Method for the blind wideband localization of one or more transmitters from a carrier that is passing by

Abstract

A method for localizing one or more sources, said source or sources being in motion relative to a network of sensors, comprises a step for the separation of the sources in order to identify the direction vectors associated with the response of the sensors to a source at a given incidence. The method comprises the following steps of associating the direction vectors a1p(1)m . . . aKp(K)m obtained for the mth transmitter and, respectively, for the instants t1 . . . tK and for the wavelengths λp(1) . . . , λp(k), and localizing the mth transmitter from the components of the vectors a1p(1)m . . . aKp(K)m measured with different wavelengths.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The invention relates to a method for localizing a plurality of ground transmitters in a broadband context from the passing by of carrier without a priori knowledge on the signals sent. The carrier may be an aircraft, a helicopter, a ship, etc.

The method is implemented, for example iteratively, during the passing by of the carrier.

2. Description of the Prior Art

The prior art describes different methods to localize one or more transmitters from a passing carrier.

FIG. 1 illustrates an example of airborne localization. The transmitter 1 is in the position (x₀,y₀,z₀) and the carrier 2 at the instant t_k is at the position (x_k,y_k,z_k) and perceives the transmitter at the incidence (θ(t_k,x0,y₀,z₀), Δ(t_k,x₀,y₀,z₀)). The angles θ(t,x₀,y₀,z₀) and Δ(t,x₀,y₀,z₀) evolve in time and depend on the position of the transmitter as well as the trajectory of the carrier. The angles θ(t,x₀,y₀,z₀) and Δ(t,x₀,y₀,z₀) are identified as can be seen in FIG. 2 relative to a network 3 of N antennas capable of being fixed under the carrier.

There are many classes of techniques used to determine the position (x_m,y_m,z_m) of the transmitters. These techniques differ especially in the parameters instantaneously estimated at the network of sensors. Thus, localising techniques can be classified under the following categories: use in direction-finding, use of the phase difference between two distant sensors, use of the measurement of the carrier frequency of the transmitter, use of the propagation times.

The patent application FR 03/13128 by the present applicant describes a method for localising one or more transmitters from a passing carrier where the direction vectors are measured in the same frequency channel and are therefore all at the same wavelength.

The method according to the invention is aimed especially at achieving a direct estimation of the positions (x_m,y_m,z_m) of each of the transmitters from a blind identification of the direction vectors of the transmitters at various instants t_k and various wavelengths λ_k.

Parametrical analysis will have the additional function of separating the different transmitters at each wavelength-instant pair (t_k,λ_p(k)). The parameters of the vectors coming from the different pairs (t_k, λ_p(k)) are then associated so that, finally, a localisation is performed on each of the transmitters.

The invention relates to a method for localizing one or more sources, said source or sources being in motion relative to a network of sensors, the method comprising a step for the separation of the sources in order to identify the direction vectors associated with the response of the sensors to a source at a given incidence. It is characterised in that it comprises at least the following steps:

• • associating the direction vectors a_1p(1)m . . . a_Kp(K)m obtained for the m^th transmitter and, respectively, for the instants t₁ . . . t_K and for the wavelengths λ_p(1) . . . , λ_p(k),
  • localizing the m^th transmitter from the components of the vectors a_1p(1)m . . . a_Kp(K)m measured for the different wavelengths.

The wideband method according to the invention offers notably the following advantages:

• • the localising of the transmitter is done by a direct method which maximises a single criterion as a function of the (x,y,z) coordinates of the transmitter,
  • it makes it possible to achieve an association of the direction vectors of the sources in the time-frequency space, making it possible especially to take EVF (Evasion de Fréquence=Frequency Evasion) and TDMA-FDMA (Time Division Multiple Access and Frequency Division Multiple Access) signals into account,
  • it can be implemented on calibrated networks or with amplitude diversity antennas such as co-localised antennas: namely antennas in a network with dipoles having a same phase centre and different orientations.

