DESCRIPTION
Title of the invention: Processing method for coherent MIMO radar using DDMA
waveforms.
[0001] The field of the invention is that of MIMO ("multiple input, multiple output")
radar with a DDMA (Doppler Division Multiple Access) waveform. The invention is
particularly suitable for millimetre radar, in particular radar operating in the W band
(frequency between 75 and 110 GHz). The invention may be applied, non-exclusively,
to airborne radar, in particular for landing assistance.
[0002] MIMO radar makes it possible to improve the detection performance of radar
in a certain number of applications, in particular airborne radar, by providing in
particular better angular resolution, and better Doppler resolution. A MIMO radar
consists of a plurality of receivers and also of a plurality of transmitters. Two types of
MIMO radar are usually distinguished: multistatic MIMO radar and coherent MIMO
radar. In multistatic MIMO radar, not all of the transmitters and/or receivers are
located at the same area, or on the same carrier. In coherent MIMO radar, the
transmitters and receivers are all co-located on the carrier: this is then referred to as
coherent MIMO radar. In the present invention, it is assumed that the radar is
coherent.
[0003] The DDMA waveform, used in the MIMO radar, is particularly advantageous,
since it exhibits good orthogonality, and therefore good isolation, between the
channels. Specifically, the signals transmitted over each of the transmit antennas are
identical, with the exception that the centre frequencies of the signals transmitted by
the various transmit antennas are slightly shifted with respect to one another, so that
these signals can be separated in the Doppler domain. Thus, each of the waveforms
is generated by applying a different phase modulation between each transmit
channel.
[0004] Next, the received signals are processed by a set of matched filters, each filter
being matched to a waveform, and isolating each transmit channel.
[0005] Figure 1 illustrates the representation of the signals received over one
particular receiver, where each signal band (vertical in the direction of the figure)
corresponds to the echoes of fixed or moving targets of each of the transmitters, after
Doppler processing. In Figure 1, the radar comprises NTX = 8 transmitters, and each
2
transmitter is shifted, in Figure 1, by FR/NTX, where FR is the pulse repetition
frequency.
[0006] The signals received through all of the receivers, and transmitted through all of
the transmitters, are next combined in order to perform DBF (digital beamforming)
processing. It is recalled that, to perform DBF processing, it is sought to optimize the
transmit/receive relationship, on each transmit path/receive path pair, by distributing
the power over the transmit antennas so as to obtain maximum energy in a given
direction.
[0007] To perform DBF processing, it is however essential to identify to which
transmitter each received signal band corresponds. The difference between the
bands of received signals is known because it is related to the phase law of the
DDMA waveform, but the absolute positioning of these bands remains to be
determined, namely to which transmitter at least one of the received signal bands
corresponds. The other transmitters may then be deduced by virtue of knowing the
phase shift of each transmitter.
[0008] Taking the transmitter having a zero phase ramp as a reference, the received
signal band is centred on the Doppler frequency related to the aiming direction and
speed of the radar. This Doppler frequency therefore has to be known with sufficient
precision so as not to confuse the various bands.
[0009] Figure 2 illustrates a vector 𝑉⃗ representing the movement of the carrier/radar,
and the aiming axis of the radar in the direction of the target C.
[0010]As is known, the value of the Doppler frequency f𝐷 is:
[0011] f𝐷 =
2
λ
. V. cosS. cosG =
2.Vr
λ
,
[0012] where V is the speed of the radar, S and G are the aiming elevation and
bearing, respectively, λ is the wavelength of the radar, and VR is the relative radial
velocity of the radar with respect to the target, which corresponds to the projection of
the speed of the radar onto the aiming axis.
[0013] The selection of the samples associated with the transmitter having the zero
phase ramp thus depends on knowing VR with increasing precision as the frequency
resolution cF and wavelength of the pulses decrease. Knowing VR is even more
3
important when the speed of the radar is unclear, due to aliasing in the Doppler
domain.
