U.S. patent application number 12/955262 was filed with the patent office on 2011-03-24 for pulse doppler coherent method and system for snr enhancement.
This patent application is currently assigned to Elta Systems Ltd.. Invention is credited to Ella Beilin, Jehezkel Grizim, Alexander Lomes, Yacov Vagman.
Application Number | 20110068969 12/955262 |
Document ID | / |
Family ID | 37672189 |
Filed Date | 2011-03-24 |
United States Patent
Application |
20110068969 |
Kind Code |
A1 |
Beilin; Ella ; et
al. |
March 24, 2011 |
PULSE DOPPLER COHERENT METHOD AND SYSTEM FOR SNR ENHANCEMENT
Abstract
A method and system for SNR enhancement in pulse-Doppler
coherent target detection. In accordance with the method of the
invention, complex signals are obtained for each of two or more
sub-intervals of the time-on-target interval, allowing simultaneous
range and Doppler measurements. A coherent integration is then
performed on the signals to generate complex-valued folded
matrices. The folded matrices are unfolded and target detection is
then performed in a process involving one or more of the unfolded
matrices. A pulse-Doppler coherent system is also provided
configured for target detection by the method of the invention.
Inventors: |
Beilin; Ella; (Ashdod,
IL) ; Grizim; Jehezkel; (Givat-Shumuel, IL) ;
Vagman; Yacov; (Rishon LeZion, IL) ; Lomes;
Alexander; (Maccabim, IL) |
Assignee: |
Elta Systems Ltd.
Ashdod
IL
|
Family ID: |
37672189 |
Appl. No.: |
12/955262 |
Filed: |
November 29, 2010 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
12081605 |
Apr 17, 2008 |
7864106 |
|
|
12955262 |
|
|
|
|
PCT/IL2006/001161 |
Oct 5, 2006 |
|
|
|
12081605 |
|
|
|
|
Current U.S.
Class: |
342/27 |
Current CPC
Class: |
G01S 13/5246 20130101;
G01S 13/227 20130101 |
Class at
Publication: |
342/27 |
International
Class: |
G01S 13/04 20060101
G01S013/04 |
Foreign Application Data
Date |
Code |
Application Number |
Oct 19, 2005 |
IL |
171464 |
May 7, 2006 |
IL |
175465 |
Claims
1. A method for target detection in a pulse-Doppler coherent
system, allowing simultaneous measurements of the target range and
Doppler parameters, comprising coherent integration of the signal
over the entire time-on-target interval consisting of two or more
portions of time differing by at least one parameter of the wave
form, the method comprising: a) transmitting a signal generated by
a transmitter; b) receiving the signal being reflected by a target
and converting the received signal into a complex signal;
performing coherent integration of said complex signal over each of
said two or more portions of time, said portions serving as
coherent processing intervals (CPIs), thus obtaining complex
matrices each corresponding to respective CPI; c) unfolding said
complex matrices; d) generating new complex matrices by complex
value interpolation of the complex matrices obtained in step (c);
e) generating phase-corrected matrices by performing a phase
correction on each of the new complex matrices obtained in step
(d); f) performing coherent integration of the phase-corrected
matrices obtained in step (e) and calculating a corresponding
real-valued matrix; g) performing target detection involving said
real-valued matrix.
2. The method according to claim 1 wherein a) the parameter of the
waveform differing the portions of time is pulse repetition
frequency (PRF) characterizing respective CPI; b) the obtained
complex signals are characterized as
x.sub.nm.sup.(l)=x.sup.(l)(tnm), where l=1 to L is a PRF index, L
is a number of PRFs used, n is a pulse number in the signal, m is a
range gate, and t.sub.n,m is a sampling time of the signal of the
range gate m of the pulse n; c) the complex matrices X.sup.l for
l=1 to L are generated by performing coherent integration of the
signals x.sub.nm.sup.(l); a d) the complex unfolded matrices
X.sub.k'm'.sup.(l), are provided by unfolding the matrices X.sup.l
for l=0 to L-1.
