FR3071621A1 - LOCATION DEVICE WITH MULTI-FREQUENCY RECEIVER - Google Patents

Abstract

The invention relates to a locating device comprising a multifrequency reception system of at least two radiolocation signals of different frequencies (L1, L2) and a signal processing means able to implement an algorithm for estimating the maximum of likelihood. The signal of each frequency (L1, L2) is connected to at least two signal processing lines, each processing line having a correlator and being arranged so that the signal processing means calculates the useful signal and signal parameters. multi-path from the signals from the processing lines. The invention applies to ground stations of reference or observation constellations of navigation satellites and can be used in mobile terminals to improve the navigation solution in urban or even airport environments.

Description

LOCATION DEVICE WITH MULTI-FREQUENCY RECEIVER

The field of the invention relates to satellite navigation systems and more particularly to radiolocation devices by calculating the propagation time of signals transmitted by satellites.

By satellite navigation system is meant here any system dedicated to wide area navigation, such as for example the existing GNSS systems (Global Navigation Satellite System) called GPS, GLONASS or the future GALILEO system, as well as all their equivalents and derivatives. Those skilled in the art are familiar with the principle of localization of satellite navigation systems. The radiofrequency signal emitted by a satellite is coded and the time taken by this signal to reach the receiver to be located is used to determine the distance between this satellite and this receiver, called pseudo-distance. The accuracy of satellite navigation systems is affected by a number of errors. These errors can be broken down into two categories: global contributions and local contributions. Mention may be made, for global contributions, of errors linked to the passage of electromagnetic waves in the ionosphere, in the troposphere, as well as errors linked to satellites (orbit errors and clock errors). With regard to local contributions, there may be mentioned the errors associated with the reflection of the signals, the errors linked to interference or the noise of the receivers.

In an urban environment, the radiolocation device is particularly affected by the phenomenon of reflection of signals on urban obstacles such as, for example, building facades. The multipath "MP" (ie all the signals reflected by the environment close to the receiver) constitute the predominant source of error for the location. Indeed, muti-paths ("MultiPath" or "MP"), will be added to the useful signal ("Line Of Sight" or "LOS" for direct path in line of sight), which will bias the estimated LOS propagation time, and ultimately cause receiver location errors. It is important to be able to delete or estimate these multipaths to improve the location accuracy of the location devices. In the single-frequency framework, the existing solutions propose, according to a first approach, to discriminate the useful signal from the multi-path signals in the time domain (respectively frequency). We can cite for example the publication Theory and Performance of Narrow Correlator Spacing in a GPS receiver, AJ Van Dierendonck, Journal of The Institute of the Navigation vol 39, no 3 fall 1992, or the publication The Multipath Estimating Delay Lock Loop: Approaching Theoretical Accuracy Limits by R. Van Nee published in 1994 in the IEEE journal pages 246-251. This first approach is faulted for situations where the multipath signals have a small delay compared to the direct signal.

A second approach consists in discriminating the useful signal from multipaths by spatial filtering of the signals received using an array of antennas. However, conventional multi-antenna algorithms (ie beamforming algorithms) suffer from the strong correlation between the direct signal and the multipaths. This effect is all the more important as the network is small.

As a third approach, we can cite the invention Method of processing multipath navigation signais in a receiver having a plurality of antennas, filed in the United States whose patent number issued is US 6,437,733, B1. This approach is based on a decomposition into singular values of the space generated by the antennas of the network so as to identify the subspace containing the targeted navigation signal and its multipaths. The delays of the various journeys are then estimated by a method of the MEDLL type ("Multipath Estimating Delay Lock Loop"). This approach is only valid if the number of antennas in the network is greater than the number of paths received by the system. As soon as we consider a small network (typically 4 antennas), this condition is defective in an urban scenario where the number of echoes easily exceeds ten.