Other features and advantages of the present information shall appear more clearly from the following description of a detailed example, given by way of an illustration that in no way restricts the scope of the invention, and from the appended figures, of which:

FIG. 1 exemplifies a localisation, by an aircraft equipped with a network of antennas, of a transmitter having a position (x₀, y₀, z₀) on the ground,

FIG. 2 shows a network of five antennas and the angles of incidence of a transmitter,

FIG. 3 is a graph showing the general operations of the method in the presence of M wideband transmitters.

For a clear understanding of the method according to the invention, the following example is given by way of illustration that in no way restricts the scope of the invention, for a system as described in FIG. 1 comprising an aircraft 2 equipped with a reception device comprising, for example, a network of N sensors (FIG. 2), and a transmitter 1 to be localised.

Before explaining the steps of the method according to the invention, the model used is defined.

Modelling

In the presence of M transmitters, the aircraft receives the vector x(t,p) at the instant t at output of the N sensors of the network and of the p^th channel having a wavelength λp.

Around the instant t_k, the vector x(t+t_k, p) sized N×1 corresponding to the mixture of the signals from the M transmitters is expressed by: $\begin{array}{cc}x\left(t+t_k,p\right)=\left[\begin{array}{c}{x}_1\left(t+t_k,p\right)\\ M\\ x_N\left(t+t_k,p\right)\end{array}\right]=\sum _{m=1}^{M}a\left({\theta }_{\mathrm{km}},{\Delta }_{\mathrm{km}},{\lambda }_p\right)s_m\left(t+t_k\right)+b\left(t+t_k\right)=A_{\mathrm{kp}}s\left(t+t_k,p\right)+b\left(t+t_k,p\right)\mathrm{for}t<\Delta t/2& \left(1\right)\end{array}$

• • where
  • b(t) is the noise vector assumed to be Gaussian,
  • a(θ,Δ,λ) is the response of the network of sensors to a source with an incidence (θ,Δ) and a wavelength λ,
  • A_kp=[a(θ_k1,Δ_k1,λ_p) . . . a(θ_kM,Δ_kM,λ_p)] corresponds to the mixing matrix, s(t)=[s₁(t) . . . s_M(t)]^T corresponds to the direction vector,
  • θ_km=θ(t_k,x_m,y_m,z_m) and Δ_km=Δ(t_k,x_m,y_m,z_m).
  • x_n(t, p) is the signal received on the n^th sensor of the carrier at output of the p^th frequency channel associated with the wavelength λ_p.

In this model, the mixing matrix A_kp depends on the instant t_k of observation as well as on the wavelength λ_p.

The above model shows that the direction vector:a_kpm=a(θ_km,Δ_km,λ_p)=a(t_k,x_p,x_m,y_m,z_m) of the m^th transmitter  (2) at the instant t_k is a known function of (t_k,λ_p) and of the position of the transmitter (x_m,y_m,z_m). The method according to the invention comprises, for example, the following steps summarised in FIG. 3

• • the parametrical estimation (PE) and the separation of the transmitters (ST) at the instants t_k and wavelength λ_p for example in identifying the vectors a_kpm for (1≦m≦M). This first step is effected by techniques of source separation and identification described, for example, in the references [2] [3],
  • the association of the parameters for the m^th transmitter, for example in associating the vectors a_1p(1)m up to a_Kp(K)m obtained at the respective pairs (instants, wavelength) (t₁λ_p(1)) . . . (t_K,λ_p(K)). The direction vectors a_kpm are considered in a (time, wavelength) space or, again, in a (time, frequency) space, the frequency being inversely proportional to the wavelength,
  • The localizing of the m^th transmitter from the vectors a_1p(1)m up to a_Kp(K)m associated, LOC-WB. Step of Association and Tracking of the Wideband Transmitters

In the presence of M sources or transmitters and after the source separation step where the direction vectors associated with a source are identified and not associated, the method gives, for the instant/wavelength pair (t_k, λ_p), the M_k signatures a_kpm for (1≦m≦M_k), signature or vector associated with a source.

