Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
~2515~S
--1--
This invention is primarily directed to an improved radar
for detecting and identifying remote objects or targets by means
of electromagnetic wave energy. The method and principles to be
described may also be applied, however, to other types of wave
energy, such as ultrasonic waves in sonar systems.
~ s in known radar systems the present method involves
the utilization of various processing techniques for extracting
useful information from the return signal or waves reflected from
the target.
The radar system described here provides optimum detec-
tion and identification of targets while minimizing the effect of
background clutter and the impairments caused by the propagation
of the electromagnetic waves through space.
This is achieved utilizing a set of new concepts for sig-
nal processing, the applicability of which is made feasible by the
development of new technology for generating microwaves and in
particular digital signal processing.
Thus, in a main aspect this invention relates to method
of detection and identification of one or more remote objects by
transmitting wave energy towards the object and receiving wave
energy reflected from the object as well as processing of infor-
mation associated with the wave energy received, wherein said pro-
cessing comprises detecting a number of separate signal parameters
pertaining to the wave energy received, said parameters together
forming a measured signature relating to the object, and the meas-
ured signature is compared with a number of prestored signatures
each comprising the same separate signal parameters relating to
~25~5~5
,
known objects being of interest to the detection and the identi-
fication, characterized in that A) said separate signal parameters
are selected to relate to a number of independent observation
domains, B) there is effected a recording of those signatures among
the prestored signatures which within predetermined tolerances cor-
respond to the measured signature, C) subsequent transmission and
reception of wave energy is used for updating the separate para-
meters and thereby the measured signature, D) the updated measured
signature is compared with the recorded prestored signatures for
updated recording of a smaller number of prestored signatures which
within predetermined tolerances correspond to the updated measured
signature, E) and steps C) and D) are repeated until there remains
a small number of recorded prestored signatures, preferably only
one such signature, which defines a small number of objects, pre-
ferably only one such object, being of interest.
In order to obtain a number of separate signal para-
meters adequate for processing as stated above, it is an advantage
to use a multifrequency radar signal transmission. More specifi-
cally, multifrequency polarimetric illumination of targets makes
it possible to determine characteristics of target signatures.
With multifrequency illumination several independent feature or
signal parameter domains of the targets are revealed. In other
words this involves the simultaneous use of orthogonal signatures
for identification of targe*s or objects of interest.
Such parameters or feature domains may be the following:
i the multifrequency response for the number of scales
~SlSa~S
-2a-
investigated (target do~n-range signature),
ii the time variations of each scale selective (see i above)
multifrequency response (vibration pattern),
iii mutual coherence of time and space variations (target
rigidity),
iv the polarization properties of each scale (target sym-
metry),
v polarimetric Doppler (target torsion).
As each of these parameter sets (domains) are indepen-
dent and thus mathematically orthogonal~ a multidimensional target
estimation is performed.
Among the parameter domains mentioned above, the multi-
frequency illumination response is considered to be fundamental.
In addition thereto, the polarization properties associated with
possible target symmetry may be very useful.
51~;~5
The aspect angle of target as seen from the radar location
at any instant is also an important magnitude which could to
some extent be considered as one of the separate signal para-
meters referred to. In this context, however, the aspect angle
is treated as a variable different from the types of signal
parameters in the domains as listed above (i ...v).
In the drawing there is an illustration of a preferred
structure of a system for carrying out the method according to
this invention.
The radar system structure shown in -the drawing comprises
a transmitter or wave-form generator 11 feeding an antenna 11A,
as well as a receiver 12 with an associated receiving antenna
12A. These co~ponents are broadband coherent microwave
components. ~he transmitter and receiver antenna
may for example be combined as a single unit. Within the an-
tenna beams there is indicated a target/to be detected and
identified. T
As will be explained later the wave-form generator 11
is preferably an adaptive wave-form generator making it possible
to adapt the wave-forms transmitted in accordance with infor-
mation which may exist or may be gained with respect to the
target or objects being out interest. Correspondingly, the
receiver 12 may be a matched receiver having characteristics
which may be selected to correspond to the particular wave-form
or signature transmitted at any instant.
