FR2737578A1 - IMPULSE DOPPLER RADAR DEVICE WITH COMPLETE DETERMINATION OF TARGET SPEED VECTOR - Google Patents

Abstract

A pulsed Doppler radar comprises a principal antenna (10) and a receiving system (11 to 15), providing distance gates (p) each subdivided by fast Fourier transformation (16) into Dopper filters (n). An anciliary antenna (20) has a second receiving system identical to the first. A module (5) establishes the average relative gain of the two antennas over selected ones of the range/Doppler gates. The signal emanating from the ancillary receiving system, weighted by this gain in a multiplier (6), is then subtracted in a comparator (172) from the signal of the principal processing system. The jammer may thus be eliminated.

Description

Pulse Doppler radar device with complete determination of the target velocity vector.

The invention relates to radar systems for allowing the tracking of a target, by estimating the trajectory thereof.

The radar is in principle airborne. The target is an aircraft, too.

The knowledge of the movement of a mobile is generally acquired by three state variables, which are its position, speed and acceleration. Each state variable has three components, for example in an orthonormal trihedron.

At present, however, the operation of the trajectory estimators is based on the measurement of the position of the target, as well as on the radial velocity of the target vis-à-vis the airborne radar, but not on the other two components of the velocity. . To access a complete knowledge of the velocity vector of the target, it is necessary to compute the derivative of the position information.

As it begins to operate, the tracking loop is initialized with position information, and radial velocity, as measured. On the other hand, the two other components of the velocity vector, and the three components of the acceleration vector are generally initialized to an arbitrary value, most often zero, for lack of measurements concerning them.

The insufficiency of this initialization has the effect of slowing the convergence of the estimator of the trajectory of the target. Those skilled in the art will immediately understand the disadvantages.

The present invention aims to provide a solution to this problem.

For this purpose, the invention uses a pulse Doppler radar device comprising radar transmission means, radar reception means, signal processing means received to derive the position and the radial velocity of a target, such as an aircraft, and means for tracking the target, by estimating the trajectory thereof.

According to a general characteristic of the invention, the transmission frequency is subject to repetitive variations, capable of increasing the discriminating power in distance from the radar, and the processing means are arranged to discriminate in distance from several bright spots of target; the device further comprises - means for comparing the actual distance of these bright points with a set of model distributions concerning different types of targets or aircraft and each associated with a determined orientation, in space, of the main axis of motion corresponding aircraft, as well as means for completely determining the target velocity vector from its measured radial velocity, the main axis orientation associated with the model distribution closest to the actual distribution.

This makes it possible to initialize and update the tracking loop, in particular its trajectory estimator, from complete speed information.

In practice, the implementation of the invention is considerably simplified, if the transmission frequency is subject, periodically, to a step modulation, each jump of which is an integer multiple of the recurrence frequency.

According to another aspect of the invention, the processing means perform, for each frequency step, a Doppler analysis at the rate of recurrence, followed by a distance analysis on the samples collected at the output of each Doppler filter, at the rate of repetition of the emission modulation. This makes it possible to remove the ambiguity between the speed and distance information, and also provides the individual distances of the bright points that the target has.

These Doppler and distance analyzes are advantageously performed by fast Fourier transforms (abbreviated as TFR or FFT).

According to yet another aspect of the invention, the comparison means operate by regenerating dummy Doppler offset signals from model distributions of distances available. Two Fourier transforms are performed on this fictitious signal. The first corresponds to the distance Fourier transform performed on the received signal, and is accompanied by a distance refinement.

The second Fourier transform, which is performed on the fictitious signal relating to the model as well as the signal received from the target, has the effect of normalizing the received signals. This makes it possible to take into account, in particular, the fact that several bright points overlap on certain incidences, hence doubled levels in the distance filters concerned.

The actual comparison is carried out by correlation between the outputs of the distance filters obtained after this double Fourier transformation, for the model on the one hand and for the target on the other hand.