At the instant t_k and at the wavelength λ_p′, the source separation step gives the M_k′ vectors b_i for (1≦i≦M_k′). The purpose of the tracking of the transmitters is especially to determine, for the m^th transmitter, the index i(m) which minimizes the difference between the vectors a_kpm and b_i(m). In this case, it will be deduced therefrom that a_k′p′m=b_i(m).

To make the association of the parameters for the m^th transmitter, a criterion of distance is defined between two vectors u and v giving: $\begin{array}{cc}d\left(u,v\right)=1-\frac{{u^{H}v}^2}{\left(u^{H}u\right)\left(v^{H}v\right)}& \left(3\right)\end{array}$ H corresponds to the transpose of the vectors u or v Thus, the index i(m) sought verifies: $\begin{array}{cc}d\left(a_{\mathrm{kpm}},b_{i\left(m\right)}\right)=\underset{1\le i\le M_{k^{\prime }}}{\min }\left[d\left(a_{\mathrm{kpm}},b_i\right)\right]& \left(4\right)\end{array}$ In this association, we consider a two-dimensional function associated with the m^th transmitter defined by:{circumflex over (β)}_m(t_k,λ_p)=d(a_kp(k)m, a_00m)  (5)

In the course of the association, there is obtained, by interpolation of the {circumflex over (β)}_m(t_k,λ_p) values for each transmitter indexed by m, a function β_m(t,λ) for 1≦m≦M. This function has the role especially of eliminating the pairs (t_k,λ_p) such that β_m(t_k,λ_p) and {circumflex over (β)}_m(t_k,λ_p) are very different: |{circumflex over (β)}_m(t_k,λ_p)—β_m(t_k,λ_p)|>η. Thus the aberrant instants which may be associated with other transmitters are eliminated.

Since the β_m(t,λ) brings into play the distance d(u,v) between the vectors u and v, it is said that u and v are close when:d(u, v)<η  (6) The value of the threshold η is chosen for example as a function of the following error model:u=v+e  (7) where e is a random variable. When the direction vectors are estimated on a duration of T samples, the law of the variable e may be approached by a Gaussian mean standard deviation law σ=1/^{{square root}{square root over (T)}}. Thus, the distance d(u, v) is proportional to a chi-2 law with (N−1) degrees of freedom (N:length of the vectors u and v). The ratio between the random variable d(u, v) and the chi-2 law is equal to σ/N. With the law of e being known, it is possible to determine the threshold η with a certain probability of false alarms. In the steps of the association of the method, a distance d_ij is defined in the time-wavelength space between the pairs (t_i,λ_p(i)) and (t_j,λ_p(j)):d_ij={square root}{square root over ((t_i−t_j)²+(λ_p(i)−λ_p(j))²)}  (8) In considering that, for each pair, (t_k,λ_p(k)), M_k vectors a_kp(k)j (1<j<M_k) have been identified, the steps of this association for K pairs (t_k,λ_p(k)) are given here below. The steps of association for K instants t_k and λ_p wavelengths are, for example, the following:

• • Step AS1 Initialization of the process at k=0, m=1 and M=1. The initial number of transmitters is determined, for example, by means of a test to detect the number of sources at the instant t0, this being a test known to those skilled in the art.
    • For all the triplets (t_k,λ_p(k),j) the initialization of a flag called flag_kj with flag_kj=0: flag_kj=0 indicates that the j^th direction vector obtained at (t_k,λ_p(k)) is not associated with any family of direction vectors.
  • Step AS2 Search for an index j and a pair (t_k,λ_p(k)) such that flag^kj=0, which expresses the fact that the association with a family of direction vectors is not achieved.
  • Step AS3 For this 1^st triplet (t_k,λ_p(k),j), flag_kj=1 is effected and link_k′i is initialized at link_k′i=0 for k′≠k and i′≠j and ind_m={k} and Φ_m={a_kp(k)j}, where Φ_m is the set of vectors associated with the m^th familly or m^th transmitter, ind_m the set of the indices k of the pairs (t_k,λ_p(k)) associated with the same transmitter, link_ki is a flag that indicates whether it has performed a test of association of the vector a_kp(k)i of of the triplet (t_k,λ_p(k),i) with the m^th family of direction vectors: link_ki=0 indicates that the association test has not been performed.
  • Step AS4 Determining the pair (t_k′,λ_p(k′)) minimizing the distance d_kk′ of the relationship (8) with (t_k,λ_p(k)) such that k∈ind_m in the time-frequency space and in which there exists at least one vector b_i=a_k′p(k′)i such that flag_k′i=0 and link_k′i=0; the vector has never been associated with a family on the one hand and there has been no test of association conducted with the m^th family on the other hand. A search is made in a set of data resulting from the separation of the sources.
  • Step AS5 By using the relationship (4) defined here above, we determine the index i(m) minimizing the difference between the vectors a_kp(k)m such that k∈ind_m and the vectors b_i identified with the instant-wavelength pairs (t_k′,λ_p(k′)) for (1≦i≦M_k′) and flag_k′i=0 and link_k′i=0.
  • Step AS6 do link_k′i(m)=1: the test of the association with m^th family has been performed.
  • Step AS7 If d(a_kp(k)m, b_i(m))≦η relationship (6) and |t_k−t_k′|<Δt_max and |λ_p(k)−λ_p(k′)|<Δλ_max then perform Φ_m={Φ_m b_i(m)}, ind_m={ind_mk′}, flag_k′i(m)=1: The vector is associated with the m^th family.
  • Step AS8 Return to the step AS4 if there is at least one doublet (t_k′,λ_p(k′)) and one index i such that the flag link_k′i=0 and flag_k′i=0.
  • Step AS9 In writing K(m)=cardinal(Φ_m), we obtain the family of vectors Φ_m={a_1p(1)m . . . a_{K(m),p(K(m)),m}} associated with the source indexed by m.
    • For each vector a_kp(k)m the estimate {circumflex over (β)}_m(t_k,λ_p) of (5) is associated and then a polynomial interpolation is made of the {circumflex over (β)}_m(t_k,λ_p) to obtain the two-dimensional interpolated function β_m(t,λ).
  • Step AS10 From the family of vectors Φ_m={a_1p(1),m . . . a_{K(M),p(K(M)),m}}, extract the J instants t_i ∈ind_j ⊂ ind_m such as the coefficients |β_m(t_i,λ_p(i))-{circumflex over (β)}_m(t_i,λ_p(i))|<η (threshold value) such that β_m(t_i,λ_p(i)) is not an aberrant point of the function β_m(t,λ)
    • It is said that there is an aberrant point when the divergence in modulus between the point {circumflex over (β)}_m(t_i,λ_p(i)) and an interpolation of the function β_m(t,λ) does not exceed a threshold η.
    • After this sorting out, the new family of pairs is Φ_m={a_k,p(k),m/k ∈ind_j}, ind_m=ind_j and K(M)=J, M←M+1 and m←M.
  • Step AS11 Return to the step AS3 if there is at least one triplet (t_k,λ_p(k),j) such that flag_kj=0.
  • Step AS12 M←M−1. After the step AS10, we are in possession of the family of vectors Φ_m={a_1p(1),m . . . a_{K(M),p(K(M)),m}} having no aberrant points. Each vector has an associated function β_m(t,λ) whose role especially is to eliminate the aberrant points which do not belong to a zone of uncertainty given by η (see equations (5)(6)(7)).
  • The steps of the method described here above have notably the following advantages:
  • The number M of transmitters is determined automatically, The following are determined for each transmitter:
  • The class of vectors Φ_m={a_1p(1),m . . . a_{K(m),p(K(m)),m}},
  • The number K(m) of direction vectors,
  • A set ind of indices indicating the pairs (t_k,λ_p(k)) associated with the vectors of the set Φ_m.
  • Managing the case of the appearance and disappearance of a transmitter,
  • Associating the transmitters in the time-wavelength space (t,λ).