Finally, another approach consists of a so-called spatiotemporal-frequency approach using an algorithm called SAGE. We know from the prior art the publication called Channel Parameter Estimation in Mobile Radio Environments Using the Sage algorithm by Bernard H. Fleury et al. or the publication Multipath Mitigation Methods Based on Diversity Algorithme, Proceedings of the 26th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS 2013), Nashville, TN, September 2013, pp. 1596-1605., De Rougerie, S., Carrie, G., Israël, J., Ries, L., Monnerat, M., Thevenon, P. This algorithm is capable of estimating all the parameters characterizing received signals ( the useful signal and the multipaths). The version of Rougerie et al. (Patents FR 2964 199 and WO 2012/025306 A1) allows the application of this algorithm in real time.

In the multi-frequency framework, solutions have also been proposed in order to take into account the effects of multi-paths. In an RTK context, the approach proposed by M. Sahmoudi and M. G. Amin in Fast Iterative Maximum-Likelihood Algorithm (FIMLA) for Multipath Mitigation in the Next Generation of GNSS Receivers, published in IEEE Trans. On Wireless Communications, vol. 7, n ° 11 in November 2008 is a three-frequency and double difference particle filtering method in order to remove phase ambiguities, estimate ionospheric biases and provide an indicator of the presence of multi-paths. However, this method does not estimate the parameters characterizing the multi-routes but detects only their presence and is only valid in an RTK context.

In GNSS Multifrequency receivers in urban environment: Theoretical analysis published in Proc. ION GNSS 2005, M. Musso, G. Géra, A. Cattoni and C. S. Regazzoni assume independent propagation channels to propose a weighted average of the estimators applied on each frequency band. This approach, noted later by M. Musso, also implicitly assumes that interband delays are known or provided by an external device and therefore does not allow an independent estimation of all the parameters characterizing the propagation channels.

Unlike this approach, N.l. Ziedan assumes in MultiFrequency Combined Processing for Direct and Multipath Signais Tracking Based on Particle Filtering published in ION GNSS 2011 that there are purely deterministic relationships between delays and phase of multipaths on two different frequencies. It offers particle filtering taking advantage of this redundancy. This approach has several drawbacks. From a modeling point of view, it relies on uncertain deterministic relationships, in particular on the phase relationships originally assumed to be equal between the LOS and the MPs. In addition, the likelihood function proposed by N.l. Ziedan implicitly assumes that the noise between the correlator outputs is white while it is not: its temporal correlation function is that of the local code.

The objective of the invention is to provide a satellite location device little or not affected by the location errors associated with multi-path phenomena. The solution of the invention is little or not affected by multi-path signals having a small delay compared to the useful signal and allows a hardware implementation requiring few computational resources in order to address localization devices for the general public. with reduced computing capacity.

To achieve this goal or others, a first aspect of the invention provides a device allowing the reception of at least two radiolocation signals of different frequencies. More specifically, the invention is a localization device comprising a multi-frequency reception system for at least two radiolocation signals of different frequencies and a signal processing means capable of implementing an algorithm for estimating the maximum likelihood, characterized in that the signal of each frequency is connected to at least two signal processing lines, each processing line comprising a correlator and being arranged so that the signal processing means calculates the useful signal and multi signal parameters - route from the signals from the processing lines.

The location device according to the invention may include one or more of the following characteristics, taken individually or in combination:

- Time offsets between the different frequencies linked to ionospheric and RF effects are corrected by a module No. 4 of the present invention.

each processing line also comprises a delay line connected by a first end to the output of the correlator and by a second end to an input of the signal processing means, the delay introduced corresponding to a multiple of an integration duration correlators.

- a first correlator of a first processing line is spaced by a fraction of the symbol period of a local code with respect to a second correlator of a second processing line.

- the maximum likelihood estimation algorithm is of the SAGE type.

- the said parameters are the propagation delay and the Doppler frequency.