Other features and advantages of the invention will appear on examining the detailed description below, and the accompanying drawings, in which - Figure 1 is a block diagram of a radar in which the invention may be Implementation; FIG. 2 is a timing diagram showing step modulation, of constant steps, of the transmission frequency; FIG. 3 is a block diagram of the processing means added to the radar according to the invention; FIG. 4 is a schematic diagram of the extractor
Doppler radar according to the invention; FIG. 5 is a schematic diagram of the distance extractor of the radar according to the invention; and FIG. 6 is a flowchart of the different steps implemented in the device according to the invention.

The attached drawings contain many information of a certain nature. Accordingly, they can not only serve to better understand the detailed description below, but also contribute to the definition of the invention, if any.

In FIG. 1, the radar comprises emission circuits supplying an antenna 14 through a circulator 12.

The antenna 14 is for example of the electronic scanning type, to allow tracking of a target. The signals received are transmitted by the circulator 12 to reception stages 16.

A generator 18 (shown separately for memory) provides a common step modulation on the one hand for the transmission frequency in the stages 10, on the other hand for the local signal that will be used for reception in the stages 16.

Conventionally, the output signal of the reception stages 16 is transmitted to members 20 performing distance processing and radial velocity. A member 22 is provided to indicate the speed of the wearer, in order to correct the radial speed if necessary.

The information obtained by the member 20 is applied to a tracking loop 24, responsible in particular for suitably implementing the transceiver antenna 14, as well as optimally managing the distance and Doppler filterings made in FIG. the organs 20, to optimize the pursuit of the target.

Some radars allow the tracking of several targets.

We will limit hereafter to the case of a single target.

According to the prior art, the tracking loop is initialized only with the distance and radial velocity information relating to a target. The other components of the velocity vector, and those of the acceleration vector are set to zero at initialization. These state variables can only reach their true values when sufficient distance information is available to derive the target velocity vector by derivation; sufficient speed information will again be needed to determine the target vector of acceleration. It is clear that all this takes a significant time, which time would be valuable in some applications.

As already indicated, the present invention will make it possible to improve things, from members which are, in FIG. 1, illustrated inside the block 3 of "special treatment".

It will already be observed that the existence of a stepwise modulation of the emission frequency is perfectly compatible with the treatment of the members 20. In fact, this modulation of the emission frequency can be processed by the methods which is well known when the frequency is varied for other reasons, for example in "frequency agility",
The possibility of increasing the distance discrimination of the radar, to distinguish several bright points within the same target, is already acquired, since the frequency of emission of a repetitive way.

However, in the general case, the treatments to be performed can reach enormous complexity.

This is why it is currently preferred, according to the invention, for the amplitude of the SF jumps of the transmission frequency to be an integer multiple of the recurrence frequency fR, which is conventionally associated with a recurrence period TR.

With reference to FIG. 2, it will be noted that the modulation of the transmission frequency comprises N steps, distinguished by an index n varying from 0 to N-1. Frequency hopping is assumed to be constant. Each frequency step (stair step) lasts K periods of recurrence of the radar.

Finally, the modulation starts at the instant TO, ending at the instant TO + N.T. In this case, this step modulation is periodic, with a period N.T.

It goes without saying that this periodicity is not necessary, and that also the steps of the stairs are not necessarily contiguous if the operation of the radar requires it.

What matters is to have, in a short enough time interval, the output signal of the reception stage 16 on each of the different bearings shown in FIG. 2.

In the following, the phase of the output signal of the reception stages 16 is noted PHIBAT (t).

FIG. 3 shows that this signal is applied firstly to a Doppler extractor 30, followed by a distance extractor 40.

In a computer memory, associated for example with the main radar computer, there is a model library 50, representative of aircraft known in the field concerned. The selection of known aircraft is to be made according to the application, and it is not useful to develop here the principles.

It will be understood that these are airplanes of the aircraft type or the like, whose main direction of movement is defined by the axis of their fuselage. The radar image of such aircraft is generally reducible to a limited number of bright dots (three or four). This is sometimes called the "radar signature". Indeed, we can go so far as to recognize an aircraft from the spatial structure of its bright spots.