[0014] For example, if the value of the frequency resolution is cF = 10 Hz, and the
frequency of the radar is FE = 3 GHz (S band, λ = 0.1 m); to know the positioning of
the signal to within a cell, the precision of VR is λ/2* cF = 0.5 m/s, which is relatively
achievable using a conventional inertial measurement unit or using GPS.
[0015] Under the same conditions, for X-band radar (FE = 10 GHz, λ = 3 cm), the
precision of VR becomes λ/2* cF = 0.15 m/s = 15 cm/s, which requires a high-quality
inertial measurement unit.
[0016] Now, still under the same conditions, for millimetre radar (W band between 75
and 110 GHz), the precision of VR becomes about 1 cm/s, which requires an inertial
measurement unit of very high precision which is not always compatible with low-cost
systems.
[0017] The article "Airborne GMTI using MIMO techniques" (J. Kantor & S.K. Davis,
2011, Technical Report 1150, Lincoln Laboratory) describes a MIMO system with a
DDMA waveform. In this article, no mention is made of the method for selecting the
samples from the various transmitters, but since the radar used is S-band radar, as
mentioned above, the required precision for the speed of the carrier is not an issue.
[0018] The object of this invention is to propose a system which not only does not
require a high degree of precision regarding knowledge of the relative radial velocity
of the radar with respect to target, but makes it possible to do without it entirely.
[0019] One subject of the invention is therefore a method for processing coherent
MIMO radar processing DDMA waveforms and comprising NTX transmitters and NRX
receivers, comprising the steps of
a) generating waveforms on at most NTX transmitters, the waveforms, modulo the
pulse repetition frequency FR, being identical from one transmitter to the next, to
within a phase ramp specific to each transmitter;
b) generating, for at least one receiver, a Range-Doppler representation of the
echoes of the transmitted waveforms, where, for each receiver, the echoes of a
transmitter over a plurality of range cells occupy at least one frequency cell in the
Doppler spectrum, called signal band, each signal band being specific to one of the
4
transmitters, the placement of the signal bands in the Doppler spectrum being
determined according to the phase ramp applied to each transmitter, the waveforms
being generated so as to leave a portion of the Doppler spectrum between two signal
bands unoccupied;
c) identifying the transmitter corresponding to each signal band, on the basis of the
Range-Doppler representation of the echoes of the transmitted waveforms.
[0020]Advantageously, step c) comprises the sub-steps of:
- calculating a signal referred to as an echo signal obtained by summing the
power of the echoes of each signal band over all of the range cells;
- generating a signal referred to as a pattern signal, over the same number of
frequency cells as the Range-Doppler representation of the echoes of the
transmitted waveforms, the transmitters being numbered in ascending order
according to the phase ramp applied to the transmitter, the pattern signal
being generated such that the transmitters are sorted, according to the
ascending order, while taking into account the unoccupied portion of the
Doppler spectrum;
- calculating a cross-correlation signal between the echo signal and the pattern
signal;
- identifying the transmitter corresponding to each signal band on the basis of
the maximum of the cross-correlation signal.
[0021]Advantageously, in the identification step, the position of the maximum of the
cross-correlation signal corresponds to the echoes of the first transmitter in the
Doppler spectrum, the first transmitter being defined as being that for which the
phase ramp is zero.
[0022]Advantageously, the unoccupied portion of the Doppler spectrum is obtained
by switching off one of the NTX transmitter, the pattern signal being non-zero for
each frequency cell k*FR/NTX, where k = {0, …, NTX -1}, and zero in the other
frequency cells.
[0023]Advantageously, the unoccupied portion of the Doppler spectrum is obtained
by determining the phase ramps such that the signal bands are spaced apart by
FR/NTX2 frequency cells in the Doppler spectrum, where NTX2 is an integer such
5
that NTX2 > NTX, the pattern signal being non-zero for each frequency cell k*
FR/NTX2, for k = {0, …, NTX -1}, and zero in the other frequency cells.
[0024]Advantageously, the Range-Doppler representation of the echoes is obtained
by means of a Fast Fourier Transform (FFT) on a plurality of acquisition points,
the number of acquisition points being supplemented by adding zeros in order to
obtain NFFT points in total, NFFT being determined such that it is an integer multiple
of the number of transmit paths NTX, if NTX is not a power of two.
