Electronic device for detecting GNSS interference(s), vehicle, method and associated computer program

Abstract

Electronic device for detecting GNSS interference(s), vehicle, associated method and computer program The present invention relates to a device (10) for detecting GNSS interference(s), comprising at least: - an inertial measurement unit (12), - a GNSS receiver (16), - a Kalman filter (20) configured to cyclically calculate an innovation vector (22) by hybridization of data supplied by: - ​​said inertial measurement unit, and - said GNSS receiver, and furthermore, at the output of said Kalman filter: - a filtering module (26) applying at least one filtering on each component of the innovation vector supplied at each calculation cycle, - an interference detection module (30) detecting an interference when during Q cycles and on a number U of components of the innovation vector, the absolute value of the output (28) of the filtering module, for at least one filtering, is greater than a predetermined threshold. Figure for abstract: Figure 1

Description

Title of the invention: Electronic device for detecting GNSS interference(s), vehicle, method and associated computer program

[0001] The present invention relates to an electronic device for detecting GNSS interference(s) suitable for being on board a vehicle suitable for moving between two geographical positions, the device comprising at least: an inertial measurement unit suitable for providing navigation measurements, a GNSS receiver for positioning measurements by satellite(s), a Kalman filter configured to cyclically calculate an innovation vector by hybridization of data provided, at the input of said Kalman filter, both by said inertial measurement unit and by said GNSS receiver.

[0002] The invention also relates to a vehicle comprising such an electronic device for detecting GNSS interference(s).

[0003] The invention also relates to a method for detecting GNSS interference(s), said method being implemented by an electronic device for detecting GNSS interference(s) suitable for being on board a vehicle suitable for moving between two geographical positions.

[0004] The invention also relates to a computer program comprising software instructions, which when executed by a computer implement such a method for detecting GNSS interference(s).

[0005] The present invention relates to the navigation of a vehicle (also called a carrier) capable of moving between two distinct geographical positions, such as a land vehicle corresponding in particular to a car, a truck, or such as an aircraft or any type of carrier.

[0006] To do this, the vehicle generally has a satellite navigation and positioning system receiver configured to determine, in particular by trilateration, a positioning (i.e. a geolocation position or a geolocation solution) of the vehicle using estimates of distances to the visible satellites of one or more satellite constellations of the satellite navigation and positioning system. Examples of satellite navigation systems are the American GPS system (from the English "Global Positioning System"), the European GALILEO system, the Russian GLONASS system, or the Chinese BEIDOU system, etc.

[0007] More specifically, the present invention relates to the detection of interferences hereinafter called “GNSS interferences” from the hybridization between the data inertial and data received by the GNSS receiver, also called inertial / GNSS hybridization.

[0008] Subsequently, “GNSS interference” means interference which consists of sending erroneous radio frequency signals to the GNSS receiver. The GNSS receiver then uses these erroneous signals to calculate the navigation data of the vehicle in which it is embedded, for example its position and speed, which will then be different from the exact (i.e. “true”) data, because of errors in the radio frequency signals.

[0009] For example, such GNSS interference occurs when, inside hangars located in airports, GNSS signal repeaters, in particular GPS, are used to propagate these signals inside the hangars, so as to be used to assist an aircraft pilot located inside the hangars to position himself.

[0010] However, when the doors of these hangars are open, the signals from these repeaters are likely to propagate outside the hangars, and to be picked up by aircraft located outside, which must use the direct GNSS signals to position themselves, and not the signals from these repeaters which then constitute interference for aircraft outside the hangars.

[0011] To remedy this, a current technique for detecting such GNSS interference is described in the standard DO-229D “Minimum Operational Performance Standards for Global Positioning System / Satellite-Based Augmentation System Airborne Equipment” and in particular, in appendix R 3 1 1, and consists of using hybridization between inertial data and GPS data so as to detect the interference using a Kalman filter.

[0012] More precisely, this current technique first calculates, during hybridization, the innovation vector which constitutes the difference between on the one hand the difference measured between the GNSS data and the inertial data, and on the other hand the difference predicted according to the hybridization model between these data.

[0013] Then, this technique then implements the comparison of the different components of this innovation vector with the expected standard deviation on each component of the innovation vector, this expected standard deviation being calculated from the covariance matrix P of the Kalman filter used for the calculation of the innovation vector, and when the absolute value of one or more components of the innovation vector are much higher than the expected standard deviation, it disables the use of the data from the GNSS receiver because there is potentially a problem with the GNSS data, coming from interference or another phenomenon.