With regard to the present invention, the need is different: it is not, at least a priori, to recognize the aircraft. It is simply a matter of associating with a structure of bright points encountered a heading, that is to say the main axis of movement that can have this aircraft. Indeed, if we can determine this course, it is possible to immediately deduce the speed vector of the aircraft, which is also known radial velocity.

It is conceivable to store the model library 50 in different ways. We could first put only relative distances of bright points, referred to a main axis of motion whose orientation would be defined a priori, for example along an axis OX of an orthonormal trihedron.

For an aircraft of a given type, one enters into library several models of bright points according to the presentation of the aircraft.

Indeed, the diffraction of electromagnetic waves on the structure of the aircraft depends on the orientation of the latter. As a result, each presentation corresponds to a distribution of the bright points in the plane reference.

In practice, a model is memorized over a range of a few degrees centered on a given presentation.

This model being chosen, each presentation corresponds to a set of carrier distances / bright points, each distance being determined by calculation to a common constant.

Finally, it is clear that the information contained in memory is recorded in a form directly compatible with the outputs of the Doppler extractor and distance, as defined below.

The module 60 of FIG. 3 then makes the comparison between the distance information relating to the target encountered and several models contained in the library 50, until a correlation is obtained, that is to say a satisfactory correspondence between the target and the model.

The output of the module 60 is then an indication of the course followed by the target, in an absolute reference definable relative to the carrier of the radar.

The module 80 then realizes the complete determination of the target velocity vector. It is recalled that we know the radial velocity of the latter with respect to the aircraft.

 It is then easy to deduce the velocity vector from it, since we know the direction of this vector, that is to say the course followed by the target.

The information thus obtained on the speed vector is used in a module 90 for initializing or updating the tracking loop, as the case may be.

It will be understood that it is now possible to fully initialize the first two state variables, that is, the position and speed of the target. Only its acceleration remains to be updated, when a sufficient number of speed information is available.

Those skilled in the art will understand that the near-instant availability of the velocity information is a means of substantially accelerating the initialization of the tracking loop. Once this initialization has been carried out, the tracking can be done, if desired, by using both the information obtained from the state variables themselves and their time derivatives, and the information that can be obtained from the means. according to the invention.

In the foregoing, it is implicitly assumed that the information contained in the model library 50 is fixed, that is to say acquired once and for all at take-off of the aircraft carrying the radar. This is indeed the simplest way to implement the invention.

However, in the event that the radar encounters a target that does not correspond perfectly to any of the models contained in its library, it is conceivable that it can, in real time, and with the aid of appropriate means of treatment, enrich its library of models from new information acquired. This enrichment can also be done offline, on the ground, from the observations made by the radar in flight.

We will now describe in more detail some of the means of the invention, essentially the modules 30, 40 and 60 of FIG.

In FIG. 4, the signal PHIBAT (t) coming from the reception stage 16 of FIG. 1 is found at the input.

This signal is applied to a comparator 31 whose output is a variable input signal as a function of time, noted
Set).

The signal SE (t) is applied to a sampler 32, operating at the rate of recurrence of the transmission pulses.

The output of the latter is the same input signal, but sampled this time according to the number of recurrences contained in each step of the transmission frequency and the number of steps of this same transmission frequency.

 I1 is therefore denoted SE (k, n).

The digital signal SE (k, n) is applied to a first Fourier transform stage denoted 33. This latter provides, at the output of its Doppler filters, noted information.
Dn (p), where p is the rank of the Doppler filters, which varies from zero.

This information is used for the distance extractor, as will be seen below.

In FIG. 4, they are applied to a Doppler refining circuit 34. One skilled in the art knows that the refining of frequencies (Doppler or distance) is a technique which makes it possible to determine the frequency of a signal with a greater accuracy than the filter analysis step, that is to say the distance between the center frequencies of two consecutive analysis filters.