3. The method according to claim 1, wherein performing coherent
integration comprises calculating a discrete Fourier transform on
one or more of the signals x.sub.nm.sup.(l) to generate one or more
signals X km ( l ) = n = 0 N ( l ) - 1 x nm ( l ) w n - 2 .pi. jkn
K , k = 1 , , K , l = 1 , , L , m = 1 , M ( l ) , ##EQU00005##
where k is an index of the Doppler frequency, K is the number of
Doppler frequencies, N is the number of pulses in the signal,
w.sub.n is a weighting factor and M.sup.(l) is the number of range
gates of the PRF 1.
4. The method according to claim 2, wherein unfolding the matrices
X.sup.l comprises: a) defining X.sub.k',m'.sup.(l) for values of m'
for which R.sub.min<mRG<R.sub.max, where
[R.sub.min,R.sub.max] is a predetermined detection region of
interest, and for values of k' for which
D.sub.min<k'PRF/K<D.sub.max, where [D.sub.min,D.sub.max,] is
a predetermined region of Doppler frequencies of interest, by
setting X.sub.k'm'.sup.(l)=X.sub.km.sup.(l), where k=k' mod K, and
m=m' mod M.
5. The method according to claim 4, wherein unfolding the matrices
X.sup.l comprises resampling the matrices X.sup.l.
6. The method according to claim 5, wherein resampling the matrices
X.sup.l comprises defining, for each pair of indices k'm', new
indices p and q by: b) dividing the interval of interest
[R.sub.min, R.sub.max] into one or more subintervals of a
predetermined length .DELTA.r; c) determining a value of p from
among all values of p for which
0.ltoreq.p.DELTA.r.ltoreq.R.sub.max-R.sub.min) such that
R.sub.p=R.sub.min+p.DELTA.r is closest to the range represented by
the range gate m; d) dividing the interval [D.sub.min,D.sub.max]
into one or more subintervals of a predetermined length .DELTA.q;
and e) determining a value of q from among all integral values of q
for which 0.ltoreq.q.DELTA.q.ltoreq.D.sub.max-D.sub.min such that
D.sub.q=D.sub.min+q.DELTA.q is closest to k.
7. The method according to claim 6, further comprising generating
one or more matrices XI.sup.(l) where XI.sub.p,q.sup.(l) is
obtained by interpolation of one or more values of
X.sub.k',m'.sup.(l) for indices k'm' in a neighborhood of the
indices p q.
8. The method according to claim 7, further comprising performing a
Doppler compensation on one or more of the matrices XI.sup.(l).
9. The method according to claim 8, wherein performing the Doppler
compensation includes calculating matrices Y.sup.(l) defined by
Y.sub.km.sup.(l)=XI.sub.km.sup.(l)e.sup.-2.pi.jD.sup.q.sup.t.sup.0.sup.(l-
).
10. The method according to claim 9, wherein performing target
detection comprises: f) calculating a real-valued matrix A where A
p , q = l = 0 L - 1 Y p , q ( l ) 2 ; ##EQU00006## and g)
determining, for each of one or more pairs of indices, whether the
value of A.sub.p,q is greater than or equal to a predetermined
threshold T, a target being detected at the location having the
associated indices p,q if the value of A.sub.p,q is greater than or
equal to T, and a target not being detected at the location having
the associated indices p,q if the value of A.sub.p,q is not greater
than or equal to T.
11. A pulse-Doppler coherent system configured for target detection
by the method of claim 1.
12. A pulse-Doppler coherent system configured for target detection
and comprising: a) a transmitter for generating and transmitting
signals, said signals characterized by their PRF, said signals
impinging on a target; b) a receiver configured to obtain the
signals reflected by the target and to convert the received signals
into complex signals, said signals corresponding to each of two or
more CPIs within a time-on-target interval, wherein each of said
CPIs are characterized by waveforms different in at least one
parameter; c) a Fast Fourier Transform (FFT) unit, for providing a
coherent integration of said signals, the output of said FFT unit
is a sequence of CPI spectra for each PRF; d) a plurality of CPI
memories for storing the respective CPI spectra for a specific PRF;
e) a plurality of interpolation units for providing a complex value
interpolation of the output of said FFT unit; f) a plurality of
unfolding units for unfolding the output of said FFT unit; g) a
phase correction unit for phase-correcting the results of the
interpolation; h) a summation unit for performing a coherent
integration of the results of the phase correction and calculating
corresponding real values; i) a decision unit for providing a
target decision.