The method of association described here above by way of an illustration that in no way restricts the scope of the invention is based on a criterion of distance of the direction vectors. Without departing from the scope of the invention, it is possible to add other criteria to it such as:

• • the signal level received in the channel considered (criterion of correlation on the level),
  • the instant of the start of transmission (front) or detection of a periodic marker (reference sequence), enabling the use of synchronisation criteria in the case of time-synchronised EVF signals (steady levels, TDMA, bursts, etc.),
  • characteristics related to the technical analysis of the signal (waveform, modulation parameters . . . ),
  • etc.

The following step is that of localising the transmitters.

The Wideband Localisation of a Transmitter

The goal of the method especially is to determine the position of the m^th transmitter from the components of the vectors a_1p(1)m up to a_Kp(K)m measured with different wavelengths.

These vectors a_kp(k)m have the particular feature of depending on the instant t_k, the wavelength λ_p(k) and the position (x_m,y_m,z_m) of the transmitter. For example, for a network formed by N=2 sensors spaced out by a distance of d in the axis of the carrier, the direction vector verifies a_kp(k)m. $\begin{array}{cc}{a}_{\mathrm{kp}\left(k\right)m}=\left[\begin{array}{c}1\\ \exp (j2\pi \frac{d}{{\lambda }_{p\left(k\right)}}\cos \left(\theta \left(t_k,x_m,y_m,z_m\right)\right)\cos \\ \left(\Delta \left(t_k,x_m,y_m,z_m\right)\right))\end{array}\right]=a\left(t_k,{\lambda }_{p\left(k\right)},x_m,y_m,z_m\right)& \left(9\right)\end{array}$

According to FIG. 1, the incidence (θ(t_k,x_m,y_m,z_m),Δ(t_k,x_m,y_m,z_m)) may be computed directly from the position (x_k,y_k,z_k) of the carrier at the instant t_k and the position (x_m,y_m,z_m) of the transmitter.

The method will, for example, build a vector b_kp(k)m from components of the vector a_kp(k)m. The vector b_kp(k)m may be a vector with a dimension (N−1)×1 in choosing the reference sensor in n=i: $\begin{array}{cc}{b}_{\mathrm{kp}\left(k\right)m}=\left[\begin{array}{c}{a}_{\mathrm{kp}\left(k\right)m}\left(1\right)/a_{\mathrm{kp}\left(k\right)m}\left(i\right)\\ M\\ a_{\mathrm{kp}\left(k\right)m}\left(i-1\right)/a_{\mathrm{kp}\left(k\right)m}\left(i\right)\\ a_{\mathrm{kp}\left(k\right)m}\left(i+1\right)/a_{\mathrm{kp}\left(k\right)m}\left(i\right)\\ M\\ a_{\mathrm{kp}\left(k\right)m}\left(N\right)/a_{\mathrm{kp}\left(k\right)m}\left(i\right)\end{array}\right]=b\left(t_k,{\lambda }_{p\left(k\right)},x_m,y_m,z_m\right)& \left(10\right)\end{array}$ where a_kp(k)m(n) is the i^th component of a_kp(k)m. Thus, in the example of the equation (9) and in fixing i=1 we get: $\begin{array}{cc}{b}_{\mathrm{kp}\left(k\right)m}=\left[\begin{array}{c}\exp (j2\pi \frac{d}{{\lambda }_{p\left(k\right)}}\cos \left(\theta \left(t_k,x_m,y_m,z_m\right)\right)\cos \\ \left(\Delta \left(t_k,x_m,y_m,z_m\right)\right))\end{array}\right]& \left(11\right)\end{array}$

It being known that the direction vectors a_kp(k)m are estimated with a certain error e_kp(k)m such that a_kp(k)m=a(t_k,λ_p(k),x_m,y_m,z_m)+e_kp(k)m. The same is true for the transformed vector b_kp(k)m of (10) at the first order when ∥e_kp(k)m∥<<1.b_kp(k)m=b(t_k,λ_p(k),x_m,y_m,z_m)+w_km  (12)

The family of localizing techniques mentioned in the prior art, using the phase shift between two sensors, requires knowledge of the phase of the vector b_kp(k)m. It being known that the vector a_kp(k)m is a function of the position (x_m,y_m,z_m) of the transmitter, the same is true for the vector b_kp(k)m.

