The present invention relates to a method for producing a set of direction finding antennas and an antenna assembly carried out according to such a method. The invention applies in particular in the field of the detection of radio signals in electronic warfare {Electronic Support English), these signals may originate from radar, telecommunication transmitters or other device radiating such a signal.
The invention relates more particularly to the finding and more particularly to a method for manufacturing an antenna array goniometry capable of measuring the direction of arrival of a radio signal. The invention also relates to a set of direction finding antennas made according to such a method.

In general, applications in electronic warfare interested in very short signals, that is especially true for radar signals. This leads to goniometers embodiments necessarily based on a set of a plurality of fixed antennas, illuminated by the interesting radio signal, will deliver a set of signals captured at the same time, this clearance being the bearer of the direction of arrival (Branch of Arrival DOA or translation) of said signal. Several estimators exist to calculate the direction of arrival. Before calculating, a need is to acquire this set of signals, which involves as many reception channels of antennas. The reception channels being relatively expensive material entities, it is important to optimize the number, that is to say, to design sets of direction finding antennas for which we seek is to maximize performance in number of antennas given either to minimize the number of data performance antennas.

The design of a set of direction-finding antennas generally must meet a specification laying down the requirements, they can be expressed in three categories:

- A first category relates to the characteristics of the antennas used to construct the set of direction finding antennas. For each antenna, it has its gain

(Amplitude and phase) depending on the direction of arrival, frequency and the radio signal of incident polarization. - A second category relates to the geometrical constraints on the supporting platform, the set of placement of direction finding antennas, including the relative positions of the antennas. These constraints describe at least the area covered by each antenna and the maximum surface assigned to accommodate the set of direction finding antennas. In fact, there is a minimum spacing between antennas.

- A third category describes the performance to be achieved by the user device in the set of direction-finding antennas. Among them, the spatial coverage, frequency and polarization, accuracy and rate of direction finding ambiguity.

Thereafter, we will speak of direction finding ambiguity. This problem exists when the set of direction-finding antennas has two similar answers, or very strongly resembling, for both arrival directions sufficiently different. This is due to the principle that a phase shift is measured at an integer times 2π close. Thus, when two antennas are spaced less than one half the length of an incident signal wave, the geometric phase shift between antenna phase centers, exceeding 2π, will be measured with ambiguity and the direction of arrival that the goniometer will provide ambiguous.

This problem is well known in the interferometers. One solution is to use an irregular arrangement of antennas for varying the angular spacing of the ambiguities of a pair of antennas to another. A judicious arrangement of antennas and finally allows to exploit the redundancy of the information measured by the various pairs of antennas to unambiguously determine the direction of arrival.

This interferometric technique, well known, uses only the phase differences between antennas regardless of the magnitude. To the extent that the amplitude would have a sense of the definition of the constituent antennas, it would be wise to use this range to improve the direction of arrival measurements.

In addition, sets of antennas goniometry interferometric bases or interferometry, are very generally antennas aligned in the desired angular measurement plane. If the direction of arrival of the radio signal is incident in a plane inclined with respect to the measurement plane, then the measurement may be very wrong. Therefore, it is necessary to compensate website a deposit interferometer when it has to work in a range of locations large enough in terms of its deposit in precision; it is then necessary to add, for example, a deposit in interferometer, a goniometer site that can also be an interferometer. In such a case, all the antennas are not directly involved in estimating two angles but indirectly correction. This type of solution is poorly adapted to size a set of direction-finding antennas in which all antennas are leveraged directly to jointly estimate the two angles. Furthermore, this technique also lends itself easily to the development of a set of direction-finding antennas where the antennas have different polarizations adapted, since the phase difference between two antennas is no longer linked to one direction arrival, but also depends on the polarization of the incident radio signal.