The signals received and detected are supplied to block
13 which comprises further detection and filtering circuits
operating according to generally known principles in order to
extract predetermined features or signal parameters from the
received wave-form. As shown by blocks 21, 22, 23, 24 and 25
these features may consist of the following parameter domains:
21 - Multifrequency signature
22 - Beat Doppler signature
23 - Mutual coherency signature
24 - Polarimetric signature
25 - Polarimetric Doppler signature.
As will be understood, these blocks 21 - 25 correspond to the
respective items i - v listed above.
~L~515~
Both analog and digital signal processing may be involved
in blocks 13 and 21 - 25, and in any case the output signals
from blocks 21 - 25 are in a digital form. There is a hori-
zontal dashed line C-C shown in the drawing, to indicate that
the functions or blocks found below that line are performed in
a suitable computer. It is to be understood that there is no
strict boundary between the more or less conventional signal
processing blocks and circuits described above and the compu-
terized functions to be explained in the following. Thus, line
C-C serves to indicate that there is at a certain level of sig-
nal processing relating to the various parameters or domains
involved, a stage where the processing is substantially taken
over by the computer in a purely digital form.
A main block or function which is typically computerized
is block 40 which is a store or target library containing known
signal parameters in one or more feature domains pertaining to
all the targets or objects being of interest for the system con-
cerned. These may for example be objects in the form of several
types of passenger aircraft which normally operate on a given
airport. The complete set of signal parameters pertaining to
a certain aircraft may be stored in block 40 for a number of
different aspect angles of such aircraft as seen from antennas
11A and 12A. Depending upon the number of signal parameters
employed, the number of different aircraft to be covered and
the number of aspect angles being relevant, the required storage
capacity in library block 40 may be very high.
In block 41 designated signature estimation there is essen-
tially performed a comparison between an actually measured signa-
ture consisting of parameter sets furnished by blocks 21 - 25
for each transmission and accompanying reception of illuminating
electromagnetic wave energy directed towards the target. Such
measured signature is then compared initially with all the
prestored signatures found in library block 40 in order to de-
termine whether one or more of such prestored signatures corres-
pond to the measured signature within predetermined tolerances.
Thus, by cross-referring to the library in this manner, the
target identity is estimated for each of the features or signal
parameters involved. The most probable target identity may be
~:~51Sf~
obtained by weighting the vote of each feature by a feature
significance factor.
At this point reference is made to blocks 26 and 27 which
relate to information extracted with respect to position and
velocity of the target. This is information being per se funda-
mental in almost all earlier and conventional radar systems.
According to the present invention such information from blocks
26 and 27 is processed further in target bearing block 30
so as to determine or estimate the aspect angle of the target.
The term aspect angle as used here, in fact comprises three
angles which together define the direction of the longitudinal
axis of a target in space.
As already mentioned above, such target aspect angle may be
introduced in block 41 together with aspect angle information
obtained from library block 40, thereby increasing the possible
amount of detail with which a target may be identified. Thus,
in a particular embodiment of this invention, having specific
advantages, the comparison of a measured signature with pre-
stored signatures is limited to prestored signatures pertàining
to the aspect angle concerned, as delivered from target bearing
block 30 through connection 31.
Thus, when estimation of target identity is completed, the
target by its own movement will cause the measured features
or signature to change if the movement is such that the aspect
angle of the target changes. This function of identity enhance-
ment by verifying aspect angle dependence is performed in block
42 which on the one hand receives data from library block 40
and on the other hand receives aspect angle data from block
30 through connection 32. Since the target signature as measured
will change in a deterministic manner as the aspect angle
changes, this results in an improved identification of the
target, based upon prestored data with respect to the target
concerned, for a range or number of aspect angles. If the
signature measured follows the expected changes, the probability
of correct identification is increased.
In connection with the above~tracking information may be
used to provide an initial knowledge of the target aspect
angle for search in the prestored signature library 40. By
~515~
tracking the target over some distance, the true bearing and
velocity of the target can be established. These data are
used to provide the aspect angle for the target identity esti-
mation, thereby reducing the search time and increasing the
probability of correct identification.
The output from block 42 may be considered as the final
product of the processing method performed by the system de-
scribed. Thus, arrow 50 indicates some form of display to be
presented to an operator or the like. The output from block 42
may, however, as indicated at 10, be fed back to the adaptive
wave-form generator 11 so as to effect possible adjustments of
the wave-form signature to ~e transmitted therefrom.