[0014] This current technique is however not satisfactory, because interference on the GNSS data can cause an error which gradually increases over time without causing, at each calculation cycle of the Kalman filter, changes in values ​​too large at the level of the components of the innovation vector, so that the values ​​of the components of this innovation vector will be of the same order of magnitude as those present in the absence of interference and will not be much higher than the standard deviation calculated from the covariance matrix P of the Kalman filter used in the calculation of the innovation vector. Thus, such interference on GNSS data will not be detected most of the time, or at least not immediately or quickly after the appearance of such interference.

[0015] The aim of the invention is then to propose an electronic device for detecting GNSS interference(s) which makes it possible to overcome the drawbacks of the current technique by increasing the efficiency and responsiveness of GNSS interference detection.

[0016] To this end, the invention relates to an electronic device for detecting GNSS interference(s), suitable for being mounted on board a vehicle suitable for moving between two geographical positions, the device comprising at least:

[0017] - an inertial measurement unit capable of providing navigation measurements,

[0018] - a GNSS receiver for positioning measurements by satellite(s),

[0019] - a Kalman filter configured to cyclically calculate an innovation vector by hybridization of data provided, as input to said Kalman filter, both by:

[0020] - said inertial measurement unit, and

[0021] - said GNS S receiver,

[0022] the device further comprising, at the output of said Kalman filter:

[0023] - a filtering module configured to apply at least one filtering on each component of the innovation vector provided at each calculation cycle by said Kalman filter,

[0024] - an interference detection module configured to detect interference when during Q cycles and on a number U of components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the output of the filtering module, for at least one filtering applied by said module, is greater than a predetermined threshold.

[0025] Thus the electronic device for detecting GNSS interference(s) according to the present invention has a particular architecture making it possible to detect interference by using not directly the components of the innovation vector but the output of a filtering on these components.

[0026] In other words, the present invention consists of using an electronic device for detecting GNSS interference(s) capable of applying one or more filterings to each component of the innovation vector, and of using the output of these filterings to detect GNSS interference.

[0027] Indeed, the present invention aims to exploit the fact that the evolution over time of the components of the innovation vector is different depending on whether an interference at level of GNSS data is present or not. The filtering used then makes it possible, specifically according to the present invention, at the level of the output returned by the filtering, to highlight this difference in evolution over time.

[0028] According to other advantageous aspects of the invention, the electronic device for detecting GNSS interference(s) comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations:

[0029] - said predetermined threshold depends on the type of filtering applied by the module of filtering and / or the current calculation cycle;

[0030] - the filtering module is configured to apply weighted sum filtering on each component of the innovation vector provided at each calculation cycle by said Kalman filter;

[0031] - said filtering module uses a finite impulse response filter whose N co efficient filtering methods, for each component of the innovation vector, are suitable for summing the last N normalized input values ​​of said filter, and for weighting said resulting sum by a predetermined weighting coefficient;

[0032] - said weighting coefficient is capable of depending on the index of said component considered in the innovation vector;

[0033] - the filtering module is configured to apply at least one filtering with co predetermined filtering efficiencies.

[0034] - said filtering module uses a finite impulse response filter, the N of which filtering coefficients, for each component of the innovation vector, are calculated with predetermined coefficients for each type of interference of a predetermined set of interference types and for each type of navigation phase of a predetermined set of navigation phase types associated with said vehicle.

[0035] The invention also relates to a vehicle comprising such an electronic device for detecting GNSS interference(s).

[0036] The invention also relates to a method for detecting GNSS interference(s), said method being implemented by an electronic device for detecting GNSS interference(s) suitable for being on board a vehicle suitable for moving between two geographical positions, the method comprising at least the following steps:

[0037] - cyclic calculation of an innovation vector, implemented by a Kalman filter of said device, by hybridization of data provided, at the input of said Kalman filter, both by an inertial measurement unit capable of providing navigation measurements and by a GNSS receiver of positioning measurements by satellite(s);

[0038] - application of at least one filtering on each component of the innovation vector provided at each calculation cycle by said Kalman filter,

[0039] - detection of interference when during Q cycles and on a number U of com components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the filtering output, for at least one filtering, is greater than a predetermined threshold.

[0040] The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement such a method for detecting GNSS interference(s) as defined above.

[0041] These characteristics and advantages of the invention will appear more clearly on reading the description which follows, given solely by way of non-limiting example, and made with reference to the appended drawings, in which:

[0042] - [Fig.l] [Fig.l] is a diagram illustrating an electronic detection device GNSS interference(s);

[0043] - [Fig.2] [Fig.2] is a representative flowchart of the detection method GNSS interference(s) according to the present invention;

[0044] - [Fig.3] [Fig.3] illustrates, under the effect of interference, the kth component normalized of the innovation vector as well as the weighted sum of this normalized component obtained at the output of the filtering implemented according to an embodiment of the invention;

[0045] - [Fig.4] [Fig.5] Figures 4 and 5 illustrate the predetermined filter coefficients, the kth normalized component of the innovation vector as well as the output of the filtering with said predetermined filtering coefficients implemented according to another embodiment of the invention and under the effect of interference occurring from separate calculation cycles.