An object of the invention is notably to correct all or part of the disadvantages of the prior art by proposing a solution for estimating the arrival direction of an incident signal in two dimensions. To this end, the invention relates to a method of manufacturing a set of direction finding antennas in two dimensions comprising at least three antennas, comprising a determining step of the optimum configuration of said set from a list of possible configurations, a configuration being defined by the gain, the pointing direction and the position within said set of each of said antennas, said phase comprises at least:

- a step of defining a network of reference antennas, said network covering a surface having a dimension in elevation and / or inversely proportional to respectively deposit a level of precision required in elevation and / or azimuth to the estimate arrival directions of the incident waves, and

comprising a plurality of antenna elements, said antenna elements being distributed in a regular mesh, the distance between two adjacent elementary antennas being substantially equal to half the wavelength associated with the maximum frequency of a frequency range of interest , the number of antennas of said array of reference antennas being greater than the number of antennas of said set, the spacing between the extreme antennas of said array being greater than or equal to the spacing between the end of said antenna assembly according to the axis deposit and / or the elevation axis,

A configuration research stage to consider from predetermined constraints to establish a list of configurations to consider,

A quantization step of the maximum level of ambiguities in each of the configurations of said list from a correlation function in order to associate to each of said configurations an evaluation quantity,

A search step of the configuration having the lowest evaluation variable, said configuration being the optimum configuration.

In one particular implementation, said plurality of direction-finding antennae being intended for direction of arrival of measures incident radio signals does not depend on the polarization of the said signals, the evaluation value associated with a configuration is equal to the maximum value of a correlation function F Horn (0 lJ 0 2 ) according to two directions of arrival where 0 ! 0 and 2 representing two directions of arrival sweeping the direction of arrival of coverage area of said pattern for a field and the direction of arrival of interest to each other, and excluding the values for which the correlation function of said reference antenna array Corre (0 1 , 0 2 ) is greater than or equal to a threshold S Ref predetermined correlation functions F Horn (0 lJ 0 2 ) and F Corre (0 1 , 0 2 ) expressing respectively from the pointing vector of said pattern and the pointing vector of said reference set.

In another possible embodiment, said antenna assembly being intended for direction of arrival of radio signals incident measures depending on the polarization of these said signals, the evaluation value assigned to a configuration is equal to the maximum value of eigenvalues of a matrix Γ * (Θ 1 , Θ 2 ) ■ Γ (Θ 1 , Θ 2 ), based on two directions of arrival where 0 ! and Θ 2 representing two directions of arrival sweeping the angular coverage area of said pattern and for one angular area of interest to each other, wherein:

Γ(Θ1, Θ2) = fa 1 Λ ' Hnorm i®2> ^-min) ^Vnorm i®2> ^-min)] où :

- Γ (Θ 1, Θ 2 ) is a square matrix 2 x 2;

- U "normal, rnin)} ^ mQ 2 χ Ν

- [U H norm (®2 > A mi n) U Vnorm {Q 2 , min )] is a matrix N x 2;

- U Hnorm (Q, TO min ) and U Vnorm (q, a min ) are two vectors forming an orthonormal basis of the plane spanned by the two pointing vectors U H (Q, A min ) and U v (Q, A min ) of the antenna array direction finding the minimum wavelength, respectively in horizontal rectilinear polarization and vertical rectilinear polarization,

- The * is the transposed and conjugate transformation.

The list of configurations to consider for example corresponds to the list of possible configurations.

In another possible implementation mode, the list of configurations to consider for example corresponds to a random selection of a predetermined number of configurations from the list of possible configurations.

The antenna array reference antennas being aligned in a grid, the positions in the possible configurations of the antennas of the set of direction finding antennas are for example aligned on said mesh.

Said array of reference antenna is for example a network of radiating elements, each antenna of said antenna assembly goniometry being made from a subnet of said network.

The invention also relates to a set of direction finding antennas obtained by such a method.