In short, such a refinement performs the following operations - looking for filters that have a peak output at the output, which is often called an "extremum" filter; - association with each extremum filter of the higher level adjacent filter, which is the filter on the right or on the left; forming the ratio of the levels between the central filter and this lateral filter; interpolation making it possible to deduce the real frequency of the incident signal considered, taking into account this ratio of proportionality.

Thus, at the output of the Doppler refining unit 34, information denoted MAXDOP is obtained, indicating the number of filters providing, after refining, an extremum, as well as ECAXAF information relating to the refined frequency values of these parameters. extrema.

An organ 35 then averages all the information obtained, the number of which is MAXDOP. The signal from this member 35 is noted ECAXDM.

 This is a mean deviation, considered as a correction to be made to the value of the Doppler frequency which was estimated during the previous cycle, that is to say the previous level of the transmission frequency.

The correction is carried out in the comparator 36, which will add this ECAXDM value to the average value previously recorded which is <XD (tn i)>.

At the output of the comparator 36 is available the frequency
Corrected Doppler noted XDM. It feeds a discrete Kalman filter 37, which gives a new estimate of the average Doppler shift for cycle n, i.e., the current cycle. This average Doppler shift is noted <XD (t)>. It serves to supply a voltage-controlled crystal oscillator, which forms the block 38 of FIG. 4, and provides an analog signal PHIX (t), which constitutes the second input of the initial comparator 31.

Those skilled in the art will understand that all this constitutes a phase loop capable of allowing the extraction of the Doppler shift in the received signal relative to the target.

 Reference is now made to FIG. 5. It contains the comparator 31 and the Fourier transforming element 33 which constitutes the filters of the Doppler extractor.

The phase loop is here simply materialized by the member 39, provided with a transfer function H (p), relative to each of the outputs Dn (p) of the module 33.

The rest of Figure 5 is the distance extractor itself.

The operation of the latter is effected not at the rate of the recurrence pulses, but at the rate of the stair steps of the transmission frequency.

First of all, at the output of each filter
Doppler, a linear integration of the n signal samples taken at the end of each Doppler analysis (on one step of the staircase). Thus, each of the members 41 realizes the sum of N successive samples from one of the Doppler filters of the Fourier transform module 33.

The output signal is noted each time SOMANP (p). The different signals of this kind are applied to another Doppler refinement module 43, which operates this time on information available at the rate of the steps, as will be remembered. The operation of the Doppler refining circuit is the same as before. Its output, denoted XDM (p), is the refined Doppler frequency determined for each extremum filter. p is the rank of the filter
Doppler concerned.

In practice, only the p = 0 channel will be extremum, because - the Doppler loop of FIG. 4 has the effect of reducing the frequencies around zero (more precisely the frequency differences around zero); - The bright points of the target are not discriminated Doppler shift, but only in distance.

In short, the XDM signal is in principle common to all bright spots, and is obtained for the p = 0 path. This results in a great simplification for the determination of this magnitude XDM, which intervenes for the determination of the distance of the bright spots.

The distance of the bright points is proportional to the "distance frequency" relative to each point and to XDM (formula 12 of the appendix). The magnitude XDM, being common at all points, does not intervene in discrimination in distance. Only the frequencies, or more precisely the "distance filters" intervene in the discrimination. Here, only the relative measure of the distance of the points is interesting, because it is a problem of recognition of the target.

Figure 5 also shows that the samples
Dn (p) from the Doppler filters of the first Fourier transformer 33 are applied to a second Fourier transformer module 45, which performs the distance analysis.

The output of this module 45 consists of signals expressed in terms of frequency, but representing distances, which for this reason are called "distance frequencies".

These signals are applied to a distance refining circuit 47, which operates as the previous ones, and therefore delivers more accurately the "distance distances1 noted
FREDIS of the different brilliant points.

The operation of the device according to the invention can be better understood by referring to the formulas given in the appendix to the present description.