[0046] [Fig.l] is an overall representation of an electronic device 10 for detecting GNSS interference(s) according to the present invention.

[0047] Such an electronic device 10 for detecting GNSS interference(s) is suitable for being mounted on board a vehicle suitable for moving between two geographical positions.

[0048] As illustrated by [Fig.l], such a device 10 firstly comprises an inertial measurement unit 12 capable of providing navigation measurements 14.

[0049] In addition, such a device 10 comprises a GNSS receiver 16 for positioning measurements 18 by satellite(s).

[0050] Such a device 10 also comprises a Kalman filter 20 configured to cyclically calculate an innovation vector 22 by hybridization of data provided, at the input of said Kalman filter, both by said inertial measurement unit 12 and by said GNSS receiver 16.

[0051] In other words, the electronic GNSS interference detection device 10 is firstly configured to use the inertial data (i.e. the measurements 14 of na vigation) and the GNSS receiver data (i.e. the 18 satellite positioning measurements) by performing a hybridization between these data, using a Kalman filter activated at different calculation cycles, these calculation cycles representing the time that passes.

[0052] Thus, the hybridization implemented by the Kalman filter 20 consists of mathematically combining the measurements 14 and 18 provided respectively by the inertial measurement unit 12 and the receiver 16 of GNSS satellite positioning measurements of said vehicle to obtain position and speed information by taking advantage of these two sources 12 and 16 of measurements 14 and 18 respectively.

[0053] Kalman filtering relies on the possibilities of modeling the evolution of the state of a physical system considered in its environment, by means of a so-called "evolution" equation (a priori estimation), and of modeling the dependency relationship existing between the states of the physical system considered and the measurements of an external sensor, by means of a so-called "observation" equation to allow a recalibration of the states of the filter (a posteriori estimation). In a Kalman filter, the effective measurement or "measurement vector" makes it possible to make an a posteriori estimate of the state of the system which is optimal in the sense that it minimizes the covariance of the error made on this estimate. The estimator part of the filter generates a posteriori estimates of the state vector of the system by using the difference observed between the effective measurement vector and its a priori prediction to generate a corrective term, called innovation.This innovation, subsequently called the innovation vector, after multiplication by a gain matrix of the Kalman filter, is, in a manner not shown, suitable for being applied to the a priori estimate of the state vector of the system and leads to obtaining the optimal a posteriori estimate.

[0054] The Kalman filtering implemented by the Kalman filter 20 is thus also capable of modeling the evolution of the errors of the inertial measurement unit 12 and of delivering, in a manner not shown, the a posteriori estimate of these errors which is used to correct the positioning and speed point of the inertial measurement unit 12.

[0055] More precisely, at each calculation cycle n, the observation vector Zn is calculated, which constitutes the difference between the inertial data 14 (i.e. navigation measurements 14) and the GNSS data 18 (i.e. positioning measurements 18 by satellite(s)).

[0056] Such an observation vector Zn corresponds, according to a first example, to the difference between the inertial position associated with the inertial data 14 and the GNSS position associated with the GNSS data 18, or according to another example to the difference between the pseudo-range associated with the satellite positioning measurements 18 corresponding to the distance between the GNSS receiver of the carrier and the satellites emitting the radio frequency signals, and the pseudo-range calculated from the inertial position of the carrier and the position of the satellites emitting the radio frequency signals. satellites emitting radio frequency signals.

[0057] The classical operation of the Kalman filter in the field of the invention is recalled below.

[0058] The Kalman filter 20 is first configured to define the state vector Xn, with Ne components, which brings together a certain number of unknown states that we wish to estimate, a component of the vector Xn being for example the error on the inertial position.

[0059] Then, the Kalman filter 20 is configured to write the relationship between the observation vector Zn, with No components, this vector being known, and the state vector Xn to be estimated in the form Zn = Hn*Xn + vn, with Hn the known observation matrix, vn the unknown measurement noise, considered as Gaussian noise, but whose variance Rn is known.

[0060] Then, the Kalman filter 20 is configured to calculate the vector Xn|n4 corresponding to the estimate of the vector Xn from the observations Zi, Z2, ... Zn [, then to calculate the innovation vector Yn = Zn - Hn*Xn|ni

[0061] This innovation vector Yn is therefore itself the difference between on the one hand Zn which corresponds to the difference obtained between the inertial data 14 and the GNSS data at cycle n, and on the other hand Hn*Xn|n_i which corresponds to the difference predicted from the estimate X nin.i. In other words, Yn corresponds to the difference predicted from the hybridization model.

