected signals received by the platform 10. Moreover,
it is well-known that for reliable signal detection the
signal sampling rate (i.e., e:quivalently, the PRF) must be
greater than the received Doppler signal bandwidth to avoid
bandwidth aliasing and thus unambiguously detect the
received signals. Thus, the' platform 10 is subject to a
constraint (the aforeme3ztioned azimuth ambiguity
constraint) wherein, to obtain substantially unambiguous
reflected signal detection in the azimuth dimension (i.e.,
in the direction of double headed arrow 40), the length 30
of the antenna 22 is directly related to the pulse
repetition frequency (PRF) which has a lower bound of
acceptable values. Accordingly, the antenna length 30 is
typically increased when other constraints lead to
selection of a PRF which is less than this lower bound.
As the incidence angle 8 between the antenna 22 and
the footprint 34 being imaged increases, the swath 18
shifts perpendicularly to the double-headed arrow 90 (in
the direction of arrow 44a) further away from directly
below the platform 10. Accordingly, the width 46 of the
swath (i.e., the longest portion of footprints 34) , and
the footprints 34 therein, increase. Thus, due to the
increase in the difference: in the signal travel time
between the nearest portion of a footprint 34 and the
furthest portion (in the range dimension), the PRF must be
reduced or the swath 18 must have a smaller swath width 46
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
9
to avoid signal ambiguity resulting from reflected signals
overlapping from different radar pulse intervals. Thus,
the platform 10 is subject to a constraint (the
aforementioned range ambiguity constraint) wherein to
obtain substantially unambiguous reflected signal detection
in the range directions (i.e., in the directions of double-
headed arrow 94), the width 26 of the antenna 22 is
directly related to PRF. Typically, this has meant an
increase in antenna width 25 to thereby decrease the swath
width 46 since the above-mentioned azimuth constraint
specifies a lower bound on the PRF.
Thus, in a conventional design of a platform 10,
either or both the width 26 and the length 30 of an antenna
22 may be unacceptably sma3-1 in order to unambiguously
detect reflected signals in both the range and azimuth
dimensions.
Mathematically, the relationships between the range
and azimuth ambiguity constraints on the PRF in relation to
the antenna dimensions may be described as follows. For an
antenna 22, ambiguous range main lobe responses can be
minimized by selecting the PRF such that
f ~ fW cost 8 [1.1]
2h(3Wsin6
where fp is the PRF, f is the carrier frequency of the
transmitted pulse, W is the antenna width 2C, h is the
satellite 10 altitude, Q" is a: factor that accounts for beam
broadening, and B is the incidence angle. Note that a flat
earth is assumed for this simple analysis. Similarly,
unambiguous azimuth responses can be minimized by selecting
the PRF such that
2~L [1.2]
CA 02318074 2000-07-12
WO 99/36797 PC'T/US98/00$36
1.0
where v is the satellite 10 speed, (3~ accounts for beam
broadening, and L is the antenna length 30. Equating these
limits leads to the standard constraint on antenna area
given by
antenna area = W. y ~ 4vh(3WaLsinB [1,3]
f c~os28
Accordingly, note that equation [1.3] is the equation that
should be modified to account for such practicalities as
earth curvature.
~,,~ axed Minimum Area Constraynt
For mathematically describing the present invention,
consider a transmitted signal S(t) which is the sum of N
sub-pulses sl (t) , defined by
N
S(t) = ~s~(t) [1.4]
i =1
with the ideal orthogonality property
f Wu~i (t' t)s;(t~)at~ = si;,'ri(t) [1.5]
where a~ is the Kronecker delta and Is(t) is the impulse
response corresponding to S~. The impulse response of a
pulse-compression radar using transmitted signals of this
form is
g(t) = A jmST (t~-t) SR(tWt1 dt~ [1.6]
where A is a complex scale factor, SR (t) is the receive
signal, ST (t) is a replica o:E the transmitted signal S (t) ,
and r is the time delay between transmission and reception.