[0025]Advantageously, the Range-Doppler representation of the echoes is obtained
by means of a Fast Fourier Transform (FFT) on pulse sequences coded over a
coherent processing interval (CPI), the number of pulses per coherent processing
interval (CPI) being adjusted so as to make it a multiple of the number of
transmitters.
[0026]Another subject of the invention is a method for processing coherent MIMO
radar using DDMA waveforms and comprising NTX transmitters and NRX
receivers, comprising the steps of:
a) generating waveforms on at most NTX transmitters, the waveforms, modulo the
pulse repetition frequency FR, being identical from one transmitter to the next, to
within a phase ramp specific to each transmitter;
b) generating, for at least one receiver, a Range-Doppler representation of the
echoes of the transmitted waveforms, where, for each receiver, the echoes of a
transmitter over a plurality of range cells occupy at least one frequency cell in the
Doppler spectrum, called signal band, each signal band being specific to one of the
transmitters, the placement of the signal bands in the Doppler spectrum being
determined according to the phase ramp applied to each transmitter, the waveforms
being generated such that the spacing between at least two consecutive signal
bands is different from the spacing between the other consecutive signal bands;
c) identifying the transmitter corresponding to each signal band, on the basis of the
Range-Doppler representation of the echoes of the transmitted waveforms.
[0027]Advantageously, the spacings between two consecutive signal bands are all
different.
6
[0028]Advantageously, the signal bands are spaced apart according to a
predetermined spacing law.
[0029]Advantageously, the spacing between each of the signal bands is determined
such that the average of the spacings is equal to FR/NTX.
[0030]Advantageously, the Range-Doppler representation of the echoes is obtained
by means of a fast Fourier transform (FFT) on a plurality of acquisition points, the
number of acquisition points being supplemented by adding zeros in order to
obtain NFFT points in total, NFFT being determined such that the spacing between
two consecutive bands is an integer multiple of FR/NFFT.
[0031]Advantageously, the Range-Doppler representation of the echoes is obtained
by means of a fast Fourier transform (FFT) on pulse sequences coded over a
coherent processing interval (CPI), the number of pulses NREC per coherent
processing interval (CPI) being determined such that the spacing between two
consecutive bands is an integer multiple of FR/NREC.
[0032]Advantageously, the method comprises an additional step of digital forming
processing, in which the waveforms are combined, in a weighted manner, with the
echoes from the various receivers.
[0033]Another subject of the invention is a MIMO radar that is able to implement the
aforementioned method.
[0034] Other features, details and advantages of the invention will become apparent
upon reading the description provided with reference to the appended drawings,
which are given by way of example and in which, respectively:
[0035] [Fig. 1] Figure 1, described above, shows a DDMA waveform according to the
prior art;
[0036] [Fig. 2] Figure 2, described above, shows the carrier vector of the radar, with
respect to a target;
[0037] [Fig. 3] Figure 3 shows the diagrams illustrating the MIMO radar processing
method according to a first embodiment of the invention, in which one of the
transmitter is switched off;
7
[0038] [Fig. 4] Figure 4 shows the diagrams illustrating the MIMO radar processing
method according to a second embodiment of the invention, in which the spacing
between the bands is modified;
[0039] [Fig. 5] Figure 5 shows the diagrams illustrating the MIMO radar processing
method according to a third embodiment of the invention, in which the spacing
between the bands is non-uniform.
[0040]According to the first and second embodiments of the invention, the method
comprises three steps.
[0041] In a first step, the radar, which comprises NTX transmitters, generates
waveforms on at most NTX transmitters. According to a first embodiment, described
in detail below, one of the NTX transmitter is switched off. According to the second
embodiment, also described in detail below, all of the transmitters are activated. The
principle of the DDMA waveform, as explained above, consists in transmitting, in
each pulse repetition interval, radar pulses that are identical from one transmitter to
the next, to within a phase ramp specific to each transmitter. The phase ramp is
created by a phase-shifter, which is a relatively straightforward component to
integrate. Next, each of the received echoes is processed by a matched filter bank,
namely one filter per waveform.