[0062] The Kalman filter 20 is also capable of generating the elements 24 on(k) with k varying between 1 and No (No being the number of components of the innovation vector Yn), each element on(k) corresponding to the expected standard deviation, in the absence of interference, for the component k of the innovation vector Yn, the calculation of on(k) using Rn as the variance of the measurement noise vn.

[0063] It should be noted that it can be demonstrated that in the absence of interference, for each value of k between 1 and No (No being the dimension of the innovation vector), Yn(k), when n varies, corresponds to a Gaussian white noise of standard deviation on(k). It should also be noted that it can be demonstrated that, under the effect of interference, for each value of k between 1 and No, the kth component of the innovation vector, when n varies, can be expressed in the form of a sum of the effect due solely to the interference on the one hand and the Gaussian white noise of standard deviation on(k) calculated by the Kalman filter 20.

[0064] Specifically according to the present invention, as illustrated by [Fig. 1], the electronic device 10 for detecting GNSS interference(s) further comprises, at the output of said Kalman filter 20, a filtering module 26 configured to apply at least one filtering to each component k of the innovation vector Yn 22 provided at each calculation cycle by said Kalman filter 20.

[0065] Indeed, as described below in relation to figures 3 to 5, such a filtering module aims to use the fact that at the start of the application of the interference, the com normalized variables of the innovation vector Yn(k) / on(k) have a behavior that does not correspond to what they have in the absence of interference.

[0066] More precisely, the filtering module 26 is configured to apply one or more filters, via one or more filters respectively, to each cycle n of calculation of the Kalman filter 20, on each component of the innovation vector.

[0067] The output of the filter of index J or of a filter of index J of the filters of the filtering module 26 is noted: SJn(k).

[0068] In the case where the filter of index J (also called filter J hereinafter) corresponds to a finite impulse response filter, which corresponds to the fact that the output of this filter J at an instant depends only on the last N values ​​of the filter input, we can write SJn(k) in the following way:

[0069] =

[0070] which amounts to:

[0071] SJn(k) = aJo,n(k) * Yn(k) + a\n(k)* Yn l(k) +... + aV1>n(k)* Yn.^(k),

[0072] with Yn(k), the kth component of the innovation vector calculated in the current cycle n, Yn.i (k) is the kth component of the innovation vector calculated in the previous cycle n-1, and aJ0>n (k), aJi n(k),... , aJx i „(k) the coefficients of the filter J, which potentially depend on said filter J considered, on the kth component of the innovation vector considered, and on the current cycle n.

[0073] Furthermore, specifically according to the present invention, as illustrated by [Fig. 1], the electronic device 10 for detecting GNSS interference(s) further comprises an interference detection module 30 configured to detect interference when during Q cycles and on a number U of components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the output 28 of the filtering module 26, for at least one filtering applied by said module, is greater than a predetermined threshold.

[0074] In other words, such an interference detection module 30 is capable of deciding that interference has been detected if, during Q consecutive calculation cycles and over a number U of components of the innovation vector, the absolute value of the output of the filtering ISJn(k)l is greater than a predetermined threshold subsequently noted ThresholdJn for the threshold associated with the filter J, and this for one or more of the filters used by the filtering module 26.

[0075] In the event of effective detection, the interference detection module 30 is capable of generating and emitting a signal 32 representative of the presence of such interference, in particular to alert the user.

[0076] As an optional addition, each predetermined threshold depends on the type of filtering applied by the filtering module 26 and / or the current calculation cycle n.

[0077] In other words, the ThresholdJn associated with the filter J depends on the filtering implemented by this filter J considered (i.e. is specific to the filter J considered) and the calculation cycle n considered and is a predefined value.

[0078] More precisely, to predefine such threshold values ​​associated respectively with each filter suitable for use within the filtering module 26, for a given filtering J, for which it is desired that it detects a given interference, when the application of this given interference is proven (for example in the test phase or on a computer simulation of the hybridization), the maximum value ISJn(k)lmax obtained by the output ISJn(k)l of the filter J is first determined, and the value ThresholdJn of the threshold associated with the filter J is by definition lower than the maximum value ISJn(k)lmax obtained by the output ISJn(k)l of the filter J.

[0079] Thus, according to the present invention, having defined the type of filtering applied by the filtering module 26, the numbers Q, U and the ThresholdJn when the filtering module 26 only implements a single filtering by means of the filter J, it is possible to analyze the interference detection performance by further determining the false alarm probability for the filtering J, this false alarm probability being determined by using the fact that each component of the innovation vector, in the absence of interference, corresponds to a white Gaussian noise, when the calculation cycle varies. The false alarm probability for the filtering J corresponds to the probability that during Q consecutive calculation cycles and on a number U of components of the innovation vector, ISJn(k)l is greater than the threshold ThresholdJn> while there is no interference, and the interference detection is all the more efficient as the false alarm probability is low.