Other features and advantages of the present invention will become apparent from reading the following description given by way of illustration and not limitation, and with reference to the accompanying drawings in which:

- Figure 1 illustrates a definition of geometric reference used and unique angles of deposit and site;

- Figure 2 shows possible steps of designing a set of direction finding antennas in two dimensions;

- Figure 3 shows an exemplary embodiment of a set of direction finding antennas in two dimensions in the case of non-dependence of the polarization;

- Figure 4 shows an exemplary embodiment of a set of direction finding antennas in two dimensions in the case of the bias dependence (in case of polarization diversity);

- Figures 5a and 5b are graphic representations of the correlation functions of the set of antennas corresponding to the configuration goniometry materialized in Figure 3 and the network reference antennas, respectively Cor (0 lJ 0 2 ) and

FcorRêf(9l> ®2) :

- Figure 6 illustrates a definition of the geometrical frame of reference used with a set of polarization diversity direction finding antennas;

- Figure 7 is a graphical representation of the generalized correlation (matrix calculation) of said plurality of polarization diversity antenna goniometry materialized in Figure 6;

- Figures 8a and 8b respectively show a configuration of a set of direction finding antennas designed according to the invention

and the graph of the correlation function illustrating the results obtained for this configuration;

- Figures 9a and 9b respectively show a configuration of a set of polarization diversity direction finding antennas designed according to the invention and the graph of generalized correlation illustrating the results obtained for this configuration.

The present invention relates to a method for producing a set of direction finding antennas capable of working in two angular dimensions, for example bearing and elevation. If necessary, the method is of course applicable with only one angular dimension.

Figure 1 points out that, for a one incoming direction, materialized by a straight arrival direction 1 1, the deposit is the angle formed by the straight line 11 0, corresponding to the projection of the line of direction of arrival on the horizontal plane and a reference axis in the horizontal plane (or lubber line, e.g., normal to a plane of alignment of the antennas). The site is the angle formed by the straight direction of arrival 11 and Projection 1 10 on the horizontal plane.

In the following, we will use the term direction of arrival symbolized by Θ, and thus generally defined by two angles, the reservoir 9 g and the site =

The set of direction-finding antennae can be realized both from conventional non-array antennas (spiral, serpentine, butterfly, horn, etc., ...) that from an array antenna in which we define a set of sub-networks, this assembly forming said set of direction finding antennas. In other words, the assembly is then made from beams formed with sub-networks of a network of elementary antennas.

The method of the invention includes a search phase of the optimal configuration of the set of direction finding antennas followed by a production phase from this optimal configuration.

In general, by configuration, comprises defining each constituent antenna within the assembly, that is to say, the gain function of the direction of arrival, frequency and polarization, position of the phase center and the pointing direction, regardless of the embodiment with conventional antennas with or formed beams. This comprehensive definition of a configuration can however be simplified as discussed further.

The method according to the invention for example comprises the following steps shown in Figure 2:

- A first step 21 of defining a reference network of antennas;

- A second step 22 of defining configurations for consideration;

- A third stage 23 of each configuration assessment for consideration by a method involving the network of reference antennas and to assess the quality of direction finding in terms of ambiguity and precision;

- A fourth step 24 of determining the best configuration;

- A fifth step 25 of embodiment of the antenna assembly goniometry for this best configuration.

The first step of defining a network of reference antennas consists in defining a plurality of K all identical elementary antennas whose phase centers are arranged regularly on a mesh surface. The distance between two adjacent antennas of the array must be substantially less than half the length minimum wavelength, the minimum wavelength X min corresponding to the maximum operating frequency f max , which is the maximum range of a frequency frequencies of interest, specific to each application. The lengths of the network, in the planes of horizontal and vertical section, are inversely proportional to the direction-finding information respectively in azimuth and in elevation. The number of antennas of the network of reference antennas is greater than the number of antennas of the set of antennas. The spacing between the extreme antennas of the array reference antennas is greater than or equal to the spacing between the extreme antennas of the set of antennas, regardless of the axis considered, site or deposit.

In general, this mesh surface is not necessarily flat, it may for example be cylindrical. However, a simplified embodiment may be a flat mesh surface.

This network reference antenna is a simple device in the method of calculation in the case where the set of direction-finding antennae is carried out with conventional antennas. For against, in the case where the set of direction-finding antennae is achieved by beams formed from subnets, the network reference antennas may specifically correspond to the array of elementary antennas which are made under the -networks generating said shaped beams.

The second step of defining patterns to be taken into considerations is to provide the third step a configuration list to evaluate such that the fourth step may select among them the best according to a criterion on size for evaluating each configuration.