[0042] The processing method according to the invention may be implemented at the
level of one of the receiver, or of a plurality thereof, or even of all of them. Application
to digital beamforming specifically requires associating all of the receivers with the
transmitters, but this application is not limiting. In general, the method according to
the invention makes it possible, in a MIMO context, to associate, even for just one
receiver, each signal band with one of the transmitters.
[0043] For this, in the second step b), a Range-Doppler representation of the echoes
of the transmitted waveforms is generated. As illustrated by Figure 1, described
above, the Range-Doppler representation comprises the frequency cells (or the
number of FFT, fast Fourier transform, points), on the abscissa and the range cells
on the ordinate. Each signal band is centred on the frequency cell of the Doppler shift
of each transmitter. The signal bands are thus spaced apart by FR/NTX, where FR is
the pulse repetition frequency (inverse of the pulse repetition interval).
8
[0044]Because of the very nature of the DDMA waveform, which introduces a phase
shift for each of the transmitter, this phase shift can be found in the Doppler spectrum,
in the spacing of the signal bands. Each of the received echoes is processed by a
matched filter bank, namely one filter per waveform. On reception, the processor of
the MIMO radar removes the phase shift which was applied on transmission.
[0045] The particularity of the method according to the invention is that the
waveforms are generated such that a portion of the Doppler spectrum is left
unoccupied between two signal bands. For a DDMA MIMO radar of the prior art, the
phase shifts applied are determined such that the signal bands are all uniformly
distributed in the Doppler spectrum. However, by leaving a portion of the spectrum
unoccupied, identification of the transmitter corresponding to each of the signal
bands is facilitated, as demonstrated below (step c) of the method).
[0046]Advantageously, step c) of the processing method comprises a first sub-step
of calculating a signal referred to as an echo signal. The echo signal is obtained by
summing, for each signal band, the power of the echoes over all of the range cells.
Thus, a signal representative of the energy per signal band is obtained. The echo
signal is thus obtained directly and easily (namely without computational complexity)
from the Range-Doppler representation.
[0047] Next, in a second sub-step, a signal referred to as a pattern signal is
generated over the same number of frequency cells as the Range-Doppler
representation of the echoes of the transmitted waveforms. The pattern signal is a
representation, in the order of the transmitters, of the shift caused by the phase ramp
applied to each transmitter. Thus, in the MIMO radar, the phase ramp which is
applied is associated with each transmitter. Next, a number is assigned to the
transmitter: n° 1 for the transmitter that has a zero phase ramp, n° 2 for the first
phase ramp, and so on. Thus, a number from 1 to NTX is assigned to each
transmitter. The pattern signal is a binary signal, the value of which is 1 (or any other
value other than zero, as long as this value is the same for all of the transmitters) for
each frequency cell corresponding to the shift caused by the slope of the various
phase laws and 0 elsewhere. Additionally, the pattern signal is also set to 0 in the
unoccupied portion of the Doppler spectrum.
9
[0048]A cross-correlation signal between the echo signal and the pattern signal is
then calculated. The cross-correlation signal corresponds to the cross-correlation
function between the echo signal and the pattern signal. Each of the two signals may
be considered as being a "comb"-type signal. The cross-correlation signal is
maximum when the two combs encounter one another.
[0049] The position of the maximum of the cross-correlation signal indicates the
position, in the Doppler spectrum, of the first transmitter. The position of the other
transmitters is deduced by shifting according to the theoretical spacing between the
signal bands, modulo the pulse repetition frequency, or modulo the number of FFT
points.
[0050] The processing described is performed systematically, on each burst of the
radar signal.
[0051] Figure 3 illustrates a first embodiment of the processing method according to
the invention. According to the first embodiment, the portion of the spectrum that is
unoccupied is produced by switching off one of the transmitter. Specifically, MIMO
radar generally has lots of transmitters, for example ten or so, and switching off one
of the transmitter affects the radar picture only by ten or so percent, in the case that
the radar comprises ten transmitters.