According to FIG. 1, it is considered that the signal coming from the reception means 36, PHIBAT (t), results from the beat between the local reception oscillator and the echo signal received from the target, with a delay due to its time to go back and forth.

Since the local oscillator and the transmitter follow the same stepwise modulation, it follows that the signal PHIBAT (t), representing a beat phase, depends only on the transmission frequency considered Fn, on the distance
D (t) of the target and the propagation velocity c of the electromagnetic waves.

The relation 2 expresses that the transmission frequency Fn varies with respect to a base value F0, by amplitude jump
SF, n being the rank of the jump considered, and varying from O to Nl.

Relation (3) expresses that the distance of the target varies from a distance OD (at time t0), taking into account the radial velocity VR of the target.

The relation (4) takes into account the fact that the distance OD is known only with a certain ambiguity. This is an unknown multiple (factor i) of the ratio between the speed of propagation of electromagnetic waves and twice the frequency of recurrence of the radar.

All this makes it possible to write the signal PHIBAT (t) in a more developed manner, according to equation (5).

The last term of this equation disappears, as being always a multiple of 2pi, because it has been seen that the jump of frequency SF is a multiple of the repetition frequency fR.

The previous term reflects that at each step change of the transmission frequency, there appears a variation of the instantaneous frequency.

However, the output of the member 20 of Figure 1 provides the average radial velocity of the target. So we can very easily compensate for this term by correcting the frequency
Doppler, from a quantity -2 / c .SF <VRO>, where <VRO> denotes the average Doppler over the time interval considered.

The Applicant then found that the expression of PHI
BAT (t) is reduced to that given by relation (6).

It is now assumed to be in a stage where the emission frequency is constant, this bearing being of rank n. In this case, the beat phase has the value PHI
BATn (t) given by relation (7). The second writing of the second term of this equation (7) involves the Doppler frequency FD (t), according to relation (8).

The Fourier transform module 33 analyzes the frequency, by phase shifting, as given in relation (9). We thus obtain directly the Doppler frequency of the target, -FD (t).

 It is now necessary to extract the distance. For this, the signal from the Doppler extractor is sampled at different times tn step change of the transmission frequency.

The module 45 performs a second Fourier transformation on this series of N samples (it is recalled that there is
N bearings).

Between the samples of rank n and n + 1, the beat phase rotates by the amount PHIBAT (tn + 1) -PHIBAT (tn).

The apparent frequency therefore has the value given by equation (10). In second writing, the second member of this equation has two terms.

The first term involves, inversely proportional to the duration T of a plateau, the ratio between the desired distance dO and the distance ambiguity AD of the analysis. It is recalled that this distance ambiguity is equal to the quotient of the speed of propagation of electromagnetic waves by twice the frequency hopping SF.

The second term represents the average Doppler shift of the target for the n rank, whose mathematical expression is given by equation (11). This signal corresponds to the magnitude XDM (p) mentioned in connection with FIG.

The relation (12) then shows how we can derive the average value of the distance d0, from the two graneurs given by the assembly of FIG.
Average Doppler XDM (p), and the frequency FREDIS distance, taken on average on the N levels of the emission frequency.

In a particular embodiment, the base transmission frequency F0 is for example 10 GHz. The recurrence frequency fR is, for example, 30 kHz.

Doppler analysis is performed for a K = 64 number of samples. The Fast Fourier transform is weighted by the Kaiser window with a beta value of 8.

The duration of a Doppler analysis is approximately 2.13 milliseconds. The analysis band is 30 kHz in frequency, for 450 m / second in speed. The analysis step is 469 Hertz in frequency, for 7 meters / second in speed.

The distance analysis is carried out on N = 32 samples, which corresponds to 32 levels of the transmission frequency.

The sampling period is 2.13 milliseconds. The Fast Fourier transform is weighted by the Kaiser window at beta = 4. Those skilled in the art will understand that this lower value of beta is made possible because bottom rejection has already been done by the first fast Fourier transformation.