Under these conditions, the localisation method consists, for example, in maximizing the following criterion of standardized vector correlation L_K(x,y,z) in the position space (x,y,z) of a transmitter. $\begin{array}{cc}{L}_K\left(x,y,z\right)=\frac{{b_K^H{v}_K\left(x,y,z\right)}^2}{\left(b_K^H{b}_K\right)\left({v_K\left(x,y,z\right)}^H{v}_K\left(x,y,z\right)\right)}\mathrm{with}{b}_K=\left[\begin{array}{c}{b}_{1m}\\ M\\ b_{\mathrm{Km}}\end{array}\right]=v_K\left(x_m,y_m,z_m\right)+w_K,v_K\left(x,y,z\right)=\left[\begin{array}{c}b\left(t_1,{\lambda }_{p\left(1\right)},x,y,z\right)\\ M\\ b\left(t_K,{\lambda }_{p\left(K\right)},x,y,z\right)\end{array}\right]\mathrm{and}{w}_K=\left[\begin{array}{c}{w}_{1m}\\ M\\ w_{\mathrm{Km}}\end{array}\right]& \left(13\right)\end{array}$ The noise vector w_K has the matrix of covariance R=E[w_K w_K^H]. Assuming that it is possible to know this matrix R, the criterion may be envisaged with a whitening technique. In these conditions, the following criterion L_K′(x,y,z) is obtained $\begin{array}{cc}{L}_K^{\prime }\left(x,y,z\right)=\frac{{b_K^H{R}^{-1}{v}_K\left(x,y,z\right)}^2}{\left(b_K^H{R}^{-1}{b}_K\right)\left({v_K\left(x,y,z\right)}^H{R}^{-1}{v}_K\left(x,y,z\right)\right)}\mathrm{with}R=E\left[w_K{w}_K^H\right]& \left(14\right)\end{array}$

The vector V_K (x,y,z) depends on the K wavelengths λ_p(1) up to λ_p(K). This is why it is said that the method achieves a broadband localisation.

The criteria L_K(x,y,z) and L_K′(x,y,z) have the advantage of enabling the implementation of a localization technique in the presence of a network of sensors calibrated in space (θ,Δ) at various wavelengths λ.

Given that, at the instant t_k, we know the analytic relationship linking the incidence (θ(t_k,x,y,z), Δ(t_k,x,y,z)) of the transmitter with its position (x,y,z), it is then possible, from the incidence (θ(t_k,x,y,z), Δ(t_k,x,y,z)), to deduce the vector a(t_k,λ_p(k),x_m,y_m,z_m)=a(θ(t_k,x,y,z) , Δ(t_k,x,y,z),λ_p(k)) in making an interpolation of the calibration table in the 3D space (θ,Δ,λ).

In an airborne context, the knowledge of the altitude h of the aircraft reduces the computation of the criterion in the search space (x,y), assuming z=h.

In the example of the equations (9) and (11) the vector v_K (x,y,z) is written as follows: $\begin{array}{cc}{v}_K\left(x,y,z\right)=\left[\begin{array}{c}\exp \left(\mathrm{j2}\pi \frac{d}{{\lambda }_{p\left(1\right)}}\cos \left(\theta \left(t_1,x,y,z\right)\right)\cos \left(\Delta \left(t_1,x,y,z\right)\right)\right)\\ M\\ \exp \left(\mathrm{j2}\pi \frac{d}{{\lambda }_{p\left(K\right)}}\cos \left(\theta \left(t_K,x,y,z\right)\right)\cos \left(\Delta \left(t_K,x,y,z\right)\right)\right)\end{array}\right]& \left(15\right)\end{array}$

In this method, it is possible to consider initialising the algorithm at K=K₀ and then recursively computing the criterion L_K(x,y,z).