[0052] In the example of Figure 3, the MIMO radar has NTX = 6 transmitters, and the
fifth transmitter is switched off.
[0053] The switching off of one of the transmitter may be performed in various ways.
For example, amplitude control, used in active antennas of radar for amplitude and
phase control, may apply a zero amplitude to the transmitter to be switched off. This
solution has the advantage of being able to be reconfigured on mission, by modifying
the amplitude control command. As an alternative, the output of the power
transmitters of the transmitter to be reached is not connected, which constitutes a
permanent solution, and one which can be modified only between two missions.
Another possibility would be to add a controlled switch upstream of each power
transmitter.
[0054] The choice of the transmitter to be switched off may be completely random,
when the number of the transmitter that is switched off is known. The choice of path
has no effect on the outcome of the process.
10
[0055] The echo signal is obtained by summing the powers of each range cell for
each signal band. In the Doppler spectrum, all of the echoes of the transmitters are
spaced apart by FR/NTX, except, by definition, in the band where the transmitter has
been switched off. In Figure 3, the echoes are received over 120 FFT points, without
this value limiting the scope of the present invention.
[0056] The pattern signal is represented over the same number of frequency cells as
the Doppler spectrum of the received echoes. The pattern signal is non-zero (it has a
value of 1 for example) for each frequency cell k*FR/NTX, where k = {0, …, NTX-1},
corresponding to a signal band, with the exception of the signal band of the switchedoff transmitter. The pattern signal is also zero in the other frequency cells. In the
example of Figure 3, the fifth transmitter (from among the six transmitters) is switched
off. Thus, the pattern signal is a comb that has a value other than 0 in the 0, FR/6,
2*FR/6, 3*FR/6 and 5*FR/6 cells, and zero elsewhere.
[0057] Lastly, the cross-correlation signal between the echo signal and the pattern
signal is calculated. In Figure 3, the cross-correlation signal, which should be read in
conjunction with the Doppler spectrum of the received echoes, is maximum at a
position corresponding, in the Doppler spectrum, to the signal band of the first
transmitter (namely the second signal band from the right of the spectrum in the
example of Figure 3). Thus, the second signal band from the right corresponds to the
echoes from the first transmitter. Next, the position of the other transmitters may be
deduced by simply shifting from left to right, circularly, in relation to the phase ramp.
In the example of Figure 3, the first signal band from the right corresponds to the
echoes of the second transmitter, the first band on the left corresponds to the echoes
of the third transmitter, and so on.
[0058] Thus, the signals from each of the transmitters is associated in the Doppler
spectrum, without using information on the radial velocity of the carrier, which
corresponds to the projection of the speed of the carrier onto the aiming axis.
[0059] The first embodiment described above is particularly suitable for DDMA MIMO
radar having a high number of transmitters, ten or so for example. Thus, the radar
efficiency is affected little by the switching off of one transmitter.
[0060]Switching off one of the transmitter may however prove to be detrimental from
the point of view of radar efficiency when the MIMO radar has few transmitters.
11
For example, if the MIMO radar comprises only four transmitters, switching off
one of the transmitter affects the radar picture by about 25%, which does not
represent optimal use of the radar.
[0061] The second embodiment of the processing method according to the invention
also makes it possible to associate the transmitters with each of the signal bands.
For this, the phase ramps are generated such that the signal bands are spaced apart
by FR/NTX2 frequency cells in the Doppler spectrum, where NTX2 is an integer such
that NTX2 > NTX. For example, NTX2 = NTX + 1. In the example of Figure 4, NTX2 =
NTX + 2, with NTX = 6 and NTX2 = 8.
[0062] The spacing thus generated also creates an empty space in the Doppler
space, like for the first embodiment. The Doppler spectrum illustrated in Figure 4
shows six signal bands spaced apart by FR/8 in the Range-Doppler representation of
the echoes of the transmitted waveforms.