The jump of the transmission frequency is SF = 9.36 MHz.

This results in a duration of the distance analysis which is approximately 68.27 milliseconds. The analysis band is 469
Hertz in frequency for 16 meters in distance. The analysis step is 14.6 Hertz in frequency for 50 centimeters in distance.

We see that this is perfectly compatible with a good discrimination of the bright spots of an aircraft. Moreover, the times required to perform the processing make it possible to access the information with a speed much greater than could be given by the simple temporal derivatives of the state variables that were available according to the prior art.

 It has been indicated above, in general, how one will be able to draw the velocity vector of the target aircraft, knowing the course of it and its radial velocity. We will see it now in more detail. The library 50 comprises aircraft models, each of which is specific to a particular target, around a given presentation, that is to say for a given heading of this target.

These models are thus defined by a distribution in x and y bright points in the reference of the aircraft carrying the radar.

The angular position in site / bearing of the target aircraft with respect to the carrier of the radar is assumed to be known. To a given model, oriented around a given heading, corresponds a distance distribution of the bright points that can be obtained at first order by a linear relation of x and y, which are the coordinates of the bright points.

 Reference is now made to Figure 6.

This first recalls the step 45 consisting in performing the fast Fourier transformation in distance for the target, the step 47 consisting in performing the distance refining of the "FREDIS" information obtained for the target, being specified here that a threshold is first made at -12dB before refining to eliminate the weak bright spots. Step 49 indicates that the obtained extremums, representative of the bright spots, are classified by increasing levels.

Step 51 consists in the choice of a model in the library 50.

Step 53 consists in adapting the bright points of the target (in IPMAX number) to be considered, given the number of bright points contained in the model.

Step 55 performs a second transformation of
Fourier "synthesized" for the target, on the selected IPMAX brilliant points.

The word "synthesized" now means that one will each time synthesize a zero Doppler shift signal, taking into account the distances retained. On this synthesized signal, we can perform fast Fourier transformations, in order to allow the analysis.

Examination of Figure 6 shows that the signal from the target has undergone its first Fourier transformation in distance (the second in all) at step 45.

The second Fourier transformation that it undergoes at 55 has the effect of normalizing the levels of the extreme filters.

The Applicant considers this normalization operation to be practically necessary. In fact, when several bright points are not discriminable, the corresponding extremum is, in the raw state, of level higher than that of the other discriminable points, the energies being canceled. Now, the recognition of the target (and especially of its course) is based on the principle of a comparison of the distribution of bright points which is purely geometrical in nature. It is therefore appropriate that the signal level does not interfere in this comparison.

In step 61, a course value is retained for the chosen model.

Step 65 consists of performing a first fast Fourier transform on a synthesized signal, as already defined. This fast Fourier transformation is the same as that performed at 45 on the signal from the target.

Thus, there is also synthesized a zero Doppler shift signal taking into account the distances of the model vis-à-vis the aircraft, given the selected heading and the approximate linear relationship expressed above. It is common to consider that bright dot patterns are "ball" models. This is the reason why step 65 of Figure 6 refers to a model possessing NPBRT (M) balls.

M is the rank of the model (its number). Not all models have the same number of balls.

Step 67 then performs a refinement of the distance frequencies obtained for the model. This refining is comparable to that of step 47 for the target.

Step 75 consists of performing a second normalization Fourier transformation on the synthesized signal relating to the model having undergone the first Fourier transformation at 65. This step 75 is comparable to step 55 performed for the target.

Outputs of the two steps 55 and 75, then draw two batteries of frequencies directly comparable distances.

This comparison is made at 76. It is indeed a correlation of the outputs of the distance filters, refined, on the one hand for the target, on the other hand for the model.

For a given model M, and a selected heading, attributed to this model, the correlation is calculated at each filter filter shift, "target distance filter" and "distance model M" batteries.

The maximum correlation can be stored in a variable marked CORMAX (cap, M).

For the different values of the heading, we obtain a network of correlation curves. The chosen model, with its course, to be the most representative of the target, is of course the one that gives the strongest correlation.