CA 02318074 2000-07-12
W4 99136797 PCTIUS98/00836
1.1
If we assume that SR (t-z) - S (t-z) and choose Sz (t) - S (t) ,
this can be written in terms of the sub-pulse format as
g(t) = A~ f ~si (t~ t)si(t~-T)dt~ [1.7]
where the orthogonality property has been applied. The
signal S tt) can thus be treated, for the purpose of
sampling the uncompressed azimuth signal of the SAR, as an
equivalent set of N independent pulses occurring within
each nominal pulse period. F'or a SAR operating at a pulse
repetition frequency of fp, the effect is to increase the
azimuth sampling frequency to Nfp while retaining the range
ambiguity properties associ~ited with a pulse repetition
rate of fP. The particular representation of the delay time
z, which depends on the processor implementation, does not
affect the analysis.
Accordingly, the N samples repeated every pulse period
represent a sampling of the azimuth waveform with an
equivalent PRF fp given by
p = N p [1.8]
where fp is called the fundamental PRF. The azimuth
ambiguity equation can now be modified assuming this change
to the sampling process, giving
fp > ~~)ii [1.9]
and leading to a modified form of the minimum area
constraint given by
4vh~i p sin6
Are tenna Area > W I [ 1.10 ]
Nfcos'~6
By processing the received signal in the manner
described above, it is thus possible to perform SAR imaging
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
1.2
using an antenna 22 significantly smaller than the minimum
size imposed by traditional theory.
The present invention c:an be utilized with various
waveform division techniques. For example, both frequency
division and code-division waveforms can be used to
implement the sub-pulse technique. The effectiveness of
the selected waveform division technique may be evaluated
by determining the degree to which it satisfies the
orthogonality condition embodied in equation [1.5].
In one embodiment of the present invention for a
frequency-division system, the sub-pulses have no common
frequency components. As a consequence, the range
resolution following range correlation corresponds to the
sub-pulse bandwidths. If, for example, the total bandwidth
B is divided equally among N sub-pulses, the range
resolution following range resolution is N times as large
as a SAR employing the full bandwidth B without sub-pulses
due to the fact that each sub-pulse has a bandwidth given
by B/N. As discussed in subsequent sections, this
resolution reduction can be recovered during the azimuth
correlation processing.
In an alternative embodiment for a code-division
system, even though common frequency components exist, the
orthogonality can be achieved by choosing waveforms for the
pulses (i.e., codes) that minimize the orthogonality
integral of equation [1.5], a.s one skilled in the art will
appreciate. Note, however, that while this approach can be
used to maintain the original bandwidth (and thus range
resolution), the trade-off is an increase in noise power in
the processed signal due to t:he presence of non-orthogonal
signals.
It is important to note that the present invention
allows for the reduction in minimum antenna area through a
decrease in antenna length, antenna width, or some
combination of length and width. However, since the sub-
CA 02318074 2000-07-12
WO 99136797 PCTIUS98/00836
:13
pulse approach does not modify the fundamental PRF, range
ambiguities are unaffected by the sub-pulse format.
Consequently, the antenna width 26 can only be reduced if
the range dimension was already oversampled or if the
fundamental PRF is reduced. 'The range resolution is larger
by the ratio of the original bandwidth to the sub-pulse
bandwidth. The antenna lengi:h 30, however, can be reduced
by an amount consistent with a beamwidth that produces a
Doppler bandwidth that is properly sampled at the total
PRF. In general, such an :increase in Doppler bandwidth
corresponds to a comparable improvement in azimuth
resolution. For example, doubling the Doppler bandwidth of
the reflected signals allows: the azimuth resolution to be
smaller by a factor of two. A typical processor will
compensate for the resolution change by processing multiple
azimuth samples to achieve an azimuth resolution matched to
the range resolution. Alternatively, the additional
samples can be used to reduce speckle (i.e., a reduction in
the statistical variation in measurements of the signal
reflection through multi-look processing where several
samples are averaged to obtain a single value for the
reflected signal measurement).
In some embodiments of t:he present invention, the sub
pulse format could potentially create range ambiguities in
a fundamental pulse interval due to time or frequency
sidelobes from the other sub-pulses. However, these
ambiguities can be minimized by one or more of the
following: (a) increasing the sub-pulse length: (b)
weighting or coding the pulses: and/or (c) including gaps
between sub-pulse bands.