13. A computer program product comprising at least one computer
readable medium having computer-executable instructions for
performing the steps of the method of claim 1 when run on a
computer.
14. A computer program product as claimed in claim 13 embodied on a
computer readable medium.
Description
[0001] This application is a continuation of U.S. application Ser.
No. 12/081,605 filed on Apr. 17, 2008, which is a continuation of
International Application No. PCT/IL2006/001161 filed on Oct. 5,
2006.
BACKGROUND
[0002] 1. Field of the Invention
[0003] The present invention relates to methods and systems for SNR
enhancement in a pulse Doppler coherent system.
[0004] 2. Background of the Invention
[0005] The pulse Doppler technique is common to most modern
surveillance and tracking radars, and ultrasound systems. This
technique employs a sequence of transmitted pulses which impinge on
a target, are reflected from the target and are collected back in
the receiver. This technique is particularly convenient when the
velocity of the target is significantly different from the velocity
of the background scatterers such as the ground, trees, foliage and
so on. Under this condition, the detection capability of the system
is maximized in terms of the signal to noise ratio (SNR) so that
the probability of detection is improved.
[0006] Most modern surveillance radars scan the surrounding space
using a relatively narrow radiation beam. The total scan time is
usually the user specified parameter of the system. The fraction of
scan time, allocated to collect target return from each beam
direction, is called time-on-target. During this fraction of time a
sequence of pulses is transmitted by the radar. The interval
between the rise of any two consecutive pulses is called the PRI
(Pulse Repetition Interval) and the rate of the pulses is called
PRF (Pulse Repetition Frequency). Detection and measurement
processes can be realized by using constant or variable PRF during
the time-on-target interval. The maximum SNR can be achieved by
coherent integration of all target returns during the entire
time-on-target interval. Prima facie, the most tempting scheme for
realization of such concept would appear to be to use a single
constant pulse repetition frequency (PRF) for transmitting pulse
sequence and utilization of target returns. However this scheme
does not support unambiguous measurement of range or velocity or
both.
[0007] Another problem related to a single PRF scheme of detection
is the problem of blind zones (blind ranges and Doppler
frequencies) in the detection map. This problem reflects the
periodic nature of transmitting and receiving in pulse radar
detection scheme and is known as the visibility problem.
[0008] One solution, known in the prior art, to both the ambiguity
and the visibility problems is to transmit two or more pulse
sequences consecutively, each sequence having a different PRF. Each
sub-interval with constant PRF provides a different "scale" of
ambiguous but simultaneous measurement of the target range and
Doppler frequency. The combination of all measurements (each with a
different PRF) during time-on-target interval allows ambiguity
resolution, but requires independent attempts of detection. In
other words, the requirement to provide simultaneous detection and
measurement of the target leads to partitioning of the
time-on-target interval to several independent sub-intervals, each
of which represents a relatively small part of the entire
time-on-target interval. The detection process in each
sub-interval, known also as "Coherent Processing Interval" and for
short CPI, can be performed optimally by using coherent
integration, but the maximum energy collected from the target
return is only a fraction of the entire energy that could be
collected during the entire time-on-target interval. Any logical or
arithmetical combination of the results of sub-interval leads to
losses and degradation in probability of detection in comparison
with coherent integration of the signal during entire
time-on-target interval.
[0009] The concept of ambiguity resolution in range is presented in
FIG. 1, showing the signals received when three pulse sequences
shown respectively as PRF1, PRF2 and PRF3 are transmitted, each
having a different PRF. The returned signals consist of a first
pulse sequence 10 having a first PRI 11, a second pulse sequence 12
having a second PRI 13, and a third pulse sequence 14 having a
third PRI 15. By using several frequencies, the unambiguous range
can be solved. This is depicted in FIG. 1, where the unambiguous
range 16 is detected at a position where pulses in the three pulse
sequences coincide. Generally, the unambiguous range and Doppler of
the target can be imagined as "coordinates" of the target detection
hit of the unfolded range-Doppler map, which covers full range of
the radar specified detection ranges and velocities (Doppler
frequencies). This map is not explicitly represented in firmware or
software of the radar, but one can think of it as sets of target
hit coordinates, each for every detected target.