It is recalled that configuration is the physical definition of a set of direction finding antennas, which assembly comprises N antennas, N being an integer greater than or equal to 2. This physical definition corresponds to each of the N antennas in the constituent the most general case, the gain function of the direction of arrival, frequency and polarization, the position of the phase center and the pointing direction. This applies regardless of the embodiment with conventional antennas or with trained beams.

The definition of a configuration can however be simplified in many cases as possible. An alternative embodiment may then lead to a configuration reducing only the positions of the phase centers of the antennas in a plane, these being all identical, arranged in a plane and pointing in the same direction.

Note that the antenna gain (depending on the direction of arrival, frequency and polarization) is a generalization of profit definition. Indeed, for the common use cases, it will not tend to employ different constituent antennas of each other, unless polarization in response to the polarization diversity of reasons.

These configurations are framed by the specification and by geometrical and technical considerations dependent fashion

embodiment of the set of direction finding antennas. For example, in the case of an embodiment with conventional antennas, the antennas to be shy or mechanically, nor hide, they will not overlap. By cons, in the case of trained harnesses, antennas could overlap since the training bundles per subnet would permit; this is a technical issue of specifications.

The network reference antenna defined in the first step, provides the regular mesh of the implantation surface of the phase centers of the component K antennas of the network, with a mesh of not substantially less than half the length of minimum wave min / 2. Phase centers of the component antennas of the set of direction finding antennas, beamforming, alignable on a half-pitch lattice d / 2 according to beamforming, may be used at half the mesh not also when the antennas are conventional.

Figure 3 illustrates an example embodiment of a plane antenna assembly goniometry 30 as well as the possible positions of each of the phase centers of the antennas thus produced with subnets 31. In order not to overload the figure, only the possible locations 31 1 for 31 0 phase centers of each of the antennas 31 have been shown.

In a first implementation mode, at this stage, we can create a list of all possible configurations of the set of direction-finding antennas, establishing all possible combinations within the constraints and specifications .

According to a second alternative mode of implementation, the configuration list to be evaluated may be determined by randomly selecting, in the set of possible configurations, a limited number of configurations relative to the screen as possible. This mode is designed to avoid too many configurations to evaluate third step, if the application is constraint in execution time. Insofar configurations are limited to the positions of the phase centers of the antennas, it is interesting that a random drawing will reproduce the configurations of statistical irregularities, making that may have, in the list as well restricted, a sufficiently irregular configuration for a level low enough direction finding ambiguity.

The third step is based on an assessment of the maximum level of goniometry ambiguities produced by each overall configuration of direction finding antennas, each evaluated configuration have been defined in the second step.

A direction finding ambiguity corresponds to the direction of arrival of measures identical for different actual arrival directions. In practice, given the imperfections of achievement of physical and measurement noise of any kind, a direction finding ambiguity corresponds to the direction of arrival measurements close enough for remote real directions of arrival.

The level of goniometry ambiguities can be evaluated by correlating the arrival directions of measurements taken by a set of direction-finding antenna in a particular field of directions of arrival, thereby eliminating field cases where the correlation the direction of arrival measurements is normal, which can be seen by correlating the direction of arrival measurement of the reference antenna array which produces an ideal response.

The correlation may be supported by a calculation of more or less generalized correlation function according to the direction of arrival measurements depend on whether the polarization of radio signals to be processed.

For this calculation, we need to distinguish two areas of arrival directions: the coverage area and the area of ​​interest. The coverage area is the area of ​​directions of arrival for which the set of direction finding antenna may receive RF signals. The area of ​​interest is given by the specification, it is no greater than the coverage area, it is usually small compared to the latter.

Practical way to calculate the correlation functions more or less generalized, we can take the angular values ​​not linearly distributed in these areas, but the angular values ​​which the sinuses are linearly distributed. This advantageously reduces the number of arrival directions taking into account, as necessary, expanding trained beams linked to misalignment.