[0080] An example of operation of the 10 electronic device for detecting GNSS interference(s).

[0081] More precisely, the method 40 for detecting GNSS interference(s) implemented by said electronic device 10 for detecting GNSS interference(s) comprises the steps described below implemented successively.

[0082] According to step 42, as indicated previously, the electronic device 10 for detecting GNSS interference(s) according to the present invention implements a cyclic calculation C of an innovation vector V, implemented by a Kalman filter 20 of said device 10, by hybridization of data provided, at the input of said Kalman filter 20, both by an inertial measurement unit 12 capable of providing navigation measurements 14 and by a GNSS receiver of satellite positioning measurements 18.

[0083] Then according to step 44, as indicated previously, the electronic device 10 for detecting GNSS interference(s), via its filtering module 26, according to the present invention applies at least one filtering on each component, represented by an arrow, of the innovation vector provided at each calculation cycle by said Kalman filter. 20. It should be noted that it is possible to apply different filters to the same component.

[0084] It should be noted that, according to two distinct embodiments, two possible types of filtering are capable of being implemented distinctly during this step 44, by way of example.

[0085] It should be noted that other types of filtering, known to those skilled in the art, are also suitable for being respectively implemented according to other associated embodiments, not shown in [Fig.2] for the sake of simplicity.

[0086] According to a first embodiment, the filtering module 26 is configured to apply, during a step 46, a weighted sum filtering, called F_S_P filtering, on each component (i.e. each component being represented by an arrow in [Fig.2]) of the innovation vector provided at each calculation cycle by said Kalman filter 20.

[0087] According to a second embodiment, the filtering module 26 is configured to apply, during a step 48, at least one filtering with predetermined filtering coefficients, called F_C_P filtering.

[0088] Whatever the embodiment 46 or 48 of the filtering step 44, the filtering output S is obtained at the end of this step.

[0089] Finally, according to step 50, the electronic device 10 for detecting GNSS interference(s) according to the present invention implements, via its detection module, a step D of detecting an interference when during Q cycles and on a number U of components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the filtering output S, for at least one filtering, is greater than a predetermined threshold and this for one or more of the filters used by the filtering module 26.

[0090] Each of the embodiments 46 and 48 is described more precisely below in relation to FIGS. 3 to 5.

[0091] According to the first embodiment, the filtering module 26 is configured to apply, during a step 46, a weighted sum filtering, called F_S_P filtering, on each component (i.e. each component being represented by an arrow in [Fig.2]) of the innovation vector V provided at each calculation cycle by said Kalman filter 20.

[0092] More precisely, according to this embodiment, a single filtering is carried out, so that it is not necessary to specify the index of this single filter for this first embodiment.

[0093] Such filtering 46 consists of using a finite impulse response filter whose N filtering coefficients, for each component of the innovation vector V, are suitable for summing the last N normalized input values ​​of said filter, and for weighting said resulting sum by a predetermined weighting coefficient.

[0094] More precisely, the coefficients of such a filter associated with the F_S_P filtering are suitable for being expressed in the following form ai>n(k) = —L- *Mk' Mk being a co-JÂ) any efficient not depending on the index i (i ranging from 0 to N-1, N being the number of filtering coefficients) and the index n but which may depend, as an optional complement, on the index k of the said component considered of the innovation vector.

[0095] Thus, the filtering output S at cycle n of component k is expressed in the following form: Sn(k) = Mk / yN~lYjk) \ = Mk / Y„(k) Y,,#) CjWÇ \• \^=o „#) / \ + „^) + ■■■ + ;

[0096] The filtering thus consists firstly in carrying out, for each component k of the innovation vector, at each calculation cycle n, the sum of the last N standardized innovations, a standardized innovation corresponding to the value of the innovation divided by the corresponding standard deviation, and secondly in weighting this sum by the coefficient Mk.

[0097] It should be noted that, in the absence of interference, the component Yn(k), when n varies, corresponds to a white Gaussian noise of standard deviation on(k), so that the normalized component of the innovation Yn(k) / on(k) is also a white Gaussian noise with a standard deviation equal to 1, and that consequently, the filtering output signal Sn(k) is also a Gaussian noise of standard deviation Mk*N1 / 2.

[0098] In Figure 3, view 52 represents, on the ordinate, the value of the normalized k component Yn(k) / on(k) of the innovation vector as a function of the value of the calculation cycle on the abscissa, in the presence of GNSS interference, appearing from the calculation cycle n=900, and view 54 represents, on the ordinate, the value of the filtering output S n(k) corresponding to the weighted sum of this normalized component by taking the number N of filtering coefficients such that N=20 and Mk=_1_.