[0010] The narrow band signal that is collected in the receiver is
usually modeled as s(t)=A(t) cos(2.pi.f.sub.ct+.PHI.(t))+N(t),
where t is time, A is the amplitude, f.sub.c is the carrier
frequency, .PHI. is the phase, and N is the noise. A basic
assumption in this model is that the bandwidth of the amplitude A
is orders of magnitude smaller than f.sub.c. The signal is
processed along the receiving channel. It is frequency
down-converted, filtered, split into two channels called the
in-phase and quadrature, de-modulated (or pulse compressed) and
digitized--not necessarily in that order. It is customary to
represent the result obtained at this stage of the processing of a
single PRF as a complex value entity:
x.sub.k.sup.l=x.sub.l(t.sub.k)=B.sup.le.sup.j(.PHI..sup.l.sup.0.sup.+2.pi-
.f.sup.d.sup.t.sup.k.sup.)+n.sup.l(t.sub.k), where l is the index
of the PRF and is related to the time interval of the measurement,
t.sub.k is the time of the specific sample, known as the "range
gate" number, B.sup.l is the amplitude which is constant within the
period of the measurement, .PHI..sup.l.sub.0 is some phase constant
within the period of the measurement, f.sub.d is the Doppler
frequency, and n.sup.l is the complex noise.
[0011] FIG. 2 shows a prior art method for target detection in a
pulse-Doppler coherent system using L different PRFs. As denoted by
20, a signal x.sub.nm.sup.(l)=x.sup.(l)(tnm) is received for each
PRF used, where l=0 to L-1 is the PRF index, n is the pulse number
in the signal, m is the range gate, and t.sub.n,m is the sampling
time of the signal of the range gate m of the pulse n, and is given
by t.sub.n,m.sup.(l)=nPRI.sup.(l)+mRG, where PRI is the pulse rate
interval and RG is the duration of a single range gate. At 22, the
signals x.sub.nm.sup.(l) are subjected to coherent integration.
This involves performing a discrete Fourier transform on the
signals x.sub.nm.sup.(l) to generate a signal spectrum for each
range gate m. The combination of all spectra for all range gates,
obtained for each CPI, composes the folded range-Doppler map given
by:
X km ( l ) = n = 0 N ( l ) - 1 x nm ( l ) w n - 2 .pi. jkn K , k =
1 , , K , l = 1 , , L , m = 1 , , M ( l ) ##EQU00001##
[0012] where N.sup.(l) is the number of pulses in the signal, k is
an index of the Doppler frequency, K is the number of Doppler
frequencies, w.sub.n is a weighting factor, and M.sup.(l) is the
number of range gates of the PRF 1. At 24, real-valued
range-Doppler maps are generated for each PRF 1, a real-valued K by
M.sup.(l) matrix P.sup.(l) is defined by setting,
P.sub.km.sup.(l)=|X.sub.km.sup.(l)|.sup.2 for each pair of indices
k and m, and at 26, the target detection is performed, whereby it
is determined whether the value P.sub.km.sup.(l) is greater than or
equal to a predetermined threshold T. If so, then at 28,
H.sub.km.sup.(l) is set to 1. If not, then at 30, H.sub.k,m.sup.(l)
is set to 0. This defines a K.times.M.sup.(l) binary matrix
H.sup.(l) for each value of l. This process is repeated for each
CPI independently, producing the sets of target hits for each CPI,
which are determined by their range-Doppler cell addresses--each
PRF defines its own (generally folded) scale of cell addressing.
Thereafter, the algorithm obviously need not record the matrices,
but rather the sets of target hits and their cell coordinates. At
32, the hit sets for each PRF are unfolded by periodically
increasing the cell addresses in range direction by a step of
ambiguous range up to the maximum instrumental range and in Doppler
direction by step of PRF up to the maximum Doppler frequency (the
unfolded target hits for each PRF can be interpreted as non-zero
values of some sparse matrices composed from zeros and ones)--the
matrices H.sup.l are subjected to a process known as "unfolding".