A second aspect of the invention proposes a method for estimating the useful signal and multi-path signal parameters originating from a radiolocation signal transmitted by a satellite by means of a location device allowing the reception of at least two signals of different frequencies, characterized in that it comprises the following stages:

a) A first step of measuring the radiolocation signal on said frequencies,

b) A second step of correlation with a local code of the signal received on said frequencies by means of correlators,

c) A third stage of construction of the temporal evolution of the outputs of the correlators by concatenation of the data successively produced in the second stage,

d) A fourth stage of construction of the signal inter-correlation function, late and in time by concatenating the data produced in the third stage,

e) A fifth step in the construction of a reference correlation function of known parameters late and in time.

f) A fifth step of estimating the useful signal and multi-path signal parameters by means of a maximum likelihood algorithm applied to the multifrequency signal by comparison of the inter-correlation function constructed from the measured signal and of the reference intercorrelation function.

Other particularities and advantages of the present invention will appear in the description below of nonlimiting exemplary embodiments, with reference to the appended drawings in which:

FIG. 1 is a general diagram of a processing chain of a GNSS "Global Navigation Satellite System", including the invention.

Figure 2 is a detailed diagram of module 1 in Figure 1.

FIG. 3 is a graph of the estimation performance of the useful parameters, namely the propagation delay and the Doppler frequency, on simulated data (GALILEO E5A E5B), by means of each of the following three methods: conventional device by treatment of a SAGE algorithm (denoted SAGE / STAP BPSK (10)), a device according to the invention (denoted SAGE / STAP BPSK (10) / BPSK (10)) and a reference device (denoted M. Musso) . This performance estimate is given as a function of the C / NO of the signal received.

FIG. 4 is a graph of the estimation performance of the useful parameters, namely the propagation delay and the Doppler frequency, on simulated data (GALILEO E5A E5B), by means of each of the following three methods: conventional device by treatment of a SAGE algorithm (denoted SAGE / STAP BPSK (10)), a device according to the invention (denoted SAGE / STAP BPSK (10) / BPSK (10)) and a reference device (denoted M. Musso) . This performance estimate is given as a function of the relative delay of a multi-path assumed to be in phase on each of the two channels.

FIG. 5 is a graph of the estimation performance of the useful parameters, namely the propagation delay and the Doppler frequency, on simulated data (GALILEO E5A E5B), by means of each of the following three methods: conventional device by treatment of a SAGE algorithm (denoted SAGE / STAP BPSK (10)), a device according to the invention (denoted SAGE / STAP BPSK (10) / BPSK (10)) and a reference device (denoted M. Musso) . This performance estimate is given as a function of the relative delay of a multi-path assumed to be in phase on one channel and in phase opposition on the other channel.

As illustrated, the raw signals received on each of the Nl bands pass through a correlator bench in order to compress the data and bring out the useful signal from the noise (module n ° 1). Their outputs are concatenated within a vector X _N l coming to supply the module n ° 2 which contains the SAGE algorithm as well as a multi-path detector. In module n ° 2, the SAGE algorithm provides an estimate of the sought parameters of the direct signal (delay ^, amplitude / _o and doppler v ₀ ), while the multi-path detector provides binary information on the presence of multi -trajets.

The parameters of the direct signal (provided by the SAGE algorithm in module n ° 2) will feed a delay tracking loop ("Delay Lock Loop" or "DLL") as well as a phase tracking loop ("Frequency and Phase Lock Loop ”or“ FPLL ”) classics (Module n ° 3). The outputs of these modules are the information sought on which the entire tracking module is controlled. The multi-path detection information (provided by the multi-path detector in module n ° 2) launches the inter-band discriminators only in the trajectory portions without multi-paths, making it possible to estimate and continue (via a DLL loop) delays between the different frequency bands (Module n ° 4).

We justify and detail in this document the originality and the relevance of this solution. It has already been demonstrated that the SAGE / STAP algorithm (patent FR 2964199) has many advantages in a multi-antenna framework (explicit estimation of multi-path parameters, reduction of memory load and computational cost, robustness to limitations RF stage, independence of the navigation system used). We show here that major adaptations are necessary before it can be used in a multi-frequency framework.