Those skilled in the art know that false studs (or "false extremums") may appear. However, their presence is not repetitive from one distance scan cycle to another, which will eliminate them.

Indeed, as and when analysis cycles, we average on the variable CORMAX, which gives a mean value CORMOY also depends on the heading and model M chosen. This results in a smoothing of the correlation curves.

The experiments conducted by the Applicant have shown an excellent operation of the device of the invention. For example, a variation of 10% on the correlation coefficient was observed, for a simple offset of 75 centimeters on a single bright spot of the model.

Of course, the present invention is not limited to the embodiment described. It extends to any variant, included in the context of the claims below.

(8) FD(t) = - 2FO VR(t)
c (9) FREBAT(t) = 2 d PHIBAT(t) = - FD(t)
dt

(12) < dO > = T.AD ( < FREDIS > + XDM) APPENDIX - FORMULAS (1) PHIBAT (t) = 4 #. Fn. D (t)
c

(8) FD (t) = - 2FO VR (t)
c (9) FREBAT (t) = 2 d PHIBAT (t) = - FD (t)
dt

(12) <dO> = T.AD (<FREDIS> + XDM)

Claims (7)

Claims. A pulsed Doppler radar device, in particular for tracking a target, and comprising radar transmission means (10), radar reception means (16), means for processing (20) signals received for deriving the position and the radial velocity of a target, such as an aircraft, and means (24) for tracking the target, by estimating the trajectory thereof, characterized in that the transmission frequency is subject to repetitive variations capable of increasing the distance discriminating power of the radar, in that the processing means (3) are arranged to discriminate in distance from several bright spots of the target, and in that the device comprises in addition means (60) for comparing the actual distance distribution of these bright spots with a set of model distributions, relating to different types of targets or aircraft, and each associated with a particular orientation, in the space, of the main axis of movement of the corresponding aircraft, as well as means (80) for completely determining the target velocity vector from its measured radial velocity, and the main axis orientation associated with the model distribution closest to the actual distribution, which makes it possible to initialize and update (90) the tracking loop (24), in particular its trajectory estimator, from complete speed information. 2. Device according to claim 1, characterized in that the transmission frequency is periodically subject to a step modulation, each jump is an integer multiple of the recurrence frequency. 3. Device according to claim 2, characterized in that the processing means perform, for each frequency step, a Doppler analysis (30) at the recurrence rate followed by a distance analysis (40) on the samples collected at the output. of each Doppler filter, at the repetition rate of the transmission modulation, so as to remove the ambiguity between the speed and distance information, which provides the individual distances of the bright points that the target comprises. 4. Device according to claim 3, characterized in that the Doppler (30) and distance (40) analyzes are performed by fast Fourier transforms. 5. Device according to one of claims 3 and 4, characterized in that the Doppler analysis is accompanied by a Doppler refinement (34), followed by an average (35), while the Doppler measurements are refreshed at the same time. using a filter of Kalman (37). 6. Device according to one of claims 3 to 5, characterized in that the distance analysis comprises, for each Doppler filter, the determination of the average (41) of the measurements made at different levels of the transmission frequency, the performing a new Doppler refining (43), the Fourier transform (45) and performing a distance refining (47). 7. Device according to one of claims 3 to 6, characterized in that the comparison means comprises - means (65) adapted to generate, for each model distribution and each course orientation, a first fictitious, offset received signal. Zero Doppler, corresponding to what would give this distance model distribution, and to perform on this first signal a Fourier transform identical to that of the signal from the target, which provides distance filter outputs relative to the model, - means for remotely refining the filter outputs relating to the model on the one hand (67), on the other hand (47), means capable of generating two other signals of Doppler zero from these refined outputs, respectively relating to the model (75) and the target (55), as well as to perform a second Fourier transformation in distance on each of these two other signals, and - means (76, 78) to compare, by correlation, the model-related filter outputs for each heading direction and the target.