[0063] Like for the first embodiment, an echo signal is created. In the second
embodiment, the value of the echo signal is the sum of the powers of the echoes
received over all of the range cells of a signal band. The echo signal therefore has
NTX positive values, spaced apart by FR/ NTX2, over NFFT frequency cells.
[0064]Additionally, the pattern signal is non-zero (it has a value of 1 for example) for
each frequency cell k*FR/NTX2, for k = {0, …, NTX -1}, and zero in the other frequency
cells. The pattern signal is represented over the same number of frequency cells as
the Doppler spectrum of the received echoes. In the example of Figure 4, the pattern
signal takes the value "1" for 0 on the abscissa, and for each frequency cell spaced
apart by FR/NTX2, and the value "0" elsewhere.
[0065] The cross-correlation signal corresponds to the cross-correlation function
between the echo signal and the pattern signal. Its maximum indicates the first
transmitter, namely that for which the phase ramp applied on transmission is zero.
[0066] In the example of Figure 4, by superposing the cross-correlation signal and the
Doppler spectrum, it can be seen that the maximum of the cross-correlation signal
corresponds to the leftmost signal band in the Doppler spectrum. The second
leftmost band corresponds to the second transmitter path, and so on. Like for the first
embodiment, the empty space created in the Doppler spectrum makes it possible to
12
identify the relative positioning of the signals, namely which transmitter they are from,
without knowing the radial velocity of the carrier with respect to the target.
[0067] The spacing, although uniform, between the DDMA signal bands (having a
value of FR/NTX or FR/NTX2, according to the first or second embodiment), is not
always equal to a whole number of frequency cells. This situation arises if NTX or
NTX2 (according to the second embodiment) is not a power of two, while the number
of points on which the FFT is performed is generally a power of two. In this case, one
option would be to take the closest frequency cell, but one and the same echo from
various transmitters would be sampled differently from one signal band to the next,
which would introduce unwanted amplitude modulation, and therefore a worse DBF
result. The other solution would be to interpolate the signals so as to put the signals
onto a sampling grid common to the various DDMA channels, or equivalently, to
perform an inverse Fourier transform, numerically apply the phase ramp opposite
that used on transmission and then redo the direct Fourier transform so as to select
the samples around the zero frequency cell. All of these operations result in obtaining
defect-free DBF, but at the cost of high computational complexity.
[0068] To overcome this, it is proposed to supplement the number of acquisition
points by adding zeros in the time domain (the "zero padding" technique), until the
number of FFT points NFFT is an integer multiple of the number of transmitters NTX.
[0069] For example, if there are NTX = 12 transmitters and NREC = 512 points on
which FFT is initially envisaged (it is recalled that TE = 1/cF = NREC/FR), an FFT of size
NFFT = 540 must be performed, which is a multiple of NTX = 12.
[0070]As an alternative, the number of pulses per coherent processing interval (CPI)
is adjusted to make it a multiple of the number of transmitters.
[0071] For example, with NTX = 12, instead of using NREC = 512, NREC = 480 or NREC
= 540 will be taken, which makes it possible to perform an FFT without zero padding
while remaining a multiple of NTx.
[0072] Figure 5 illustrates a third embodiment of the invention.
[0073]Step a) of the method according to the third embodiment of the invention is
identical to step a) of the first two embodiments.
13
[0074] In step b), a Range-Doppler representation of the echoes of the transmitted
waveforms is generated for at least one receive path. For each receiver, the echoes
of a transmitter over a plurality of range cells occupy at least one frequency cell in the
Doppler spectrum (signal band). Each signal band is specific to one of the transmitter.
The placement of the signal bands in the Doppler spectrum is determined according
to the phase ramp applied to each transmitter. What more particularly characterizes
the third embodiment is the spacing between each of the signal bands: the spacing
between each of the signal bands is non-uniform: there are at least two signal bands
whose spacing (between these two bands) is different from the spacing between the
other signal bands. In particular, all of the spacings between two consecutive signal
bands are different, but this is not a required condition. This non-uniform spacing
makes it possible to easily identify to which transmitter the echoes correspond,
without it being necessary to switch off one of the transmitters. This embodiment
therefore does not affect the radar efficiency.