A SAR image processing system utilizing the present
invention includes two subsystems:
(2.1) a SAR sensor system, typically located on a
spacecraft, aircraft, or other platform, and used for
generating radar signals and receiving them after having
CA 02318074 2000-07-12
WO 99136797 PCT/US98100836
~L4
been reflected from a body such as, for example, the body
14 ~ and
(2.2) a SAR processor system, used (a) for processing
data received from the SAR sensor system, (b) for
generating images of the bode and (c) for generating other
information which can be derived from data corresponding to
signals reflected by the body and received by the SAR
sensor system. Note that the SAR processor system is
generally not located on t:he platform 10 with the SAR
sensor system due to t:he considerable amount of
computational hardware required for image processing.
However, there are embodiments for which the SAR sensor
system and the SAR processor system may be located on the
same platform 10, as one ski7.led in the art would be aware
of .
Fig. 2 shows a block d:Lagram of a SAR sensor system
204 embodying the present invention. The blocks in this
diagram may be implemented as distinct computational units,
consisting of electronics hardware and associated software,
although in some cases the functionality of a number of
such computational units have been combined. The SAR
sensor system 204 can be logically divided on the basis of
two functions: a signal transmission function and a signal
reception function. The signal transmission function
includes the functionality for generating and transmitting
a SAR radar signal. The signal reception function includes
the functionality for receiving and storing a reflected SAR
radar signal. The hardwar~e/software units for the two
functions are generally combined within a SAR sensor system
on a single platform 10, using substantially different
hardware/software units for the two functions but employing
common hardware/software units for some purposes, as is
described hereinbelow. Note, however, that in some
embodiments, the transmission and reception functions are
performed on separate platforms 10. In such cases,
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
1.5
otherwise common hardware/software units are, of course,
reproduced on each platform.
In the single platform embodiment of a SAR sensor
system 204 shown in Fig. 2, the timing and control unit 208
provides reference signals, timing functions, and control
functions for the entire SAF; sensor system. The exciter
unit 212 modulates and filters a reference signal. The
transmitter unit 216 amplifiers and filters the signal from
the exciter unit 212. The anl:enna unit 220, which includes
antenna 22, transmits the signal from the transmitter unit
224 and receives signals reflected from the body 14. The
receiver unit 224 amplifies, filters and frequency shifts
the signal from the antenna unit 220. The digital
electronics unit 228 digitizes and processes the signal
from the receiver unit 224. '.Phe data transmission unit 236
transmits data received from either the digital electronics
unit 228 or the data storage unit 232. In this embodiment,
the signal processing and storage is performed
substantially using radio frequency(RF) and digital signal
processing electronics. In alternate embodiments, some
portion of the processing and storage may be performed
using optical techniques.
Fig. 2 also shows a block diagram of the internal
structure of the timing and control unit 208. The units
within the timing and control. unit 208 may be described as
follows : The master control7~~er unit 248 provides overall
control for the SAR sensor 204 including control and
coordination of the constituent units. The nominal PRF
pulse and timing controller unit 252 generates and
distributes control information concerning the fundamental
pulse characteristics. The sub-PRF pulse and timing
controller unit 256 uses the information generated by unit
252 and additional information from unit 208 to generate
and distribute control information concerning the sub-pulse
characteristics. This control information is used to
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
1.6
control the radio frequency (RF) signal generated by the
stable local oscillator unit (240) after it has been
frequency multiplied in the frequency multiplier unit 244.
The timing and control unit 208 also includes an antenna
controller unit 260, which controls antenna functions such
as beam steering angle, and a miscellaneous timing and
command controller unit 264, which controls the receiver
224, digital electronics unit 228, and assorted
miscellaneous functions. Note that at least one of these
units internal to the timing and control unit 208 is
entirely novel and unanticipai:ed in conventional SAR timing
and control units: i.e., the sub-PRF pulse and timing
controller unit 256. This new unit, which may be
implemented in hardware or as a software module, provides
the control functions, tim5_ng, and waveform parameters
necessary to generate sub-pulses with a frequency, format
and timing differing from that of pulses at the nominal or
conventional PRF. Determination and generation of these
control and timing parameters involves algorithms
fundamentally different from those used for determining
nominal PRF and timing. For example, determination of the
nominal PRF pulse and timing parameters depends
substantially on knowledge of the radar frequency, the
platform speed/altitude, the antenna dimensions, and data
window timing issues. Determination of sub-PRF pulse and
timing parameters depends in addition on signal
orthogonality, signal-to-noise concerns, and more
constrained data window timing issues such as transmit
interference and nadir interference. Thus, pulse and
timing calculations for both the nominal PRF and sub-PRF
are required to achieve the desired radar waveform and
control functions must be generated and issued at both the
PRF and sub-PRF levels.