The algorithm described in patent FR 2964199 makes it possible to estimate the set of variablesψ; = [^, T _r! , v _rt ] ^r , where and v _rl are respectively the amplitude, the relative delay (compared to the absolute delay of the direct path τ ₀ ) and the relative doppler (compared to the absolute doppler of the direct path v ₀ ) of each multi -traject I, the index 0 being reserved for the direct journey. For this, it is necessary to be able to decompose all of these state variables on hidden subspaces linked to each of the multi-paths /. These hidden subspaces are by nature independent of each other, which constitutes a fundamental assumption of the algorithm described in patent FR 2,964,199.

In the multi-frequency framework (where the frequencies ^, ..., are considered), the set of variables that we seek to estimate becomes Ψ / = [ψ / ^Α 1 = 1, ... Λ, ' ^OÙ , V, Î Γ

The are the amplitudes of each multi-path I on the band l. and are all independent from each other.

The dopplers v% are connected by a simple relation of proportionality between them involving the ratio of the wavelengths: = ν ^ λ _ί / λ _ι .

If the band Li is taken as a reference (ie the frequency indicated by 1), the delays τ% can be written from the delay on the band L-ι, and the time shift between the bands Lt and Lj. This time difference between the two bands, noted Δ ⁷ ;, is linked to atmospheric effects and to the RF reception chains. It is therefore independent of multi-paths (ie of the index /). By noting the delay of each multi-path on the band L1r _ri = rf, we can then write:

~ ^τ rl ⁺

Taking the frequency l_i as a reference, the set of state variables is therefore actually written for each multi-path: Ψζ = y with n _rl = rf, v _rl = v “, the delay and the relative doppler on Li.

The Δ ^ intervening in the definition of all the multi-paths, it is no longer possible to decompose the variables to be estimated on independent subspaces linked to each multi-path. However, this is a fundamental assumption of the SAGE STAP algorithm (patent FR 2964199) necessary for its application.

To circumvent this problem, we propose to separate the estimation of the parameters of the pathsψ _ζ =, T _rl , v _rÎ J from the estimation of the inter-band biases | δ ^, ..., Δ ^]. Module n ° 4 makes it possible to estimate the inter-band biases [δ ^, ..., Δ ^ ^£ ] in the case without multi-paths. It therefore uses the outputs of module n ° 2 which informs about the presence of multi-paths at a given time. The overall inter-band delay is therefore reliably estimated on portions of trajectories without multipaths. These estimates can then be used in all sections of the trajectory because it has been shown that the drift time constant of these biases is low (of the order of several minutes outside of extraordinary ionospheric events). Module n ° 1 (correlator bench) thus applies not to raw multi-frequency signals but to multi-frequency signals corrected for these inter-band biases. Consequently, the variables to be estimated for the SAGE algorithm (module n ° 2) are \ | / z = [yf ¹ , ... ,, Trl, v _rl J. The subspaces defined by each multi-path are quite independent between them and the SAGE algorithm (module n ° 2) described below can effectively apply.

We now detail in the following description the four modules according to the invention, as shown diagrammatically in FIG. 1.

Module 1 is also shown schematically in Figure 2 to facilitate its understanding.

The expression y (t) of the signal brought back into baseband at the output of the antenna can be in the form:

Zl y (O = Ë / = 0 yf ¹ xexp (2j ^ v / ^il i) xc1 (rr _/ ^il ) + b (r) rf ^NL xexp (2; ^ vz ⁱ '21 / Â _JV /) xc ₁ ( iT _/ ^{£ 1} ) where is the measurement noise on each band and c, denotes the signal spreading sequence on the band /. This expression involves delays and absolute Doppler frequencies on the Li band of each multipath:

T = ^ - r _rl

V! = VV _rl

Let us first consider the architecture of Figure 2 with an odd number P of correlators numbered by the index p such that pe [- (Pl) / 2 (Pl) / 2] and a number N of taps (Taps) post-correlation. Each correlation is performed over a duration Tint. Let us first consider the output Rk, t of a correlator for the path 7 of unit amplitude on the band k. The correlator is controlled on delay and Doppler of the direct path L-band _v n ^'th correlation output corresponding to the p Tap for the band k is:

aA v Ί ¹ _v -Ρ ^τ Μί-τ)

Δ ^, ν, ι, Ρ, η) = - x I ¹

Tnt (»-ΐ) η _ω x ^ex PL ^_ 2 jnvfa! X _k tJx exp | 2 jπνλ ^! X _k t \ dt

The correlation output with the local code can be rewritten:

, v _rl , p, n} = r _k (r _rl -pT _e -A ^ xsinckv ^ / λρΓ ^] x exp [- 2 jnv A!] βχρ [/π.ν,Λ /] where η /.) denotes the autocorrelation function of the code on the band k.

The independent amplitude and phase terms of n and p are now gathered in the complex amplitude 7 / * of the path /. The correlator output can therefore be in the form:

^R kj (ï _rl

yl

YÏ * = sinckv ^! X _k T ^] exp [jKvA / with:

R _k ^ rl ~ ^Δ 4> Vrl, P, ») = r _k & rl ~ P ^T e ~) ^{X eX} Pl · λΐ ^ Α ¹ 1

The time correlation vector R _fc of the signal is then constructed as follows:

First, the multi-correlator outputs are concatenated in a column vector to reconstruct a sampled inter-correlation function of the signal received on the band k.

These inter-correlation functions obtained for different post-correlation Taps are always concatenated in a column vector so as to plot the temporal evolution of the inter-correlation functions. This temporal evolution will make it possible to characterize the relative Doppler of the echoes.

Relative Doppler shifts are assumed to be constant over the processing time NT _in t. The relative delays according to the post-correlation Taps evolve according to:

The temporal correlation vector R. is therefore of the form:

r (tlï ^Λ £ ^ *> Ζ, 1 2 ^jl I r (t '' γΖ, Ι ^ 4 '* ^ rl 52

-A'v ^L rl, 2 ^ L ^ rl '~>

\

RÎT -Δ ⁴ v ~ (Rk ^τ Η, Ν

Or :

r _k (ï _r i, x ~ + ^(P ₂ T _e ) x exp [- 2jnv _rl \! À _k T _M ]

^r k (? rl, l ^ L, 2 ^{X eX} PÎ J ZCV Âj / À _k ] ^r k (? ri, 2 H - T _e ) xexp [- 2j7LV _rl À ^ / X _k 2.T _int ] ^r k (Jri, 2 ~ AΖζ) x exp [- Ij / tv! Z _k 2, T _int ] ^r k (fri, NL,) xexp [- 2jw _rl ^ / λ, ΝΤ ^]

-Ιλτρχι

The model of the signal received post-correlation for the multi-frequency multi-correlator architecture proposed can finally be put in the following vectorized form:

r-i

1 = 0 with

Χ (ψΓΧ) = 'JjVZW _£ xl

QWM θΛΤΜ 'II' χ (ψΧ) = θΤ / ΛίΙ θΝΜ X θΛΤίκΙ ... θΛΓΛύ VT · / ^- Α /> Κ / / NPN _l * N _l

or again where ψ / represents the vector of the parameters associated with the path Ί and X, X / and ^> NP, Tint are column vectors of dimension 2NP and O _NP * i represents a zero column vector of dimension NP. The vector A _a represents the time offsets between the band Li and the bands. The choice of the number of correlators P has been fixed in order to cover the entire domain 10 of the delays [~ T _C , T _c ] corresponding to ± 1 code chip. As recalled in patent FR 2964199, the number of correlators which offers the best compromise between complexity and performance is therefore:

^P op _t = ceil \ lxT _c xF _e ]

Module No. 2, as shown diagrammatically in FIG. 1, comprises at least two sub-modules called the “SAGE” sub-module and the multi-path detection sub-module. We will now detail module n ° 2, starting with the SAGE sub-module.