[0075] The third embodiment comprises a third step c) of identifying the transmitter
corresponding to each signal band, on the basis of the Range-Doppler representation
of the echoes of the transmitted waveforms. This step is identical to step c) of the first
two embodiments.
[0076] The Doppler spectrum illustrated in figure 5 shows six signal bands, the
spacing of which is non-uniform in the Range-Doppler representation of the echoes
of the transmitted waveforms. In Figure 5, the spacing varies from FR/8 to FR/6; these
values are given only by way of example.
[0077] In the third embodiment, the value of the echo signal is the sum of the powers
of the echoes received over all of the range cells of a signal band. The echo signal
therefore has NTX positive values.
[0078]Additionally, the pattern signal is non-zero (it has a value of 1 for example) for
each frequency cell f(k)*FR/NTX, for k = {0, …, NTX-1}, and zero in the other
frequency cells, and f() is a function representative of the non-uniform spacing. The
function f() may therefore define a spacing law. For example, the spacing law may be
linear or logarithmic. The pattern signal is represented over the same number of
frequency cells as the Doppler spectrum of the received echoes.
14
[0079]Advantageously, a spacing between each of the signal bands is determined
such that the average of the spacings is equal to FR/NTX.
[0080] The cross-correlation signal corresponds to the cross-correlation function
between the echo signal and the pattern signal. Its maximum indicates the first
transmitter, namely that for which the phase ramp applied on transmission is zero.
[0081] In the example of Figure 5, by superposing the cross-correlation signal and the
Doppler spectrum, it can be seen that the maximum of the cross-correlation signal
corresponds to the leftmost signal band in the Doppler spectrum. The second
leftmost band corresponds to the second transmitter, and so on.
[0082] Like for the first two embodiments, it is also possible to supplement the
number of FFT acquisition points by adding zeros ("zero padding") up to NFFT points,
or to adjust the number of pulses NREC per coherent processing interval (CPI). In this
third embodiment with non-uniform spacing, le number of FFT points NFFT, or,
respectively, the number of pulses NREC, should be such that the frequency
difference between two consecutive bands, related to the phase ramps of the DDMA
modulation, is a whole number of frequency cells, i.e. integer multiple of FR/NFFT or of
FR/NREC, respectively.
[0083] The method according to the invention is particularly suitable for the millimetre
band (W band), for automotive radar applications or applications of radar on board
aeroplanes or drones, for the detection of fixed or moving target relative to the carrier.

CLAIMS
1. Method for processing coherent MIMO radar processing DDMA waveforms
and comprising NTX transmitters and NRX receivers, comprising the steps of
- a) generating waveforms on at most NTX transmitters, the waveforms,
modulo the pulse repetition frequency FR, being identical from one
transmitter to the next, to within a phase ramp specific to each
transmitter;
- b) generating, for at least one receiver, a range-Doppler representation
of the echoes of the transmitted waveforms, where, for each receiver,
the echoes of a transmitter over a plurality of range cells occupy at
least one frequency cell in the Doppler spectrum, called signal band,
each signal band being specific to one of the transmitters, the
placement of the signal bands in the Doppler spectrum being
determined according to the phase ramp applied to each transmitter,
the waveforms being generated so as to leave a portion of the Doppler
spectrum between two signal bands unoccupied;
- c) identifying the transmitter corresponding to each signal band, on the
basis of the Range-Doppler representation of the echoes of the
transmitted waveforms.
2. Method according to Claim 1, wherein step c) comprises the sub-steps of:
- calculating a signal referred to as an echo signal obtained by summing
the power of the echoes of each signal band over all of the range cells;
- generating a signal referred to as a pattern signal, over the same
number of frequency cells as the Range-Doppler representation of the
echoes of the transmitted waveforms, the transmitters being numbered
in ascending order according to the phase ramp applied to the
transmitter, the pattern signal being generated such that the transmitter
are sorted, according to the ascending order, while taking into account
the unoccupied portion of the Doppler spectrum;
- calculating a cross-correlation signal between the echo signal and the
pattern signal;
- identifying the transmitter corresponding to each signal band on the
basis of the maximum of the cross-correlation signal.