In this process, the dimensions of each matrix H.sup.l are
increased by defining H.sub.k',m'.sup.(l) for values of m' for
which Rmin<m'RG<Rmax where [R.sub.min,R.sub.max] is a
predetermined detection region of interest, and for values of k'
for which Dmin<k'PRF/K<Dmax, Where [Dmin,Dmax] is a
predetermined region of Doppler frequencies of interest, by setting
H.sub.k'm'.sup.(l)=H.sub.km.sup.(l), where k=k' mod K, and m=m' mod
M. In step 34, the matrices H.sup.l are resampled by defining, for
each pair of indices k, m, new indices p and q, as follows. The
range of interest is divided into subintervals of a predetermined
length .DELTA.r. A value of p is found from among all allowed
values of p (i.e. integral values of p for which
0.ltoreq.p.ltoreq..DELTA.r.ltoreq.R.sub.max-R.sub.min) such that
R.sub.p=R.sub.min+p.DELTA.r is closest to the range represented by
the range gate m. The interval [D.sub.min,D.sub.max] is divided
into subintervals of a predetermined length .DELTA.d. A value of q
is found from among all allowed values of q (i.e. integral values
of q for which
0.ltoreq.q.ltoreq..DELTA.d.ltoreq.D.sub.max-D.sub.min) such that
D.sub.q=D.sub.min+q.DELTA.d is closest to k. This generates at 36
new binary matrices U.sup.l where U.sub.p,q.sup.l=H.sub.k,m.sup.l,
wherein the indices p,q correspond to the indices k,m. The sum A of
the unfolded matrices is then calculated at 37, where
A p , q = l = 0 L - 1 U p , q l . ##EQU00002##
At 38, it is determined, for each pair of indices, whether the sum
A.sub.p,q is greater than or equal to a predetermined threshold A.
If so, then at 40 a target is detected at the location having the
associated indices p,q, and the process terminates. If not, then at
42 it is determined that a target is not detected at the location
having the associated indices p, q, and the process terminates.
[0013] To summarize, the following observations are made:
[0014] 1. Although target coherency is maintained for all of the
pulses transmitted within the time-on-target interval, in known
methods, only the signal received within a single CPI is integrated
coherently.
[0015] 2. The effectiveness of the integration depends on the
coherence of the signal. The notion of coherence means that the
relative phases are constant within the period of the measurement
(up to some relatively small noisy contribution) or they vary in a
predictable manner. Normally this requirement implies that the
radar contributes a phase and amplitude that are essentially
constant, at least within the time of measurement, and that the
contribution of the target to phase variation is mainly due to its
motion. The greater the signal-to-noise ratio of a target, the
greater is its maximal detection range. Thus, increasing the
coherent integration interval to the whole period when a target is
illuminated by the antenna (time-on-target interval), the maximum
possible signal-to-noise ratio is obtained, and, as a result, the
maximum detection range.
[0016] Although theoretically two PRFs are sufficient to resolve
ambiguity, the required number of PRFs is actually higher. This is
due to the fact that some range gates are blind in each PRF. In the
simplified representation of FIG. 1, these are the ranges,
corresponding to the time during which the system is transmitting
and cannot receive. This was referred to above as the problem of
visibility. The number of PRFs used typically varies from 2 to 8
depending on the level of visibility that is required. However, the
amount of time that can be allocated to the integration procedure
of each PRF is reduced as the number of PRFs is increased. Since
the signal-to-noise ratio is proportional to the coherent
integration interval duration, as the number of PRFs is increased,
the signal to noise ratio of each PRF decreases. This impairs the
effectiveness of the conventional technique.
SUMMARY
[0017] An object of the present invention is to maximize the
results of integration procedure to the extent allowed solely by
the coherence of the target.
[0018] In its first aspect, the present invention provides a method
for target SNR enhancement in a pulse-Doppler coherent system,
while allowing simultaneous measurements of the target kinematical
parameters. The method may be used, for example, in surveillance
and tracking radar or an ultrasound system. A sequence of
transmitted pulses, reflected from targets and collected by the
radar, is processed. In accordance with the invention, the
processing includes a two-step coherent integration procedure. This
is in contrast to the prior art methods in which a one-step
coherent integration procedure is followed by a detection decision.
The non-coherent combination, i.e. binary integration of the
results of detection, for each CPI, as is done in the prior art
methods, is avoided in the method of the present invention.