In the present invention, one or several of the units
of the SAR sensor system 204 is modified to achieve the
CA 02318074 2000-07-12
WO 99/36797 PCTNS98/00836
17
proposed improvement as compared with a SAR sensor that
does not employ sub-pulse, according to the present
invention. In one embodiment, the antenna unit 220 is
modified: (a) by reducing the antenna 22 aperture area, or
(b) by introducing electronic or mechanical devices which
provide for transmitting anti receiving out of only some
portion of the otherwise unmodified aperture. In a second
embodiment, the antenna unit 220 is modified in a similar
manner and the timing, duration and modulation (but not the
total bandwidth of the transmitted signals) of the pulses
is also modified. In particular, the pulse format may be
modified to provide a pulse sequence with two distinct
pulse frequencies, one being the nominal pulse frequency of
a traditional SAR system and another being a higher
frequency generated by introducing additional pulses ("sub-
pulses") in the pulse stream. This change in the pulse
format requires modification of the timing and control unit
208. In a third embodiment, the total bandwidth of the
transmitted signal is also modified. This may require
further modification of all units, including the
transmitter unit 216 and the antenna unit 220, to
accommodate the increased bandwidth and provide correct
bandpass characteristics.
Figs. 3A and 3B show a flow diagram of the operation
of the SAR sensor system. The desired timing and duration
of the radar pulses is first calculated (step 304).
Subsequently, the desired timing and duration of sub-pulses
is calculated (step 308). The values calculated in steps
304 and 308 are then used (~:tep 316)to generate pulse and
sub-pulse signals from the stable oscillator signal
generated in step 312. Theae pulses and sub-pulses are
then modulated (step 320), translated to an intermediate
frequency and bandpass filtered (step 324), then translated
to a carrier frequency and again bandpass filtered (step
328). The pulse signals are then amplified (step 332) and
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
1.8
transmitted using the antenna 22 (step 336). The signals
reflected from the body 14 are: received through the antenna
22, amplified (step 344) and translated to an intermediate
frequency and bandpass filtered (step 348). Subsequently,
the signals are translated to a video frequency and
bandpass filtered (step 352). The signals are then
digitized (step 356} and processed to achieve the desired
digital format, and then either stored or transmitted (step
360) .
Fig. 4 shows a block diagram for an embodiment of a
SAR processor system 404 fo:r the present invention. In
this embodiment, the SAR processor system 404 includes
digital computation devices and software which may be
implemented within a singles computer or using multiple
computers. The nominal control parameter generator 408
uses sensor data output by the SAR sensor system 204 as
well as ancillary data to calculate parameters that are
used to control processing of the sensor data. The nominal
control parameter generator 408 provides pulse correlation
control parameters to the sub-pulse control parameter
generator 410. The sub-pulse control parameter generator
410 provides the control parameters required to perform the
processing at the sub-pulse level. This control
information is derived from tlhe data itself as well as from
known information about the SAR sensor system and the
transmitted waveforms. The sub-pulse control parameters
are calculated using substantially different algorithms
than those used to calculate the nominal control
parameters, but incorporate parameter information generated
by the nominal control parameter generator 408. For
example, the following differences in algorithms are
necessary: the sub-pulse control parameters must account
for differences between the sub-pulses contained in each
pulse sequence, such differences being typically frequency,
phase, or code differences used to distinguish the sub-
CA 02318074 2000-07-12
WO 99136797 PCT/US98/00836
19
pulse from each other. The sub-pulse control parameters
also have to account for differences in the data window,
which is typically modified to accommodate the sub-pulse
signals. The range correlation processor 412 employs the
control parameters to perform a range correlation on the
sensor data, allowing the data to be transformed or
"compressed" from a large time-bandwidth product format to
a nearly unity time-bandwidth product format as one skilled
in the art will understand. The sub-pulse segregation