Here we assume that the inter-band delays have already been estimated by module no. 4 and that they are compensated for at the level of the correlator bank of the module no. 2. The entry model is now:

£ -1 _

Χ (ψ) ⁼ ^ i ^ - ^ rlWrl) ^Χ Ύ / ^^ ΝΡΝ ^, Τ ^

1 = 0 with

Ψ / = [y /, / Y! ^{1 1} = 1 ^ '^ rlWrlÎ and ψ = [ψθ ···] ^ _{+2) ΐχ1} X

The subspaces defined by each multi-path are well independent of each other. We can therefore apply the iterative process of the SAGE algorithm to estimate the parameters (see patent FR 2964199). The log likelihood 10 (to within a constant) associated with the hidden subspace I is then:

Α _ζ (ψ /) = (χ _ζ -R (u _/ , v _{r /} ) xy _; ) ⁱⁱ C- ¹ (x _/ -R (u, v _ri ) xy ₍ )

With X, = [χ ^ ^{ι) Γ} ·· path estimation ”/ from E-STEP (see SAGE algorithm patent FR 2964199), and C the covariance matrix associated with the signal model above:

xb ^w ^^ ^V NPN _l , τ) = C = Ληί ' NP ^ ®NP ®NP ζ (θ) r / Te) n (P.te) r (Te) ^ (0) η (Te) ; r / Te) ζ (0) r (Te) r. (Te) ζ (0) r _t (Te) / (PTE) ... r _t (Te) ζ (0). P

l / v is the identity matrix of dimension Λ / and Onp the zero square matrix of dimension NP. We note that the autocorrelation function of postcorrelation noise is no longer that of white noise but that of the local precorrelation code. Finally, ^ represents the noise power at the output of the bank 5 of correlators observed on each frequency band k. To simplify the cost function, one can use the quadratic relation between γ _ζ βΐΛ _ζ (ψ _ζ ), as well as the diagonal structures by block of the various matrices. So we have :

A _z (T _rZ , v _rf )! | R ^ (T _rZ , v _rZ ) xX _z ^(it) | tÎ σ ² Rf (T _rf) v _rf ) xR _t (r _rf , v _ri ) with:

t ί ^{γ x} x exp [- 2jnv _rl \ / ^ NT ^]

We will now detail the multi-path detection (MP) sub-module. Multipath detection is obtained by a test of χ ² on the difference between the observed signal and the estimate that we have of the direct path, this difference being normalized by the post correlation noise power and its covariance:

ΛΧ = (X - X „(“ „)) C- '(X - Χ„ (ψ „))

Under the HO hypothesis, we have no multipath. In the case of a Gaussian channel, AX then follows a law of χ ² centered at N * P * N \ _ degrees of freedom. Under hypothesis H1, multi-paths are present. ΔΧ then follows a law of χ ² not centered at N * P * Ni_ degrees of freedom. From a probability of false alarm (typically Pf _a = 10 ' ³ ), we can detect if we are under the hypothesis HO or H1.

We will now detail the module n ° 3:

This module contains the standard DLL and FPLL tracking loops well known by those skilled in the art [Kaplan, Understanding GPS: principles and applications, 1 ed. Norwood, MA 02062, USA: Artech House, Inc, 1996], This module takes as input the estimation of the delay, Doppler and relative phase between the received signal and the reference code, and outputs the necessary commands to control the correlations in module n ° 2.