[0019] In one preferred embodiment of the invention, the received
signals are subjected to a first coherent integration step by
carrying out, for example, a discrete Fourier transform on the
signals. This generates complex valued folded matrices (one for
each PRF used). The folded matrices are unfolded and resampled. New
matrices are then generated by complex value interpolation of the
original matrices. To account for target motion, a Doppler phase
correction is necessary for each of the interpolated matrices.
Consecutive CPI-s have specific delays relative to the first one.
These delays entail that each cell of the unfolded consecutive
CPI-s be shifted by the phase which is determined by Doppler
frequency of the cell and time delay of the CPI containing this
cell. These phase shifts can be calculated and their effect can be
compensated for each cell. The Doppler phase corrected matrices are
then summed in a second coherent integration step. The resulting
matrix is composed of the cells containing coherent sums of
unfolded range-Doppler matrices received from different CPIs. It
covers all radar specified ranges and Doppler frequencies. The
resulting matrix is then converted into a real-valued matrix A,
taking magnitudes of each cell. Detection is then performed in each
cell of the matrix.
[0020] The decision regarding target detection is taken only after
integration of the entire signal, collected during time-on-target
interval, without any intermediate logical decisions.
[0021] In a second aspect, the present invention provides a
pulse-Doppler coherent system configured for target detection by
the method of the invention.
BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to understand the invention and to see how it may
be carried out in practice, a preferred embodiment will now be
described, by way of non-limiting example only, with reference to
the accompanying drawings, in which
[0023] FIG. 1 shows a prior art method for resolution of ambiguity
of target detection;
[0024] FIG. 2 shows a prior art method for target detection in a
pulse-Doppler coherent system;
[0025] FIG. 3 shows a method for target detection in a
pulse-Doppler coherent system in accordance with one embodiment of
the invention; and
[0026] FIG. 4 is a block diagram showing functionality of a system
that implements the method of the invention.
DETAILED DESCRIPTION OF EMBODIMENTS
[0027] FIG. 3 shows a method for target detection in a
pulse-Doppler coherent system, in accordance with one embodiment of
the invention. At 50, a signal x.sub.nm.sup.(l)=x.sup.(l)(tnm) is
received for each PRF used, where l=1 to L is the PRF index, L is
the number of PRFs used, n is the pulse number in the signal, m is
the range gate, and t.sub.nm is the sampling time of the signal of
the range gate m of the pulse n, given by
t.sub.n,m.sup.(l)=nPRI.sup.(l)+mRG+t.sub.l, where PRI is the pulse
rate interval, RG is the duration of a single range gate and
t.sub.l is the start time of the l-th CPI counted from some
reference point--for example the beginning of the 1-st CPI. At 52,
the signals x.sub.nm.sup.(l) are subjected to coherent integration
in which a discrete Fourier transform is performed on the signals
x.sub.nm.sup.(l) to generate
X km ( l ) = n = 0 N ( l ) - 1 x nm ( l ) w n - 2 .pi. jkn K , k =
0 , , K - 1 , l = 0 , , L - 1 , m = 0 , , M ( l ) - 1 ,
##EQU00003##
a signal where k is an index of the Doppler frequency, K is the
number of Doppler frequencies, N is the number of pulses in the
signal, w.sub.n is a weighting factor and M.sup.(l) is the number
of range gates of the PRF l. At 54, the complex matrices X.sup.l
are unfolded by defining X.sub.k',m'.sup.(l) for values of m' for
which R.sub.min<m'R.sub.G<R.sub.max, where
[R.sub.min,R.sub.max] is a predetermined detection region of
interest, and for values of k' for which
D.sub.min<k'PRF/K<D.sub.max, where [D.sub.min,D.sub.max] is a
predetermined region of Doppler frequencies of interest, by setting
X.sub.k'm'.sup.(l)=X.sub.km.sup.(l), where k=k' mod K, and m=m' mod
M. At 56, the matrices X.sup.l are resampled by defining, for each
pair of indices k'm', new indices p and q, as follows. The range of
interest is divided into subintervals of a predetermined length
.DELTA.r. A value of p is found from among all allowed values of p
(i.e. integral values of p for which
0.ltoreq.p.DELTA.r.ltoreq.R.sub.max-R.sub.min) such that
R.sub.p=R.sub.min+p.DELTA.r is closest to the range represented by
the range gate m. The interval [D.sub.min,D.sub.max] is divided
into subintervals of a predetermined length dd. A value of q is
found from among all allowed values of q (i.e. integral values of q
for which 0.ltoreq.q.DELTA.d.ltoreq.D.sub.max-D.sub.min) such that
D.sub.q=D.sub.min+q.DELTA.d is closest to k. New matrices
XI.sup.(l) are generated at 58 where XI.sub.p,q.sup.(l) is obtained
by complex value interpolation of one or more values of
XI.sub.k',m'.sup.(l) for indices k'm' in a neighborhood of the
indices p q. Any method of interpolation may be used in accordance
with the invention. The interpolation may be linear interpolation
or higher order interpolation. At 60, a Doppler phase correction is
performed on each of the matrices XI.sup.(l) to yield matrices
Y.sup.(l) defined by
Y.sub.pq.sup.(l)=XI.sub.pq.sup.(l)e.sup.-2.pi.jD.sup.q.sup.t.sup.l.