process 416 iterates the range correlation procedure for
each of the N sub-pulses or sub-pulse sequences within a
nominal pulse period and provides appropriate frequency and
phase corrections to make the sub-pulses appear
substantially identical subsequent to range correlation
processing. Because the reflected signals corresponding to
each sub-pulse are only distinguishable by virtue of the
orthogonality properties of the sub-pulse waveform, range
correlation processing must be done separately for each
sub-pulse. The range compressed sub-pulses are then
recombined to achieve a sample of the azimuth signal with
N times the sample rate associated with the fundamental
PRF. Subsequently, the corner turn memory 420 stores and
translates the sensor data from range processing to azimuth
processing. The azimuth correlation processor 424 employs
the control parameters to perform an azimuth correlation on
the sensor data, allowing t:he data to be transformed or
"compressed" using a standard technique in radar signal
processing from a large time-bandwidth product to a near
unity time-bandwidth product as one skilled in the art will
understand. The sub-pulse resampling processor 428
corrects the two-dimensional data for resolution and
geometry artifacts introduced by the use of the sub-pulse
waveform. For example, it is typically necessary at this
stage to coherently combine: data samples in the azimuth
dimension so as to match t:he azimuth resolution to the
CA 02318074 2000-07-12
WO 99136797 PCTIUS98100836
range resolution within the image. The multi-look
processor 432 processes the two-dimensional data set
generated by the previous processors to generate an image
of the desired resolution and characteristics.
5 Additionally, it is important to note that in an
alternative embodiment, the SAR processor system 404 may
consist substantially of optical computation devices.
Note that the present invention requires substantial
modification to a conventional SAR processor system to
10 accommodate changes to the data introduced as a result of
modifications to the SAR sensor system 204. The nominal
control parameter generator 408 must be modified through
the addition of the sub-pulse control parameter generator
410 to provide reference functions of a different format
15 than used in a conventional processor as a result of the
changes in timing, duration,. modulation and bandwidth of
the transmitted signals. Both the range correlation
processor 412 and the azimuth correlation processor 424
require corresponding modifications. These modifications
20 are typically software changes that account for the
differences in pulse format. In particular, sub-pulse data
must be modified in an appropriate fashion within the range
correlation processor 412 and the sub-pulse segregation
processor 416 to generate corresponding range compressed
pulses that appear to the azimuth correlation processor 424
as a substantially equivalent series of pulse samples of
the azimuth signal. The azimuth correlation processor 424
must also be modified to accommodate these pulse samples
which are received at a higher rate and with possibly
different formats than in a standard configuration. In
some embodiments, the increased azimuth beamwidth 42 of the
transmitted signals, resulting from the reduction in
antenna 22 aperture, also requires modification of the
algorithms used to provide t:he azimuth correlation within
the azimuth correlation processor 424.
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
21
Fig. 5 shows a flow diagram of the operation of the
SAR processor system 404 for the present invention. The
data received from the SAR sensor system 204 are used to
generate control parameters for the pulse and sub-pulse
signal data (steps 508 and 512). These parameters are used
to control the range correlation process of step 516 which
generates range compressed data that are accumulated and
corrected through the process of step 520. The data are
transformed using a corner turn process (step 524) and, in
conjunction with the control parameters, processed through
an azimuth correlation (stE:p 528) to generate a two-
dimensional data set which is compressed in both range and
azimuth. Subsequently, the data output from the azimuth
correlation step is provided to step 532 for resampling to
remove artifacts introduced through use of the sub-pulse
waveform, such as differences between the range and azimuth
resolutions. In step 53Ei, multiple data looks are
processed to achieve the desired data resolution and image
quality. Finally, in step 540, an image corresponding to
the reflected signal distribution is output.