[0099] In view 52, ​​before the application of the interference (i.e. before the cycle n=900 on the abscissa), the normalized innovation corresponds to a white noise of standard deviation 1, then at the start of the application of the interference, between n=900 and n= 1000 on the abscissa, the normalized innovation has a behavior which does not correspond to a white noise, finally after n= 1000, the normalized innovation again has a behavior which corresponds to a white noise.

[0100] It should be noted that the value of the normalized k component Yn(k) / on(k) is a dimensionless quantity because it is the innovation component divided by the corresponding standard deviation.

[0101] This figure also illustrates that in the absence of interference (i.e. before the application of the interference) the signal Sn(k) is a Gaussian noise of standard deviation 1, and that under the effect of the interference the maximum reached by the weighted sum, of the order of 8, on view 54 corresponding to the output of the weighted sum filtering, is much higher than the maximum reached by the normalized component worth 3.4 on view 52.

[0102] Thus, if we consider the case where Q=1 and U=1, which amounts to the fact that the interference is detected if for a component the output of the filter is greater than the threshold over a calculation cycle, we see the advantage of using the output of the filter to detect the interference and not simply the normalized component of the innovation.

[0103] If, to detect the interference, we simply use the normalized component of the innovation, as currently carried out according to the state of the art recalled previously, we therefore need a threshold less than or equal to 3.4 to detect the interference, whereas by using the output of the filter associated with the F_S_P filtering to detect the interference, in other words with a weighted sum as filtering, we need a threshold less than or equal to 8 to detect the interference, so that the output of the filter according to the present invention makes it possible to detect interference with thresholds that are significantly higher than with the simple normalized component of the innovation.

[0104] In the absence of interference, the output of the filter as the normalized component being Gaussian noises of standard deviation 1, the fact of being able to use a higher detection threshold makes it possible to drastically reduce the probability of associated false alarm.

[0105] According to the example illustrated by [Fig. 3], the probability of false alarm associated, according to the current state of the art, with a threshold of 3.4 would in fact have been equal to 6.7*10 4, which amounts to the fact that a Gaussian noise of standard deviation 1 has a probability of 6.7*10 4 of exceeding (in absolute value) the value 3.4, whereas the probability of false alarm associated with a threshold of 8 obtained according to the present invention is advantageously equal to 1.2*10 A which amounts to the fact that a Gaussian noise of standard deviation 1 has a probability of 1.2*10 15 of exceeding (in absolute value) the value 8.

[0106] According to the second embodiment, the filtering module 26 is configured to apply, during a step 48 of [Fig.2], at least one filtering with predetermined filtering coefficients, called F_C_P filtering.

[0107] In this example, it is possible to use one or more filters, for example F filters, the jth filter being of index J. Each of the F filters being a finite impulse response filter with N coefficients, the equation of which is expressed in the following form for the filter J of index J:

[0108] S'„(k) = rM 1 = / 4 * 1 m ) \— + ^r + - + ^r) k,n

[0109] with MJk>n being any coefficient not depending on the index i (i ranging from 0 to Nl, N being the number of filtering coefficients) but which may depend on the index k of the component considered of the innovation vector, of the index n of the calculation cycle and of the index J of the filter considered in the filtering module 26 and o-( for i between 0 and Nl) of the pre-determined coefficients The coefficients aJi>n(k) of this filter are therefore expressed in the following form: aJin(k) = * MJkn hence the name of such a filter namely filter with predetermined filter coefficients.

[0110] In the absence of interference, the kth component Yn(k) of the innovation vector, when n varies, is a white Gaussian noise of standard deviation on(k), and the output Sn(k) of such a filter F_C_P according to the second embodiment with predetermined filtering coefficients, when n varies, is a Gaussian noise of standard deviation / „az-1 ( c'y \+^2 MJk>n, the unit in which the coefficients c / are expressed being that in which the innovation is expressed.

[0111] As an optional addition, the N coefficients c / of such a filter with predetermined filtering coefficients, for each component of the innovation vector, are predetermined for each type of interference of a predetermined set of interference types and for each type of navigation phase of a predetermined set of navigation phase types associated with said vehicle.

[0112] Indeed, it is established that, for a given inertial / GNSS hybrid navigation system (i.e. implementing inertial / GNSS hybridization) in a given navigation phase, for a given interference L, on the calculation cycles which follow the start of the interference, the effect due to the interference on the innovation can approximately be written in the form Y= aL* un with a time signal having a maximum equal to 1, a varying as a function of the calculation cycle n considered, and capable of depending on the navigation phase considered while not depending on the component considered in the innovation vector, nor on the interference considered but just on its type.