Thus, L_K(x,y,z) is computed recursively as follows: $\begin{array}{cc}{L}_{K+1}\left(x,y,z\right)=\frac{{{\alpha }_{K+1}\left(x,y,z\right)}^2}{{\beta }_{K+1}{\gamma }_{K+1}\left(x,y,z\right)}\mathrm{where}{\alpha }_{K+1}\left(x,y,z\right)={\alpha }_K\left(x,y,z\right)+b_{K+1p\left(K+1\right)m}^{H}b\left(t_{K+1,}{\lambda }_{p\left(K+1\right)},x,y,z\right){\gamma }_{K+1}\left(x,y,z\right)={\gamma }_K\left(x,y,z\right)+{b\left(t_{K+1},{\lambda }_{p\left(K+1\right)},x,y,z\right)}^{H}b\left(t_{K+1},{\lambda }_{p\left(K+1\right)},x,y,z\right){\beta }_{K+1}={\beta }_K+b_{K+1p\left(K+1\right)m}^H{b}_{K+1p\left(K+1\right)m}& \left(16\right)\end{array}$

when the vectors b(t_K+1,λ_p(K+1),x,y,z) and b_kp(k)m are constant standards equal to ρ the relationship of recurrence of the equation (16) becomes: $\begin{array}{cc}{L}_{K+1}\left(x,y,z\right)=\frac{{{\alpha }_{K+1}\left(x,y,z\right)}^2}{{{\beta }^2\left(K+1\right)}^2}\mathrm{Where}\text{}{\alpha }_{K+1}\left(x,y,z\right)={\alpha }_K\left(x,y,z\right)+b_{K+1p\left(K+1\right)m}^{H}b\left(t_{K+1},{\lambda }_{p\left(K+1\right)},x,y,z\right)& \left(17\right)\end{array}$

The method has been described up to this point in assuming that the transmitters have fixed positions. It can easily be extended to the case of moving targets with a speed vector (v_xm,v_ym,v_zm) for which there is a model of evolution as described in the patent application FR 03/13128.

The method according to the invention can be applied to a very large number of measurements. In this case, the value of K is diminished in order to reduce the numerical complexity of the computations.

The method provides, for example, for the performance of the following processing operations on the elementary measurements:

• • decimation of the pairs (t_k, λ_p(k)),
  • filtering (smoothing of the measurements which are the direction vectors) and sub-sampling,
  • merging on a defined duration (extraction by association of direction vectors to produce a measurement of synthesis).

Bibliography

• 1—R O. SCHMIDT—November 1981—A signal subspace approach to multiple emitter location and spectral estimation
• 2—J. F. CARDOSO A. SOULOUMIAC—December 1993—Blind beamforming for non-gaussian signals IEE Proceedings-F, Vol. 140, No. 6, pp. 362-370
• 3—P. COMON—April 1994—Independent Component Analysis, a new concept Elsevier—Signal Processing, Vol 36, no. 3, pp 287-314

Claims

1. A method for localizing one or more sources, being in motion relative to a network of sensors the method comprising the following steps: separating the sources in order to identify the direction vectors associated with the response of the sensors to a source at a given incidence; associating the direction vectors a1p(1)m . . . aKp(K)m obtained for the mth transmitter and, respectively, for the instants t1 . . . tK and for the wavelengths λp(1) . . . , λp(k), and localizing the mth transmitter from the components of the vectors a1p(1)m . . . aKp(K)m measured with different wavelengths.