EXAMPLE.: FREQUENCY-DIVISION WAVEFORM
Consider a sub-pulse radar system for the present
invention that is designed to use linear FM (chirp) signals
at multiple frequencies for implementing orthogonal sub-
pulses. Assuming the SAR transmitter is designed such that
the bandwidth of the transmitted chirp signal is a factor
of N times that required to achieve the desired range
resolution (for simplicity, let N be an integer) , then if
pulse compression is performed separately on each of the N
sub-bands, the result is a set of N samples of the SAR
azimuth waveform for each pulse.
To obtain image data derived from the two level signal
sampling technique of the present invention, the frequency
offsets of the sub-pulses must be removed. This can be
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
22
done with conventional SAR pulse hardware, requiring only
knowledge of the transmitted signal characteristics (i.e.,
such characteristics as frequency, phase and amplitude of
the signals as a function of time). Further, such
conventional signal pulse processing allows for the
received signal to be frequency translated or demodulated
using an appropriate oscillator frequency fm prior to on-
board digitization and data storage, as one skilled in the
art will understand. Accordingly,: assuming that the
starting phase of each sub-pulse is the same, the primary
modification to the conventional pulse processing involves
adjusting the range compression algorithm to cross
correlate the received signal with replicas of each sub-
pulse separately, as one skilled in the art will
appreciate.
Note that this change is easily implemented since it
requires only an adjustment of the center frequency and
phase of the correlator reference function, as is discussed
below.
For a pulse centered at f - fo with a large time-
bandwidth product, the system impulse response is given as
a function of time t for the nt° pulse by [2]
f
gn ( t) = Anexp [ j2n (fo -fm) t] exp [-,j4nRn-°,sinc ( u) [2.1]
c
where .~" is an overall complex amplitude, R" is the range to
the body for the center of the nt'' pulse, c is the speed of
light, fa is the center frequency of the transmitted signal,
and
a = Ktp (t-2Rn/C) [2.2J
with K the chirp rate and rp the pulse length. When a
frequency offset f~= fo+ afj is included, this becomes
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
23
f
gn~~(t) = Anexp[j2n(fo - m + aifo;i t]exp [-j9z~Rn ° (1 + a1)] sinc(n)
c
[2.3]
where crj= afjlfo with af~ the frequency offset of the ith sub-
band.
The processing modification involves adjusting the
standard demodulation step to account for the frequency and
phase offsets indicated in Equation [2.3]. This is
performed through multiplication by the demodulation
function
Di ( t) = exp [-j2n (fo - m + aifo) t]
and the phase function
Qn~ i = eXp~j 4IIRn f ~ ai~
C
where the value of R" corresponds to a fixed point in each
resolution element.
As discussed by Harger in a previously cited
reference, this frequency and phase adjustment is only
exactly valid for the point within the resolution element
corresponding to R~. At other- points within the resolution
element, a residual phase erx-or exists. For example, with
a sub-pulse sequence which employs only two sub-pulses
offset from each other by their bandwidth, the residual
phase error is ~n at the edges of the resolution element if
the adjustment is made at the center of the element. As a
result of this pulse-to-pulse phase difference, the
summation over pulses performed during azimuth correlation
processing results in signal cancellation for signals
reflected from objects not located at the center of the
element. In the example de:>cribed here, the cancellation
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
24
is exact at the edge of the element (with lesser
cancellation moving towards the element center). This
coupling between the azimuth processing and range
resolution does not occur in standard SAR processing. When
worked out, the cancellation for an N sub-pulse format
produces the exact range impulse response as a pulse with
N times the bandwidth of each sub-pulse.
All of the quantities wised in these steps are known
during the pulse processing stage, so the frequency
dependent effects of the sub-pulse transmit format can be
effectively eliminated. The result is the desired range-
compressed signal
f
gn(s) = Aoexp [-j4nRn c,sinc(u)
which is similar to the functional format for standard
pulse processing but with pulses occurring at N times the
rate in normal pulse processing.