[0113] Typically, if we consider the type of interference where the error in the radio frequency signals causes the appearance of an error in the calculated position increasing linearly over time, one will not depend on the direction coefficient giving the increase, as a function of time, of the error in the calculated position.

[0114] aL is a coefficient which depends on the component considered in the innovation vector and on the interference considered (on the value of the errors applied to the radiofrequency signals), without depending on the navigation phase considered or on the calculation cycle n. For example, in the case of the type of interference where the error on the radiofrequency signals causes the appearance of an error on the calculated position increasing linearly over time, aL depends on the director coefficient giving the increase (as a function of time) of the error on the calculated position.

[0115] According to this optional addition, a first list of possible types of interference is established, comprising for example a type of interference consisting of the appearance of a fixed error on the calculated position, another type of interference consisting of the appearance of an error increasing linearly over time on the calculated position, etc.

[0116] In addition, a second list of the different navigation phases of the carrier using GNSS data and inertial data is also established.

[0117] From these two lists, for each type of interference and each navigation phase, we determine predetermined coefficients c / from the evolution that we obtain on a component of the innovation vector just after having applied the interference of the type considered, when the carrier is in the navigation phase considered.

[0118] In this case, the application of the interference of the type considered is effectively proven and the coefficients c / are suitable for being determined either in the test phase or on a computer simulation of the hybridization implemented by the Kalman filter 20, considering that advantageously, for each navigation phase considered, there are as many filters as there are types of interference considered.

[0119] More precisely, for a given navigation phase, the coefficients c / are suitable, for example, for being predetermined, for each possible type of interference, by designating, for example, J the interference of the type considered, firstly by applying this interference J to an inertia / GNSS hybridization at the calculation cycle ni, then by selecting a component k0 of the innovation vector by recording the N successive values ​​Yni(k0), Yni+i(k0), ..., Yni+N4(ko) of the component k0, so that:

[0120] c'0= Ynl+N_!(ko), ch= Ynl+N_2(k0), ..., cV1= Ynl(k0).

[0121] Such coefficients associated with said given navigation phase are then used, when the carrier (i.e. the vehicle) is in said given navigation phase, and this by considering all possible types of interference, to define, for each component k of the innovation vector, the coefficients aJi>n(k) of the filter J according to the formula aJin(k) = * MJkn filter which will be applied to the component k of the innovation vector and which will contribute, via its output S, to the subsequent detection of the presence or absence of this type of interference J. Thus, for each component of the innovation vector, there are, according to this second embodiment, as many filters as there are types of interference considered.

[0122] Such a second embodiment 48 based on the use of at least one filtering with predetermined filtering coefficients, called F_C_P filtering, has the advantage that for each navigation phase considered and each type of interference considered, the method for calculating the predetermined coefficients c / is extracted from the method which itself defines a filter, which allows, when the wearer is in the navigation phase considered and for the type of interference considered, for a given false alarm probability (the false alarm probability being the probability of detecting interference when there is none) to obtain the minimum probability of non-detection (the non-detection probability being the probability of not detecting this type of interference when it is present).

[0123] Figures 4 and 5 are associated with such a second embodiment 48 by considering for simplicity a single filter J with predetermined filtering coefficients, [Fig.4] representing the effect of an interference appearing from the calculation cycle n=900, while [Fig.5] represents the effect of an interference appearing from the calculation cycle n=1200.

[0124] In [Fig.4] view 58 represents, again as in view 52 of the [Fig.3], on the ordinate the value of the normalized k component Yn(k) / on(k) of the innovation vector as a function of the value of the calculation cycle on the abscissa, in the presence of GNSS interference, (already illustrated by [Fig.3]) appearing from the calculation cycle n=900.

[0125] View 60 of [Fig.4] represents, in the presence of GNSS interference, appearing from the calculation cycle n=900, on the ordinate the value of the filtering output SJn(k) of the filter J with the predetermined filtering coefficients (the coefficients c / used for this filter are illustrated by view 56 of [Fig.4]) and:

[0126] MJkn= / r / 2 cji pz \~V2 such that in the absence of interference, the output of the — + -4— + a--4±_ U*) 2 filter is a Gaussian noise of standard deviation 1.

[0127] In Figures 4 and 5, the same filter J with predetermined filtering coefficients is applied, so that views 56 of [Fig.4] then 62 of [Fig.5], representing the coefficients c / of this filter J as a function of the index i on the abscissa, are identical.

[0128] In [Fig.5] view 64 represents on the ordinate the value of the normalized k component Yn(k) / on(k) of the innovation vector as a function of the value of the calculation cycle on the abscissa, in the presence of GNSS interference, appearing from the calculation cycle n=1200.