We will now detail module n ° 4:

To estimate inter-band delays, we suggest using /V/.-1 DLL-type tracking loops, known to those skilled in the art. The loop n ° k takes as input the difference between the discriminator on the frequency Li and the discriminator on the frequency L _k :

ηΛ _{= £} L _k _ _e L _t ^L k where ε ⁴ is the normalized discriminator on the band L _k . These discriminators are well known to those skilled in the art. As the inter-band delay phenomena have long correlation times, we propose to use DLL loops with orders lower than 2 and bands lower than 0.5Hz. However, these loops are not robust to multipaths. Thus, we propose to use a multi-path detector from module n ° 2 in order to cut the loop in the presence of MP or fading. In the case where the loop is cut, we propose to keep the last estimate until the return of a less degraded situation.

For a location device comprising the architecture described in this document, the longest calculations (module no. 2) are carried out with a matrix of reduced size compared to a conventional architecture using the baseband signal. The matrix size is equal to N _L ^X N ^X P, that is to say the number N _L of frequencies, multiplied by the number N of delay lines, multiplied by the number P of correlators. To fix the ideas, in the single-frequency framework, with equivalent estimation performance, we reduce the size of the vectors to be processed by a factor of 500 compared to direct processing on signals not compressed by the bench of correlators.

In terms of performance, the proposed invention makes it possible to reduce noise by a factor compared to single frequency processing (processing of the GALILEO E5a and E5b bands). In addition, the proposed invention improves performance in the case of low power compared to methods from the literature [Musso, Maristella, Géra, Gianluca, Cattoni, Andrea, Regazzoni, Carlo S., GNSS Multifrequency Receivers in Urban Environment: Theoretical Analysis, Proceedings of the 18th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2005), Long Beach, CA, September 2005, pp. 2661-2669). In the presence of multipaths in phase or in phase opposition, the proposed invention clearly improves the rejection performance of multipaths compared to the single frequency case and compared to the methods of the literature.

The invention applies to ground reference or observation stations for constellations of navigation satellites. This invention can be used in mobile terminals to improve the navigation solution in urban or even airport environments.

Claims (7)

1. Locating device comprising a multifrequency reception system of at least two radiolocation signals of different frequencies (L1, L2) and a signal processing means (pP) capable of implementing an algorithm for estimating the maximum of likelihood, characterized in that the signal of each frequency (L1, L2) is connected to at least two signal processing lines, each processing line comprising a correlator (C11, C12) and being arranged so that the processing means of signals (μΡ) calculates the parameters of useful signal and multi-path signal from signals from the processing lines. 2. Locating device according to claim 1, characterized in that time offsets between the different frequencies are corrected by a module No. 4. 3. Locating device according to claim 1, characterized in that each processing line also comprises a delay line (R11m) connected by a first end to the output of the correlator (C11) and by a second end to an input of the means signal processing (μΡ), the delay introduced corresponding to a multiple of an integration time of the correlators. 4. Locating device according to claim 1, characterized in that a first correlator (C11) of a first processing line is spaced by a fraction of the symbol period of a local code with respect to a second correlator ( C12) of a second processing line. 5. Device according to claim 1, characterized in that the algorithm for estimating the maximum likelihood is of SAGE type. 6. Device according to claim 1, characterized in that said parameters are the delay and the doppler frequency. 7. Method for estimating useful signal and multi-path signal parameters originating from a radiolocation signal emitted by a satellite by means of a location device allowing the reception of at least two signals of different frequencies, characterized in that it comprises the following stages: a) A first step of measuring the radiolocation signal on said frequencies, b) A second step of correlation with a local code of the signal received on said frequencies by means of correlators, c) A third stage of construction of the temporal evolution of the outputs of the correlators by concatenation of the data successively produced in the second stage, d) A fourth stage of construction of the signal inter-correlation function, late and in time by concatenating the data produced in the third stage, e) A fifth step in the construction of a reference correlation function of known parameters late and in time. f) A fifth step of estimating the useful signal and multi-path signal parameters by means of a maximum likelihood algorithm applied to the multi-frequency signal by comparison of the inter-correlation function constructed from the measured signal and of the reference intercorrelation function.