Signal Space Implementations
The transmitted pulses i.n this example describing the
modified PRF approach of the present invention can be
implemented in any of several formats. Fig. 6 shows a
general representation of one category of pulse format
embodiments wherein the sub-pulses use a linear FM
modulation as described in 'this example. Each of the N
sub-pulses have center frequency fE, bandwidth B~, duration
zr, and start time tl. A number of factors must be
considered when selecting values for these parameters. The
fundamental pulse length, given by
~N
Lri:=1 T j i
and the fundamental PRF are determined, in part, by the
desired relation between peak and average transmitted power
CA 02318074 2000-07-12
WO 99136797 PCT/US98/00836
~5
as well as the capabilities of the transmitter hardware.
The fundamental PRF must also satisfy the range ambiguity
constraint give by Equation [1.1]. These ambiguity
considerations are the same as those used in a standard
radar without the modified PRF of the present invention.
In addition, unless multiple receive channels are
used, the pulse format must be selected so that the
standard transmit window and nadir return reference
constraints are met, as one skilled in the art will
understand and as is described in the Curlander reference
cited above. Additionally, note that these constraints are
illustrated in Figs. 7A and 7B. That is, the transmit
window constraint shown in Fig. 7A can be described as
follows : reflected pulses must arrive at the platform 10
during intervals that fall between transmit windows (the
transmit window has length given by the sum of the pulse
length rF and twice the pulse protect window rte).
Accordingly, this constraint: implies that each sub-pulse
receive window should be located in a time period that is
between the transmission windows of sub-pulses. Moreover,
as shown in Fig. 78, the nadir constraint implies that
nadir pulse returns from preceding pulses do not occur
within a current window for pulse reception. That is, Fig.
7B may be described as follows: reflected pulses must
arrive at the platform during intervals that differ from
intervals when nadir-reflected pulses arrive. However,
unlike traditional radars, the nadir constraint condition
must only be met for sub-pulses with the same frequency
being received in a particular pulse reception window (as
long as the nadir pulse return does not saturate the
receiver front-end, as one skilled in the art will
appreciate).
Three particular representations of this pulse format
may be implemented in typical radars. These are shown in
Figs. 8A, 8B and 8C. In Fig. 8A, a graph of chirp pulses
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
2. 6
804 are shown wherein each chirp pulse has N (illustrated
for 1~4) sub-pulses that are contiguous in both frequency
and time, as in a standard chirp pulse. This approach has
the advantage of being fully consistent with existing chirp
radars. Moreover, separation into sub-pulses may be
implemented entirely in the image processor without any
modification of radar hardware. The result is an
interlaced sampling in which N sub-pulses are transmitted
within the time period 808 defined substantially by the
entire contiguous chirp pulse length and substantially no
sub-pulses are transmitted during the long time gap 812
between pulses. However, such interlacing of sampling is
particularly sensitive to noise in the sampled signal.
The pulse format representation of Fig. 8B is a
modification of the first approach represented in Fig. 8A.
However, the sub-pulses are separated in time as shown by
the sub-pulse graphs 820 so that a non-transmit window B24
is provided for each sub-pul~;e. This pulse format reduces
the sensitivity to noise. However, it also decreases the
size of the maximum period between sub-pulses available for
the receive data window in any pulse period.
The pulse format representation of Fig. 8C is a
further modification of the format of Fig. 8B wherein the
sub-pulses are spaced equally over the fundamental pulse
period 828, wherein each sub-pulse is provided with a
longer non-transmit window 824. This approach minimizes
the sensitivity to noise in the sampled signal, but it also
severely limits the maximum i:nterpulse period available for
the receive data window and thus the available swath width
46. However, these undesirable effects can be mitigated
through system design trade-offs. In one embodiment, such
undesirable effects may be mitigated by allocating some of
the reduced antenna area to a width reduction, thus
lowering the minimum fundamental PRF and increasing the
potential data window. In an alternative embodiment, the
CA 02318074 2000-07-12
WO 99/36797 PCT/US98/00836
~7
undesirable effects may be mitigated by multiple data sub-
windows, typically one for each sub-pulse. While this
results in a striped image (i.e., ranges corresponding to
the sub-pulse transmit periods between data sub-windows
will have no data), the striping may be removed by
combining images from multiple looks performed at slightly
different PRF's, as one skilled in the art will appreciate.