The first case is when the direction of arrival measurements made with the set of direction finding antennas do not depend on the polarization of radio signals incident. In this case, the correlation is expressed by a simple correlation function. The maximum level of ambiguity of a set of direction finding antennas, associated with a given configuration, corresponds to the maximum value of the correlation function of said set max & l) & 2 (F Horn Q 1 , Q 2 )) where 0 ! 0 and 2 are two directions of arrival sweeping the coverage area for one and the area of interest to the other (the allocation of areas to 0 ! and 0 2 is indifferent), and excluding values for which the correlation function of the reference antenna array Corre (0 1 , 0 2 ) is greater than or equal to a threshold S Ref predetermined. Generally speaking, the result is between 0 and 1 terminal included. A preferred value of the threshold S ref is 0.5.

Correlation functions F Horn (0 lJ 0 2 ) and F Corre (0 1 , 0 2 ) are expressed respectively from the pointing vector (steering vector or translation) of the set of U direction finding antennas ( Q, A min ) and the pointing vector U of the reference antenna array Ref (Q, A min ):

f CO r (0i 0 2 ) =
■ U {Q 2 , min ) | 2 a

FcorRéf (®l> Θ 2 ) = | ^ Re / (®l ^ - mm) " ^ ref ^ 2> ^ -min) \ where the sign * is the transposed and combined processing.

Generally speaking, a pointing vector of a group G of

P antennas, U G {Q, X), is a unit vector comprising P components, including the p-th component is proportional to the response of the p-th antenna, amplitude and phase A G p {Q, X) D = G p {Q, X) ■ e OM c p u (0) where, as shown in Figure 1:

- D G p ®, X) is the radiation pattern or gain (amplitude and phase) of the p-th antenna of the group G in the arrival direction 0 and the wavelength λ;

- M G p is the spatial position of the phase center of the p-th antenna of the group G with respect to an origin O;

- u (0) is the unit vector along the direction of arrival 0; - e i-- ° M G , v -um represents the term of the phase shift related to the position of the p-th antenna phase center in the group G. In general, the U pointing vector G ®, X) can be expressed as the ratio of vector A G {Q, X), the p-th component A is G p ®, X), at its Euclidean norm || i4 G (0, l) || which itself can be expressed as follows || i4 G (0, / l) || ^ = A * G {Q, X) ■ A G (Q, X), A * G (Q, X) is the conjugate transposed vector of the vector A G ®, X).

The pointing vector U {Q min ) is obtained by application of the above to the N antennas of the set of direction finding antennas.

The pointing vector U Ref (q, a min ) is obtained by application of the above to K antennas of the reference antenna array. Given the generally low directivity antennas network reference antennas in a work alternative embodiment, the gains D Ref k ®, X) can be replaced by 1. In another implementation variant, these gains D ref k {Q, X) may be replaced by the weighting coefficients P k different from one antenna to another, in order not to penalize a configuration of the set of direction finding antennas providing a level of lesser ambiguities at the cost of a slight deterioration of the accuracy of direction of arrival.

The second case is when the direction of arrival measurements made with the set of direction finding antennas depend on the polarization of the radio signals for both polarization diversity reasons incident radio signals for response reasons polarization of the component antennas. When the set of direction finding antennas must be capable of processing a signal incident polarization diversity, use of the component antennas may form a decomposition base of the polarization, which is preferably orthogonal. For example, polarization of the antennas are conventionally used suitable rectilinear horizontal and vertical rectilinear matched polarization antennas. But it can also be polarization antennas right circular adapted and suited left circular polarization antennas.

In this case, the correlation level of the set of direction-finding antennae is assessed with the matrix product Γ * (Θ 1 , Θ 2 ) ■ Γ (Θ 1 , Θ 2 ) and

maximum level of ambiguity is the largest eigenvalue of the matrix product:

¾a [C * (Th 1 , Th 2 ) - C (Th 1 , Th 2 )]

with

1>

OR

Vpmax own means maximum value;

Γ (Θ 1 , Θ 2 ) is a square matrix 2 x 2;

is a matrix 2 x N;

- [U H norm (®2 > A mi n) U Vnorm {Q 2 , min )] is a matrix N x 2;