[0129] View 66 of Figure 5 represents, in the presence of GNSS interference, appearing from the calculation cycle n=1200, on the ordinate the value of the filtering output SJn(k) of the filter J with the predetermined filtering coefficients illustrated by view 56 of Figure 4 and MJk>n = / r / 2 yz \ "V2 such that in the absence of interference, the filter output is Gaussian noise of standard deviation 1.

[0130] In the two cases illustrated by figures 4 and 5 associated with cycles of appearance interference n=900 and n=1200, it is first noted that in the absence of interference, the output of the filtering as the normalized component are Gaussian noises of standard deviation 1. In addition, it is noted that the maximum, of the order of 9 on [Fig.4] and 11 on [Fig.5], reached by the output, in absolute value, of the filtering implemented specifically according to the present invention, on views 60 of [Fig.4] as 66 of [Fig.5], under the effect of the interference is clearly higher than the maximum, of the order of 3.4 on [Fig.4] and 3.5 on [Fig.5], reached by the normalized component of the innovation, in absolute value, on views 58 of [Fig.4] and 64 of [Fig.5].

[0131] We can thus see the advantage of using the output of the filter of the filtering module 26 to detect the interference and not simply the standardized component of the innovation as proposed according to the current state of the art.

[0132] Indeed, by using the output of the filter, it is therefore possible to detect interference with thresholds significantly higher than the simple standardized component of the innovation, which advantageously makes it possible to reduce the probability of associated false alarms.

[0133] Those skilled in the art will understand that the invention is not limited to the embodiments described, nor to the particular examples of the description, the embodiments and variants mentioned above being suitable for being combined with each other to generate new embodiments of the invention.

[0134] The present invention thus proposes a particular architecture of an electronic device for detecting GNSS interference(s) not using the components of the innovation vector directly but using the output of a filtering of these components, which makes it possible to exploit the fact that the evolution over time of the components of the innovation vector is different depending on whether interference of the GNSS data is present or not. The filtering 44 used in fact makes it possible to highlight this difference in evolution over time, at the level of the output returned by the filtering.

Claims

Claims

1. Electronic device (10) for detecting GNSS interference(s), suitable for being on board a vehicle suitable for moving between two geographical positions, the device comprising at least: - an inertial measurement unit (12) suitable for providing navigation measurements (14), - a GNSS receiver (16) for satellite positioning measurements (18), - a Kalman filter (20) configured to cyclically calculate an innovation vector (22) by hybridization of data supplied, at the input of said Kalman filter, both by: - ​​said inertial measurement unit (12), and - said GNSS receiver (16), the device (10) being characterized in that it further comprises, at the output of said Kalman filter: - a filtering module (26) configured to apply at least one filtering to each component of the innovation vector supplied at each calculation cycle by said Kalman filter,- an interference detection module (30) configured to detect interference when, during Q cycles and on a number U of components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the output (28) of the filtering module, for at least one filtering applied by said module, is greater than a predetermined threshold, the filtering module (26) being configured to apply at least one filtering with predetermined filtering coefficients, using a finite impulse response filter, the N filtering coefficients of which, for each component of the innovation vector, are calculated with said predetermined filtering coefficients, said filtering coefficients being predetermined for each type of interference of a predetermined set of interference types and for each type of navigation phase of a predetermined set of navigation phase types associated with said vehicle.,

2. Device (10) according to claim 1, wherein said predetermined threshold depends on the type of filtering applied by the filtering module (26) and / or the current calculation cycle.

3. Vehicle suitable for moving between two geographical positions, said

4.

5. vehicle being characterized in that it comprises an electronic device (10) for detecting GNSS interference(s) according to any one of the preceding claims. Method for detecting GNSS interference(s), said method being implemented by an electronic device (10) for detecting GNSS interference(s) suitable for being on board a vehicle suitable for moving between two geographical positions, the method comprising at least the following steps: - cyclic calculation of an innovation vector, implemented by a Kalman filter of said device, by hybridization of data provided, at the input of said Kalman filter, both by an inertial measurement unit capable of providing navigation measurements (14) and by a GNSS receiver of positioning measurements (18) by satellite(s); - applying at least one filtering to each component of the innovation vector provided at each calculation cycle by said Kalman filter, with predetermined filtering coefficients, using a finite impulse response filter, the N filtering coefficients of which, for each component of the innovation vector, are calculated with said predetermined filtering coefficients, said filtering coefficients being predetermined for each type of interference of a predetermined set of interference types and for each type of navigation phase of a predetermined set of navigation phase types associated with said vehicle; - detection of interference when during Q cycles and on a number U of components of the innovation vector, Q and U each being an integer greater than or equal to one, the absolute value of the filtering output, for at least one filtering, is greater than a predetermined threshold. Computer program comprising software instructions which, when executed by a computer, implement the method for detecting interference(s) according to claim 4.

Images