3. Method according to Claim 2, wherein, in the identification step, the position of
the maximum of the cross-correlation signal corresponds to the echoes of the
first transmitter in the Doppler spectrum, the first transmitter being defined as
being that for which the phase ramp is zero.
16
4. Method according to either of Claims 2 and 3, wherein the unoccupied portion
of the Doppler spectrum is obtained by switching off one of the NTX
transmitters, the pattern signal being non-zero for each frequency cell
k*FR/NTX, where k = {0, …, NTX -1}, and zero in the other frequency cells.
5. Method according to either of Claims 2 and 3, wherein the unoccupied portion
of the Doppler spectrum is obtained by determining the phase ramps such that
the signal bands are spaced apart by FR/NTX2 frequency cells in the Doppler
spectrum, where NTX2 is an integer such that NTX2 > NTX, the pattern signal
being non-zero for each frequency cell k* FR/NTX2, for k = {0, …, NTX -1}, and
zero in the other frequency cells.
6. Method according to one of the preceding claims, wherein the Range-Doppler
representation of the echoes is obtained by means of a fast Fourier transform
(FFT) on a plurality of acquisition points, the number of acquisition points
being supplemented by adding zeros in order to obtain NFFT points in total,
NFFT being determined such that it is an integer multiple of the number of
transmitters NTX, if NTX is not a power of two.
7. Method according to one of Claims 1 to 5, wherein the Range-Doppler
representation of the echoes is obtained by means of a fast Fourier transform
(FFT) on pulse sequences coded over a coherent processing interval (CPI),
the number of pulses per coherent processing interval (CPI) being adjusted so
as to make it a multiple of the number of transmitters.
8. Method for processing coherent MIMO radar processing DDMA waveforms
and comprising NTX transmitters and NRX receivers, comprising the steps of
- a) generating waveforms on at most NTX transmitters, the waveforms,
modulo the pulse repetition frequency FR being identical from one
transmitter to the next, to within a phase ramp specific to each
transmitter;
- b) generating, for at least one receiver, a Range-Doppler representation
of the echoes of the transmitted waveforms, where, for each receiver,
the echoes of a transmitter over a plurality of range cells occupy at
least one frequency cell in the Doppler spectrum, called signal band,
each signal band being specific to one of the transmitters, the
placement of the signal bands in the Doppler spectrum being
determined according to the phase ramp applied to each transmitter,
the waveforms being generated such that the spacing between at least
two consecutive signal bands is different from the spacing between the
other consecutive signal bands;
17
- c) identifying the transmitter corresponding to each signal band, on the
basis of the Range-Doppler representation of the echoes of the
transmitted waveforms.
9. Method according to Claim 8, wherein the spacings between two consecutive
signal bands are all different.
10.Method according to either of Claims 8 and 9, wherein the signal bands are
spaced apart according to a predetermined spacing law.
11.Method according to one of Claims 8 to 10, wherein the spacing between each
of the signal bands is determined such that the average of the spacings is
equal to FR/NTX.
12.Method according to one of Claims 8 to 11, wherein the Range-Doppler
representation of the echoes is obtained by means of a fast Fourier transform
(FFT) on a plurality of acquisition points, the number of acquisition points
being supplemented by adding zeros in order to obtain NFFT points in total,
NFFT being determined such that the spacing between two consecutive bands
is an integer multiple of FR/NFFT.
13.Method according to one of Claims 8 to 11, wherein the Range-Doppler
representation of the echoes is obtained by means of a fast Fourier transform
(FFT) on pulse sequences coded over a coherent processing interval (CPI),
the number of pulses NREC per coherent processing interval (CPI) being
determined such that the spacing between two consecutive bands is an
integer multiple of FR/NREC.
14.Method according to one of the preceding claims, comprising an additional
step of digital forming processing, in which the waveforms are combined, in a
weighted manner, with the echoes from the various receivers.
15.MIMO radar, characterized in that it is able to implement the method according
to any one of Claims 1 to 14.