The display represented by arrow 50 may be a cathode ray
tube screen arranged for a presentation of the PPI type, indi-
cating for exa~ple a map or contours of the field surveyed,
targets detected in this field, a listing showing the number
of targets and their types, the sectors in which the respec-
tive targets are positioned, etc.
From the structure and functions described with reference
to the system in the drawing, it appears that the radar ope-
rates in a number of modes. It has a detection mode, a
tracking mode and an identification mode. In the detection
and tracking mode, enhanced sensitivity can be achieved using
the matched illumination technique. This is accomplished by
transmitting a waveform with a spectrum which is the complex
conjugate of the transfer function of the expected target for
the assumed look angle. This process of transmitting the com-
plex conjugate waveform is repeated to ensure that the approp-
riate waveform for the given target at the given look angle
is transmitted.
After the target is detected, located and its course de-
termined, the identification process is started. The time re-
quired for an identification will depend on the number of para-
meters required. Some features, such as scale selection
vibration pattern, requires an integration time whereby the
target will move a distance equal to a few times its own length.
During identification, the target is illuminated with a co-
` ~;2 515~5
herent comb of frequencies. The maximal frequency separationgives the minimum scale size investigated, and the minimum fre-
quency separation determines the largest scale, which will
correspond to the size of the largest target expected. The co-
herency of this waveform is such that the relative phase jitter
between any two frequencies after a delay corresponding to the
maximum distance to the target, is less than a fraction of a
wave period.
Upon reception, the in-phase and quadrature components of
each frequency line is detected in a homodyne detection circuit.
The bandwidth of this circuit is equivalent to the maximum
Doppler shift expected. For the chosen integration time, the
RMS power (variance) of the return on each frequency line is
determined. Using complex multiplication, two and two returns
are combined to give the scale selective return, the scale
corresponding to the frequency separation of the two frequency
components combined.
The amplitude of these products, normalized with the ampli-
tude of each of the two frequency lines gives the multifre-
quency response (signature).
Through spectral analysis of the time series associated
with each of the scale returns, the beat Doppler signature
is calculated. This gives the scale selective vibration
pattern. For targets like aircraft the Doppler shift will be
proportional to the frequency separation and the speed of the
target, whereas for clutter like the sea surface, the Doppler
shift will be a function of the square root of the frequency
separation only.
The mutual coherency filter gives the rigidity of the
target through time-space coherency filtering.
By using two separate transmitting and receiving channels,
the polarization signature of each scale is provided in the
form of a scattering matrix.
The polarimetric Doppler signature reveals the time
variations of the symmetry properties of the target as it moves.
This signature is revealed using spectral analysis of each
individual element of the scattering matrix and also correlation
between the three independent elements of the scattering matrix.
~s~s
The complete set of identification parameters is obtained
over the integration period wherein the radar is illuminating
the target. This integration period might in the case of
a pulsed radar or angular scanning radar be interrupted,
provided the sampling frequency is high enough to reveal the
vibrations of the target.
Thus, in subsequent transmission and reception of electro-
magnetic wave energy the separate parameters and thereby the
measured signature are updated. The updated measured signature
is compared with the previously recorded prestored signatures
having resulted from the previous transmission/reception cycle,
for updated recording of a smaller number of prestored signa-
tures which within predetermined tolerances correspond to the
updated measured signature.
The process whereby the measured signature is compared
with the prestored signature can be any of the methods described
in estimation theory. In the first approximation, an estimate
of the look angle of the target (being unknown at this point),
is calculated so that only the set of signatures for this look
angle for all the targets in the target library is used.
By repeating these cycles or method steps in analogy with
the principles of successive approximations, there will finally
remain a very small number of possible targets, and preferably
only one and the correct target being identified with a high
degree of certainty.
In the process of selecting the one target, only the set
of possible targets selected in the previous iteration is
further investigated. Furthermore, for each updated signature
a new updated estimation of the look angle is calculated, so
that the reference signature for the possible target investi-
gated in the new iteration is the signature corresponding to
the new look angle of the target. Through the coupling of the
aspect angle information and signature measurements increased
probability of correct identification is accomplished.
Experts in this field will understand that the method de-
scribed here may be introduced in existing coherent radar in-
stallations by having these retrofitted with the required signal
processing and identification means.