- U Hnorm (Q, TO min ) and U Vnorm (q, a min ) are two vectors forming an orthonormal basis of the plane spanned by the two U pointing vectors H (Q, A min ) and U v ®, min ) of the set of direction-finding antenna to the minimum wavelength, respectively in a horizontal linear polarization (H) (corresponding to an electric field collinear with the vector Û H Q) of FIG 4, wherein said vector is a unit vector orthogonal the direction of arrival line defined by Θ and located in the local horizontal plane of the set of direction finding antennas) and straight vertical polarization (V) (corresponding to a co-linear with the electric field vector u v {& ) of FIG 4, wherein said vector is a unit vector orthogonal to direction of arrival of the line defined by Θ and located in the local vertical plane, comprising the right direction of arrival of the set of direction finding antennas ), which may take for example the form suiva nt:

- The sign * is the transposed and combined processing.

U pointing vectors H (Q, A min ) and U v ®, min ) are unit vectors comprising N components since they correspond to the set of direction finding antenna that has N antennas, their nth

components are proportional to the responses, amplitude and phase of the n-th antennas respectively horizontal polarization

_4 Η <η (Θ, λ) = D H n {Q, X) ■ and i-- ° Mn um and in polarization vertical V n ®, X) = D V n {Q, X). ^ ▼ buttons Ο ιι'ι ' (Θ) , where de manière similaire au premier cas et comme illustré by the figure 4:

- D H n ®, X) and D V n {Q, X) are the radiation patterns or gains (amplitude and phase) of the n-th antenna of the set of direction-finding antenna in the direction of arrival Θ, the wavelength λ and respectively in horizontal rectilinear polarization and vertical rectilinear polarization;

- M n is the spatial position of the phase center of the n-th antenna of the set of direction finding antennas with respect to an origin 0;

- u (0) is the unit vector along the direction of arrival Θ;

- e i-- ° Mn um term represents the phase shift related to the position of the phase of the n-th antenna center in the set of direction finding antennas.

The fourth step of determining the best configuration is to retain the configuration of the set of direction finding antennas having the maximum level of the lower ambiguities among those calculated in the third step, and less than a threshold S max predetermined. This threshold ensures that the maximum level of ambiguity is sufficiently low for the quality of the direction finding and, if necessary if it is not, to start the method according to the invention from the first step necessarily releasing constraints such as, for example, the number of direction finding antennas N, that is to say by increasing it.

The preferred values of the threshold S max is less than or equal to 0.9.

Further explanation is provided below on the basis of non-limiting examples illustrated by FIGS.

Figures 5a and 5b illustrate the phenomenon of goniometry ambiguities by the graphical representation of the correlation function.

To facilitate interpretation, the direction of arrival Θ is restricted to deposit 9 g , the site 9 S is assumed to be zero. As suggested earlier, the deposit scales are also sine of the deposit. The coverage area in azimuth is from -90 to 90 degrees, and the field of interest of deposit ranges from -9 gi = -15 degrees to 9 gi = 15 degrees.

Figure 5a corresponds to the correlation function FCOR (.®i > ®2) FCOR = {^ gl , 9 g2 ) of the configuration of the set of direction finding antennas previously described with Figure 3. In this configuration, the N direction-finding antennas have the same radiation pattern pointed in the same direction and are regularly spaced according to a pitch G according to the y-axis (horizontal) and is therefore naturally a set of ambiguous antennas to the length of At minimum wave min . This results in a multitude of lines 51 in addition to the right 50, which itself is not considered as a place of ambiguity. Indeed, Figure 5b corresponds to the correlation function F CORREF (®i > Θ 2 ) = FcorRéf (d gl , 9 g2 ) of the network reference antennas, it illustrates its robustness to ambiguities, that is to -dire the best possible result in terms of rejection of ambiguities. Only the pairs (9 gl , 9 g2 ) belonging to the right 50 passing through the origin, 9 gl = 9 g2 , provide a value of F CORREF (9 gl , 9 g2 ) equal to 1, which is absolutely normal.