2. The method according to claim 1 wherein, in considering that, for each pair, (tk,λp(k)), Mk vectors akp(k)j (1<j<Mk) have been identified, the step of association for K pairs (tk,λp(k)) comprises the following steps: 1) resetting of the process at k=0, m=1 and M=1 and, for all the triplets (tk,λp(k),j), resetting a flag of association with a transmitter flagkj at flagkj=0; 2) searching for an index j and a pair (tk,λp(k)) such that the flag, flagkj=0; 3) for this 1st triplet (tk,λp(k),j) obtained at the step 2, effecting flagkj=1 and resetting a test flag of association with the transmitter of this triplet linkk′i=0 for k′≠k and i′≠j and indm={k} and Φm={akp(k)j}; 4) determining the pair (tk′,λp(k′)) minimizing the distance dkk with (tk,λp(k)) such that k∈indm in the time-frequency space and in which there exists at least one vector bi=ak′p(k′)i such that flagk′i=0 and linkk′i=0; 5) in using the relationship (4) defined here above, determining the index i(m) that minimizes the difference between the vectors akp(k)m such that k∈indm and the vectors bi identified with the instant-wavelength pairs (tk′,λp(k′)) for (1≦i≦Mk′) and flagk′i=0 and linkk′i=0; 6) doing linkk′i(m)=1: the test of association has been performed; 7) if d(akp(k)m, bi(m))≦η and |tk−tk′|<Δtmax then: Φm={Φmbi(m)}, indm={indmk′}, flagk′i(m)=1; 8) if there is at least one doublet (tk′,λp(k′)) and one index i such that linkk′i=0, then repeating the steps from the step 4; 9) defining the family of vectors Φm={a1p(1)m . . . aK(m),p(K(m,)),m} associated with the source indexed by m in writing K(m)=cardinal(Φm); and 10) from the family of vectors Φm={a1p(1),m . . . aK(M),p(K(M)),m}, extracting the J instants ti ∈indj ⊂ indm which correspond to aberrant points located outside a defined zone. 11) returning to the step 3) if there is at least one triplet (tk,λp(k),j) such that flagkj=0.

3: The method according to claim 1 wherein the localizing step comprises the following steps: maximizing a criterion of standardized vector correlation LK(x,y,z) in the position space (x,y,z) of a transmitter with LK⁡(x,y,z)=bKH⁢vK⁡(x,y,z)2(bKH⁢bK)⁢(vK⁡(x,y,z)H⁢vK⁡(x,y,z))⁢ ⁢withbK=[b1⁢m⋮bKm]=vK⁡(xm,ym,zm)+wK,vK⁡(x,y,z)=[b⁡(t1,λp⁡(1),x,y,z)⋮b⁡(tK,λp⁡(K),x,y,z)]⁢ ⁢andwK=[w1⁢m⋮wKm]where wk is the noise vector for all the positions (x, y, z) of a transmitter.

4: The method according to claim 3, wherein the vector bK comprises a noise-representing vector whose components are functions of the components of the vectors a1m . . . aKm.

5: A method according to claim 3, comprising a step in which the matrix of covariance R=E[wKwKH] of the noise vector is determined, and wherein the following criterion is maximized

6: The method according to claim 5, wherein the assessment of the criterion LK(x,y,z) and/or the criterion LK′(x,y,z) is recursive.

7: The method according to claim 1, comprising a step to compare the maximum values with a threshold value.

8: The method according to claim 1, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

9. The method according to claim 2, comprising a step to compare the maximum values with a threshold value.

10. The method according to claim 3, comprising a step to compare the maximum values with a threshold value.

11. The method according to claim 4, comprising a step to compare the maximum values with a threshold value.

12. The method according to claim 5, comprising a step to compare the maximum values with a threshold value.

13. The method according to claim 6, comprising a step to compare the maximum values with a threshold value.

14. The method according to claim 2, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

15. The method according to claim 3, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

16. The method according to claim 4, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

17. The method according to claim 5, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

18. The method according to claim 6, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.

19. The method according to claim 7, wherein the transmitters to be localized are mobile and wherein the vector considered is parametrized by the position of the transmitter to be localized and the speed vector.