[0028] A real-valued matrix A is then calculated at 62, where
A p , q = l = 0 L - 1 Y p , q ( l ) 2 . ##EQU00004##
Since the Y.sub.pq.sup.(l) are complex values, the calculation of A
is a coherent integration step. At 64, it is determined, for each
pair of indices, whether the sum A.sub.p,q is greater than or equal
to a predetermined threshold T. If so, then at 66 a target is
detected at the location having the associated indices p, q, and
the process terminates. If not, then at 68 it is determined that a
target is not detected at the location having the associated
indices p,q, and the process terminates.
[0029] FIG. 4 is a block diagram showing functionality of a system
80 for implementing coherent integration of multiple CPI-s
according to the method of the invention as described above with
reference to FIG. 3. The system 80 includes a transmitter 81 having
a Tx antenna for tracking an object 82 and a digital receiver 83
having an Rx antenna for receiving an echo signal reflected by the
object. An FFT unit 84 is coupled to an output of the receiver 83,
and a plurality of CPI memories 86-89 is coupled to an output of
the FFT unit 84. A like plurality of interpolation units 90-93 is
coupled to the CPI memories, and a like plurality of unfolding
units 94-97 is coupled to respective outputs of the interpolation
units. A phase correction unit 98 is coupled to the unfolding
units, a summation unit 99 is coupled to an output of the phase
correction unit and a detection decision unit 100 is coupled to an
output of the phase correction unit.
[0030] The transmitter 81 generates and transmits via the Tx
antenna sequences of signals characterized by their PRF values. An
electromagnetic wave reaches the object 82 and its echo returns to
the digital receiver 83 via its Rx antenna. Without loss of
generality the Tx and Rx antennae can be implemented by the same
physical device. The receiver 83 is matched to the transmitted
signals and digitized samples are fed to the FFT unit 84, whose
output is a sequence of CPI spectra for each PRF. The spectrum of
each CPI is stored in a respective one of the memories 86-89 so
that each memory stores the respective CPI spectra for a specific
PRF. The received signals are shifted slightly from CPI to CPI with
respect to the range-Doppler cells. It is to be noted that the
terms `grid`, `map` and `cells` are equivalent and are used
interchangeably throughout the specification. The interpolation
units 90-93 serve to align the CPI signals with respect to the
aforementioned cells. The results of the interpolation are unfolded
by the unfolding units 94-97 by repeating the CPI-s range-Doppler
maps up to the instrumented values of range and velocity. Each
unfolded map is then phase-corrected by the phase correction unit
98 by multiplying the contents of each cell of the map by a
respective complex exponent. The phase of each complex exponent is
proportional to the product of the Doppler frequency and the time
of the beginning of the appropriate CPI measured with respect to
some reference time. Successful operation of the system 80 requires
accurate estimation of these times. The phased corrected maps are
fed to the summation unit 99 which generates a single map, which is
fed to the detection unit 100. The detection unit 100 calculates
the absolute value of each cell of the resulting map and compares
the resulting absolute values to respective thresholds to provide
target detection decision.
[0031] It will also be understood that the system according to the
invention may be a suitably programmed computer. Likewise, the
invention contemplates a computer program being readable by a
computer for executing the method of the invention. The invention
further contemplates a machine-readable memory tangibly embodying a
program of instructions executable by the machine for executing the
method of the invention.
* * * * *