Note that these graphical representations of correlation functions, the thickness of the lines 50, 51 reflects the achievable accuracy of the estimated arrival direction. More right 50, 51 is thinner, the estimation of the deposit is accurate. Indeed, the thickness of these lines reflects the speed at which the pointing vectors décorrèlent gradually as it spreads arrival directions. The accuracy of direction of arrival direct result of this speed decorrelation, itself linked to the geometric dimensions of the set of direction finding antennas.

Note also in Figure 5a that the spacing 52 between

X

lines 50, 51 is regular and is precisely - ^. This is because two deposits 9 gl and 9 g2 exhibit similar pointing vectors

X

when the difference in their sinuses is an integer number of times - ≡L . This

^ G similarity to the non-zero integer, creates ambiguities that are normal here given the geometry of the configuration and result in the highest value of the correlation function, that is to say 1.

6 shows an exemplary embodiment of a direction finding antenna array 70 and polarization diversity and the possible positions of the antennas 71, 72. As in Figure 3, not to overload the figure, only the possible locations 730 phase center 73 of each of the antennas goniometry 71, 72 have been shown. The materialized configuration is directly inspired by the materialized configuration of the set of antennas monopolarization of Figure 3 and each antenna 71, 72 is aligned in a regular mesh. In this example, the set of polarization diversity direction finding antennas 70 comprises twice as many antennas distributed over the same surface area to achieve the same accuracy of that direction of arrival in the configuration evidenced in Figure 3. Half antenna 71 has a suitable horizontal rectilinear polarization and the other half of the antennas 72 has a suitable vertical rectilinear polarization.

In general, the antennae 71 are adapted polarization orthogonal to that of the antennas 72 and the antennas 71, 72 may be arranged in any manner to form the set of direction finding antennas provided to use as many antennas 71 that antenna 72.

According to a particular embodiment, the antennas forming the set of direction-finding antennae can be arranged in alternating checkerboard an antenna 71 and an antenna 72. Advantageously, this allows bipolarisations architecture checkerboard:

- to homogenize the likelihood of interception of the incident signals on the set of direction finding antennas according to their polarization;

- To provide a joint estimation of the direction of arrival and polarization of the intercepted signal;

- To optimize the accuracy of direction of arrival in elevation and azimuth.

According to another embodiment, the set of direction finding antennas 70 to polarization diversity antenna consists of double, so-called polarization and comprising two antennas adapted orthogonal polarizations whose phase centers are coincident near the imperfections. In this case, the number of antennas bipolarisations is identical to that of an antenna assembly 30 monopolarization.

Figure 7 corresponds to the correlation function FCOR (.®i > ®2) FCOR = {^ gl , E g2 ) for configuring materialized in Figure 6. The regular arrangement of the antennas 71, 72 causes maximum level of ambiguity. For coverage deposit areas and interest as in Figure 5a, Figure 7 shows half of straight, which is normal because the spacing according to the y-axis (horizontal) between two successive antennae is reduced in a two report.

Figure 8a shows an example of configuration of a set of direction finding antennas 80 designed according to the invention. This configuration was chosen from a list of ten thousand possible configurations obtained by a series of random draws. For a domain direction of arrival of interest comprising deposits between -15 and +15 degrees and sites between -10 and +10 degrees, the correlation function F Horn (0 lJ 0 2 ) is less than 0.75 . 8b graphically depicts the correlation function FCOR (®i> ®2) = {FCOR Pgi> Qg2) at zero site for deposits of interest between -15 and +15 degrees and cover deposits between -90 and + 90 degrees, the value of this function has a value less than 0.7 outside the right 50. the comparison of figures 5a and 8b allows to highlight the significant reduction in the level of all the ambiguities of antennas goniometry, the accuracy of direction of arrival being unchanged.

Figure 9a shows an example of configuration of a direction finding antenna array 90 and polarization diversity designed according to the invention. This configuration was chosen from a list of one million possible configurations obtained by a series of random draws. For a domain direction of arrival of interest comprising deposits between -15 and +15 degrees and sites between -10 and +10 degrees, the correlation function Cor (0 1 , 0 2 ) is less than 0.85 . 9b graphically represents the correlation function F Horn (0 1 , 0 2 ) = F Horn (0 5l , 0 52 ) at zero site for deposits of interest between -15 and +15 degrees and cover deposits between -90 and +90 degrees, the value of this function has a value less than 0.5 outside the right 50. the comparison of figures 7 and 9b allows to highlight the significant reduction of

level ambiguities of the set of direction finding antennas, the accuracy of direction of arrival being unchanged.