Siq~Zal Power Considerations
The effect of the sub-pulse processing of the present
invention on the transmit power required to achieve a given
signal-to-noise performance depends on the invention
embodiment. However, in general, the signal-to-noise
performance for the present. invention differs from the
performance of standard signal processing algorithms as a
result of four factors: 1) modification of the number of
pulses integrated during range processing, 2) modification
of the number of pulses integrated during azimuth
processing, 3) modification of the antenna gain, and 4)
modification of the resolution or the bandwidth.
For example, consider a radar system with a fixed
bandwidth B, fixed peak power, and fixed pulse length
wherein the system is modified to operate at a sub-pulse
factor of N. For such a radar system, a reduction in
antenna length by the factor N and an increase in signal-
to-noise by a factor N can be obtained by absorbing a loss
in resolution by the same factor. The reduction in antenna
area decreases the two-way gain by a factor of IVz. Since
the bandwidth is the same, the noise power is unchanged.
The number of samples integrated in the range dimension is
reduced by a factor of N, but is offset by the decrease in
range resolution by the same factor. The number of samples
integrated in the azimuth dimension is increased by a
factor Nz, due to both the increased sample rate and the
increased integration time (i.e., the increased azimuth
CA 02318074 2000-07-12
WO 99/36797 PCTIUS98/00836
28
beamwidth). The reduction in azimuth resolution introduces
an additional factor of N. The gain loss of Nl is thus
offset by the azimuth integration gain of I~, giving a total
gain of N.
Note that if, instead, 'the bandwidth is increased by
the sub-pulse factor N and t:he resolution remains fixed,
the antenna length can be decreased by the factor N with a
resulting signal-to-noise loss of N.
E~rther note that reducing the bandwidth of the noise
limiting passband does not provide any advantage in terms
of signal-to-noise. While this does reduce the noise
power, it also increases the correlation time of the noise
signal by a factor N, so coherent integration gain is
obtained only for sampling a.t a bandwidth less than e/N.
As a result, the decrease in noise power is offset by the
required decrease in range sampling rate.
E~;, .
As an example, consider a spaceborne SAR designed to
produce 75 m resolution (slant range) images of the earth
on a global basis from an altitude of 700 km: The radar
operates at one of L, C-, or X-band (taken here to be 1.25,
5.3, and 9.6 GHz). In order to achieve large swaths as
well as short repeat times between images, an incidence
angle capability of 55° is desired. Using standard pulse
processing, the ambiguity :relationship implies minimum
aperture areas of approximately 49.6 m2, 11.6 m2 and 6.4 m2
for the three bands (these: values will vary somewhat
depending on the particular design). A typical
configuration for such an antenna would be a length of 19
m and widths of 2.6 m, 0.6 m, and 0.3 m. Even at X-band,
a physically large antenna is required to achieve this
global imaging mission given standard pulse processing.
By incorporating sub-pulse processing of the present
invention, considerably smaller antennas can be used to
CA 02318074 2000-07-12
WO 99/36797 PCTNS98/00836
29
achieve the same performance. Using a bandwidth of 10 MHz
and a sub-pulse bandwidth of 2 MHz (5 sub-pulses per
fundamental pulse interval), the antenna area can be
reduced by a factor of five. The antennas could be
configured with lengths of 3.8 m and widths of 2.6 m, 0.6
m, and 0.3 m. Even at L-band, the antenna size (3.8 m by
2.6 m) is relatively manageable on a small spacecraft. The
signal-to-noise of this system operating with 75 m
resolution would be a factor of 7 dB better than the full
19 m antenna operating at 15 m resolution. For the same
sensitivity, then, the transmit peak and average powers
could be reduced by a factor of five.
The foregoing discussion of the invention has been
presented for purposes of illustration and description.
Further, the description i~; not intended to limit the
invention to the form disc:Losed herein. Consequently,
variation and modification commensurate with the above
teachings, within the skill and knowledge of the relevant
art, are within the scope of- the present invention. The
embodiment described hereinabove is further intended to
explain the best mode presently known of practicing the
invention and to enable others skilled in the art to
utilize the invention as such,. or in other embodiments, and
with the various modifications required by their particular
application or uses of the irwention. It is intended that
the appended claims be construed to include alternative
embodiments to the extent permitted by the prior art.