CLAIMS
A method of manufacturing a set of direction finding antenna (70, 80, 90) in two dimensions comprising at least three antennas, characterized in that, comprising a determining step of the optimum configuration of said set from a list of possible configurations , a pattern being defined by the gain, the pointing direction and the position within said set of each of said antennas, said phase comprises at least:

- a step (21) for defining a network of reference antennas, said network covering a surface having a dimension in elevation and / or inversely proportional to respectively deposit a level of precision required in elevation and / or azimuth to estimating directions of arrival of the incident waves, and comprising a plurality of antenna elements, said antenna elements being distributed in a regular mesh, the distance between two adjacent elementary antennas being substantially equal to half the wavelength associated the maximum frequency of a frequency range of interest, the number of antennas of said array being greater than the number of antennas of said set, the spacing between the extreme antennas of said array being greater than or equal to the spacing between the extreme antennas of said set according to the bearing axis and / or the elevation axis,

- a step (22) Research configurations to consider from predetermined constraints to establish a list of configurations to consider,

- a step (23) quantizing the maximum level of ambiguities of each of the configurations of said list from a correlation function in order to associate to each of said configurations an evaluation quantity,

- a step (24) to search the configuration having the lowest evaluation variable, said configuration being the optimum configuration.

The method of claim 1 wherein said set of direction-finding antennae being intended for direction of arrival of measures incident radio signals does not depend on the polarization of the said signals, the evaluation value assigned to a configuration is equal the maximum value of a correlation function F Horn (0 lJ 0 2 ) according to two directions of arrival where 0 ! 0 and 2 representing two directions of arrival sweeping the direction of arrival of coverage area of said pattern for a field and the direction of arrival of interest to each other, and excluding the values for which the correlation function of said reference antenna array Corre (0 1 , 0 2 ) is greater than or equal to a threshold s Ref predetermined correlation functions Cor (0 1 , 0 2 ) and PcorRéf (0i, 0 2 ) s 'expressing respectively from the pointing vector of said pattern and the pointing vector of said reference set.

The method of claim 1 wherein said plurality of antennas being adapted to radio signal incoming direction measures incident polarization dependent of these said signals, the evaluation value assigned to a configuration is equal to the maximum value the eigenvalues of a matrix Γ values * (0 1; 0 2 ) ■ Γ (0 1 , 0 2 ), based on two directions of arrival where 0 ! 0 and 2 representing two directions of arrival sweeping the angular coverage area of said pattern and for one angular area of interest to each other, wherein:

C (Th 1 , Th 2 ) =

or :

Γ (0 1; 0 2 ) is a square matrix 2 x 2;

U Hnorm ®1> ^-min

is a matrix 2 x N;

-Uvnorm (.®l> ^min

[U H norm (®2 > A mi n) U Vnorm {Q 2 , min )] is an N x 2 matrix;

U H norm (®, ^ min) and U Vnorm (Q, A min ) are two vectors forming an orthonormal basis of the plane spanned by the two U pointing vectors H (Q, A min ) and U v ®, min ) of the set of direction-finding antenna to the minimum wavelength, respectively

horizontal rectilinear polarization and vertical linear polarization,

- The sign * is the transposed and conjugate transformation.

Method according to one of the preceding claims wherein the list of configurations to be considered is the complete list of possible configurations.

Method according to one of claims 1 to 3 wherein the list of configurations to be considered is a random selection of a predetermined number of patterns from the list of possible configurations.

Method according to one of the preceding claims wherein the reference antenna array antennas being aligned in a grid, the positions in the possible configurations of the antennas of the set of direction finding antennas are aligned on said mesh.

Method according to one of the preceding claims wherein said array of reference antenna is an array of radiating elements, each antenna of said antenna assembly goniometry being made from a subnet of said network.

Goniometry set of antennas, characterized in that it is produced by the method according to any preceding claim.